From fab230f059993102e9aa63c8e7837d0bee1793c0 Mon Sep 17 00:00:00 2001 From: arnavkapoor Date: Tue, 4 Aug 2020 17:12:45 +0530 Subject: [PATCH 1/4] updating structure for ordinal support --- .../supplementary_translation_data/af.json | 22 ++- .../supplementary_translation_data/ak.json | 22 ++- .../supplementary_translation_data/am.json | 22 ++- .../supplementary_translation_data/ar.json | 22 ++- .../supplementary_translation_data/az.json | 22 ++- .../supplementary_translation_data/be.json | 22 ++- .../supplementary_translation_data/bg.json | 22 ++- .../supplementary_translation_data/bs.json | 22 ++- .../supplementary_translation_data/ca.json | 22 ++- .../supplementary_translation_data/ccp.json | 22 ++- .../supplementary_translation_data/chr.json | 22 ++- .../supplementary_translation_data/cs.json | 22 ++- .../supplementary_translation_data/cy.json | 22 ++- .../supplementary_translation_data/da.json | 22 ++- .../supplementary_translation_data/de-CH.json | 22 ++- .../supplementary_translation_data/de.json | 22 ++- .../supplementary_translation_data/ee.json | 22 ++- .../supplementary_translation_data/el.json | 22 ++- .../supplementary_translation_data/en-IN.json | 22 ++- .../supplementary_translation_data/en.json | 26 +++- .../supplementary_translation_data/eo.json | 22 ++- .../es-419.json | 22 ++- .../supplementary_translation_data/es.json | 50 +++--- .../supplementary_translation_data/et.json | 22 ++- .../supplementary_translation_data/fa-AF.json | 22 ++- .../supplementary_translation_data/fa.json | 22 ++- .../supplementary_translation_data/ff.json | 22 ++- .../supplementary_translation_data/fi.json | 22 ++- .../supplementary_translation_data/fil.json | 22 ++- .../supplementary_translation_data/fo.json | 22 ++- .../supplementary_translation_data/fr-BE.json | 22 ++- .../supplementary_translation_data/fr-CH.json | 22 ++- .../supplementary_translation_data/fr.json | 22 ++- .../supplementary_translation_data/ga.json | 22 ++- .../supplementary_translation_data/he.json | 22 ++- .../supplementary_translation_data/hi.json | 22 ++- .../supplementary_translation_data/hr.json | 22 ++- .../supplementary_translation_data/hu.json | 22 ++- .../supplementary_translation_data/hy.json | 22 ++- .../supplementary_translation_data/id.json | 22 ++- .../supplementary_translation_data/is.json | 22 ++- .../supplementary_translation_data/it.json | 22 ++- .../supplementary_translation_data/ja.json | 22 ++- .../supplementary_translation_data/ka.json | 22 ++- .../supplementary_translation_data/kl.json | 22 ++- .../supplementary_translation_data/km.json | 22 ++- .../supplementary_translation_data/ko.json | 22 ++- .../supplementary_translation_data/ky.json | 22 ++- .../supplementary_translation_data/lb.json | 22 ++- .../supplementary_translation_data/lo.json | 22 ++- .../supplementary_translation_data/lrc.json | 22 ++- .../supplementary_translation_data/lt.json | 22 ++- .../supplementary_translation_data/lv.json | 22 ++- .../supplementary_translation_data/mk.json | 22 ++- .../supplementary_translation_data/ms.json | 22 ++- .../supplementary_translation_data/mt.json | 22 ++- .../supplementary_translation_data/my.json | 22 ++- .../supplementary_translation_data/nb.json | 22 ++- .../supplementary_translation_data/nl.json | 22 ++- .../supplementary_translation_data/nn.json | 22 ++- .../supplementary_translation_data/pl.json | 22 ++- .../supplementary_translation_data/pt-PT.json | 22 ++- .../supplementary_translation_data/pt.json | 22 ++- .../supplementary_translation_data/qu.json | 22 ++- .../supplementary_translation_data/ro.json | 22 ++- .../supplementary_translation_data/root.json | 22 ++- .../supplementary_translation_data/ru.json | 145 ++++++++++-------- .../supplementary_translation_data/se.json | 22 ++- .../supplementary_translation_data/sk.json | 22 ++- .../supplementary_translation_data/sl.json | 22 ++- .../supplementary_translation_data/sq.json | 22 ++- .../sr-Latn.json | 22 ++- .../supplementary_translation_data/sr.json | 22 ++- .../supplementary_translation_data/su.json | 22 ++- .../supplementary_translation_data/sv.json | 22 ++- .../supplementary_translation_data/sw.json | 22 ++- .../supplementary_translation_data/ta.json | 22 ++- .../supplementary_translation_data/th.json | 22 ++- .../supplementary_translation_data/tr.json | 22 ++- .../supplementary_translation_data/uk.json | 22 ++- .../supplementary_translation_data/vi.json | 22 ++- .../yue-Hans.json | 22 ++- .../supplementary_translation_data/yue.json | 22 ++- .../zh-Hant.json | 22 ++- .../supplementary_translation_data/zh.json | 22 ++- scripts/modify_supplementary_data_files.py | 124 +++++++++++++++ 86 files changed, 1561 insertions(+), 588 deletions(-) create mode 100644 scripts/modify_supplementary_data_files.py diff --git a/number_parser_data/supplementary_translation_data/af.json b/number_parser_data/supplementary_translation_data/af.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/af.json +++ b/number_parser_data/supplementary_translation_data/af.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/ak.json b/number_parser_data/supplementary_translation_data/ak.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/ak.json +++ b/number_parser_data/supplementary_translation_data/ak.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/am.json b/number_parser_data/supplementary_translation_data/am.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/am.json +++ b/number_parser_data/supplementary_translation_data/am.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/ar.json b/number_parser_data/supplementary_translation_data/ar.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/ar.json +++ b/number_parser_data/supplementary_translation_data/ar.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/az.json b/number_parser_data/supplementary_translation_data/az.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/az.json +++ b/number_parser_data/supplementary_translation_data/az.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/be.json b/number_parser_data/supplementary_translation_data/be.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/be.json +++ b/number_parser_data/supplementary_translation_data/be.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/bg.json b/number_parser_data/supplementary_translation_data/bg.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/bg.json +++ b/number_parser_data/supplementary_translation_data/bg.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/bs.json b/number_parser_data/supplementary_translation_data/bs.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/bs.json +++ b/number_parser_data/supplementary_translation_data/bs.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/ca.json b/number_parser_data/supplementary_translation_data/ca.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/ca.json +++ b/number_parser_data/supplementary_translation_data/ca.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/ccp.json b/number_parser_data/supplementary_translation_data/ccp.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/ccp.json +++ b/number_parser_data/supplementary_translation_data/ccp.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/chr.json b/number_parser_data/supplementary_translation_data/chr.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/chr.json +++ b/number_parser_data/supplementary_translation_data/chr.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/cs.json b/number_parser_data/supplementary_translation_data/cs.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/cs.json +++ b/number_parser_data/supplementary_translation_data/cs.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/cy.json b/number_parser_data/supplementary_translation_data/cy.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/cy.json +++ b/number_parser_data/supplementary_translation_data/cy.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/da.json b/number_parser_data/supplementary_translation_data/da.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/da.json +++ b/number_parser_data/supplementary_translation_data/da.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/de-CH.json b/number_parser_data/supplementary_translation_data/de-CH.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/de-CH.json +++ b/number_parser_data/supplementary_translation_data/de-CH.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/de.json b/number_parser_data/supplementary_translation_data/de.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/de.json +++ b/number_parser_data/supplementary_translation_data/de.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/ee.json b/number_parser_data/supplementary_translation_data/ee.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/ee.json +++ b/number_parser_data/supplementary_translation_data/ee.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/el.json b/number_parser_data/supplementary_translation_data/el.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/el.json +++ b/number_parser_data/supplementary_translation_data/el.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/en-IN.json b/number_parser_data/supplementary_translation_data/en-IN.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/en-IN.json +++ b/number_parser_data/supplementary_translation_data/en-IN.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/en.json b/number_parser_data/supplementary_translation_data/en.json index 87b3aea..e56908b 100644 --- a/number_parser_data/supplementary_translation_data/en.json +++ b/number_parser_data/supplementary_translation_data/en.json @@ -1,14 +1,24 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "quintillion": 1e18, - "sextillion": 1e21 + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "quintillion": 1e+18, + "sextillion": 1e+21 + } + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, "SKIP_TOKENS": [ "and", "-" - ] + ], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/eo.json b/number_parser_data/supplementary_translation_data/eo.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/eo.json +++ b/number_parser_data/supplementary_translation_data/eo.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/es-419.json b/number_parser_data/supplementary_translation_data/es-419.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/es-419.json +++ b/number_parser_data/supplementary_translation_data/es-419.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/es.json b/number_parser_data/supplementary_translation_data/es.json index 3f67238..d286927 100644 --- a/number_parser_data/supplementary_translation_data/es.json +++ b/number_parser_data/supplementary_translation_data/es.json @@ -1,26 +1,36 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": { - "diecisiete": 17, - "dieciocho": 18, - "diecinueve": 19, - "veintisiete": 27, - "veintiocho": 28, - "veintinueve": 29 + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": { + "diecisiete": 17, + "dieciocho": 18, + "diecinueve": 19, + "veintisiete": 27, + "veintiocho": 28, + "veintinueve": 29 + }, + "TENS": {}, + "HUNDREDS": { + "cien": 100, + "ciento": 100 + }, + "BIG_POWERS_OF_TEN": { + "millón": 1000000, + "millardo": 1000000000, + "billón": 1000000000000, + "trilón": 1000000000000000000, + "trillones": 1000000000000000000 + } }, - "TENS": {}, - "HUNDREDS": { - "cien": 100, - "ciento": 100 - }, - "BIG_POWERS_OF_TEN": { - "millón": 1000000, - "millardo": 1000000000, - "billón": 1000000000000, - "trilón": 1000000000000000000, - "trillones": 1000000000000000000 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, "SKIP_TOKENS": [ "y" - ] + ], + "IS_LONG": true } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/et.json b/number_parser_data/supplementary_translation_data/et.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/et.json +++ b/number_parser_data/supplementary_translation_data/et.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/fa-AF.json b/number_parser_data/supplementary_translation_data/fa-AF.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/fa-AF.json +++ b/number_parser_data/supplementary_translation_data/fa-AF.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/fa.json b/number_parser_data/supplementary_translation_data/fa.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/fa.json +++ b/number_parser_data/supplementary_translation_data/fa.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/ff.json b/number_parser_data/supplementary_translation_data/ff.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/ff.json +++ b/number_parser_data/supplementary_translation_data/ff.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/fi.json b/number_parser_data/supplementary_translation_data/fi.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/fi.json +++ b/number_parser_data/supplementary_translation_data/fi.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/fil.json b/number_parser_data/supplementary_translation_data/fil.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/fil.json +++ b/number_parser_data/supplementary_translation_data/fil.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/fo.json b/number_parser_data/supplementary_translation_data/fo.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/fo.json +++ b/number_parser_data/supplementary_translation_data/fo.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/fr-BE.json b/number_parser_data/supplementary_translation_data/fr-BE.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/fr-BE.json +++ b/number_parser_data/supplementary_translation_data/fr-BE.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/fr-CH.json b/number_parser_data/supplementary_translation_data/fr-CH.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/fr-CH.json +++ b/number_parser_data/supplementary_translation_data/fr-CH.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/fr.json b/number_parser_data/supplementary_translation_data/fr.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/fr.json +++ b/number_parser_data/supplementary_translation_data/fr.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/ga.json b/number_parser_data/supplementary_translation_data/ga.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/ga.json +++ b/number_parser_data/supplementary_translation_data/ga.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/he.json b/number_parser_data/supplementary_translation_data/he.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/he.json +++ b/number_parser_data/supplementary_translation_data/he.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/hi.json b/number_parser_data/supplementary_translation_data/hi.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/hi.json +++ b/number_parser_data/supplementary_translation_data/hi.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/hr.json b/number_parser_data/supplementary_translation_data/hr.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/hr.json +++ b/number_parser_data/supplementary_translation_data/hr.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/hu.json b/number_parser_data/supplementary_translation_data/hu.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/hu.json +++ b/number_parser_data/supplementary_translation_data/hu.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/hy.json b/number_parser_data/supplementary_translation_data/hy.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/hy.json +++ b/number_parser_data/supplementary_translation_data/hy.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/id.json b/number_parser_data/supplementary_translation_data/id.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/id.json +++ b/number_parser_data/supplementary_translation_data/id.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/is.json b/number_parser_data/supplementary_translation_data/is.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/is.json +++ b/number_parser_data/supplementary_translation_data/is.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/it.json b/number_parser_data/supplementary_translation_data/it.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/it.json +++ b/number_parser_data/supplementary_translation_data/it.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/ja.json b/number_parser_data/supplementary_translation_data/ja.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/ja.json +++ b/number_parser_data/supplementary_translation_data/ja.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/ka.json b/number_parser_data/supplementary_translation_data/ka.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/ka.json +++ b/number_parser_data/supplementary_translation_data/ka.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/kl.json b/number_parser_data/supplementary_translation_data/kl.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/kl.json +++ b/number_parser_data/supplementary_translation_data/kl.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/km.json b/number_parser_data/supplementary_translation_data/km.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/km.json +++ b/number_parser_data/supplementary_translation_data/km.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/ko.json b/number_parser_data/supplementary_translation_data/ko.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/ko.json +++ b/number_parser_data/supplementary_translation_data/ko.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/ky.json b/number_parser_data/supplementary_translation_data/ky.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/ky.json +++ b/number_parser_data/supplementary_translation_data/ky.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/lb.json b/number_parser_data/supplementary_translation_data/lb.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/lb.json +++ b/number_parser_data/supplementary_translation_data/lb.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/lo.json b/number_parser_data/supplementary_translation_data/lo.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/lo.json +++ b/number_parser_data/supplementary_translation_data/lo.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/lrc.json b/number_parser_data/supplementary_translation_data/lrc.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/lrc.json +++ b/number_parser_data/supplementary_translation_data/lrc.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/lt.json b/number_parser_data/supplementary_translation_data/lt.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/lt.json +++ b/number_parser_data/supplementary_translation_data/lt.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/lv.json b/number_parser_data/supplementary_translation_data/lv.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/lv.json +++ b/number_parser_data/supplementary_translation_data/lv.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/mk.json b/number_parser_data/supplementary_translation_data/mk.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/mk.json +++ b/number_parser_data/supplementary_translation_data/mk.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/ms.json b/number_parser_data/supplementary_translation_data/ms.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/ms.json +++ b/number_parser_data/supplementary_translation_data/ms.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/mt.json b/number_parser_data/supplementary_translation_data/mt.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/mt.json +++ b/number_parser_data/supplementary_translation_data/mt.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/my.json b/number_parser_data/supplementary_translation_data/my.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/my.json +++ b/number_parser_data/supplementary_translation_data/my.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/nb.json b/number_parser_data/supplementary_translation_data/nb.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/nb.json +++ b/number_parser_data/supplementary_translation_data/nb.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/nl.json b/number_parser_data/supplementary_translation_data/nl.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/nl.json +++ b/number_parser_data/supplementary_translation_data/nl.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/nn.json b/number_parser_data/supplementary_translation_data/nn.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/nn.json +++ b/number_parser_data/supplementary_translation_data/nn.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/pl.json b/number_parser_data/supplementary_translation_data/pl.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/pl.json +++ b/number_parser_data/supplementary_translation_data/pl.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/pt-PT.json b/number_parser_data/supplementary_translation_data/pt-PT.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/pt-PT.json +++ b/number_parser_data/supplementary_translation_data/pt-PT.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/pt.json b/number_parser_data/supplementary_translation_data/pt.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/pt.json +++ b/number_parser_data/supplementary_translation_data/pt.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/qu.json b/number_parser_data/supplementary_translation_data/qu.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/qu.json +++ b/number_parser_data/supplementary_translation_data/qu.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/ro.json b/number_parser_data/supplementary_translation_data/ro.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/ro.json +++ b/number_parser_data/supplementary_translation_data/ro.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/root.json b/number_parser_data/supplementary_translation_data/root.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/root.json +++ b/number_parser_data/supplementary_translation_data/root.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/ru.json b/number_parser_data/supplementary_translation_data/ru.json index e6a07d3..d4815b4 100644 --- a/number_parser_data/supplementary_translation_data/ru.json +++ b/number_parser_data/supplementary_translation_data/ru.json @@ -1,72 +1,81 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": { - "сто": 100, - "ста": 100, - "двести": 200, - "двухсот": 200, - "двумстам": 200, - "двухстах": 200, - "двумястами": 200, - "триста": 300, - "трехсот": 300, - "тремстам": 300, - "трехстах": 300, - "тремястами": 300, - "четыреста": 400, - "четырехсот": 400, - "четыремстам": 400, - "четырехстах": 400, - "четырьмястами": 400, - "пятьсот": 500, - "пятисот": 500, - "пятистам": 500, - "пятистах": 500, - "пятьюстами": 500, - "шестьсот": 600, - "шестисот": 600, - "шестистам": 600, - "шестистах": 600, - "шестьюстами": 600, - "семьсот": 700, - "семисот": 700, - "семистам": 700, - "семистах": 700, - "семьюстами": 700, - "восемьсот": 800, - "восьмисот": 800, - "восьмистам": 800, - "восьмистах": 800, - "восемьюстами": 800, - "девятьсот": 900, - "девятисот": 900, - "девятистам": 900, - "девятистах": 900, - "девятьюстами": 900, - "двасти": 200 + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": { + "сто": 100, + "ста": 100, + "двести": 200, + "двухсот": 200, + "двумстам": 200, + "двухстах": 200, + "двумястами": 200, + "триста": 300, + "трехсот": 300, + "тремстам": 300, + "трехстах": 300, + "тремястами": 300, + "четыреста": 400, + "четырехсот": 400, + "четыремстам": 400, + "четырехстах": 400, + "четырьмястами": 400, + "пятьсот": 500, + "пятисот": 500, + "пятистам": 500, + "пятистах": 500, + "пятьюстами": 500, + "шестьсот": 600, + "шестисот": 600, + "шестистам": 600, + "шестистах": 600, + "шестьюстами": 600, + "семьсот": 700, + "семисот": 700, + "семистам": 700, + "семистах": 700, + "семьюстами": 700, + "восемьсот": 800, + "восьмисот": 800, + "восьмистам": 800, + "восьмистах": 800, + "восемьюстами": 800, + "девятьсот": 900, + "девятисот": 900, + "девятистам": 900, + "девятистах": 900, + "девятьюстами": 900, + "двасти": 200 + }, + "BIG_POWERS_OF_TEN": { + "квинтиллион": 1e+18, + "квинтиллиона": 1e+18, + "квинтиллионам": 1e+18, + "квинтиллионами": 1e+18, + "квинтиллионах": 1e+18, + "квинтиллионе": 1e+18, + "квинтиллионов": 1e+18, + "квинтиллионом": 1e+18, + "квинтиллиону": 1e+18, + "секстиллион": 1e+21, + "секстиллиона": 1e+21, + "секстиллионам": 1e+21, + "секстиллионами": 1e+21, + "секстиллионах": 1e+21, + "секстиллионе": 1e+21, + "секстиллионов": 1e+21, + "секстиллионом": 1e+21, + "секстиллиону": 1e+21 + } }, - "BIG_POWERS_OF_TEN": { - "квинтиллион": 1e18, - "квинтиллиона": 1e18, - "квинтиллионам": 1e18, - "квинтиллионами": 1e18, - "квинтиллионах": 1e18, - "квинтиллионе": 1e18, - "квинтиллионов": 1e18, - "квинтиллионом": 1e18, - "квинтиллиону": 1e18, - - "секстиллион": 1e21, - "секстиллиона": 1e21, - "секстиллионам": 1e21, - "секстиллионами": 1e21, - "секстиллионах": 1e21, - "секстиллионе": 1e21, - "секстиллионов": 1e21, - "секстиллионом": 1e21, - "секстиллиону": 1e21 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/se.json b/number_parser_data/supplementary_translation_data/se.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/se.json +++ b/number_parser_data/supplementary_translation_data/se.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/sk.json b/number_parser_data/supplementary_translation_data/sk.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/sk.json +++ b/number_parser_data/supplementary_translation_data/sk.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/sl.json b/number_parser_data/supplementary_translation_data/sl.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/sl.json +++ b/number_parser_data/supplementary_translation_data/sl.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/sq.json b/number_parser_data/supplementary_translation_data/sq.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/sq.json +++ b/number_parser_data/supplementary_translation_data/sq.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/sr-Latn.json b/number_parser_data/supplementary_translation_data/sr-Latn.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/sr-Latn.json +++ b/number_parser_data/supplementary_translation_data/sr-Latn.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/sr.json b/number_parser_data/supplementary_translation_data/sr.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/sr.json +++ b/number_parser_data/supplementary_translation_data/sr.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/su.json b/number_parser_data/supplementary_translation_data/su.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/su.json +++ b/number_parser_data/supplementary_translation_data/su.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/sv.json b/number_parser_data/supplementary_translation_data/sv.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/sv.json +++ b/number_parser_data/supplementary_translation_data/sv.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/sw.json b/number_parser_data/supplementary_translation_data/sw.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/sw.json +++ b/number_parser_data/supplementary_translation_data/sw.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/ta.json b/number_parser_data/supplementary_translation_data/ta.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/ta.json +++ b/number_parser_data/supplementary_translation_data/ta.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/th.json b/number_parser_data/supplementary_translation_data/th.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/th.json +++ b/number_parser_data/supplementary_translation_data/th.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/tr.json b/number_parser_data/supplementary_translation_data/tr.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/tr.json +++ b/number_parser_data/supplementary_translation_data/tr.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/uk.json b/number_parser_data/supplementary_translation_data/uk.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/uk.json +++ b/number_parser_data/supplementary_translation_data/uk.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/vi.json b/number_parser_data/supplementary_translation_data/vi.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/vi.json +++ b/number_parser_data/supplementary_translation_data/vi.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/yue-Hans.json b/number_parser_data/supplementary_translation_data/yue-Hans.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/yue-Hans.json +++ b/number_parser_data/supplementary_translation_data/yue-Hans.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/yue.json b/number_parser_data/supplementary_translation_data/yue.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/yue.json +++ b/number_parser_data/supplementary_translation_data/yue.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/zh-Hant.json b/number_parser_data/supplementary_translation_data/zh-Hant.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/zh-Hant.json +++ b/number_parser_data/supplementary_translation_data/zh-Hant.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/number_parser_data/supplementary_translation_data/zh.json b/number_parser_data/supplementary_translation_data/zh.json index 0de4bdd..8a8ba3a 100644 --- a/number_parser_data/supplementary_translation_data/zh.json +++ b/number_parser_data/supplementary_translation_data/zh.json @@ -1,8 +1,18 @@ { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": false } \ No newline at end of file diff --git a/scripts/modify_supplementary_data_files.py b/scripts/modify_supplementary_data_files.py new file mode 100644 index 0000000..8274e94 --- /dev/null +++ b/scripts/modify_supplementary_data_files.py @@ -0,0 +1,124 @@ +""" +The raw CLDR data was retreived on 25th Jun , 2020 from the following link +https://github.com/unicode-cldr/cldr-rbnf +""" + +import os +import json +import re +from collections import OrderedDict + +SUPPLEMENTARY_PATH = "../number_parser_data/supplementary_translation_data/" +f = ["UNIT_NUMBERS","DIRECT_NUMBERS","TENS","HUNDREDS","BIG_POWERS_OF_TEN"] +tokens = "SKIP_TOKENS" + +for file_name in os.listdir(SUPPLEMENTARY_PATH): + if file_name == "en.json": + continue + new_dic = {"NUMBERS": {} , "ORDINAL_NUMBERS": {}} + full_supplementary_path = os.path.join(SUPPLEMENTARY_PATH, file_name) + with open(full_supplementary_path, 'r') as source: + data = json.load(source) + for i in range(5): + new_dic["NUMBERS"][f[i]] = data[f[i]] + new_dic["ORDINAL_NUMBERS"][f[i]] = {} + + new_dic[tokens] = data[tokens] + new_dic["IS_LONG"] = False + # print(new_dic,file_name) + # break + mfinal = json.dumps(new_dic, indent=4, ensure_ascii=False) + with open(full_supplementary_path, 'w') as ff: + ff.write(mfinal) + # break + +# VALID_KEYS = ["spellout-cardinal", "spellout-numbering"] +# def _is_valid(key): +# """Identifying whether the given key of the source language file needs to be extracted.""" +# is_valid = False +# for valid_key in VALID_KEYS: +# if valid_key in key: +# is_valid = True +# return is_valid + +# def _count_zero(number): +# """Counting the number of zeroes in the given number.""" +# zero_count = 0 +# while number > 9: +# if number % 10 == 0: +# zero_count += 1 +# number /= 10 +# else: +# break +# return zero_count + + +# fset = {} + + +# PATH_TO_FILE = "/home/arnav/GSOC_2020/number-parser/number_parser_data/raw_cldr_translation_data/" +# with open(PATH_TO_FILE + "ru.json") as f: +# data = json.load(f) +# requisite_data = data['rbnf']['rbnf']['SpelloutRules'] +# for keys, vals in requisite_data.items(): +# if _is_valid(keys): +# for key, val in vals.items(): +# try: +# if _count_zero((int)(key)) == 2: +# if key[0] == "1": +# continue +# get_vals = val.split("<") +# get_val = get_vals[1] +# suffix = get_vals[-1].split("[")[0] +# rq1 = get_val[1:] +# rq2 = rq1.replace("feminine", "masculine") +# rq3 = rq1.replace("masculine", "feminine") + +# fd = (int)(key) // 100 +# d1 = requisite_data["%" + rq1] +# d2 = requisite_data["%" + rq2] +# d3 = requisite_data["%" + rq3] + +# if(get_vals[-1] == "сти[ >>];"): +# for i in range(fd,3): +# prefix = d3[str(i)] +# prefix = prefix[:-1] +# fset[prefix+suffix] = i*100 +# prefix = d2[str(i)] +# prefix = prefix[:-1] +# fset[prefix+suffix] = i*100 + + +# elif (get_vals[-1] == "ста[ >>];"): +# for i in range(fd,5): +# prefix = d2[str(i)] +# prefix = prefix[:-1] +# fset[prefix+suffix] = i*100 +# # print(i*100,prefix+suffix) +# else: +# for i in range(fd,10): +# prefix = d2[str(i)] +# prefix = prefix[:-1] +# fset[prefix+suffix] = i*100 +# # print(i*100,prefix+suffix) +# # print(d2) +# except: +# pass + +# fdict = OrderedDict() +# mvals = (sorted(fset.items(), key=lambda x: x[1])) + +# for each in mvals: +# fdict[each[0]] = each[1] + +# RT_PATH = "/home/arnav/GSOC_2020/number-parser/number_parser_data/supplementary_translation_data" +# mfinal = {} +# with open(RT_PATH + "/ru.json" ) as f: +# data = json.load(f) +# data["HUNDREDS"].update(fdict) +# mfinal = json.dumps(data, indent=4, ensure_ascii=False) +# print(mfinal) +# with open(RT_PATH + "/ru.json","w" ) as ff: +# ff.write(mfinal) + + From 00bc9a3cdb2a4f528a9ff871bb497e09edb109f3 Mon Sep 17 00:00:00 2001 From: arnavkapoor Date: Tue, 4 Aug 2020 19:46:10 +0530 Subject: [PATCH 2/4] updating parser for ordinal number support --- number_parser/parser.py | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-) diff --git a/number_parser/parser.py b/number_parser/parser.py index f648acc..934feb5 100644 --- a/number_parser/parser.py +++ b/number_parser/parser.py @@ -4,7 +4,6 @@ SENTENCE_SEPARATORS = [".", ","] SUPPORTED_LANGUAGES = ['en', 'es', 'hi', 'ru'] RE_BUG_LANGUAGES = ['hi'] -LONG_SCALES_LANGUAGES = ['es'] class LanguageData: @@ -22,18 +21,18 @@ def __init__(self, language): if language not in SUPPORTED_LANGUAGES: raise ValueError(f'"{language}" is not a supported language') language_info = getattr(import_module('number_parser.data.' + language), 'info') - self.unit_numbers = _normalize_dict(language_info["UNIT_NUMBERS"]) - self.direct_numbers = _normalize_dict(language_info["DIRECT_NUMBERS"]) - self.tens = _normalize_dict(language_info["TENS"]) - self.hundreds = _normalize_dict(language_info["HUNDREDS"]) - self.big_powers_of_ten = _normalize_dict(language_info["BIG_POWERS_OF_TEN"]) + self.unit_numbers = _normalize_dict(language_info["NUMBERS"]["UNIT_NUMBERS"]) + self.direct_numbers = _normalize_dict(language_info["NUMBERS"]["DIRECT_NUMBERS"]) + self.tens = _normalize_dict(language_info["NUMBERS"]["TENS"]) + self.hundreds = _normalize_dict(language_info["NUMBERS"]["HUNDREDS"]) + self.big_powers_of_ten = _normalize_dict(language_info["NUMBERS"]["BIG_POWERS_OF_TEN"]) self.skip_tokens = language_info["SKIP_TOKENS"] self.all_numbers = {**self.unit_numbers, **self.direct_numbers, **self.tens, **self.hundreds, **self.big_powers_of_ten} self.unit_and_direct_numbers = {**self.unit_numbers, **self.direct_numbers} self.maximum_group_value = 100 - if language in LONG_SCALES_LANGUAGES: + if language_info["IS_LONG"]: self.maximum_group_value = 10000 @@ -175,6 +174,7 @@ def parse_number(input_string, language='en'): if token in lang_data.skip_tokens and index != 0: continue return None + print(normalized_tokens) number_built = _build_number(normalized_tokens, lang_data) if len(number_built) == 1: return int(number_built[0]) From 14221303c8c20f2c6322b409c40887ccc1b91915 Mon Sep 17 00:00:00 2001 From: arnavkapoor Date: Tue, 4 Aug 2020 19:46:25 +0530 Subject: [PATCH 3/4] updated structure of py files --- number_parser/data/af.py | 72 +-- number_parser/data/ak.py | 48 +- number_parser/data/am.py | 54 ++- number_parser/data/ar.py | 72 +-- number_parser/data/az.py | 72 +-- number_parser/data/be.py | 142 +++--- number_parser/data/bg.py | 122 ++--- number_parser/data/bs.py | 116 ++--- number_parser/data/ca.py | 98 ++-- number_parser/data/ccp.py | 88 ++-- number_parser/data/chr.py | 94 ++-- number_parser/data/cs.py | 116 ++--- number_parser/data/cy.py | 68 +-- number_parser/data/da.py | 78 +-- number_parser/data/de-CH.py | 72 +-- number_parser/data/de.py | 72 +-- number_parser/data/ee.py | 26 +- number_parser/data/el.py | 144 +++--- number_parser/data/en-IN.py | 94 ++-- number_parser/data/en.py | 98 ++-- number_parser/data/eo.py | 52 +- number_parser/data/es.py | 162 ++++--- number_parser/data/et.py | 64 +-- number_parser/data/fa-AF.py | 92 ++-- number_parser/data/fa.py | 108 +++-- number_parser/data/ff.py | 62 ++- number_parser/data/fi.py | 604 ++++++++++++------------ number_parser/data/fil.py | 58 ++- number_parser/data/fo.py | 102 ++-- number_parser/data/fr-BE.py | 90 ++-- number_parser/data/fr-CH.py | 90 ++-- number_parser/data/fr.py | 82 ++-- number_parser/data/ga.py | 88 ++-- number_parser/data/he.py | 110 +++-- number_parser/data/hi.py | 236 ++++----- number_parser/data/hr.py | 130 ++--- number_parser/data/hu.py | 64 +-- number_parser/data/hy.py | 70 +-- number_parser/data/id.py | 60 ++- number_parser/data/is.py | 108 +++-- number_parser/data/it.py | 94 ++-- number_parser/data/ja.py | 54 ++- number_parser/data/ka.py | 92 ++-- number_parser/data/kl.py | 76 +-- number_parser/data/km.py | 58 ++- number_parser/data/ko.py | 90 ++-- number_parser/data/ky.py | 72 +-- number_parser/data/lb.py | 86 ++-- number_parser/data/lo.py | 44 +- number_parser/data/lrc.py | 108 +++-- number_parser/data/lt.py | 80 ++-- number_parser/data/lv.py | 70 +-- number_parser/data/mk.py | 98 ++-- number_parser/data/ms.py | 60 ++- number_parser/data/mt.py | 118 ++--- number_parser/data/my.py | 44 +- number_parser/data/nb.py | 98 ++-- number_parser/data/nl.py | 72 +-- number_parser/data/nn.py | 96 ++-- number_parser/data/pl.py | 284 +++++------ number_parser/data/pt-PT.py | 130 ++--- number_parser/data/pt.py | 130 ++--- number_parser/data/qu.py | 54 ++- number_parser/data/ro.py | 72 +-- number_parser/data/root.py | 22 +- number_parser/data/ru.py | 434 ++++++++--------- number_parser/data/se.py | 54 ++- number_parser/data/sk.py | 104 ++-- number_parser/data/sl.py | 110 +++-- number_parser/data/sq.py | 66 +-- number_parser/data/sr-Latn.py | 106 +++-- number_parser/data/sr.py | 106 +++-- number_parser/data/su.py | 48 +- number_parser/data/sv.py | 94 ++-- number_parser/data/sw.py | 58 ++- number_parser/data/ta.py | 106 +++-- number_parser/data/th.py | 44 +- number_parser/data/tr.py | 72 +-- number_parser/data/uk.py | 136 +++--- number_parser/data/vi.py | 50 +- number_parser/data/yue-Hans.py | 62 ++- number_parser/data/yue.py | 62 ++- number_parser/data/zh-Hant.py | 64 +-- number_parser/data/zh.py | 64 +-- scripts/write_complete_language_data.py | 43 +- 85 files changed, 4558 insertions(+), 3705 deletions(-) diff --git a/number_parser/data/af.py b/number_parser/data/af.py index 5aa6e2d..c7e0126 100644 --- a/number_parser/data/af.py +++ b/number_parser/data/af.py @@ -1,35 +1,45 @@ info = { - "UNIT_NUMBERS": { - "nul": 0, - "een": 1, - "twee": 2, - "drie": 3, - "vier": 4, - "vyf": 5, - "ses": 6, - "sewe": 7, - "agt": 8, - "nege": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "nul": 0, + "een": 1, + "twee": 2, + "drie": 3, + "vier": 4, + "vyf": 5, + "ses": 6, + "sewe": 7, + "agt": 8, + "nege": 9 + }, + "DIRECT_NUMBERS": { + "tien": 10, + "elf": 11, + "twaalf": 12, + "dertien": 13, + "veertien": 14, + "vyftien": 15, + "sestien": 16, + "sewentien": 17, + "agttien": 18, + "negentien": 19 + }, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "miljoen": 1000000, + "miljard": 1000000000, + "biljoen": 1000000000000, + "biljard": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "tien": 10, - "elf": 11, - "twaalf": 12, - "dertien": 13, - "veertien": 14, - "vyftien": 15, - "sestien": 16, - "sewentien": 17, - "agttien": 18, - "negentien": 19 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "miljoen": 1000000, - "miljard": 1000000000, - "biljoen": 1000000000000, - "biljard": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/ak.py b/number_parser/data/ak.py index fa4a189..d4b004b 100644 --- a/number_parser/data/ak.py +++ b/number_parser/data/ak.py @@ -1,23 +1,33 @@ info = { - "UNIT_NUMBERS": { - "hwee": 0, - "biako": 1, - "koro": 1, - "abien": 2, - "abiasa": 3, - "anan": 4, - "anum": 5, - "asia": 6, - "asuon": 7, - "awɔtwe": 8, - "akron": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "hwee": 0, + "biako": 1, + "koro": 1, + "abien": 2, + "abiasa": 3, + "anan": 4, + "anum": 5, + "asia": 6, + "asuon": 7, + "awɔtwe": 8, + "akron": 9 + }, + "DIRECT_NUMBERS": {}, + "TENS": { + "aduonu": 20, + "aduasa": 30 + }, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "DIRECT_NUMBERS": {}, - "TENS": { - "aduonu": 20, - "aduasa": 30 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/am.py b/number_parser/data/am.py index f3f2d05..8c27c63 100644 --- a/number_parser/data/am.py +++ b/number_parser/data/am.py @@ -1,26 +1,36 @@ info = { - "UNIT_NUMBERS": { - "ባዶ": 0, - "አንድ": 1, - "ሁለት": 2, - "ሦስት": 3, - "አራት": 4, - "አምስት": 5, - "ስድስት": 6, - "ሰባት": 7, - "ስምንት": 8, - "ዘጠኝ": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "ባዶ": 0, + "አንድ": 1, + "ሁለት": 2, + "ሦስት": 3, + "አራት": 4, + "አምስት": 5, + "ስድስት": 6, + "ሰባት": 7, + "ስምንት": 8, + "ዘጠኝ": 9 + }, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "መቶ": 100, + "ሺ": 1000, + "ሚሊዮን": 1000000, + "ቢሊዮን": 1000000000, + "ቲሪሊዮን": 1000000000000, + "ቈዲሪሊዮን": 1000000000000000 + } }, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "መቶ": 100, - "ሺ": 1000, - "ሚሊዮን": 1000000, - "ቢሊዮን": 1000000000, - "ቲሪሊዮን": 1000000000000, - "ቈዲሪሊዮን": 1000000000000000 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/ar.py b/number_parser/data/ar.py index 6e01a62..2b2dc1a 100644 --- a/number_parser/data/ar.py +++ b/number_parser/data/ar.py @@ -1,35 +1,45 @@ info = { - "UNIT_NUMBERS": { - "صفر": 0, - "واحد": 1, - "واحدة": 1, - "إثنان": 2, - "إثنتان": 2, - "ثلاثة": 3, - "أربعة": 4, - "خمسة": 5, - "ستة": 6, - "سبعة": 7, - "ثمانية": 8, - "تسعة": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "صفر": 0, + "واحد": 1, + "واحدة": 1, + "إثنان": 2, + "إثنتان": 2, + "ثلاثة": 3, + "أربعة": 4, + "خمسة": 5, + "ستة": 6, + "سبعة": 7, + "ثمانية": 8, + "تسعة": 9 + }, + "DIRECT_NUMBERS": { + "عشرة": 10, + "إحدى عشر": 11, + "إثنا عشر": 12, + "إثنتا عشرة": 12 + }, + "TENS": {}, + "HUNDREDS": { + "مائتان": 200 + }, + "BIG_POWERS_OF_TEN": { + "مائة": 100, + "آلاف": 1000, + "مليون": 1000000, + "مليار": 1000000000, + "ترليون": 1000000000000, + "كوادرليون": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "عشرة": 10, - "إحدى عشر": 11, - "إثنا عشر": 12, - "إثنتا عشرة": 12 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": {}, - "HUNDREDS": { - "مائتان": 200 - }, - "BIG_POWERS_OF_TEN": { - "مائة": 100, - "آلاف": 1000, - "مليون": 1000000, - "مليار": 1000000000, - "ترليون": 1000000000000, - "كوادرليون": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/az.py b/number_parser/data/az.py index adea8e3..ccc845c 100644 --- a/number_parser/data/az.py +++ b/number_parser/data/az.py @@ -1,35 +1,45 @@ info = { - "UNIT_NUMBERS": { - "sıfır": 0, - "bir": 1, - "iki": 2, - "üç": 3, - "dörd": 4, - "beş": 5, - "altı": 6, - "yeddi": 7, - "səkkiz": 8, - "doqquz": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "sıfır": 0, + "bir": 1, + "iki": 2, + "üç": 3, + "dörd": 4, + "beş": 5, + "altı": 6, + "yeddi": 7, + "səkkiz": 8, + "doqquz": 9 + }, + "DIRECT_NUMBERS": {}, + "TENS": { + "iyirmi": 20, + "otuz": 30, + "qırx": 40, + "əlli": 50, + "atmış": 60, + "yetmiş": 70, + "səqsən": 80, + "doxsan": 90 + }, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "yüz": 100, + "min": 1000, + "milyon": 1000000, + "milyard": 1000000000, + "trilyon": 1000000000000, + "katrilyon": 1000000000000000 + } }, - "DIRECT_NUMBERS": {}, - "TENS": { - "iyirmi": 20, - "otuz": 30, - "qırx": 40, - "əlli": 50, - "atmış": 60, - "yetmiş": 70, - "səqsən": 80, - "doxsan": 90 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "yüz": 100, - "min": 1000, - "milyon": 1000000, - "milyard": 1000000000, - "trilyon": 1000000000000, - "katrilyon": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/be.py b/number_parser/data/be.py index 7fce197..89b59a7 100644 --- a/number_parser/data/be.py +++ b/number_parser/data/be.py @@ -1,70 +1,80 @@ info = { - "UNIT_NUMBERS": { - "нуль": 0, - "адзiн": 1, - "адна": 1, - "адно": 1, - "два": 2, - "дзве": 2, - "тры": 3, - "чатыры": 4, - "пяць": 5, - "шэсць": 6, - "сем": 7, - "восем": 8, - "дзевяць": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "нуль": 0, + "адзiн": 1, + "адна": 1, + "адно": 1, + "два": 2, + "дзве": 2, + "тры": 3, + "чатыры": 4, + "пяць": 5, + "шэсць": 6, + "сем": 7, + "восем": 8, + "дзевяць": 9 + }, + "DIRECT_NUMBERS": { + "дзесяць": 10, + "адзінаццаць": 11, + "дванаццаць": 12, + "трынаццаць": 13, + "чатырнаццаць": 14, + "пятнаццаць": 15, + "шаснаццаць": 16, + "сямнаццаць": 17, + "васямнаццаць": 18, + "дзевятнаццаць": 19 + }, + "TENS": { + "дваццаць": 20, + "трыццаць": 30, + "сорак": 40, + "пяцьдзесят": 50, + "пяцьдзясят": 50, + "шэсцьдзесят": 60, + "семдзесят": 70, + "восемдзесят": 80, + "дзевяноста": 90 + }, + "HUNDREDS": { + "дзвесце": 200, + "трыста": 300, + "чатырыста": 400, + "пяцьсот": 500, + "шэсцьсот": 600, + "семсот": 700, + "сямсот": 700, + "васямсот": 800, + "восемсот": 800, + "дзевяцьсот": 900 + }, + "BIG_POWERS_OF_TEN": { + "тысяч": 1000, + "тысяча": 1000, + "тысячы": 1000, + "мільён": 1000000, + "мільёнаў": 1000000, + "мільёны": 1000000, + "мільярд": 1000000000, + "мільярдаў": 1000000000, + "мільярды": 1000000000, + "трыльён": 1000000000000, + "трыльёны": 1000000000000, + "трылёнаў": 1000000000000, + "квадрыльён": 1000000000000000, + "квадрыльёнаў": 1000000000000000, + "квадрыльёны": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "дзесяць": 10, - "адзінаццаць": 11, - "дванаццаць": 12, - "трынаццаць": 13, - "чатырнаццаць": 14, - "пятнаццаць": 15, - "шаснаццаць": 16, - "сямнаццаць": 17, - "васямнаццаць": 18, - "дзевятнаццаць": 19 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": { - "дваццаць": 20, - "трыццаць": 30, - "сорак": 40, - "пяцьдзесят": 50, - "пяцьдзясят": 50, - "шэсцьдзесят": 60, - "семдзесят": 70, - "восемдзесят": 80, - "дзевяноста": 90 - }, - "HUNDREDS": { - "дзвесце": 200, - "трыста": 300, - "чатырыста": 400, - "пяцьсот": 500, - "шэсцьсот": 600, - "семсот": 700, - "сямсот": 700, - "васямсот": 800, - "восемсот": 800, - "дзевяцьсот": 900 - }, - "BIG_POWERS_OF_TEN": { - "тысяч": 1000, - "тысяча": 1000, - "тысячы": 1000, - "мільён": 1000000, - "мільёнаў": 1000000, - "мільёны": 1000000, - "мільярд": 1000000000, - "мільярдаў": 1000000000, - "мільярды": 1000000000, - "трыльён": 1000000000000, - "трыльёны": 1000000000000, - "трылёнаў": 1000000000000, - "квадрыльён": 1000000000000000, - "квадрыльёнаў": 1000000000000000, - "квадрыльёны": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/bg.py b/number_parser/data/bg.py index c6c15ff..bc14471 100644 --- a/number_parser/data/bg.py +++ b/number_parser/data/bg.py @@ -1,60 +1,70 @@ info = { - "UNIT_NUMBERS": { - "нула": 0, - "един": 1, - "едно": 1, - "два": 2, - "две": 2, - "три": 3, - "четири": 4, - "пет": 5, - "шест": 6, - "седем": 7, - "осем": 8, - "девет": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "нула": 0, + "един": 1, + "едно": 1, + "два": 2, + "две": 2, + "три": 3, + "четири": 4, + "пет": 5, + "шест": 6, + "седем": 7, + "осем": 8, + "девет": 9 + }, + "DIRECT_NUMBERS": { + "десет": 10, + "единадесет": 11, + "дванадесет": 12, + "тринадесет": 13, + "четиринадесет": 14, + "петнадесет": 15, + "шестнадесет": 16, + "седемнадесет": 17, + "осемнадесет": 18, + "деветнадесет": 19 + }, + "TENS": { + "двадесет": 20, + "тридесет": 30, + "четиридесет": 40, + "петдесет": 50, + "шестдесет": 60, + "седемдесет": 70, + "осемдесет": 80, + "деветдесет": 90 + }, + "HUNDREDS": { + "двеста": 200, + "триста": 300, + "четиристотин": 400, + "петстотин": 500, + "шестстотин": 600, + "седемстотин": 700, + "осемстотин": 800, + "деветстотин": 900 + }, + "BIG_POWERS_OF_TEN": { + "хиляди": 1000, + "милион": 1000000, + "милиона": 1000000, + "милиард": 1000000000, + "милиарда": 1000000000, + "трилион": 1000000000000, + "трилиона": 1000000000000, + "квадрилион": 1000000000000000, + "квадрилиона": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "десет": 10, - "единадесет": 11, - "дванадесет": 12, - "тринадесет": 13, - "четиринадесет": 14, - "петнадесет": 15, - "шестнадесет": 16, - "седемнадесет": 17, - "осемнадесет": 18, - "деветнадесет": 19 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": { - "двадесет": 20, - "тридесет": 30, - "четиридесет": 40, - "петдесет": 50, - "шестдесет": 60, - "седемдесет": 70, - "осемдесет": 80, - "деветдесет": 90 - }, - "HUNDREDS": { - "двеста": 200, - "триста": 300, - "четиристотин": 400, - "петстотин": 500, - "шестстотин": 600, - "седемстотин": 700, - "осемстотин": 800, - "деветстотин": 900 - }, - "BIG_POWERS_OF_TEN": { - "хиляди": 1000, - "милион": 1000000, - "милиона": 1000000, - "милиард": 1000000000, - "милиарда": 1000000000, - "трилион": 1000000000000, - "трилиона": 1000000000000, - "квадрилион": 1000000000000000, - "квадрилиона": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/bs.py b/number_parser/data/bs.py index 2d92e39..bed13f3 100644 --- a/number_parser/data/bs.py +++ b/number_parser/data/bs.py @@ -1,57 +1,67 @@ info = { - "UNIT_NUMBERS": { - "nula": 0, - "jedan": 1, - "jedinica": 1, - "jedno": 1, - "dva": 2, - "dve": 2, - "tri": 3, - "četiri": 4, - "pet": 5, - "šest": 6, - "sedam": 7, - "osam": 8, - "devet": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "nula": 0, + "jedan": 1, + "jedinica": 1, + "jedno": 1, + "dva": 2, + "dve": 2, + "tri": 3, + "četiri": 4, + "pet": 5, + "šest": 6, + "sedam": 7, + "osam": 8, + "devet": 9 + }, + "DIRECT_NUMBERS": { + "deset": 10, + "jedenaest": 11, + "dvanaest": 12, + "trinaest": 13, + "četrnaest": 14, + "petnaest": 15, + "šestnaest": 16, + "sedamnaest": 17, + "osamnaest": 18, + "devetnaest": 19 + }, + "TENS": { + "dvadeset": 20, + "trideset": 30, + "četrdeset": 40, + "pedeset": 50, + "šezdeset": 60, + "sedamdeset": 70, + "osamdeset": 80, + "devedeset": 90 + }, + "HUNDREDS": { + "dvesta": 200, + "trista": 300, + "četristo": 400, + "petsto": 500, + "šesto": 600, + "sedamsto": 700, + "osamsto": 800, + "devetsto": 900 + }, + "BIG_POWERS_OF_TEN": { + "hiljada": 1000, + "milion": 1000000, + "miliard": 1000000000, + "bilion": 1000000000000, + "biliard": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "deset": 10, - "jedenaest": 11, - "dvanaest": 12, - "trinaest": 13, - "četrnaest": 14, - "petnaest": 15, - "šestnaest": 16, - "sedamnaest": 17, - "osamnaest": 18, - "devetnaest": 19 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": { - "dvadeset": 20, - "trideset": 30, - "četrdeset": 40, - "pedeset": 50, - "šezdeset": 60, - "sedamdeset": 70, - "osamdeset": 80, - "devedeset": 90 - }, - "HUNDREDS": { - "dvesta": 200, - "trista": 300, - "četristo": 400, - "petsto": 500, - "šesto": 600, - "sedamsto": 700, - "osamsto": 800, - "devetsto": 900 - }, - "BIG_POWERS_OF_TEN": { - "hiljada": 1000, - "milion": 1000000, - "miliard": 1000000000, - "bilion": 1000000000000, - "biliard": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/ca.py b/number_parser/data/ca.py index db3c23f..92e3a1f 100644 --- a/number_parser/data/ca.py +++ b/number_parser/data/ca.py @@ -1,48 +1,58 @@ info = { - "UNIT_NUMBERS": { - "zero": 0, - "u": 1, - "un": 1, - "una": 1, - "dos": 2, - "dues": 2, - "tres": 3, - "quatre": 4, - "cinc": 5, - "sis": 6, - "set": 7, - "vuit": 8, - "nou": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "zero": 0, + "u": 1, + "un": 1, + "una": 1, + "dos": 2, + "dues": 2, + "tres": 3, + "quatre": 4, + "cinc": 5, + "sis": 6, + "set": 7, + "vuit": 8, + "nou": 9 + }, + "DIRECT_NUMBERS": { + "deu": 10, + "onze": 11, + "dotze": 12, + "tretze": 13, + "catorze": 14, + "quinze": 15, + "setze": 16, + "disset": 17, + "divuit": 18, + "dinou": 19 + }, + "TENS": { + "vint": 20, + "trenta": 30, + "quaranta": 40, + "cinquanta": 50, + "seixanta": 60, + "setanta": 70, + "vuitanta": 80, + "noranta": 90 + }, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "mil": 1000, + "milions": 1000000, + "miliards": 1000000000, + "bilions": 1000000000000, + "biliards": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "deu": 10, - "onze": 11, - "dotze": 12, - "tretze": 13, - "catorze": 14, - "quinze": 15, - "setze": 16, - "disset": 17, - "divuit": 18, - "dinou": 19 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": { - "vint": 20, - "trenta": 30, - "quaranta": 40, - "cinquanta": 50, - "seixanta": 60, - "setanta": 70, - "vuitanta": 80, - "noranta": 90 - }, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "mil": 1000, - "milions": 1000000, - "miliards": 1000000000, - "bilions": 1000000000000, - "biliards": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/ccp.py b/number_parser/data/ccp.py index 6d233c9..45a2be1 100644 --- a/number_parser/data/ccp.py +++ b/number_parser/data/ccp.py @@ -1,43 +1,53 @@ info = { - "UNIT_NUMBERS": { - "𑄥𑄪𑄚𑄳𑄠𑄴𑄧": 0, - "𑄆𑄇𑄴": 1, - "𑄘𑄨": 2, - "𑄖𑄨𑄚𑄴": 3, - "𑄌𑄳𑄆𑄬𑄢𑄴": 4, - "𑄛𑄌𑄴": 5, - "𑄍𑄧": 6, - "𑄥𑄖𑄴": 7, - "𑄃𑄖𑄳𑄠𑄴𑄧": 8, - "𑄚𑄧": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "𑄥𑄪𑄚𑄳𑄠𑄴𑄧": 0, + "𑄆𑄇𑄴": 1, + "𑄘𑄨": 2, + "𑄖𑄨𑄚𑄴": 3, + "𑄌𑄳𑄆𑄬𑄢𑄴": 4, + "𑄛𑄌𑄴": 5, + "𑄍𑄧": 6, + "𑄥𑄖𑄴": 7, + "𑄃𑄖𑄳𑄠𑄴𑄧": 8, + "𑄚𑄧": 9 + }, + "DIRECT_NUMBERS": { + "𑄘𑄧𑄌𑄴": 10, + "𑄆𑄉𑄢𑄧": 11, + "𑄝𑄢𑄳𑄦𑄧": 12, + "𑄖𑄬𑄢𑄳𑄦𑄧": 13, + "𑄌𑄮𑄖𑄴𑄙𑄮": 14, + "𑄛𑄧𑄚𑄴𑄘𑄳𑄢𑄧": 15, + "𑄥𑄪𑄣𑄮": 16, + "𑄥𑄧𑄖𑄴𑄧𑄢𑄧": 17, + "𑄃𑄘𑄧𑄢𑄳𑄦𑄧": 18, + "𑄃𑄪𑄚𑄴𑄮𑄌𑄴": 19 + }, + "TENS": { + "𑄇𑄪𑄢𑄨": 20, + "𑄖𑄳𑄢𑄨𑄌𑄴": 30, + "𑄌𑄣𑄨𑄨𑄌𑄴": 40, + "𑄛𑄧𑄚𑄴𑄎𑄌𑄴": 50, + "𑄦𑄬𑄖𑄴": 60, + "𑄦𑄮𑄖𑄴𑄪𑄢𑄴": 70, + "𑄃𑄎𑄨": 80, + "𑄚𑄧𑄛𑄴𑄝𑄰": 90 + }, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "𑄦𑄎𑄢𑄴": 1000, + "𑄣𑄇𑄴": 100000, + "𑄇𑄪𑄖𑄨": 10000000 + } }, - "DIRECT_NUMBERS": { - "𑄘𑄧𑄌𑄴": 10, - "𑄆𑄉𑄢𑄧": 11, - "𑄝𑄢𑄳𑄦𑄧": 12, - "𑄖𑄬𑄢𑄳𑄦𑄧": 13, - "𑄌𑄮𑄖𑄴𑄙𑄮": 14, - "𑄛𑄧𑄚𑄴𑄘𑄳𑄢𑄧": 15, - "𑄥𑄪𑄣𑄮": 16, - "𑄥𑄧𑄖𑄴𑄧𑄢𑄧": 17, - "𑄃𑄘𑄧𑄢𑄳𑄦𑄧": 18, - "𑄃𑄪𑄚𑄴𑄮𑄌𑄴": 19 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": { - "𑄇𑄪𑄢𑄨": 20, - "𑄖𑄳𑄢𑄨𑄌𑄴": 30, - "𑄌𑄣𑄨𑄨𑄌𑄴": 40, - "𑄛𑄧𑄚𑄴𑄎𑄌𑄴": 50, - "𑄦𑄬𑄖𑄴": 60, - "𑄦𑄮𑄖𑄴𑄪𑄢𑄴": 70, - "𑄃𑄎𑄨": 80, - "𑄚𑄧𑄛𑄴𑄝𑄰": 90 - }, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "𑄦𑄎𑄢𑄴": 1000, - "𑄣𑄇𑄴": 100000, - "𑄇𑄪𑄖𑄨": 10000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/chr.py b/number_parser/data/chr.py index ccc4957..8df1e81 100644 --- a/number_parser/data/chr.py +++ b/number_parser/data/chr.py @@ -1,46 +1,56 @@ info = { - "UNIT_NUMBERS": { - "ꮭ ꭺꮝꮧ": 0, - "ꮠꮼ": 1, - "ꮤꮅ": 2, - "ꮶꭲ": 3, - "ꮕꭹ": 4, - "ꭿꮝꭹ": 5, - "ꮡꮣꮅ": 6, - "ꭶꮅꮙꭹ": 7, - "ꮷꮑꮃ": 8, - "ꮠꮑꮃ": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "ꮭ ꭺꮝꮧ": 0, + "ꮠꮼ": 1, + "ꮤꮅ": 2, + "ꮶꭲ": 3, + "ꮕꭹ": 4, + "ꭿꮝꭹ": 5, + "ꮡꮣꮅ": 6, + "ꭶꮅꮙꭹ": 7, + "ꮷꮑꮃ": 8, + "ꮠꮑꮃ": 9 + }, + "DIRECT_NUMBERS": { + "ꮝꭺꭿ": 10, + "ꮜꮪ": 11, + "ꮤꮅꮪ": 12, + "ꮶꭶꮪ": 13, + "ꮒꭶꮪ": 14, + "ꭿꮝꭶꮪ": 15, + "ꮣꮃꮪ": 16, + "ꭶꮅꮖꮪ": 17, + "ꮑꮃꮪ": 18, + "ꮠꮑꮃꮪ": 19 + }, + "TENS": { + "ꮤꮅꮝꭺ": 20, + "ꮶꭲꮝꭺ": 30, + "ꮕꭹꮝꭺ": 40, + "ꭿꮝꭹꮝꭺ": 50, + "ꮡꮣꮅꮝꭺ": 60, + "ꭶꮅꮖꮝꭺ": 70, + "ꮷꮑꮃꮝꭺ": 80, + "ꮠꮑꮃꮝꭺ": 90 + }, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "ꮝꭺꭿꮵꮖ": 100, + "ꭲꮿꭶᏼꮅ": 1000, + "ꭲᏻꮖꮧꮕꮣ": 1000000, + "ꭲꮿꮤꮃꮧꮕꮫ": 1000000000, + "ꭲꮿꮶꭰꮧꮕꮫ": 1000000000000, + "ꭲꮿꮕꭶꮧꮕꮫ": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "ꮝꭺꭿ": 10, - "ꮜꮪ": 11, - "ꮤꮅꮪ": 12, - "ꮶꭶꮪ": 13, - "ꮒꭶꮪ": 14, - "ꭿꮝꭶꮪ": 15, - "ꮣꮃꮪ": 16, - "ꭶꮅꮖꮪ": 17, - "ꮑꮃꮪ": 18, - "ꮠꮑꮃꮪ": 19 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": { - "ꮤꮅꮝꭺ": 20, - "ꮶꭲꮝꭺ": 30, - "ꮕꭹꮝꭺ": 40, - "ꭿꮝꭹꮝꭺ": 50, - "ꮡꮣꮅꮝꭺ": 60, - "ꭶꮅꮖꮝꭺ": 70, - "ꮷꮑꮃꮝꭺ": 80, - "ꮠꮑꮃꮝꭺ": 90 - }, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "ꮝꭺꭿꮵꮖ": 100, - "ꭲꮿꭶᏼꮅ": 1000, - "ꭲᏻꮖꮧꮕꮣ": 1000000, - "ꭲꮿꮤꮃꮧꮕꮫ": 1000000000, - "ꭲꮿꮶꭰꮧꮕꮫ": 1000000000000, - "ꭲꮿꮕꭶꮧꮕꮫ": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/cs.py b/number_parser/data/cs.py index ac67136..bd07ab8 100644 --- a/number_parser/data/cs.py +++ b/number_parser/data/cs.py @@ -1,57 +1,67 @@ info = { - "UNIT_NUMBERS": { - "nula": 0, - "jeden": 1, - "jedna": 1, - "jedno": 1, - "dva": 2, - "dvě": 2, - "tři": 3, - "čtyři": 4, - "pět": 5, - "šest": 6, - "sedm": 7, - "osm": 8, - "devět": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "nula": 0, + "jeden": 1, + "jedna": 1, + "jedno": 1, + "dva": 2, + "dvě": 2, + "tři": 3, + "čtyři": 4, + "pět": 5, + "šest": 6, + "sedm": 7, + "osm": 8, + "devět": 9 + }, + "DIRECT_NUMBERS": { + "deset": 10, + "jedenáct": 11, + "dvanáct": 12, + "třináct": 13, + "čtrnáct": 14, + "patnáct": 15, + "šestnáct": 16, + "sedmnáct": 17, + "osmnáct": 18, + "devatenáct": 19 + }, + "TENS": { + "padesát": 50, + "šedesát": 60, + "sedmdesát": 70, + "osmdesát": 80, + "devadesát": 90 + }, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "set": 100, + "sta": 100, + "stě": 100, + "tisíc": 1000, + "tisíce": 1000, + "milión": 1000000, + "milióny": 1000000, + "miliónů": 1000000, + "miliarda": 1000000000, + "miliardy": 1000000000, + "miliardů": 1000000000, + "bilión": 1000000000000, + "bilióny": 1000000000000, + "biliónů": 1000000000000, + "biliarda": 1000000000000000, + "biliardy": 1000000000000000, + "biliardů": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "deset": 10, - "jedenáct": 11, - "dvanáct": 12, - "třináct": 13, - "čtrnáct": 14, - "patnáct": 15, - "šestnáct": 16, - "sedmnáct": 17, - "osmnáct": 18, - "devatenáct": 19 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": { - "padesát": 50, - "šedesát": 60, - "sedmdesát": 70, - "osmdesát": 80, - "devadesát": 90 - }, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "set": 100, - "sta": 100, - "stě": 100, - "tisíc": 1000, - "tisíce": 1000, - "milión": 1000000, - "milióny": 1000000, - "miliónů": 1000000, - "miliarda": 1000000000, - "miliardy": 1000000000, - "miliardů": 1000000000, - "bilión": 1000000000000, - "bilióny": 1000000000000, - "biliónů": 1000000000000, - "biliarda": 1000000000000000, - "biliardy": 1000000000000000, - "biliardů": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/cy.py b/number_parser/data/cy.py index 6515f60..aba0a5a 100644 --- a/number_parser/data/cy.py +++ b/number_parser/data/cy.py @@ -1,33 +1,43 @@ info = { - "UNIT_NUMBERS": { - "dim": 0, - "un": 1, - "dau": 2, - "dwy": 2, - "tair": 3, - "tri": 3, - "pedair": 4, - "pedwar": 4, - "pum": 5, - "pump": 5, - "chwe": 6, - "chwech": 6, - "saith": 7, - "wyth": 8, - "naw": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "dim": 0, + "un": 1, + "dau": 2, + "dwy": 2, + "tair": 3, + "tri": 3, + "pedair": 4, + "pedwar": 4, + "pum": 5, + "pump": 5, + "chwe": 6, + "chwech": 6, + "saith": 7, + "wyth": 8, + "naw": 9 + }, + "DIRECT_NUMBERS": {}, + "TENS": { + "dau ddeg": 20 + }, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "cant": 100, + "mil": 1000, + "miliwn": 1000000, + "biliwn": 1000000000, + "triliwn": 1000000000000, + "kwadriliwn": 1000000000000000 + } }, - "DIRECT_NUMBERS": {}, - "TENS": { - "dau ddeg": 20 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "cant": 100, - "mil": 1000, - "miliwn": 1000000, - "biliwn": 1000000000, - "triliwn": 1000000000000, - "kwadriliwn": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/da.py b/number_parser/data/da.py index 08a2166..3859f83 100644 --- a/number_parser/data/da.py +++ b/number_parser/data/da.py @@ -1,38 +1,48 @@ info = { - "UNIT_NUMBERS": { - "nul": 0, - "en": 1, - "et": 1, - "to": 2, - "tre": 3, - "fire": 4, - "fem": 5, - "seks": 6, - "syv": 7, - "otte": 8, - "ni": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "nul": 0, + "en": 1, + "et": 1, + "to": 2, + "tre": 3, + "fire": 4, + "fem": 5, + "seks": 6, + "syv": 7, + "otte": 8, + "ni": 9 + }, + "DIRECT_NUMBERS": { + "ti": 10, + "elleve": 11, + "tolv": 12, + "tretten": 13, + "fjorten": 14, + "femten": 15, + "seksten": 16, + "sytten": 17, + "atten": 18, + "nitten": 19 + }, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "tusind": 1000, + "tusinde": 1000, + "millioner": 1000000, + "milliarder": 1000000000, + "billioner": 1000000000000, + "billiarder": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "ti": 10, - "elleve": 11, - "tolv": 12, - "tretten": 13, - "fjorten": 14, - "femten": 15, - "seksten": 16, - "sytten": 17, - "atten": 18, - "nitten": 19 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "tusind": 1000, - "tusinde": 1000, - "millioner": 1000000, - "milliarder": 1000000000, - "billioner": 1000000000000, - "billiarder": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/de-CH.py b/number_parser/data/de-CH.py index 0a489b7..d6ba532 100644 --- a/number_parser/data/de-CH.py +++ b/number_parser/data/de-CH.py @@ -1,35 +1,45 @@ info = { - "UNIT_NUMBERS": { - "null": 0, - "ein": 1, - "eine": 1, - "einen": 1, - "einer": 1, - "eines": 1, - "eins": 1, - "zwei": 2, - "drei": 3, - "vier": 4, - "fünf": 5, - "sechs": 6, - "sieben": 7, - "acht": 8, - "neun": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "null": 0, + "ein": 1, + "eine": 1, + "einen": 1, + "einer": 1, + "eines": 1, + "eins": 1, + "zwei": 2, + "drei": 3, + "vier": 4, + "fünf": 5, + "sechs": 6, + "sieben": 7, + "acht": 8, + "neun": 9 + }, + "DIRECT_NUMBERS": { + "zehn": 10, + "elf": 11, + "zwölf": 12, + "sechzehn": 16, + "siebzehn": 17 + }, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "Millionen": 1000000, + "Milliarden": 1000000000, + "Billionen": 1000000000000, + "Billiarden": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "zehn": 10, - "elf": 11, - "zwölf": 12, - "sechzehn": 16, - "siebzehn": 17 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "Millionen": 1000000, - "Milliarden": 1000000000, - "Billionen": 1000000000000, - "Billiarden": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/de.py b/number_parser/data/de.py index 0a489b7..d6ba532 100644 --- a/number_parser/data/de.py +++ b/number_parser/data/de.py @@ -1,35 +1,45 @@ info = { - "UNIT_NUMBERS": { - "null": 0, - "ein": 1, - "eine": 1, - "einen": 1, - "einer": 1, - "eines": 1, - "eins": 1, - "zwei": 2, - "drei": 3, - "vier": 4, - "fünf": 5, - "sechs": 6, - "sieben": 7, - "acht": 8, - "neun": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "null": 0, + "ein": 1, + "eine": 1, + "einen": 1, + "einer": 1, + "eines": 1, + "eins": 1, + "zwei": 2, + "drei": 3, + "vier": 4, + "fünf": 5, + "sechs": 6, + "sieben": 7, + "acht": 8, + "neun": 9 + }, + "DIRECT_NUMBERS": { + "zehn": 10, + "elf": 11, + "zwölf": 12, + "sechzehn": 16, + "siebzehn": 17 + }, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "Millionen": 1000000, + "Milliarden": 1000000000, + "Billionen": 1000000000000, + "Billiarden": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "zehn": 10, - "elf": 11, - "zwölf": 12, - "sechzehn": 16, - "siebzehn": 17 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "Millionen": 1000000, - "Milliarden": 1000000000, - "Billionen": 1000000000000, - "Billiarden": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/ee.py b/number_parser/data/ee.py index 5fe6a5d..12d12a8 100644 --- a/number_parser/data/ee.py +++ b/number_parser/data/ee.py @@ -1,11 +1,21 @@ info = { - "UNIT_NUMBERS": { - "ɖekeo": 0, - "ɖeka": 1 + "NUMBERS": { + "UNIT_NUMBERS": { + "ɖekeo": 0, + "ɖeka": 1 + }, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/el.py b/number_parser/data/el.py index 3a7d65d..f8a9509 100644 --- a/number_parser/data/el.py +++ b/number_parser/data/el.py @@ -1,71 +1,81 @@ info = { - "UNIT_NUMBERS": { - "μηδέν": 0, - "ένα": 1, - "ένας": 1, - "μία": 1, - "δύο": 2, - "τρία": 3, - "τρεις": 3, - "τέσσερα": 4, - "τέσσερις": 4, - "πέντε": 5, - "έξι": 6, - "επτά": 7, - "οκτώ": 8, - "εννέα": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "μηδέν": 0, + "ένα": 1, + "ένας": 1, + "μία": 1, + "δύο": 2, + "τρία": 3, + "τρεις": 3, + "τέσσερα": 4, + "τέσσερις": 4, + "πέντε": 5, + "έξι": 6, + "επτά": 7, + "οκτώ": 8, + "εννέα": 9 + }, + "DIRECT_NUMBERS": { + "δέκα": 10, + "έντεκα": 11, + "δώδεκα": 12 + }, + "TENS": { + "είκοσι": 20, + "τριάντα": 30, + "σαράντα": 40, + "πενήντα": 50, + "εξήντα": 60, + "εβδομήντα": 70, + "ογδόντα": 80, + "εννενήντα": 90 + }, + "HUNDREDS": { + "διακόσια": 200, + "διακόσιες": 200, + "διακόσιοι": 200, + "τριακόσια": 300, + "τριακόσιες": 300, + "τριακόσιοι": 300, + "τετρακόσια": 400, + "τετρακόσιες": 400, + "τετρακόσιοι": 400, + "πεντακόσια": 500, + "πεντακόσιες": 500, + "πεντακόσιοι": 500, + "εξακόσια": 600, + "εξακόσιες": 600, + "εξακόσιοι": 600, + "επτακόσια": 700, + "επτακόσιες": 700, + "επτακόσιοι": 700, + "οκτακόσια": 800, + "οκτακόσιες": 800, + "οκτακόσιοι": 800, + "εννιακόσια": 900, + "εννιακόσιες": 900, + "εννιακόσιοι": 900 + }, + "BIG_POWERS_OF_TEN": { + "χίλιάδες": 1000, + "εκατομμύρια": 1000000, + "εκατομμύριο": 1000000, + "δισεκατομμύρια": 1000000000, + "δισεκατομμύριο": 1000000000, + "τρισεκατομμύρια": 1000000000000, + "τρισεκατομμύριο": 1000000000000, + "τετράκις εκατομμύρια": 1000000000000000, + "τετράκις εκατομμύριο": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "δέκα": 10, - "έντεκα": 11, - "δώδεκα": 12 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": { - "είκοσι": 20, - "τριάντα": 30, - "σαράντα": 40, - "πενήντα": 50, - "εξήντα": 60, - "εβδομήντα": 70, - "ογδόντα": 80, - "εννενήντα": 90 - }, - "HUNDREDS": { - "διακόσια": 200, - "διακόσιες": 200, - "διακόσιοι": 200, - "τριακόσια": 300, - "τριακόσιες": 300, - "τριακόσιοι": 300, - "τετρακόσια": 400, - "τετρακόσιες": 400, - "τετρακόσιοι": 400, - "πεντακόσια": 500, - "πεντακόσιες": 500, - "πεντακόσιοι": 500, - "εξακόσια": 600, - "εξακόσιες": 600, - "εξακόσιοι": 600, - "επτακόσια": 700, - "επτακόσιες": 700, - "επτακόσιοι": 700, - "οκτακόσια": 800, - "οκτακόσιες": 800, - "οκτακόσιοι": 800, - "εννιακόσια": 900, - "εννιακόσιες": 900, - "εννιακόσιοι": 900 - }, - "BIG_POWERS_OF_TEN": { - "χίλιάδες": 1000, - "εκατομμύρια": 1000000, - "εκατομμύριο": 1000000, - "δισεκατομμύρια": 1000000000, - "δισεκατομμύριο": 1000000000, - "τρισεκατομμύρια": 1000000000000, - "τρισεκατομμύριο": 1000000000000, - "τετράκις εκατομμύρια": 1000000000000000, - "τετράκις εκατομμύριο": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/en-IN.py b/number_parser/data/en-IN.py index 6be25ad..e3ea3c8 100644 --- a/number_parser/data/en-IN.py +++ b/number_parser/data/en-IN.py @@ -1,46 +1,56 @@ info = { - "UNIT_NUMBERS": { - "zero": 0, - "one": 1, - "two": 2, - "three": 3, - "four": 4, - "five": 5, - "six": 6, - "seven": 7, - "eight": 8, - "nine": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "zero": 0, + "one": 1, + "two": 2, + "three": 3, + "four": 4, + "five": 5, + "six": 6, + "seven": 7, + "eight": 8, + "nine": 9 + }, + "DIRECT_NUMBERS": { + "ten": 10, + "eleven": 11, + "twelve": 12, + "thirteen": 13, + "fourteen": 14, + "fifteen": 15, + "sixteen": 16, + "seventeen": 17, + "eighteen": 18, + "nineteen": 19 + }, + "TENS": { + "twenty": 20, + "thirty": 30, + "forty": 40, + "fifty": 50, + "sixty": 60, + "seventy": 70, + "eighty": 80, + "ninety": 90 + }, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "hundred": 100, + "thousand": 1000, + "lakh": 100000, + "crore": 10000000, + "trillion": 1000000000000, + "quadrillion": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "ten": 10, - "eleven": 11, - "twelve": 12, - "thirteen": 13, - "fourteen": 14, - "fifteen": 15, - "sixteen": 16, - "seventeen": 17, - "eighteen": 18, - "nineteen": 19 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": { - "twenty": 20, - "thirty": 30, - "forty": 40, - "fifty": 50, - "sixty": 60, - "seventy": 70, - "eighty": 80, - "ninety": 90 - }, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "hundred": 100, - "thousand": 1000, - "lakh": 100000, - "crore": 10000000, - "trillion": 1000000000000, - "quadrillion": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/en.py b/number_parser/data/en.py index 687a833..00b7194 100644 --- a/number_parser/data/en.py +++ b/number_parser/data/en.py @@ -1,51 +1,61 @@ info = { - "UNIT_NUMBERS": { - "zero": 0, - "one": 1, - "two": 2, - "three": 3, - "four": 4, - "five": 5, - "six": 6, - "seven": 7, - "eight": 8, - "nine": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "zero": 0, + "one": 1, + "two": 2, + "three": 3, + "four": 4, + "five": 5, + "six": 6, + "seven": 7, + "eight": 8, + "nine": 9 + }, + "DIRECT_NUMBERS": { + "ten": 10, + "eleven": 11, + "twelve": 12, + "thirteen": 13, + "fourteen": 14, + "fifteen": 15, + "sixteen": 16, + "seventeen": 17, + "eighteen": 18, + "nineteen": 19 + }, + "TENS": { + "twenty": 20, + "thirty": 30, + "forty": 40, + "fifty": 50, + "sixty": 60, + "seventy": 70, + "eighty": 80, + "ninety": 90 + }, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "hundred": 100, + "thousand": 1000, + "million": 1000000, + "billion": 1000000000, + "trillion": 1000000000000, + "quadrillion": 1000000000000000, + "quintillion": 1000000000000000000, + "sextillion": 1000000000000000000000 + } }, - "DIRECT_NUMBERS": { - "ten": 10, - "eleven": 11, - "twelve": 12, - "thirteen": 13, - "fourteen": 14, - "fifteen": 15, - "sixteen": 16, - "seventeen": 17, - "eighteen": 18, - "nineteen": 19 - }, - "TENS": { - "twenty": 20, - "thirty": 30, - "forty": 40, - "fifty": 50, - "sixty": 60, - "seventy": 70, - "eighty": 80, - "ninety": 90 - }, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "hundred": 100, - "thousand": 1000, - "million": 1000000, - "billion": 1000000000, - "trillion": 1000000000000, - "quadrillion": 1000000000000000, - "quintillion": 1000000000000000000, - "sextillion": 1000000000000000000000 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, "SKIP_TOKENS": [ "-", "and" - ] + ], + "IS_LONG": 0 } diff --git a/number_parser/data/eo.py b/number_parser/data/eo.py index e5a4c6f..b1c32fa 100644 --- a/number_parser/data/eo.py +++ b/number_parser/data/eo.py @@ -1,25 +1,35 @@ info = { - "UNIT_NUMBERS": { - "nulo": 0, - "unu": 1, - "du": 2, - "tri": 3, - "kvar": 4, - "kvin": 5, - "ses": 6, - "sep": 7, - "ok": 8, - "naŭ": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "nulo": 0, + "unu": 1, + "du": 2, + "tri": 3, + "kvar": 4, + "kvin": 5, + "ses": 6, + "sep": 7, + "ok": 8, + "naŭ": 9 + }, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "mil": 1000, + "milionoj": 1000000, + "miliardoj": 1000000000, + "bilionoj": 1000000000000, + "biliardoj": 1000000000000000 + } }, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "mil": 1000, - "milionoj": 1000000, - "miliardoj": 1000000000, - "bilionoj": 1000000000000, - "biliardoj": 1000000000000000 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/es.py b/number_parser/data/es.py index 5ce35d1..3a034bf 100644 --- a/number_parser/data/es.py +++ b/number_parser/data/es.py @@ -1,82 +1,92 @@ info = { - "UNIT_NUMBERS": { - "cero": 0, - "un": 1, - "una": 1, - "uno": 1, - "dos": 2, - "tres": 3, - "cuatro": 4, - "cinco": 5, - "seis": 6, - "siete": 7, - "ocho": 8, - "nueve": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "cero": 0, + "un": 1, + "una": 1, + "uno": 1, + "dos": 2, + "tres": 3, + "cuatro": 4, + "cinco": 5, + "seis": 6, + "siete": 7, + "ocho": 8, + "nueve": 9 + }, + "DIRECT_NUMBERS": { + "diez": 10, + "once": 11, + "doce": 12, + "trece": 13, + "catorce": 14, + "quince": 15, + "dieciséis": 16, + "diecisiete": 17, + "dieciocho": 18, + "diecinueve": 19, + "veinte": 20, + "veintiuna": 21, + "veintiuno": 21, + "veintiún": 21, + "veintidós": 22, + "veintitrés": 23, + "veinticuatro": 24, + "veinticinco": 25, + "veintiséis": 26, + "veintisiete": 27, + "veintiocho": 28, + "veintinueve": 29 + }, + "TENS": { + "treinta": 30, + "cuarenta": 40, + "cincuenta": 50, + "sesenta": 60, + "setenta": 70, + "ochenta": 80, + "noventa": 90 + }, + "HUNDREDS": { + "cien": 100, + "ciento": 100, + "doscientas": 200, + "doscientos": 200, + "trescientas": 300, + "trescientos": 300, + "cuatrocientas": 400, + "cuatrocientos": 400, + "quinientas": 500, + "quinientos": 500, + "seiscientas": 600, + "seiscientos": 600, + "setecientas": 700, + "setecientos": 700, + "ochocientas": 800, + "ochocientos": 800, + "novecientas": 900, + "novecientos": 900 + }, + "BIG_POWERS_OF_TEN": { + "mil": 1000, + "millones": 1000000, + "millón": 1000000, + "millardo": 1000000000, + "billones": 1000000000000, + "billón": 1000000000000, + "trillones": 1000000000000000000, + "trilón": 1000000000000000000 + } }, - "DIRECT_NUMBERS": { - "diez": 10, - "once": 11, - "doce": 12, - "trece": 13, - "catorce": 14, - "quince": 15, - "dieciséis": 16, - "diecisiete": 17, - "dieciocho": 18, - "diecinueve": 19, - "veinte": 20, - "veintiuna": 21, - "veintiuno": 21, - "veintiún": 21, - "veintidós": 22, - "veintitrés": 23, - "veinticuatro": 24, - "veinticinco": 25, - "veintiséis": 26, - "veintisiete": 27, - "veintiocho": 28, - "veintinueve": 29 - }, - "TENS": { - "treinta": 30, - "cuarenta": 40, - "cincuenta": 50, - "sesenta": 60, - "setenta": 70, - "ochenta": 80, - "noventa": 90 - }, - "HUNDREDS": { - "cien": 100, - "ciento": 100, - "doscientas": 200, - "doscientos": 200, - "trescientas": 300, - "trescientos": 300, - "cuatrocientas": 400, - "cuatrocientos": 400, - "quinientas": 500, - "quinientos": 500, - "seiscientas": 600, - "seiscientos": 600, - "setecientas": 700, - "setecientos": 700, - "ochocientas": 800, - "ochocientos": 800, - "novecientas": 900, - "novecientos": 900 - }, - "BIG_POWERS_OF_TEN": { - "mil": 1000, - "millones": 1000000, - "millón": 1000000, - "millardo": 1000000000, - "billones": 1000000000000, - "billón": 1000000000000, - "trillones": 1000000000000000000, - "trilón": 1000000000000000000 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, "SKIP_TOKENS": [ "y" - ] + ], + "IS_LONG": 1 } diff --git a/number_parser/data/et.py b/number_parser/data/et.py index a1fd9cc..54e2f0f 100644 --- a/number_parser/data/et.py +++ b/number_parser/data/et.py @@ -1,31 +1,41 @@ info = { - "UNIT_NUMBERS": { - "null": 0, - "üks": 1, - "kaks": 2, - "kolm": 3, - "neli": 4, - "viis": 5, - "kuus": 6, - "seitse": 7, - "kaheksa": 8, - "üheksa": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "null": 0, + "üks": 1, + "kaks": 2, + "kolm": 3, + "neli": 4, + "viis": 5, + "kuus": 6, + "seitse": 7, + "kaheksa": 8, + "üheksa": 9 + }, + "DIRECT_NUMBERS": { + "kümme": 10 + }, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "tuhat": 1000, + "miljon": 1000000, + "miljonit": 1000000, + "miljard": 1000000000, + "miljardit": 1000000000, + "biljon": 1000000000000, + "biljonit": 1000000000000, + "biljard": 1000000000000000, + "biljardit": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "kümme": 10 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "tuhat": 1000, - "miljon": 1000000, - "miljonit": 1000000, - "miljard": 1000000000, - "miljardit": 1000000000, - "biljon": 1000000000000, - "biljonit": 1000000000000, - "biljard": 1000000000000000, - "biljardit": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/fa-AF.py b/number_parser/data/fa-AF.py index fdee3fe..0bf53af 100644 --- a/number_parser/data/fa-AF.py +++ b/number_parser/data/fa-AF.py @@ -1,45 +1,55 @@ info = { - "UNIT_NUMBERS": { - "صفر": 0, - "یک": 1, - "دو": 2, - "سه": 3, - "چهار": 4, - "پنج": 5, - "شش": 6, - "هفت": 7, - "هشت": 8, - "نه": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "صفر": 0, + "یک": 1, + "دو": 2, + "سه": 3, + "چهار": 4, + "پنج": 5, + "شش": 6, + "هفت": 7, + "هشت": 8, + "نه": 9 + }, + "DIRECT_NUMBERS": { + "ده": 10, + "یازده": 11, + "دوازده": 12, + "سیزده": 13, + "چهارده": 14, + "پانزده": 15, + "شانزده": 16, + "هفده": 17, + "هجده": 18, + "نزده": 19 + }, + "TENS": { + "بیست": 20, + "سی": 30, + "چهل": 40, + "پنجاه": 50, + "شصت": 60, + "هفتاد": 70, + "هشتاد": 80, + "نود": 90 + }, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "هزار": 1000, + "میلیون": 1000000, + "میلیارد": 1000000000, + "بیلیون": 1000000000000, + "بیلیارد": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "ده": 10, - "یازده": 11, - "دوازده": 12, - "سیزده": 13, - "چهارده": 14, - "پانزده": 15, - "شانزده": 16, - "هفده": 17, - "هجده": 18, - "نزده": 19 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": { - "بیست": 20, - "سی": 30, - "چهل": 40, - "پنجاه": 50, - "شصت": 60, - "هفتاد": 70, - "هشتاد": 80, - "نود": 90 - }, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "هزار": 1000, - "میلیون": 1000000, - "میلیارد": 1000000000, - "بیلیون": 1000000000000, - "بیلیارد": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/fa.py b/number_parser/data/fa.py index b0160be..db5fb7b 100644 --- a/number_parser/data/fa.py +++ b/number_parser/data/fa.py @@ -1,53 +1,63 @@ info = { - "UNIT_NUMBERS": { - "صفر": 0, - "یک": 1, - "دو": 2, - "سه": 3, - "چهار": 4, - "پنج": 5, - "شش": 6, - "هفت": 7, - "هشت": 8, - "نه": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "صفر": 0, + "یک": 1, + "دو": 2, + "سه": 3, + "چهار": 4, + "پنج": 5, + "شش": 6, + "هفت": 7, + "هشت": 8, + "نه": 9 + }, + "DIRECT_NUMBERS": { + "ده": 10, + "یازده": 11, + "دوازده": 12, + "سیزده": 13, + "چهارده": 14, + "پانزده": 15, + "شانزده": 16, + "هفده": 17, + "هجده": 18, + "نوزده": 19 + }, + "TENS": { + "بیست": 20, + "سی": 30, + "چهل": 40, + "پنجاه": 50, + "شصت": 60, + "هفتاد": 70, + "هشتاد": 80, + "نود": 90 + }, + "HUNDREDS": { + "دویست": 200, + "سیصد": 300, + "چهارصد": 400, + "پانصد": 500, + "ششصد": 600, + "هفتصد": 700, + "هشتصد": 800, + "نهصد": 900 + }, + "BIG_POWERS_OF_TEN": { + "هزار": 1000, + "میلیون": 1000000, + "میلیارد": 1000000000, + "هزار میلیارد": 1000000000000 + } }, - "DIRECT_NUMBERS": { - "ده": 10, - "یازده": 11, - "دوازده": 12, - "سیزده": 13, - "چهارده": 14, - "پانزده": 15, - "شانزده": 16, - "هفده": 17, - "هجده": 18, - "نوزده": 19 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": { - "بیست": 20, - "سی": 30, - "چهل": 40, - "پنجاه": 50, - "شصت": 60, - "هفتاد": 70, - "هشتاد": 80, - "نود": 90 - }, - "HUNDREDS": { - "دویست": 200, - "سیصد": 300, - "چهارصد": 400, - "پانصد": 500, - "ششصد": 600, - "هفتصد": 700, - "هشتصد": 800, - "نهصد": 900 - }, - "BIG_POWERS_OF_TEN": { - "هزار": 1000, - "میلیون": 1000000, - "میلیارد": 1000000000, - "هزار میلیارد": 1000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/ff.py b/number_parser/data/ff.py index 7eaf7b0..900fd7c 100644 --- a/number_parser/data/ff.py +++ b/number_parser/data/ff.py @@ -1,30 +1,40 @@ info = { - "UNIT_NUMBERS": { - "ɓolum": 0, - "gooto": 1, - "goʼo": 1, - "ɗiɗi": 2, - "ɗiɗo": 2, - "tati": 3, - "tato": 3, - "nawi": 4, - "nawo": 4, - "jowi": 5, - "njowo": 5, - "jeegomo": 6, - "jeegoʼo": 6, - "jeeɗiɗi": 7, - "jeeɗiɗo": 7, - "jeetati": 8, - "jeetato": 8, - "jeenawi": 9, - "jeenawo": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "ɓolum": 0, + "gooto": 1, + "goʼo": 1, + "ɗiɗi": 2, + "ɗiɗo": 2, + "tati": 3, + "tato": 3, + "nawi": 4, + "nawo": 4, + "jowi": 5, + "njowo": 5, + "jeegomo": 6, + "jeegoʼo": 6, + "jeeɗiɗi": 7, + "jeeɗiɗo": 7, + "jeetati": 8, + "jeetato": 8, + "jeenawi": 9, + "jeenawo": 9 + }, + "DIRECT_NUMBERS": {}, + "TENS": { + "noogas": 20 + }, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "DIRECT_NUMBERS": {}, - "TENS": { - "noogas": 20 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/fi.py b/number_parser/data/fi.py index 5ab6946..d47f2eb 100644 --- a/number_parser/data/fi.py +++ b/number_parser/data/fi.py @@ -1,301 +1,311 @@ info = { - "UNIT_NUMBERS": { - "nolla": 0, - "nollaa": 0, - "nollaan": 0, - "nollaksi": 0, - "nollalla": 0, - "nollalle": 0, - "nollalta": 0, - "nollan": 0, - "nollana": 0, - "nollassa": 0, - "nollasta": 0, - "nollia": 0, - "nollien": 0, - "nolliin": 0, - "nolliksi": 0, - "nollilla": 0, - "nollille": 0, - "nollilta": 0, - "nollina": 0, - "nollissa": 0, - "nollista": 0, - "yhdeksi": 1, - "yhdelle": 1, - "yhdellä": 1, - "yhdeltä": 1, - "yhden": 1, - "yhdessä": 1, - "yhdestä": 1, - "yhteen": 1, - "yhtenä": 1, - "yhtä": 1, - "yksi": 1, - "yksien": 1, - "yksiin": 1, - "yksiksi": 1, - "yksille": 1, - "yksillä": 1, - "yksiltä": 1, - "yksinä": 1, - "yksissä": 1, - "yksistä": 1, - "yksiä": 1, - "kahdeksi": 2, - "kahdella": 2, - "kahdelle": 2, - "kahdelta": 2, - "kahden": 2, - "kahdessa": 2, - "kahdesta": 2, - "kahta": 2, - "kahteen": 2, - "kahtena": 2, - "kaksi": 2, - "kaksia": 2, - "kaksien": 2, - "kaksiin": 2, - "kaksiksi": 2, - "kaksilla": 2, - "kaksille": 2, - "kaksilta": 2, - "kaksina": 2, - "kaksissa": 2, - "kaksista": 2, - "kolme": 3, - "kolmea": 3, - "kolmeen": 3, - "kolmeksi": 3, - "kolmella": 3, - "kolmelle": 3, - "kolmelta": 3, - "kolmen": 3, - "kolmena": 3, - "kolmessa": 3, - "kolmesta": 3, - "kolmia": 3, - "kolmien": 3, - "kolmiin": 3, - "kolmiksi": 3, - "kolmilla": 3, - "kolmille": 3, - "kolmilta": 3, - "kolmina": 3, - "kolmissa": 3, - "kolmista": 3, - "neljien": 4, - "neljiin": 4, - "neljiksi": 4, - "neljille": 4, - "neljillä": 4, - "neljiltä": 4, - "neljinä": 4, - "neljissä": 4, - "neljistä": 4, - "neljiä": 4, - "neljä": 4, - "neljäksi": 4, - "neljälle": 4, - "neljällä": 4, - "neljältä": 4, - "neljän": 4, - "neljänä": 4, - "neljässä": 4, - "neljästä": 4, - "neljää": 4, - "neljään": 4, - "viideksi": 5, - "viidelle": 5, - "viidellä": 5, - "viideltä": 5, - "viiden": 5, - "viidessä": 5, - "viidestä": 5, - "viisi": 5, - "viisien": 5, - "viisiin": 5, - "viisiksi": 5, - "viisille": 5, - "viisillä": 5, - "viisiltä": 5, - "viisinä": 5, - "viisissä": 5, - "viisistä": 5, - "viisiä": 5, - "viiteen": 5, - "viitenä": 5, - "viittä": 5, - "kuudeksi": 6, - "kuudella": 6, - "kuudelle": 6, - "kuudelta": 6, - "kuuden": 6, - "kuudessa": 6, - "kuudesta": 6, - "kuusi": 6, - "kuusia": 6, - "kuusien": 6, - "kuusiin": 6, - "kuusiksi": 6, - "kuusilla": 6, - "kuusille": 6, - "kuusilta": 6, - "kuusina": 6, - "kuusissa": 6, - "kuusista": 6, - "kuuteen": 6, - "kuutena": 6, - "kuutta": 6, - "seitsemien": 7, - "seitsemiin": 7, - "seitsemiksi": 7, - "seitsemille": 7, - "seitsemillä": 7, - "seitsemiltä": 7, - "seitseminä": 7, - "seitsemissä": 7, - "seitsemistä": 7, - "seitsemiä": 7, - "seitsemäksi": 7, - "seitsemälle": 7, - "seitsemällä": 7, - "seitsemältä": 7, - "seitsemän": 7, - "seitsemänä": 7, - "seitsemässä": 7, - "seitsemästä": 7, - "seitsemää": 7, - "seitsemään": 7, - "kahdeksaa": 8, - "kahdeksaan": 8, - "kahdeksaksi": 8, - "kahdeksalla": 8, - "kahdeksalle": 8, - "kahdeksalta": 8, - "kahdeksan": 8, - "kahdeksana": 8, - "kahdeksassa": 8, - "kahdeksasta": 8, - "kahdeksia": 8, - "kahdeksien": 8, - "kahdeksiin": 8, - "kahdeksiksi": 8, - "kahdeksilla": 8, - "kahdeksille": 8, - "kahdeksilta": 8, - "kahdeksina": 8, - "kahdeksissa": 8, - "kahdeksista": 8, - "yhdeksien": 9, - "yhdeksiin": 9, - "yhdeksiksi": 9, - "yhdeksille": 9, - "yhdeksillä": 9, - "yhdeksiltä": 9, - "yhdeksinä": 9, - "yhdeksissä": 9, - "yhdeksistä": 9, - "yhdeksiä": 9, - "yhdeksäksi": 9, - "yhdeksälle": 9, - "yhdeksällä": 9, - "yhdeksältä": 9, - "yhdeksän": 9, - "yhdeksänä": 9, - "yhdeksässä": 9, - "yhdeksästä": 9, - "yhdeksää": 9, - "yhdeksään": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "nolla": 0, + "nollaa": 0, + "nollaan": 0, + "nollaksi": 0, + "nollalla": 0, + "nollalle": 0, + "nollalta": 0, + "nollan": 0, + "nollana": 0, + "nollassa": 0, + "nollasta": 0, + "nollia": 0, + "nollien": 0, + "nolliin": 0, + "nolliksi": 0, + "nollilla": 0, + "nollille": 0, + "nollilta": 0, + "nollina": 0, + "nollissa": 0, + "nollista": 0, + "yhdeksi": 1, + "yhdelle": 1, + "yhdellä": 1, + "yhdeltä": 1, + "yhden": 1, + "yhdessä": 1, + "yhdestä": 1, + "yhteen": 1, + "yhtenä": 1, + "yhtä": 1, + "yksi": 1, + "yksien": 1, + "yksiin": 1, + "yksiksi": 1, + "yksille": 1, + "yksillä": 1, + "yksiltä": 1, + "yksinä": 1, + "yksissä": 1, + "yksistä": 1, + "yksiä": 1, + "kahdeksi": 2, + "kahdella": 2, + "kahdelle": 2, + "kahdelta": 2, + "kahden": 2, + "kahdessa": 2, + "kahdesta": 2, + "kahta": 2, + "kahteen": 2, + "kahtena": 2, + "kaksi": 2, + "kaksia": 2, + "kaksien": 2, + "kaksiin": 2, + "kaksiksi": 2, + "kaksilla": 2, + "kaksille": 2, + "kaksilta": 2, + "kaksina": 2, + "kaksissa": 2, + "kaksista": 2, + "kolme": 3, + "kolmea": 3, + "kolmeen": 3, + "kolmeksi": 3, + "kolmella": 3, + "kolmelle": 3, + "kolmelta": 3, + "kolmen": 3, + "kolmena": 3, + "kolmessa": 3, + "kolmesta": 3, + "kolmia": 3, + "kolmien": 3, + "kolmiin": 3, + "kolmiksi": 3, + "kolmilla": 3, + "kolmille": 3, + "kolmilta": 3, + "kolmina": 3, + "kolmissa": 3, + "kolmista": 3, + "neljien": 4, + "neljiin": 4, + "neljiksi": 4, + "neljille": 4, + "neljillä": 4, + "neljiltä": 4, + "neljinä": 4, + "neljissä": 4, + "neljistä": 4, + "neljiä": 4, + "neljä": 4, + "neljäksi": 4, + "neljälle": 4, + "neljällä": 4, + "neljältä": 4, + "neljän": 4, + "neljänä": 4, + "neljässä": 4, + "neljästä": 4, + "neljää": 4, + "neljään": 4, + "viideksi": 5, + "viidelle": 5, + "viidellä": 5, + "viideltä": 5, + "viiden": 5, + "viidessä": 5, + "viidestä": 5, + "viisi": 5, + "viisien": 5, + "viisiin": 5, + "viisiksi": 5, + "viisille": 5, + "viisillä": 5, + "viisiltä": 5, + "viisinä": 5, + "viisissä": 5, + "viisistä": 5, + "viisiä": 5, + "viiteen": 5, + "viitenä": 5, + "viittä": 5, + "kuudeksi": 6, + "kuudella": 6, + "kuudelle": 6, + "kuudelta": 6, + "kuuden": 6, + "kuudessa": 6, + "kuudesta": 6, + "kuusi": 6, + "kuusia": 6, + "kuusien": 6, + "kuusiin": 6, + "kuusiksi": 6, + "kuusilla": 6, + "kuusille": 6, + "kuusilta": 6, + "kuusina": 6, + "kuusissa": 6, + "kuusista": 6, + "kuuteen": 6, + "kuutena": 6, + "kuutta": 6, + "seitsemien": 7, + "seitsemiin": 7, + "seitsemiksi": 7, + "seitsemille": 7, + "seitsemillä": 7, + "seitsemiltä": 7, + "seitseminä": 7, + "seitsemissä": 7, + "seitsemistä": 7, + "seitsemiä": 7, + "seitsemäksi": 7, + "seitsemälle": 7, + "seitsemällä": 7, + "seitsemältä": 7, + "seitsemän": 7, + "seitsemänä": 7, + "seitsemässä": 7, + "seitsemästä": 7, + "seitsemää": 7, + "seitsemään": 7, + "kahdeksaa": 8, + "kahdeksaan": 8, + "kahdeksaksi": 8, + "kahdeksalla": 8, + "kahdeksalle": 8, + "kahdeksalta": 8, + "kahdeksan": 8, + "kahdeksana": 8, + "kahdeksassa": 8, + "kahdeksasta": 8, + "kahdeksia": 8, + "kahdeksien": 8, + "kahdeksiin": 8, + "kahdeksiksi": 8, + "kahdeksilla": 8, + "kahdeksille": 8, + "kahdeksilta": 8, + "kahdeksina": 8, + "kahdeksissa": 8, + "kahdeksista": 8, + "yhdeksien": 9, + "yhdeksiin": 9, + "yhdeksiksi": 9, + "yhdeksille": 9, + "yhdeksillä": 9, + "yhdeksiltä": 9, + "yhdeksinä": 9, + "yhdeksissä": 9, + "yhdeksistä": 9, + "yhdeksiä": 9, + "yhdeksäksi": 9, + "yhdeksälle": 9, + "yhdeksällä": 9, + "yhdeksältä": 9, + "yhdeksän": 9, + "yhdeksänä": 9, + "yhdeksässä": 9, + "yhdeksästä": 9, + "yhdeksää": 9, + "yhdeksään": 9 + }, + "DIRECT_NUMBERS": { + "kymmeneen": 10, + "kymmeneksi": 10, + "kymmenelle": 10, + "kymmenellä": 10, + "kymmeneltä": 10, + "kymmenen": 10, + "kymmenenä": 10, + "kymmenessä": 10, + "kymmenestä": 10, + "kymmenien": 10, + "kymmeniin": 10, + "kymmeniksi": 10, + "kymmenille": 10, + "kymmenillä": 10, + "kymmeniltä": 10, + "kymmeninä": 10, + "kymmenissä": 10, + "kymmenistä": 10, + "kymmeniä": 10, + "kymmentä": 10 + }, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "miljoona": 1000000, + "miljoonaa": 1000000, + "miljoonaan": 1000000, + "miljoonaksi": 1000000, + "miljoonalla": 1000000, + "miljoonalle": 1000000, + "miljoonalta": 1000000, + "miljoonan": 1000000, + "miljoonana": 1000000, + "miljoonassa": 1000000, + "miljoonasta": 1000000, + "miljoonia": 1000000, + "miljoonien": 1000000, + "miljooniin": 1000000, + "miljooniksi": 1000000, + "miljoonilla": 1000000, + "miljoonille": 1000000, + "miljoonilta": 1000000, + "miljoonina": 1000000, + "miljoonissa": 1000000, + "miljoonista": 1000000, + "miljardeihin": 1000000000, + "miljardeiksi": 1000000000, + "miljardeilla": 1000000000, + "miljardeille": 1000000000, + "miljardeilta": 1000000000, + "miljardeina": 1000000000, + "miljardeissa": 1000000000, + "miljardeista": 1000000000, + "miljardeja": 1000000000, + "miljardi": 1000000000, + "miljardia": 1000000000, + "miljardien": 1000000000, + "miljardiin": 1000000000, + "miljardiksi": 1000000000, + "miljardilla": 1000000000, + "miljardille": 1000000000, + "miljardilta": 1000000000, + "miljardin": 1000000000, + "miljardina": 1000000000, + "miljardissa": 1000000000, + "miljardista": 1000000000, + "biljoona": 1000000000000, + "biljoonaa": 1000000000000, + "biljoonaan": 1000000000000, + "biljoonaksi": 1000000000000, + "biljoonalla": 1000000000000, + "biljoonalle": 1000000000000, + "biljoonalta": 1000000000000, + "biljoonan": 1000000000000, + "biljoonana": 1000000000000, + "biljoonassa": 1000000000000, + "biljoonasta": 1000000000000, + "biljoonia": 1000000000000, + "biljoonien": 1000000000000, + "biljooniin": 1000000000000, + "biljooniksi": 1000000000000, + "biljoonilla": 1000000000000, + "biljoonille": 1000000000000, + "biljoonilta": 1000000000000, + "biljoonina": 1000000000000, + "biljoonissa": 1000000000000, + "biljoonista": 1000000000000 + } }, - "DIRECT_NUMBERS": { - "kymmeneen": 10, - "kymmeneksi": 10, - "kymmenelle": 10, - "kymmenellä": 10, - "kymmeneltä": 10, - "kymmenen": 10, - "kymmenenä": 10, - "kymmenessä": 10, - "kymmenestä": 10, - "kymmenien": 10, - "kymmeniin": 10, - "kymmeniksi": 10, - "kymmenille": 10, - "kymmenillä": 10, - "kymmeniltä": 10, - "kymmeninä": 10, - "kymmenissä": 10, - "kymmenistä": 10, - "kymmeniä": 10, - "kymmentä": 10 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "miljoona": 1000000, - "miljoonaa": 1000000, - "miljoonaan": 1000000, - "miljoonaksi": 1000000, - "miljoonalla": 1000000, - "miljoonalle": 1000000, - "miljoonalta": 1000000, - "miljoonan": 1000000, - "miljoonana": 1000000, - "miljoonassa": 1000000, - "miljoonasta": 1000000, - "miljoonia": 1000000, - "miljoonien": 1000000, - "miljooniin": 1000000, - "miljooniksi": 1000000, - "miljoonilla": 1000000, - "miljoonille": 1000000, - "miljoonilta": 1000000, - "miljoonina": 1000000, - "miljoonissa": 1000000, - "miljoonista": 1000000, - "miljardeihin": 1000000000, - "miljardeiksi": 1000000000, - "miljardeilla": 1000000000, - "miljardeille": 1000000000, - "miljardeilta": 1000000000, - "miljardeina": 1000000000, - "miljardeissa": 1000000000, - "miljardeista": 1000000000, - "miljardeja": 1000000000, - "miljardi": 1000000000, - "miljardia": 1000000000, - "miljardien": 1000000000, - "miljardiin": 1000000000, - "miljardiksi": 1000000000, - "miljardilla": 1000000000, - "miljardille": 1000000000, - "miljardilta": 1000000000, - "miljardin": 1000000000, - "miljardina": 1000000000, - "miljardissa": 1000000000, - "miljardista": 1000000000, - "biljoona": 1000000000000, - "biljoonaa": 1000000000000, - "biljoonaan": 1000000000000, - "biljoonaksi": 1000000000000, - "biljoonalla": 1000000000000, - "biljoonalle": 1000000000000, - "biljoonalta": 1000000000000, - "biljoonan": 1000000000000, - "biljoonana": 1000000000000, - "biljoonassa": 1000000000000, - "biljoonasta": 1000000000000, - "biljoonia": 1000000000000, - "biljoonien": 1000000000000, - "biljooniin": 1000000000000, - "biljooniksi": 1000000000000, - "biljoonilla": 1000000000000, - "biljoonille": 1000000000000, - "biljoonilta": 1000000000000, - "biljoonina": 1000000000000, - "biljoonissa": 1000000000000, - "biljoonista": 1000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/fil.py b/number_parser/data/fil.py index b936ea9..426d7a1 100644 --- a/number_parser/data/fil.py +++ b/number_parser/data/fil.py @@ -1,28 +1,38 @@ info = { - "UNIT_NUMBERS": { - "walâ": 0, - "isá": 1, - "dalawá": 2, - "tatló": 3, - "ápat": 4, - "limá": 5, - "anim": 6, - "pitó": 7, - "waló": 8, - "siyám": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "walâ": 0, + "isá": 1, + "dalawá": 2, + "tatló": 3, + "ápat": 4, + "limá": 5, + "anim": 6, + "pitó": 7, + "waló": 8, + "siyám": 9 + }, + "DIRECT_NUMBERS": { + "sampû": 10 + }, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "daán": 100, + "libó": 1000, + "milyón": 1000000, + "bilyón": 1000000000, + "trilyón": 1000000000000, + "katrilyón": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "sampû": 10 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "daán": 100, - "libó": 1000, - "milyón": 1000000, - "bilyón": 1000000000, - "trilyón": 1000000000000, - "katrilyón": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/fo.py b/number_parser/data/fo.py index 2647853..ea349dd 100644 --- a/number_parser/data/fo.py +++ b/number_parser/data/fo.py @@ -1,50 +1,60 @@ info = { - "UNIT_NUMBERS": { - "null": 0, - "ein": 1, - "eitt": 1, - "tveir": 2, - "tvey": 2, - "tvær": 2, - "tríggir": 3, - "tríggjar": 3, - "trý": 3, - "fýre": 4, - "fimm": 5, - "seks": 6, - "sjey": 7, - "átta": 8, - "níggju": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "null": 0, + "ein": 1, + "eitt": 1, + "tveir": 2, + "tvey": 2, + "tvær": 2, + "tríggir": 3, + "tríggjar": 3, + "trý": 3, + "fýre": 4, + "fimm": 5, + "seks": 6, + "sjey": 7, + "átta": 8, + "níggju": 9 + }, + "DIRECT_NUMBERS": { + "tíggju": 10, + "ellivu": 11, + "tólv": 12, + "trettan": 13, + "fjúrtan": 14, + "fímtan": 15, + "sekstan": 16, + "seytan": 17, + "átjan": 18, + "nítjan": 19 + }, + "TENS": { + "tjúgo": 20, + "tríati": 30, + "fýrati": 40, + "fimmti": 50, + "seksti": 60, + "sjeyti": 70, + "áttati": 80, + "níti": 90 + }, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "tusin": 1000, + "millióner": 1000000, + "milliarder": 1000000000, + "billióner": 1000000000000, + "billiarder": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "tíggju": 10, - "ellivu": 11, - "tólv": 12, - "trettan": 13, - "fjúrtan": 14, - "fímtan": 15, - "sekstan": 16, - "seytan": 17, - "átjan": 18, - "nítjan": 19 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": { - "tjúgo": 20, - "tríati": 30, - "fýrati": 40, - "fimmti": 50, - "seksti": 60, - "sjeyti": 70, - "áttati": 80, - "níti": 90 - }, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "tusin": 1000, - "millióner": 1000000, - "milliarder": 1000000000, - "billióner": 1000000000000, - "billiarder": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/fr-BE.py b/number_parser/data/fr-BE.py index 14112c4..32bcf43 100644 --- a/number_parser/data/fr-BE.py +++ b/number_parser/data/fr-BE.py @@ -1,44 +1,54 @@ info = { - "UNIT_NUMBERS": { - "zéro": 0, - "un": 1, - "une": 1, - "deux": 2, - "trois": 3, - "quatre": 4, - "cinq": 5, - "six": 6, - "sept": 7, - "huit": 8, - "neuf": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "zéro": 0, + "un": 1, + "une": 1, + "deux": 2, + "trois": 3, + "quatre": 4, + "cinq": 5, + "six": 6, + "sept": 7, + "huit": 8, + "neuf": 9 + }, + "DIRECT_NUMBERS": { + "dix": 10, + "onze": 11, + "douze": 12, + "treize": 13, + "quatorze": 14, + "quinze": 15, + "seize": 16 + }, + "TENS": { + "vingt": 20, + "trente": 30, + "quarante": 40, + "cinquante": 50, + "soixante": 60, + "septante": 70, + "quatre-vingt": 80, + "nonante": 90 + }, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "cent": 100, + "mille": 1000, + "millions": 1000000, + "milliards": 1000000000, + "billions": 1000000000000, + "billiards": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "dix": 10, - "onze": 11, - "douze": 12, - "treize": 13, - "quatorze": 14, - "quinze": 15, - "seize": 16 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": { - "vingt": 20, - "trente": 30, - "quarante": 40, - "cinquante": 50, - "soixante": 60, - "septante": 70, - "quatre-vingt": 80, - "nonante": 90 - }, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "cent": 100, - "mille": 1000, - "millions": 1000000, - "milliards": 1000000000, - "billions": 1000000000000, - "billiards": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/fr-CH.py b/number_parser/data/fr-CH.py index 6626c30..562ecf7 100644 --- a/number_parser/data/fr-CH.py +++ b/number_parser/data/fr-CH.py @@ -1,44 +1,54 @@ info = { - "UNIT_NUMBERS": { - "zéro": 0, - "un": 1, - "une": 1, - "deux": 2, - "trois": 3, - "quatre": 4, - "cinq": 5, - "six": 6, - "sept": 7, - "huit": 8, - "neuf": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "zéro": 0, + "un": 1, + "une": 1, + "deux": 2, + "trois": 3, + "quatre": 4, + "cinq": 5, + "six": 6, + "sept": 7, + "huit": 8, + "neuf": 9 + }, + "DIRECT_NUMBERS": { + "dix": 10, + "onze": 11, + "douze": 12, + "treize": 13, + "quatorze": 14, + "quinze": 15, + "seize": 16 + }, + "TENS": { + "vingt": 20, + "trente": 30, + "quarante": 40, + "cinquante": 50, + "soixante": 60, + "septante": 70, + "huitante": 80, + "nonante": 90 + }, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "cent": 100, + "mille": 1000, + "millions": 1000000, + "milliards": 1000000000, + "billions": 1000000000000, + "billiards": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "dix": 10, - "onze": 11, - "douze": 12, - "treize": 13, - "quatorze": 14, - "quinze": 15, - "seize": 16 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": { - "vingt": 20, - "trente": 30, - "quarante": 40, - "cinquante": 50, - "soixante": 60, - "septante": 70, - "huitante": 80, - "nonante": 90 - }, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "cent": 100, - "mille": 1000, - "millions": 1000000, - "milliards": 1000000000, - "billions": 1000000000000, - "billiards": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/fr.py b/number_parser/data/fr.py index 1a4b3d9..988308c 100644 --- a/number_parser/data/fr.py +++ b/number_parser/data/fr.py @@ -1,40 +1,50 @@ info = { - "UNIT_NUMBERS": { - "zéro": 0, - "un": 1, - "une": 1, - "deux": 2, - "trois": 3, - "quatre": 4, - "cinq": 5, - "six": 6, - "sept": 7, - "huit": 8, - "neuf": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "zéro": 0, + "un": 1, + "une": 1, + "deux": 2, + "trois": 3, + "quatre": 4, + "cinq": 5, + "six": 6, + "sept": 7, + "huit": 8, + "neuf": 9 + }, + "DIRECT_NUMBERS": { + "dix": 10, + "onze": 11, + "douze": 12, + "treize": 13, + "quatorze": 14, + "quinze": 15, + "seize": 16 + }, + "TENS": { + "vingt": 20, + "trente": 30, + "quarante": 40, + "cinquante": 50 + }, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "cent": 100, + "mille": 1000, + "millions": 1000000, + "milliards": 1000000000, + "billions": 1000000000000, + "billiards": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "dix": 10, - "onze": 11, - "douze": 12, - "treize": 13, - "quatorze": 14, - "quinze": 15, - "seize": 16 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": { - "vingt": 20, - "trente": 30, - "quarante": 40, - "cinquante": 50 - }, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "cent": 100, - "mille": 1000, - "millions": 1000000, - "milliards": 1000000000, - "billions": 1000000000000, - "billiards": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/ga.py b/number_parser/data/ga.py index a83484b..94b114a 100644 --- a/number_parser/data/ga.py +++ b/number_parser/data/ga.py @@ -1,43 +1,53 @@ info = { - "UNIT_NUMBERS": { - "a náid": 0, - "náid": 0, - "a haon": 1, - "aon": 1, - "a dó": 2, - "dhá": 2, - "dó": 2, - "a trí": 3, - "trí": 3, - "a ceathair": 4, - "ceathair": 4, - "ceithre": 4, - "a cúig": 5, - "cúig": 5, - "a sé": 6, - "sé": 6, - "a seacht": 7, - "seacht": 7, - "a hocht": 8, - "ocht": 8, - "a naoi": 9, - "naoi": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "a náid": 0, + "náid": 0, + "a haon": 1, + "aon": 1, + "a dó": 2, + "dhá": 2, + "dó": 2, + "a trí": 3, + "trí": 3, + "a ceathair": 4, + "ceathair": 4, + "ceithre": 4, + "a cúig": 5, + "cúig": 5, + "a sé": 6, + "sé": 6, + "a seacht": 7, + "seacht": 7, + "a hocht": 8, + "ocht": 8, + "a naoi": 9, + "naoi": 9 + }, + "DIRECT_NUMBERS": { + "a deich": 10, + "deich": 10 + }, + "TENS": { + "fiche": 20, + "tríocha": 30, + "daichead": 40, + "caoga": 50, + "seasca": 60, + "seachtó": 70, + "ochtó": 80, + "nócha": 90 + }, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "DIRECT_NUMBERS": { - "a deich": 10, - "deich": 10 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": { - "fiche": 20, - "tríocha": 30, - "daichead": 40, - "caoga": 50, - "seasca": 60, - "seachtó": 70, - "ochtó": 80, - "nócha": 90 - }, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/he.py b/number_parser/data/he.py index 798b047..df4a127 100644 --- a/number_parser/data/he.py +++ b/number_parser/data/he.py @@ -1,54 +1,64 @@ info = { - "UNIT_NUMBERS": { - "אפס": 0, - "אחד": 1, - "אחת": 1, - "שני": 2, - "שניים": 2, - "שתי": 2, - "שתיים": 2, - "שלוש": 3, - "שלושה": 3, - "ארבע": 4, - "ארבעה": 4, - "חמישה": 5, - "חמש": 5, - "שישה": 6, - "שש": 6, - "שבע": 7, - "שבעה": 7, - "שמונה": 8, - "תשע": 9, - "תשעה": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "אפס": 0, + "אחד": 1, + "אחת": 1, + "שני": 2, + "שניים": 2, + "שתי": 2, + "שתיים": 2, + "שלוש": 3, + "שלושה": 3, + "ארבע": 4, + "ארבעה": 4, + "חמישה": 5, + "חמש": 5, + "שישה": 6, + "שש": 6, + "שבע": 7, + "שבעה": 7, + "שמונה": 8, + "תשע": 9, + "תשעה": 9 + }, + "DIRECT_NUMBERS": { + "עשר": 10, + "עשרה": 10, + "אחד עשר": 11, + "אחת עשרה": 11, + "שניים עשר": 12, + "שתים עשרה": 12 + }, + "TENS": { + "עשרים": 20, + "שלושים": 30, + "ארבעים": 40, + "חמישים": 50, + "שישים": 60, + "שבעים": 70, + "שמונים": 80, + "תשעים": 90 + }, + "HUNDREDS": { + "מאתיים": 200 + }, + "BIG_POWERS_OF_TEN": { + "מאות": 100, + "אלפים": 1000, + "מיליון": 1000000, + "מיליארד": 1000000000, + "ביליון": 1000000000000, + "טריליון": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "עשר": 10, - "עשרה": 10, - "אחד עשר": 11, - "אחת עשרה": 11, - "שניים עשר": 12, - "שתים עשרה": 12 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": { - "עשרים": 20, - "שלושים": 30, - "ארבעים": 40, - "חמישים": 50, - "שישים": 60, - "שבעים": 70, - "שמונים": 80, - "תשעים": 90 - }, - "HUNDREDS": { - "מאתיים": 200 - }, - "BIG_POWERS_OF_TEN": { - "מאות": 100, - "אלפים": 1000, - "מיליון": 1000000, - "מיליארד": 1000000000, - "ביליון": 1000000000000, - "טריליון": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/hi.py b/number_parser/data/hi.py index e711696..d97f0d1 100644 --- a/number_parser/data/hi.py +++ b/number_parser/data/hi.py @@ -1,117 +1,127 @@ info = { - "UNIT_NUMBERS": { - "शून्य": 0, - "एक": 1, - "दो": 2, - "तीन": 3, - "चार": 4, - "पाँच": 5, - "छह": 6, - "सात": 7, - "आठ": 8, - "नौ": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "शून्य": 0, + "एक": 1, + "दो": 2, + "तीन": 3, + "चार": 4, + "पाँच": 5, + "छह": 6, + "सात": 7, + "आठ": 8, + "नौ": 9 + }, + "DIRECT_NUMBERS": { + "दस": 10, + "ग्यारह": 11, + "बारह": 12, + "तेरह": 13, + "चौदह": 14, + "पन्द्रह": 15, + "सोलह": 16, + "सत्रह": 17, + "अठारह": 18, + "उन्नीस": 19, + "बीस": 20, + "इक्कीस": 21, + "बाईस": 22, + "तेईस": 23, + "चौबीस": 24, + "पच्चीस": 25, + "छब्बीस": 26, + "सत्ताईस": 27, + "अट्ठाईस": 28, + "उनतीस": 29, + "तीस": 30, + "इकतीस": 31, + "बत्तीस": 32, + "तैंतीस": 33, + "चौंतीस": 34, + "पैंतीस": 35, + "छत्तीस": 36, + "सैंतीस": 37, + "अड़तीस": 38, + "उनतालीस": 39, + "चालीस": 40, + "इकतालीस": 41, + "बयालीस": 42, + "तैंतालीस": 43, + "चौवालीस": 44, + "पैंतालीस": 45, + "छियालीस": 46, + "सैंतालीस": 47, + "अड़तालीस": 48, + "उनचास": 49, + "पचास": 50, + "इक्यावन": 51, + "बावन": 52, + "तिरेपन": 53, + "चौवन": 54, + "पचपन": 55, + "छप्पन": 56, + "सत्तावन": 57, + "अट्ठावन": 58, + "उनसठ": 59, + "साठ": 60, + "इकसठ": 61, + "बासठ": 62, + "तिरेसठ": 63, + "चौंसठ": 64, + "पैंसठ": 65, + "छियासठ": 66, + "सड़सठ": 67, + "अड़सठ": 68, + "उनहत्तर": 69, + "सत्तर": 70, + "इकहत्तर": 71, + "बहत्तर": 72, + "तिहत्तर": 73, + "चौहत्तर": 74, + "पचहत्तर": 75, + "छिहत्तर": 76, + "सतहत्तर": 77, + "अठहत्तर": 78, + "उनासी": 79, + "अस्सी": 80, + "इक्यासी": 81, + "बयासी": 82, + "तिरासी": 83, + "चौरासी": 84, + "पचासी": 85, + "छियासी": 86, + "सत्तासी": 87, + "अट्ठासी": 88, + "नवासी": 89, + "नब्बे": 90, + "इक्यानबे": 91, + "बानबे": 92, + "तिरानबे": 93, + "चौरानबे": 94, + "पंचानबे": 95, + "छियानबे": 96, + "सत्तानबे": 97, + "अट्ठानबे": 98, + "निन्यानबे": 99 + }, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "सौ": 100, + "हज़ार": 1000, + "लाख": 100000, + "करोड़": 10000000, + "अरब": 1000000000, + "खरब": 100000000000 + } }, - "DIRECT_NUMBERS": { - "दस": 10, - "ग्यारह": 11, - "बारह": 12, - "तेरह": 13, - "चौदह": 14, - "पन्द्रह": 15, - "सोलह": 16, - "सत्रह": 17, - "अठारह": 18, - "उन्नीस": 19, - "बीस": 20, - "इक्कीस": 21, - "बाईस": 22, - "तेईस": 23, - "चौबीस": 24, - "पच्चीस": 25, - "छब्बीस": 26, - "सत्ताईस": 27, - "अट्ठाईस": 28, - "उनतीस": 29, - "तीस": 30, - "इकतीस": 31, - "बत्तीस": 32, - "तैंतीस": 33, - "चौंतीस": 34, - "पैंतीस": 35, - "छत्तीस": 36, - "सैंतीस": 37, - "अड़तीस": 38, - "उनतालीस": 39, - "चालीस": 40, - "इकतालीस": 41, - "बयालीस": 42, - "तैंतालीस": 43, - "चौवालीस": 44, - "पैंतालीस": 45, - "छियालीस": 46, - "सैंतालीस": 47, - "अड़तालीस": 48, - "उनचास": 49, - "पचास": 50, - "इक्यावन": 51, - "बावन": 52, - "तिरेपन": 53, - "चौवन": 54, - "पचपन": 55, - "छप्पन": 56, - "सत्तावन": 57, - "अट्ठावन": 58, - "उनसठ": 59, - "साठ": 60, - "इकसठ": 61, - "बासठ": 62, - "तिरेसठ": 63, - "चौंसठ": 64, - "पैंसठ": 65, - "छियासठ": 66, - "सड़सठ": 67, - "अड़सठ": 68, - "उनहत्तर": 69, - "सत्तर": 70, - "इकहत्तर": 71, - "बहत्तर": 72, - "तिहत्तर": 73, - "चौहत्तर": 74, - "पचहत्तर": 75, - "छिहत्तर": 76, - "सतहत्तर": 77, - "अठहत्तर": 78, - "उनासी": 79, - "अस्सी": 80, - "इक्यासी": 81, - "बयासी": 82, - "तिरासी": 83, - "चौरासी": 84, - "पचासी": 85, - "छियासी": 86, - "सत्तासी": 87, - "अट्ठासी": 88, - "नवासी": 89, - "नब्बे": 90, - "इक्यानबे": 91, - "बानबे": 92, - "तिरानबे": 93, - "चौरानबे": 94, - "पंचानबे": 95, - "छियानबे": 96, - "सत्तानबे": 97, - "अट्ठानबे": 98, - "निन्यानबे": 99 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "सौ": 100, - "हज़ार": 1000, - "लाख": 100000, - "करोड़": 10000000, - "अरब": 1000000000, - "खरब": 100000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/hr.py b/number_parser/data/hr.py index 6f880d9..3cc2e90 100644 --- a/number_parser/data/hr.py +++ b/number_parser/data/hr.py @@ -1,64 +1,74 @@ info = { - "UNIT_NUMBERS": { - "nula": 0, - "jedan": 1, - "jedna": 1, - "jedno": 1, - "dva": 2, - "dvije": 2, - "tri": 3, - "četiri": 4, - "pet": 5, - "šest": 6, - "sedam": 7, - "osam": 8, - "devet": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "nula": 0, + "jedan": 1, + "jedna": 1, + "jedno": 1, + "dva": 2, + "dvije": 2, + "tri": 3, + "četiri": 4, + "pet": 5, + "šest": 6, + "sedam": 7, + "osam": 8, + "devet": 9 + }, + "DIRECT_NUMBERS": { + "deset": 10, + "jedanaest": 11, + "dvanaest": 12, + "trinaest": 13, + "četrnaest": 14, + "petnaest": 15, + "šesnaest": 16, + "sedamnaest": 17, + "osamnaest": 18, + "devetnaest": 19 + }, + "TENS": { + "dvadeset": 20, + "trideset": 30, + "četrdeset": 40, + "pedeset": 50, + "šezdeset": 60, + "sedamdeset": 70, + "osamdeset": 80, + "devedeset": 90 + }, + "HUNDREDS": { + "dvjesto": 200, + "tristo": 300, + "četiristo": 400, + "petsto": 500, + "šeststo": 600, + "sedamsto": 700, + "osamsto": 800, + "devetsto": 900 + }, + "BIG_POWERS_OF_TEN": { + "tisuća": 1000, + "tisuće": 1000, + "milijun": 1000000, + "milijuna": 1000000, + "milijarda": 1000000000, + "milijarde": 1000000000, + "milijardi": 1000000000, + "bilijun": 1000000000000, + "bilijuna": 1000000000000, + "bilijarda": 1000000000000000, + "bilijarde": 1000000000000000, + "bilijardi": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "deset": 10, - "jedanaest": 11, - "dvanaest": 12, - "trinaest": 13, - "četrnaest": 14, - "petnaest": 15, - "šesnaest": 16, - "sedamnaest": 17, - "osamnaest": 18, - "devetnaest": 19 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": { - "dvadeset": 20, - "trideset": 30, - "četrdeset": 40, - "pedeset": 50, - "šezdeset": 60, - "sedamdeset": 70, - "osamdeset": 80, - "devedeset": 90 - }, - "HUNDREDS": { - "dvjesto": 200, - "tristo": 300, - "četiristo": 400, - "petsto": 500, - "šeststo": 600, - "sedamsto": 700, - "osamsto": 800, - "devetsto": 900 - }, - "BIG_POWERS_OF_TEN": { - "tisuća": 1000, - "tisuće": 1000, - "milijun": 1000000, - "milijuna": 1000000, - "milijarda": 1000000000, - "milijarde": 1000000000, - "milijardi": 1000000000, - "bilijun": 1000000000000, - "bilijuna": 1000000000000, - "bilijarda": 1000000000000000, - "bilijarde": 1000000000000000, - "bilijardi": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/hu.py b/number_parser/data/hu.py index 3cdfd63..1ad9340 100644 --- a/number_parser/data/hu.py +++ b/number_parser/data/hu.py @@ -1,31 +1,41 @@ info = { - "UNIT_NUMBERS": { - "nulla": 0, - "egy": 1, - "kettő": 2, - "két": 2, - "három": 3, - "négy": 4, - "öt": 5, - "hat": 6, - "hét": 7, - "nyolc": 8, - "kilenc": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "nulla": 0, + "egy": 1, + "kettő": 2, + "két": 2, + "három": 3, + "négy": 4, + "öt": 5, + "hat": 6, + "hét": 7, + "nyolc": 8, + "kilenc": 9 + }, + "DIRECT_NUMBERS": { + "tíz": 10, + "húsz": 20 + }, + "TENS": { + "harminc": 30, + "negyven": 40, + "ötven": 50, + "hatvan": 60, + "hetven": 70, + "nyolcvan": 80, + "kilencven": 90 + }, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "DIRECT_NUMBERS": { - "tíz": 10, - "húsz": 20 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": { - "harminc": 30, - "negyven": 40, - "ötven": 50, - "hatvan": 60, - "hetven": 70, - "nyolcvan": 80, - "kilencven": 90 - }, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/hy.py b/number_parser/data/hy.py index 9bc9dc2..9d15f91 100644 --- a/number_parser/data/hy.py +++ b/number_parser/data/hy.py @@ -1,34 +1,44 @@ info = { - "UNIT_NUMBERS": { - "զրո": 0, - "մեկ": 1, - "երկու": 2, - "երեք": 3, - "չորս": 4, - "հինգ": 5, - "վեց": 6, - "յոթ": 7, - "ութ": 8, - "ինը": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "զրո": 0, + "մեկ": 1, + "երկու": 2, + "երեք": 3, + "չորս": 4, + "հինգ": 5, + "վեց": 6, + "յոթ": 7, + "ութ": 8, + "ինը": 9 + }, + "DIRECT_NUMBERS": {}, + "TENS": { + "քսան": 20, + "երեսուն": 30, + "քառասուն": 40, + "հիսուն": 50, + "վաթսուն": 60, + "յոթանասուն": 70, + "ութսուն": 80, + "իննասուն": 90 + }, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "հազար": 1000, + "միլիօն": 1000000, + "միլիար": 1000000000, + "բիլիօն": 1000000000000, + "բիլիար": 1000000000000000 + } }, - "DIRECT_NUMBERS": {}, - "TENS": { - "քսան": 20, - "երեսուն": 30, - "քառասուն": 40, - "հիսուն": 50, - "վաթսուն": 60, - "յոթանասուն": 70, - "ութսուն": 80, - "իննասուն": 90 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "հազար": 1000, - "միլիօն": 1000000, - "միլիար": 1000000000, - "բիլիօն": 1000000000000, - "բիլիար": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/id.py b/number_parser/data/id.py index f4368ea..7a774d7 100644 --- a/number_parser/data/id.py +++ b/number_parser/data/id.py @@ -1,29 +1,39 @@ info = { - "UNIT_NUMBERS": { - "kosong": 0, - "satu": 1, - "dua": 2, - "tiga": 3, - "empat": 4, - "lima": 5, - "enam": 6, - "tujuh": 7, - "delapan": 8, - "sembilan": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "kosong": 0, + "satu": 1, + "dua": 2, + "tiga": 3, + "empat": 4, + "lima": 5, + "enam": 6, + "tujuh": 7, + "delapan": 8, + "sembilan": 9 + }, + "DIRECT_NUMBERS": { + "sepuluh": 10, + "sebelas": 11 + }, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "ratus": 100, + "ribu": 1000, + "juta": 1000000, + "miliar": 1000000000, + "triliun": 1000000000000, + "kuadriliun": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "sepuluh": 10, - "sebelas": 11 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "ratus": 100, - "ribu": 1000, - "juta": 1000000, - "miliar": 1000000000, - "triliun": 1000000000000, - "kuadriliun": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/is.py b/number_parser/data/is.py index fc2f2a6..57032f7 100644 --- a/number_parser/data/is.py +++ b/number_parser/data/is.py @@ -1,53 +1,63 @@ info = { - "UNIT_NUMBERS": { - "núll": 0, - "ein": 1, - "einn": 1, - "eitt": 1, - "tveir": 2, - "tvær": 2, - "tvö": 2, - "þrjár": 3, - "þrjú": 3, - "þrír": 3, - "fjórar": 4, - "fjórir": 4, - "fjögur": 4, - "fimm": 5, - "sex": 6, - "sjó": 7, - "átta": 8, - "níu": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "núll": 0, + "ein": 1, + "einn": 1, + "eitt": 1, + "tveir": 2, + "tvær": 2, + "tvö": 2, + "þrjár": 3, + "þrjú": 3, + "þrír": 3, + "fjórar": 4, + "fjórir": 4, + "fjögur": 4, + "fimm": 5, + "sex": 6, + "sjó": 7, + "átta": 8, + "níu": 9 + }, + "DIRECT_NUMBERS": { + "tíu": 10, + "ellefu": 11, + "tólf": 12, + "þrettán": 13, + "fjórtán": 14, + "fimmtán": 15, + "sextán": 16, + "sautján": 17, + "átján": 18, + "nítján": 19 + }, + "TENS": { + "tuttugu": 20, + "þrjátíu": 30, + "fjörutíu": 40, + "fimmtíu": 50, + "sextíu": 60, + "sjötíu": 70, + "áttatíu": 80, + "níutíu": 90 + }, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "þúsund": 1000, + "milliónur": 1000000, + "milliarður": 1000000000, + "billiónur": 1000000000000, + "billiarður": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "tíu": 10, - "ellefu": 11, - "tólf": 12, - "þrettán": 13, - "fjórtán": 14, - "fimmtán": 15, - "sextán": 16, - "sautján": 17, - "átján": 18, - "nítján": 19 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": { - "tuttugu": 20, - "þrjátíu": 30, - "fjörutíu": 40, - "fimmtíu": 50, - "sextíu": 60, - "sjötíu": 70, - "áttatíu": 80, - "níutíu": 90 - }, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "þúsund": 1000, - "milliónur": 1000000, - "milliarður": 1000000000, - "billiónur": 1000000000000, - "billiarður": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/it.py b/number_parser/data/it.py index f203619..10d1969 100644 --- a/number_parser/data/it.py +++ b/number_parser/data/it.py @@ -1,46 +1,56 @@ info = { - "UNIT_NUMBERS": { - "zero": 0, - "un": 1, - "una": 1, - "uno": 1, - "due": 2, - "tre": 3, - "quattro": 4, - "cinque": 5, - "sei": 6, - "sette": 7, - "otto": 8, - "nove": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "zero": 0, + "un": 1, + "una": 1, + "uno": 1, + "due": 2, + "tre": 3, + "quattro": 4, + "cinque": 5, + "sei": 6, + "sette": 7, + "otto": 8, + "nove": 9 + }, + "DIRECT_NUMBERS": { + "dieci": 10, + "undici": 11, + "dodici": 12, + "tredici": 13, + "quattordici": 14, + "quindici": 15, + "sedici": 16, + "diciassette": 17, + "diciotto": 18, + "diciannove": 19 + }, + "TENS": { + "vent": 20, + "trent": 30, + "quarant": 40, + "cinquant": 50, + "sessant": 60, + "settant": 70, + "ottant": 80, + "novant": 90 + }, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "milioni": 1000000, + "miliardi": 1000000000, + "bilioni": 1000000000000, + "biliardi": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "dieci": 10, - "undici": 11, - "dodici": 12, - "tredici": 13, - "quattordici": 14, - "quindici": 15, - "sedici": 16, - "diciassette": 17, - "diciotto": 18, - "diciannove": 19 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": { - "vent": 20, - "trent": 30, - "quarant": 40, - "cinquant": 50, - "sessant": 60, - "settant": 70, - "ottant": 80, - "novant": 90 - }, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "milioni": 1000000, - "miliardi": 1000000000, - "bilioni": 1000000000000, - "biliardi": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/ja.py b/number_parser/data/ja.py index f9046c1..28e56f6 100644 --- a/number_parser/data/ja.py +++ b/number_parser/data/ja.py @@ -1,25 +1,35 @@ info = { - "UNIT_NUMBERS": { - "〇": 0, - "零": 0, - "一": 1, - "元": 1, - "壱": 1, - "二": 2, - "弐": 2, - "三": 3, - "参": 3, - "四": 4, - "五": 5, - "伍": 5, - "六": 6, - "七": 7, - "八": 8, - "九": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "〇": 0, + "零": 0, + "一": 1, + "元": 1, + "壱": 1, + "二": 2, + "弐": 2, + "三": 3, + "参": 3, + "四": 4, + "五": 5, + "伍": 5, + "六": 6, + "七": 7, + "八": 8, + "九": 9 + }, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/ka.py b/number_parser/data/ka.py index 3f377fa..98fbac5 100644 --- a/number_parser/data/ka.py +++ b/number_parser/data/ka.py @@ -1,45 +1,55 @@ info = { - "UNIT_NUMBERS": { - "ნული": 0, - "ერთი": 1, - "ორი": 2, - "სამი": 3, - "ოთხი": 4, - "ხუთი": 5, - "ექვსი": 6, - "შვიდი": 7, - "რვა": 8, - "ცხრა": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "ნული": 0, + "ერთი": 1, + "ორი": 2, + "სამი": 3, + "ოთხი": 4, + "ხუთი": 5, + "ექვსი": 6, + "შვიდი": 7, + "რვა": 8, + "ცხრა": 9 + }, + "DIRECT_NUMBERS": { + "ათი": 10, + "თერთმეტი": 11, + "თორმეტი": 12, + "ცამეტი": 13, + "თოთხმეტი": 14, + "თხუთმეტი": 15, + "თექვსმეტი": 16, + "ჩვიდმეტი": 17, + "თრვამეტი": 18, + "ცხრამეტი": 19 + }, + "TENS": {}, + "HUNDREDS": { + "ორას": 200, + "სამას": 300, + "ოთხას": 400, + "ხუთას": 500, + "ექვსას": 600, + "შვიდას": 700, + "რვაას": 800, + "ცხრაას": 900 + }, + "BIG_POWERS_OF_TEN": { + "ათას": 1000, + "მილიონ": 1000000, + "მილიარდ": 1000000000, + "ბილიონ": 1000000000000, + "ბილიარდ": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "ათი": 10, - "თერთმეტი": 11, - "თორმეტი": 12, - "ცამეტი": 13, - "თოთხმეტი": 14, - "თხუთმეტი": 15, - "თექვსმეტი": 16, - "ჩვიდმეტი": 17, - "თრვამეტი": 18, - "ცხრამეტი": 19 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": {}, - "HUNDREDS": { - "ორას": 200, - "სამას": 300, - "ოთხას": 400, - "ხუთას": 500, - "ექვსას": 600, - "შვიდას": 700, - "რვაას": 800, - "ცხრაას": 900 - }, - "BIG_POWERS_OF_TEN": { - "ათას": 1000, - "მილიონ": 1000000, - "მილიარდ": 1000000000, - "ბილიონ": 1000000000000, - "ბილიარდ": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/kl.py b/number_parser/data/kl.py index 702ed2c..b544fdb 100644 --- a/number_parser/data/kl.py +++ b/number_parser/data/kl.py @@ -1,37 +1,47 @@ info = { - "UNIT_NUMBERS": { - "nuulu": 0, - "ataaseq": 1, - "marluk": 2, - "pingasut": 3, - "sisamat": 4, - "tallimat": 5, - "arfinillit": 6, - "arfineq-marluk": 7, - "arfineq-pingasut": 8, - "arfineq-sisamat": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "nuulu": 0, + "ataaseq": 1, + "marluk": 2, + "pingasut": 3, + "sisamat": 4, + "tallimat": 5, + "arfinillit": 6, + "arfineq-marluk": 7, + "arfineq-pingasut": 8, + "arfineq-sisamat": 9 + }, + "DIRECT_NUMBERS": { + "qulit": 10, + "aqqanilit": 11, + "aqqaneq-marluk": 12, + "aqqaneq-pingasut": 13, + "aqqaneq-sisamat": 14, + "aqqaneq-tallimat": 15, + "arfersanillit": 16, + "arfersaneq-marluk": 17, + "arfersaneq-pingasut": 18, + "arfersaneq-sisamat": 19 + }, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "uutritillit": 100, + "tuusintillit": 1000, + "millionillit": 1000000, + "milliardillit": 1000000000, + "billioniillit": 1000000000000, + "billiardillit": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "qulit": 10, - "aqqanilit": 11, - "aqqaneq-marluk": 12, - "aqqaneq-pingasut": 13, - "aqqaneq-sisamat": 14, - "aqqaneq-tallimat": 15, - "arfersanillit": 16, - "arfersaneq-marluk": 17, - "arfersaneq-pingasut": 18, - "arfersaneq-sisamat": 19 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "uutritillit": 100, - "tuusintillit": 1000, - "millionillit": 1000000, - "milliardillit": 1000000000, - "billioniillit": 1000000000000, - "billiardillit": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/km.py b/number_parser/data/km.py index c130bb5..7f339ae 100644 --- a/number_parser/data/km.py +++ b/number_parser/data/km.py @@ -1,28 +1,38 @@ info = { - "UNIT_NUMBERS": { - "សូន្យ": 0, - "មួយ": 1, - "ពីរ": 2, - "បី": 3, - "បួន": 4, - "ប្រាំ": 5, - "ប្រាំមួយ": 6, - "ប្រាំពីរ": 7, - "ប្រាំបី": 8, - "ប្រាំបួន": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "សូន្យ": 0, + "មួយ": 1, + "ពីរ": 2, + "បី": 3, + "បួន": 4, + "ប្រាំ": 5, + "ប្រាំមួយ": 6, + "ប្រាំពីរ": 7, + "ប្រាំបី": 8, + "ប្រាំបួន": 9 + }, + "DIRECT_NUMBERS": {}, + "TENS": { + "ម្ភៃ": 20, + "សាមសិប": 30, + "សែសិប": 40, + "ហាសិប": 50, + "ហុកសិប": 60, + "ចិតសិប": 70, + "ប៉ែតសិប": 80, + "កៅសិប": 90 + }, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "DIRECT_NUMBERS": {}, - "TENS": { - "ម្ភៃ": 20, - "សាមសិប": 30, - "សែសិប": 40, - "ហាសិប": 50, - "ហុកសិប": 60, - "ចិតសិប": 70, - "ប៉ែតសិប": 80, - "កៅសិប": 90 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/ko.py b/number_parser/data/ko.py index 015c2ea..9870739 100644 --- a/number_parser/data/ko.py +++ b/number_parser/data/ko.py @@ -1,44 +1,54 @@ info = { - "UNIT_NUMBERS": { - "공": 0, - "영": 0, - "일": 1, - "하나": 1, - "한": 1, - "두": 2, - "둘": 2, - "이": 2, - "삼": 3, - "세": 3, - "셋": 3, - "네": 4, - "넷": 4, - "사": 4, - "다섯": 5, - "오": 5, - "여섯": 6, - "육": 6, - "일곱": 7, - "칠": 7, - "여덟": 8, - "팔": 8, - "구": 9, - "아홉": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "공": 0, + "영": 0, + "일": 1, + "하나": 1, + "한": 1, + "두": 2, + "둘": 2, + "이": 2, + "삼": 3, + "세": 3, + "셋": 3, + "네": 4, + "넷": 4, + "사": 4, + "다섯": 5, + "오": 5, + "여섯": 6, + "육": 6, + "일곱": 7, + "칠": 7, + "여덟": 8, + "팔": 8, + "구": 9, + "아홉": 9 + }, + "DIRECT_NUMBERS": { + "스무": 20 + }, + "TENS": { + "스물": 20, + "서른": 30, + "마흔": 40, + "쉰": 50, + "예순": 60, + "일흔": 70, + "여든": 80, + "아흔": 90 + }, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "DIRECT_NUMBERS": { - "스무": 20 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": { - "스물": 20, - "서른": 30, - "마흔": 40, - "쉰": 50, - "예순": 60, - "일흔": 70, - "여든": 80, - "아흔": 90 - }, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/ky.py b/number_parser/data/ky.py index 6d554be..54d4b1c 100644 --- a/number_parser/data/ky.py +++ b/number_parser/data/ky.py @@ -1,35 +1,45 @@ info = { - "UNIT_NUMBERS": { - "нөл": 0, - "бир": 1, - "эки": 2, - "үч": 3, - "төрт": 4, - "беш": 5, - "алты": 6, - "жети": 7, - "сегиз": 8, - "тогуз": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "нөл": 0, + "бир": 1, + "эки": 2, + "үч": 3, + "төрт": 4, + "беш": 5, + "алты": 6, + "жети": 7, + "сегиз": 8, + "тогуз": 9 + }, + "DIRECT_NUMBERS": {}, + "TENS": { + "жыйырма": 20, + "отуз": 30, + "кырк": 40, + "элүү": 50, + "алтымыш": 60, + "жетимиш": 70, + "сексен": 80, + "токсон": 90 + }, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "жүз": 100, + "миң": 100000, + "миллион": 1000000, + "миллиард": 1000000000, + "триллион": 1000000000000, + "квадриллион": 1000000000000000 + } }, - "DIRECT_NUMBERS": {}, - "TENS": { - "жыйырма": 20, - "отуз": 30, - "кырк": 40, - "элүү": 50, - "алтымыш": 60, - "жетимиш": 70, - "сексен": 80, - "токсон": 90 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "жүз": 100, - "миң": 100000, - "миллион": 1000000, - "миллиард": 1000000000, - "триллион": 1000000000000, - "квадриллион": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/lb.py b/number_parser/data/lb.py index 79d7020..69b85aa 100644 --- a/number_parser/data/lb.py +++ b/number_parser/data/lb.py @@ -1,42 +1,52 @@ info = { - "UNIT_NUMBERS": { - "null": 0, - "een": 1, - "eent": 1, - "eng": 1, - "zwee": 2, - "zwou": 2, - "dräi": 3, - "véier": 4, - "fënnef": 5, - "sechs": 6, - "siwen": 7, - "aacht": 8, - "néng": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "null": 0, + "een": 1, + "eent": 1, + "eng": 1, + "zwee": 2, + "zwou": 2, + "dräi": 3, + "véier": 4, + "fënnef": 5, + "sechs": 6, + "siwen": 7, + "aacht": 8, + "néng": 9 + }, + "DIRECT_NUMBERS": { + "zéng": 10, + "eelef": 11, + "zwielef": 12, + "dräizéng": 13, + "véierzéng": 14, + "fofzéng": 15, + "siechzéng": 16, + "siwwenzéng": 17, + "uechtzéng": 18, + "nonzéng": 19 + }, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "Millioun": 1000000, + "Milliounen": 1000000, + "Milliard": 1000000000, + "Milliarden": 1000000000, + "Billioun": 1000000000000, + "Billiounen": 1000000000000, + "Billiard": 1000000000000000, + "Billiarden": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "zéng": 10, - "eelef": 11, - "zwielef": 12, - "dräizéng": 13, - "véierzéng": 14, - "fofzéng": 15, - "siechzéng": 16, - "siwwenzéng": 17, - "uechtzéng": 18, - "nonzéng": 19 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "Millioun": 1000000, - "Milliounen": 1000000, - "Milliard": 1000000000, - "Milliarden": 1000000000, - "Billioun": 1000000000000, - "Billiounen": 1000000000000, - "Billiard": 1000000000000000, - "Billiarden": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/lo.py b/number_parser/data/lo.py index 4d4fae0..8158ede 100644 --- a/number_parser/data/lo.py +++ b/number_parser/data/lo.py @@ -1,21 +1,31 @@ info = { - "UNIT_NUMBERS": { - "ศูนย์": 0, - "ໜຶ່ງ": 1, - "ສອງ": 2, - "ສາມ": 3, - "ສີ່": 4, - "ຫ້າ": 5, - "ຫົກ": 6, - "ເຈັດ": 7, - "ແປດ": 8, - "ເກົ້າ": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "ศูนย์": 0, + "ໜຶ່ງ": 1, + "ສອງ": 2, + "ສາມ": 3, + "ສີ່": 4, + "ຫ້າ": 5, + "ຫົກ": 6, + "ເຈັດ": 7, + "ແປດ": 8, + "ເກົ້າ": 9 + }, + "DIRECT_NUMBERS": {}, + "TENS": { + "ຊາວ": 20 + }, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "DIRECT_NUMBERS": {}, - "TENS": { - "ຊາວ": 20 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/lrc.py b/number_parser/data/lrc.py index 840a583..cc8f9c4 100644 --- a/number_parser/data/lrc.py +++ b/number_parser/data/lrc.py @@ -1,53 +1,63 @@ info = { - "UNIT_NUMBERS": { - "صفر": 0, - "یٱک": 1, - "دۏ": 2, - "ساٛ": 3, - "چار": 4, - "پنج": 5, - "شٱش": 6, - "هفت": 7, - "هشت": 8, - "نۏ": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "صفر": 0, + "یٱک": 1, + "دۏ": 2, + "ساٛ": 3, + "چار": 4, + "پنج": 5, + "شٱش": 6, + "هفت": 7, + "هشت": 8, + "نۏ": 9 + }, + "DIRECT_NUMBERS": { + "دٱ": 10, + "یازدٱ": 11, + "دۊۋازدٱ": 12, + "سینزٱ": 13, + "چاردٱ": 14, + "پۊمزٱ": 15, + "شۊمزٱ": 16, + "هاٛبدٱ": 17, + "هیژدٱ": 18, + "نۊزدٱ": 19 + }, + "TENS": { + "بیست": 20, + "سی": 30, + "چاٛهل": 40, + "پنجا": 50, + "شٱصد": 60, + "هفتاد": 70, + "هشتاد": 80, + "نٱۋد": 90 + }, + "HUNDREDS": { + "داٛۋیسد": 200, + "سیصد": 300, + "چارصد": 400, + "پۊمصد": 500, + "شٱشصد": 600, + "هفصد": 700, + "هشصد": 800, + "نۏصد": 900 + }, + "BIG_POWERS_OF_TEN": { + "هزار": 1000, + "ماٛلیۊن": 1000000, + "میلیارد": 1000000000, + "هزار میلیاد": 1000000000000 + } }, - "DIRECT_NUMBERS": { - "دٱ": 10, - "یازدٱ": 11, - "دۊۋازدٱ": 12, - "سینزٱ": 13, - "چاردٱ": 14, - "پۊمزٱ": 15, - "شۊمزٱ": 16, - "هاٛبدٱ": 17, - "هیژدٱ": 18, - "نۊزدٱ": 19 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": { - "بیست": 20, - "سی": 30, - "چاٛهل": 40, - "پنجا": 50, - "شٱصد": 60, - "هفتاد": 70, - "هشتاد": 80, - "نٱۋد": 90 - }, - "HUNDREDS": { - "داٛۋیسد": 200, - "سیصد": 300, - "چارصد": 400, - "پۊمصد": 500, - "شٱشصد": 600, - "هفصد": 700, - "هشصد": 800, - "نۏصد": 900 - }, - "BIG_POWERS_OF_TEN": { - "هزار": 1000, - "ماٛلیۊن": 1000000, - "میلیارد": 1000000000, - "هزار میلیاد": 1000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/lt.py b/number_parser/data/lt.py index 6ab7bea..97c2e73 100644 --- a/number_parser/data/lt.py +++ b/number_parser/data/lt.py @@ -1,39 +1,49 @@ info = { - "UNIT_NUMBERS": { - "nulis": 0, - "viena": 1, - "vienas": 1, - "du": 2, - "dvi": 2, - "tris": 3, - "trys": 3, - "keturi": 4, - "keturias": 4, - "penki": 5, - "penkias": 5, - "šeši": 6, - "šešias": 6, - "septyni": 7, - "septynias": 7, - "aštuoni": 8, - "aštuonias": 8, - "devyni": 9, - "devynias": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "nulis": 0, + "viena": 1, + "vienas": 1, + "du": 2, + "dvi": 2, + "tris": 3, + "trys": 3, + "keturi": 4, + "keturias": 4, + "penki": 5, + "penkias": 5, + "šeši": 6, + "šešias": 6, + "septyni": 7, + "septynias": 7, + "aštuoni": 8, + "aštuonias": 8, + "devyni": 9, + "devynias": 9 + }, + "DIRECT_NUMBERS": { + "dešimt": 10, + "vienuolika": 11, + "dvylika": 12, + "trylika": 13 + }, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "šimtai": 100, + "milijonų": 1000000, + "milijardų": 1000000000, + "trilijonų": 1000000000000, + "kvadrilijonų": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "dešimt": 10, - "vienuolika": 11, - "dvylika": 12, - "trylika": 13 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "šimtai": 100, - "milijonų": 1000000, - "milijardų": 1000000000, - "trilijonų": 1000000000000, - "kvadrilijonų": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/lv.py b/number_parser/data/lv.py index bfb25d2..79acf9b 100644 --- a/number_parser/data/lv.py +++ b/number_parser/data/lv.py @@ -1,34 +1,44 @@ info = { - "UNIT_NUMBERS": { - "nulle": 0, - "viena": 1, - "viens": 1, - "divas": 2, - "divi": 2, - "trīs": 3, - "četras": 4, - "četri": 4, - "piecas": 5, - "pieci": 5, - "sešas": 6, - "seši": 6, - "septiņas": 7, - "septiņi": 7, - "astoņas": 8, - "astoņi": 8, - "deviņas": 9, - "deviņi": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "nulle": 0, + "viena": 1, + "viens": 1, + "divas": 2, + "divi": 2, + "trīs": 3, + "četras": 4, + "četri": 4, + "piecas": 5, + "pieci": 5, + "sešas": 6, + "seši": 6, + "septiņas": 7, + "septiņi": 7, + "astoņas": 8, + "astoņi": 8, + "deviņas": 9, + "deviņi": 9 + }, + "DIRECT_NUMBERS": { + "desmit": 10 + }, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "miljoni": 1000000, + "miljardi": 1000000000, + "biljoni": 1000000000000, + "biljardi": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "desmit": 10 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "miljoni": 1000000, - "miljardi": 1000000000, - "biljoni": 1000000000000, - "biljardi": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/mk.py b/number_parser/data/mk.py index d67f84b..204f534 100644 --- a/number_parser/data/mk.py +++ b/number_parser/data/mk.py @@ -1,48 +1,58 @@ info = { - "UNIT_NUMBERS": { - "нула": 0, - "еден": 1, - "една": 1, - "едно": 1, - "два": 2, - "две": 2, - "три": 3, - "четири": 4, - "пет": 5, - "шест": 6, - "седум": 7, - "осум": 8, - "девет": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "нула": 0, + "еден": 1, + "една": 1, + "едно": 1, + "два": 2, + "две": 2, + "три": 3, + "четири": 4, + "пет": 5, + "шест": 6, + "седум": 7, + "осум": 8, + "девет": 9 + }, + "DIRECT_NUMBERS": { + "десет": 10, + "единаесет": 11, + "дванаесет": 12, + "тринаесет": 13, + "четиринаесет": 14, + "петнаесет": 15, + "шеснаесет": 16, + "седумнаесет": 17, + "осумнаесет": 18, + "деветнаесет": 19 + }, + "TENS": { + "дваесет": 20, + "триесет": 30, + "четириесет": 40, + "педесет": 50, + "шеесет": 60, + "седумдесет": 70, + "осумдесет": 80, + "деведесет": 90 + }, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "илјада": 1000, + "милион": 1000000, + "милијарда": 1000000000, + "билион": 1000000000000, + "билијарда": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "десет": 10, - "единаесет": 11, - "дванаесет": 12, - "тринаесет": 13, - "четиринаесет": 14, - "петнаесет": 15, - "шеснаесет": 16, - "седумнаесет": 17, - "осумнаесет": 18, - "деветнаесет": 19 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": { - "дваесет": 20, - "триесет": 30, - "четириесет": 40, - "педесет": 50, - "шеесет": 60, - "седумдесет": 70, - "осумдесет": 80, - "деведесет": 90 - }, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "илјада": 1000, - "милион": 1000000, - "милијарда": 1000000000, - "билион": 1000000000000, - "билијарда": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/ms.py b/number_parser/data/ms.py index 929f225..39e5565 100644 --- a/number_parser/data/ms.py +++ b/number_parser/data/ms.py @@ -1,29 +1,39 @@ info = { - "UNIT_NUMBERS": { - "kosong": 0, - "satu": 1, - "dua": 2, - "tiga": 3, - "empat": 4, - "lima": 5, - "enam": 6, - "tujuh": 7, - "lapan": 8, - "sembilan": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "kosong": 0, + "satu": 1, + "dua": 2, + "tiga": 3, + "empat": 4, + "lima": 5, + "enam": 6, + "tujuh": 7, + "lapan": 8, + "sembilan": 9 + }, + "DIRECT_NUMBERS": { + "sepuluh": 10, + "sebelas": 11 + }, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "ratus": 100, + "ribu": 1000, + "juta": 1000000, + "bilion": 1000000000, + "trilion": 1000000000000, + "kuadrilion": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "sepuluh": 10, - "sebelas": 11 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "ratus": 100, - "ribu": 1000, - "juta": 1000000, - "bilion": 1000000000, - "trilion": 1000000000000, - "kuadrilion": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/mt.py b/number_parser/data/mt.py index fab6c63..acacff0 100644 --- a/number_parser/data/mt.py +++ b/number_parser/data/mt.py @@ -1,58 +1,68 @@ info = { - "UNIT_NUMBERS": { - "żero": 0, - "waħda": 1, - "wieħed": 1, - "tnejn": 2, - "żewġ": 2, - "tliet": 3, - "tlieta": 3, - "erbaʼ": 4, - "erbgħa": 4, - "ħames": 5, - "ħamsa": 5, - "sitt": 6, - "sitta": 6, - "sebaʼ": 7, - "sebgħa": 7, - "tmien": 8, - "tmienja": 8, - "disaʼ": 9, - "disgħa": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "żero": 0, + "waħda": 1, + "wieħed": 1, + "tnejn": 2, + "żewġ": 2, + "tliet": 3, + "tlieta": 3, + "erbaʼ": 4, + "erbgħa": 4, + "ħames": 5, + "ħamsa": 5, + "sitt": 6, + "sitta": 6, + "sebaʼ": 7, + "sebgħa": 7, + "tmien": 8, + "tmienja": 8, + "disaʼ": 9, + "disgħa": 9 + }, + "DIRECT_NUMBERS": { + "għaxar": 10, + "għaxra": 10, + "ħdax": 11, + "ħdax-il": 11, + "tnax": 12, + "tnax-il": 12, + "tlettax": 13, + "tlettax-il": 13, + "erbatax": 14, + "erbatax-il": 14, + "ħmistax": 15, + "ħmistax-il": 15, + "sittax": 16, + "sittax-il": 16, + "sbatax": 17, + "sbatax-il": 17, + "tmintax": 18, + "tmintax-il": 18, + "dsatax": 19, + "dsatax-il": 19 + }, + "TENS": {}, + "HUNDREDS": { + "mitejn": 200 + }, + "BIG_POWERS_OF_TEN": { + "mija": 100, + "elef": 1000, + "miljuni": 1000000, + "biljuni": 1000000000, + "triljuni": 1000000000000, + "kvadriljuni": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "għaxar": 10, - "għaxra": 10, - "ħdax": 11, - "ħdax-il": 11, - "tnax": 12, - "tnax-il": 12, - "tlettax": 13, - "tlettax-il": 13, - "erbatax": 14, - "erbatax-il": 14, - "ħmistax": 15, - "ħmistax-il": 15, - "sittax": 16, - "sittax-il": 16, - "sbatax": 17, - "sbatax-il": 17, - "tmintax": 18, - "tmintax-il": 18, - "dsatax": 19, - "dsatax-il": 19 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": {}, - "HUNDREDS": { - "mitejn": 200 - }, - "BIG_POWERS_OF_TEN": { - "mija": 100, - "elef": 1000, - "miljuni": 1000000, - "biljuni": 1000000000, - "triljuni": 1000000000000, - "kvadriljuni": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/my.py b/number_parser/data/my.py index 3482b3e..86c978c 100644 --- a/number_parser/data/my.py +++ b/number_parser/data/my.py @@ -1,21 +1,31 @@ info = { - "UNIT_NUMBERS": { - "သုည": 0, - "တစ်": 1, - "နှစ်": 2, - "သုံး": 3, - "လေး": 4, - "ငါး": 5, - "ခြောက်": 6, - "ခုနှစ်": 7, - "ရှစ်": 8, - "ကိုး": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "သုည": 0, + "တစ်": 1, + "နှစ်": 2, + "သုံး": 3, + "လေး": 4, + "ငါး": 5, + "ခြောက်": 6, + "ခုနှစ်": 7, + "ရှစ်": 8, + "ကိုး": 9 + }, + "DIRECT_NUMBERS": { + "ဆယ်": 10 + }, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "DIRECT_NUMBERS": { - "ဆယ်": 10 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/nb.py b/number_parser/data/nb.py index 143950e..011d194 100644 --- a/number_parser/data/nb.py +++ b/number_parser/data/nb.py @@ -1,48 +1,58 @@ info = { - "UNIT_NUMBERS": { - "null": 0, - "ei": 1, - "ett": 1, - "én": 1, - "to": 2, - "tre": 3, - "fire": 4, - "fem": 5, - "seks": 6, - "sju": 7, - "åtte": 8, - "ni": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "null": 0, + "ei": 1, + "ett": 1, + "én": 1, + "to": 2, + "tre": 3, + "fire": 4, + "fem": 5, + "seks": 6, + "sju": 7, + "åtte": 8, + "ni": 9 + }, + "DIRECT_NUMBERS": { + "ti": 10, + "elleve": 11, + "tolv": 12, + "tretten": 13, + "fjorten": 14, + "femten": 15, + "seksten": 16, + "sytten": 17, + "atten": 18, + "nitten": 19 + }, + "TENS": { + "tjue": 20, + "tretti": 30, + "førti": 40, + "femti": 50, + "seksti": 60, + "sytti": 70, + "åtti": 80, + "nitti": 90 + }, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "hundre": 100, + "tusen": 1000, + "millioner": 1000000, + "milliarder": 1000000000, + "billioner": 1000000000000, + "billiarder": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "ti": 10, - "elleve": 11, - "tolv": 12, - "tretten": 13, - "fjorten": 14, - "femten": 15, - "seksten": 16, - "sytten": 17, - "atten": 18, - "nitten": 19 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": { - "tjue": 20, - "tretti": 30, - "førti": 40, - "femti": 50, - "seksti": 60, - "sytti": 70, - "åtti": 80, - "nitti": 90 - }, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "hundre": 100, - "tusen": 1000, - "millioner": 1000000, - "milliarder": 1000000000, - "billioner": 1000000000000, - "billiarder": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/nl.py b/number_parser/data/nl.py index 0616144..3f8bcb6 100644 --- a/number_parser/data/nl.py +++ b/number_parser/data/nl.py @@ -1,35 +1,45 @@ info = { - "UNIT_NUMBERS": { - "nul": 0, - "een": 1, - "twee": 2, - "drie": 3, - "vier": 4, - "vijf": 5, - "zes": 6, - "zeven": 7, - "acht": 8, - "negen": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "nul": 0, + "een": 1, + "twee": 2, + "drie": 3, + "vier": 4, + "vijf": 5, + "zes": 6, + "zeven": 7, + "acht": 8, + "negen": 9 + }, + "DIRECT_NUMBERS": { + "tien": 10, + "elf": 11, + "twaalf": 12, + "dertien": 13, + "veertien": 14, + "vijftien": 15, + "zestien": 16, + "zeventien": 17, + "achttien": 18, + "negentien": 19 + }, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "miljoen": 1000000, + "miljard": 1000000000, + "biljoen": 1000000000000, + "biljard": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "tien": 10, - "elf": 11, - "twaalf": 12, - "dertien": 13, - "veertien": 14, - "vijftien": 15, - "zestien": 16, - "zeventien": 17, - "achttien": 18, - "negentien": 19 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "miljoen": 1000000, - "miljard": 1000000000, - "biljoen": 1000000000000, - "biljard": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/nn.py b/number_parser/data/nn.py index 2620e05..f4b6531 100644 --- a/number_parser/data/nn.py +++ b/number_parser/data/nn.py @@ -1,47 +1,57 @@ info = { - "UNIT_NUMBERS": { - "null": 0, - "eitt": 1, - "éin": 1, - "to": 2, - "tre": 3, - "fire": 4, - "fem": 5, - "seks": 6, - "sju": 7, - "åtte": 8, - "ni": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "null": 0, + "eitt": 1, + "éin": 1, + "to": 2, + "tre": 3, + "fire": 4, + "fem": 5, + "seks": 6, + "sju": 7, + "åtte": 8, + "ni": 9 + }, + "DIRECT_NUMBERS": { + "ti": 10, + "elleve": 11, + "tolv": 12, + "tretten": 13, + "fjorten": 14, + "femten": 15, + "seksten": 16, + "sytten": 17, + "atten": 18, + "nitten": 19 + }, + "TENS": { + "tjue": 20, + "tretti": 30, + "førti": 40, + "femti": 50, + "seksti": 60, + "sytti": 70, + "åtti": 80, + "nitti": 90 + }, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "hundre": 100, + "tusen": 1000, + "millionar": 1000000, + "milliardar": 1000000000, + "billionar": 1000000000000, + "biliardar": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "ti": 10, - "elleve": 11, - "tolv": 12, - "tretten": 13, - "fjorten": 14, - "femten": 15, - "seksten": 16, - "sytten": 17, - "atten": 18, - "nitten": 19 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": { - "tjue": 20, - "tretti": 30, - "førti": 40, - "femti": 50, - "seksti": 60, - "sytti": 70, - "åtti": 80, - "nitti": 90 - }, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "hundre": 100, - "tusen": 1000, - "millionar": 1000000, - "milliardar": 1000000000, - "billionar": 1000000000000, - "biliardar": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/pl.py b/number_parser/data/pl.py index 4c3ea54..8b94245 100644 --- a/number_parser/data/pl.py +++ b/number_parser/data/pl.py @@ -1,141 +1,151 @@ info = { - "UNIT_NUMBERS": { - "zera": 0, - "zerem": 0, - "zero": 0, - "zeru": 0, - "zerze": 0, - "dziesięciu": 1, - "dziesięć": 1, - "jeden": 1, - "jedna": 1, - "jednego": 1, - "jednej": 1, - "jednemu": 1, - "jedno": 1, - "jednym": 1, - "jedną": 1, - "dwa": 2, - "dwadzieścia": 2, - "dwaj": 2, - "dwie": 2, - "dwiema": 2, - "dwoma": 2, - "dwudziestu": 2, - "dwóch": 2, - "dwóm": 2, - "trzech": 3, - "trzej": 3, - "trzem": 3, - "trzema": 3, - "trzy": 3, - "trzydziestu": 3, - "trzydzieści": 3, - "czterdziestu": 4, - "czterdzieści": 4, - "czterech": 4, - "czterej": 4, - "czterem": 4, - "czterema": 4, - "cztery": 4, - "pięcioma": 5, - "pięciu": 5, - "pięć": 5, - "pięćdziesiąt": 5, - "pięćdziesięciu": 5, - "sześcioma": 6, - "sześciu": 6, - "sześć": 6, - "sześćdziesiąt": 6, - "sześćdziesięciu": 6, - "siedem": 7, - "siedemdziesiąt": 7, - "siedemdziesięciu": 7, - "siedmioma": 7, - "siedmiu": 7, - "osiem": 8, - "osiemdziesiąt": 8, - "osiemdziesięciu": 8, - "ośmioma": 8, - "ośmiu": 8, - "dziewięcioma": 9, - "dziewięciu": 9, - "dziewięć": 9, - "dziewięćdziesiąt": 9, - "dziewięćdziesięciu": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "zera": 0, + "zerem": 0, + "zero": 0, + "zeru": 0, + "zerze": 0, + "dziesięciu": 1, + "dziesięć": 1, + "jeden": 1, + "jedna": 1, + "jednego": 1, + "jednej": 1, + "jednemu": 1, + "jedno": 1, + "jednym": 1, + "jedną": 1, + "dwa": 2, + "dwadzieścia": 2, + "dwaj": 2, + "dwie": 2, + "dwiema": 2, + "dwoma": 2, + "dwudziestu": 2, + "dwóch": 2, + "dwóm": 2, + "trzech": 3, + "trzej": 3, + "trzem": 3, + "trzema": 3, + "trzy": 3, + "trzydziestu": 3, + "trzydzieści": 3, + "czterdziestu": 4, + "czterdzieści": 4, + "czterech": 4, + "czterej": 4, + "czterem": 4, + "czterema": 4, + "cztery": 4, + "pięcioma": 5, + "pięciu": 5, + "pięć": 5, + "pięćdziesiąt": 5, + "pięćdziesięciu": 5, + "sześcioma": 6, + "sześciu": 6, + "sześć": 6, + "sześćdziesiąt": 6, + "sześćdziesięciu": 6, + "siedem": 7, + "siedemdziesiąt": 7, + "siedemdziesięciu": 7, + "siedmioma": 7, + "siedmiu": 7, + "osiem": 8, + "osiemdziesiąt": 8, + "osiemdziesięciu": 8, + "ośmioma": 8, + "ośmiu": 8, + "dziewięcioma": 9, + "dziewięciu": 9, + "dziewięć": 9, + "dziewięćdziesiąt": 9, + "dziewięćdziesięciu": 9 + }, + "DIRECT_NUMBERS": { + "dziesięcioma": 10, + "dziesięciu": 10, + "dziesięć": 10, + "jedenastoma": 11, + "jedenastu": 11, + "jedenaście": 11, + "dwanaście": 12, + "dwunastoma": 12, + "dwunastu": 12, + "trzynastoma": 13, + "trzynastu": 13, + "trzynaście": 13, + "czternastoma": 14, + "czternastu": 14, + "czternaście": 14, + "piętnastoma": 15, + "piętnastu": 15, + "piętnaście": 15, + "szesnastoma": 16, + "szesnastu": 16, + "szesnaście": 16, + "siedemnastoma": 17, + "siedemnastu": 17, + "siedemnaście": 17, + "osiemnastoma": 18, + "osiemnastu": 18, + "osiemnaście": 18, + "dziewiętnastoma": 19, + "dziewiętnastu": 19, + "dziewiętnaście": 19 + }, + "TENS": { + "dwudziestoma": 20, + "trzydziestoma": 30, + "czterdziestoma": 40, + "pięćdziesięcioma": 50, + "sześćdziesięcioma": 60, + "siedemdziesięcioma": 70, + "osiemdziesięcioma": 80, + "dziewięćdziesięcioma": 90 + }, + "HUNDREDS": { + "dwieście": 200, + "dwustu": 200 + }, + "BIG_POWERS_OF_TEN": { + "tysiącach": 1000, + "tysiącami": 1000, + "tysiące": 1000, + "tysiącom": 1000, + "tysięcy": 1000, + "milionach": 1000000, + "milionami": 1000000, + "milionom": 1000000, + "miliony": 1000000, + "milionów": 1000000, + "miliardach": 1000000000, + "miliardami": 1000000000, + "miliardom": 1000000000, + "miliardy": 1000000000, + "miliardów": 1000000000, + "bilionach": 1000000000000, + "bilionami": 1000000000000, + "bilionom": 1000000000000, + "biliony": 1000000000000, + "bilionów": 1000000000000, + "biliardach": 1000000000000000, + "biliardami": 1000000000000000, + "biliardom": 1000000000000000, + "biliardy": 1000000000000000, + "biliardów": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "dziesięcioma": 10, - "dziesięciu": 10, - "dziesięć": 10, - "jedenastoma": 11, - "jedenastu": 11, - "jedenaście": 11, - "dwanaście": 12, - "dwunastoma": 12, - "dwunastu": 12, - "trzynastoma": 13, - "trzynastu": 13, - "trzynaście": 13, - "czternastoma": 14, - "czternastu": 14, - "czternaście": 14, - "piętnastoma": 15, - "piętnastu": 15, - "piętnaście": 15, - "szesnastoma": 16, - "szesnastu": 16, - "szesnaście": 16, - "siedemnastoma": 17, - "siedemnastu": 17, - "siedemnaście": 17, - "osiemnastoma": 18, - "osiemnastu": 18, - "osiemnaście": 18, - "dziewiętnastoma": 19, - "dziewiętnastu": 19, - "dziewiętnaście": 19 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": { - "dwudziestoma": 20, - "trzydziestoma": 30, - "czterdziestoma": 40, - "pięćdziesięcioma": 50, - "sześćdziesięcioma": 60, - "siedemdziesięcioma": 70, - "osiemdziesięcioma": 80, - "dziewięćdziesięcioma": 90 - }, - "HUNDREDS": { - "dwieście": 200, - "dwustu": 200 - }, - "BIG_POWERS_OF_TEN": { - "tysiącach": 1000, - "tysiącami": 1000, - "tysiące": 1000, - "tysiącom": 1000, - "tysięcy": 1000, - "milionach": 1000000, - "milionami": 1000000, - "milionom": 1000000, - "miliony": 1000000, - "milionów": 1000000, - "miliardach": 1000000000, - "miliardami": 1000000000, - "miliardom": 1000000000, - "miliardy": 1000000000, - "miliardów": 1000000000, - "bilionach": 1000000000000, - "bilionami": 1000000000000, - "bilionom": 1000000000000, - "biliony": 1000000000000, - "bilionów": 1000000000000, - "biliardach": 1000000000000000, - "biliardami": 1000000000000000, - "biliardom": 1000000000000000, - "biliardy": 1000000000000000, - "biliardów": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/pt-PT.py b/number_parser/data/pt-PT.py index 800bc6e..5a93dcf 100644 --- a/number_parser/data/pt-PT.py +++ b/number_parser/data/pt-PT.py @@ -1,64 +1,74 @@ info = { - "UNIT_NUMBERS": { - "zero": 0, - "um": 1, - "uma": 1, - "dois": 2, - "duas": 2, - "três": 3, - "quatro": 4, - "cinco": 5, - "seis": 6, - "sete": 7, - "oito": 8, - "nove": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "zero": 0, + "um": 1, + "uma": 1, + "dois": 2, + "duas": 2, + "três": 3, + "quatro": 4, + "cinco": 5, + "seis": 6, + "sete": 7, + "oito": 8, + "nove": 9 + }, + "DIRECT_NUMBERS": { + "dez": 10, + "onze": 11, + "doze": 12, + "treze": 13, + "catorze": 14, + "quinze": 15, + "dezasseis": 16, + "dezassete": 17, + "dezoito": 18, + "dezanove": 19 + }, + "TENS": { + "vinte": 20, + "trinta": 30, + "quarenta": 40, + "cinquenta": 50, + "sessenta": 60, + "setenta": 70, + "oitenta": 80, + "noventa": 90 + }, + "HUNDREDS": { + "duzentas": 200, + "duzentos": 200, + "trezentas": 300, + "trezentos": 300, + "quatrocentas": 400, + "quatrocentos": 400, + "quinhentas": 500, + "quinhentos": 500, + "seiscentas": 600, + "seiscentos": 600, + "setecentas": 700, + "setecentos": 700, + "oitocentas": 800, + "oitocentos": 800, + "novecentas": 900, + "novecentos": 900 + }, + "BIG_POWERS_OF_TEN": { + "mil": 1000, + "milhões": 1000000, + "biliões": 1000000000, + "triliões": 1000000000000, + "quatriliões": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "dez": 10, - "onze": 11, - "doze": 12, - "treze": 13, - "catorze": 14, - "quinze": 15, - "dezasseis": 16, - "dezassete": 17, - "dezoito": 18, - "dezanove": 19 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": { - "vinte": 20, - "trinta": 30, - "quarenta": 40, - "cinquenta": 50, - "sessenta": 60, - "setenta": 70, - "oitenta": 80, - "noventa": 90 - }, - "HUNDREDS": { - "duzentas": 200, - "duzentos": 200, - "trezentas": 300, - "trezentos": 300, - "quatrocentas": 400, - "quatrocentos": 400, - "quinhentas": 500, - "quinhentos": 500, - "seiscentas": 600, - "seiscentos": 600, - "setecentas": 700, - "setecentos": 700, - "oitocentas": 800, - "oitocentos": 800, - "novecentas": 900, - "novecentos": 900 - }, - "BIG_POWERS_OF_TEN": { - "mil": 1000, - "milhões": 1000000, - "biliões": 1000000000, - "triliões": 1000000000000, - "quatriliões": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/pt.py b/number_parser/data/pt.py index 3bf9f52..dd34221 100644 --- a/number_parser/data/pt.py +++ b/number_parser/data/pt.py @@ -1,64 +1,74 @@ info = { - "UNIT_NUMBERS": { - "zero": 0, - "um": 1, - "uma": 1, - "dois": 2, - "duas": 2, - "três": 3, - "quatro": 4, - "cinco": 5, - "seis": 6, - "sete": 7, - "oito": 8, - "nove": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "zero": 0, + "um": 1, + "uma": 1, + "dois": 2, + "duas": 2, + "três": 3, + "quatro": 4, + "cinco": 5, + "seis": 6, + "sete": 7, + "oito": 8, + "nove": 9 + }, + "DIRECT_NUMBERS": { + "dez": 10, + "onze": 11, + "doze": 12, + "treze": 13, + "catorze": 14, + "quinze": 15, + "dezesseis": 16, + "dezessete": 17, + "dezoito": 18, + "dezenove": 19 + }, + "TENS": { + "vinte": 20, + "trinta": 30, + "quarenta": 40, + "cinquenta": 50, + "sessenta": 60, + "setenta": 70, + "oitenta": 80, + "noventa": 90 + }, + "HUNDREDS": { + "duzentas": 200, + "duzentos": 200, + "trezentas": 300, + "trezentos": 300, + "quatrocentas": 400, + "quatrocentos": 400, + "quinhentas": 500, + "quinhentos": 500, + "seiscentas": 600, + "seiscentos": 600, + "setecentas": 700, + "setecentos": 700, + "oitocentas": 800, + "oitocentos": 800, + "novecentas": 900, + "novecentos": 900 + }, + "BIG_POWERS_OF_TEN": { + "mil": 1000, + "milhões": 1000000, + "bilhões": 1000000000, + "trilhões": 1000000000000, + "quatrilhões": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "dez": 10, - "onze": 11, - "doze": 12, - "treze": 13, - "catorze": 14, - "quinze": 15, - "dezesseis": 16, - "dezessete": 17, - "dezoito": 18, - "dezenove": 19 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": { - "vinte": 20, - "trinta": 30, - "quarenta": 40, - "cinquenta": 50, - "sessenta": 60, - "setenta": 70, - "oitenta": 80, - "noventa": 90 - }, - "HUNDREDS": { - "duzentas": 200, - "duzentos": 200, - "trezentas": 300, - "trezentos": 300, - "quatrocentas": 400, - "quatrocentos": 400, - "quinhentas": 500, - "quinhentos": 500, - "seiscentas": 600, - "seiscentos": 600, - "setecentas": 700, - "setecentos": 700, - "oitocentas": 800, - "oitocentos": 800, - "novecentas": 900, - "novecentos": 900 - }, - "BIG_POWERS_OF_TEN": { - "mil": 1000, - "milhões": 1000000, - "bilhões": 1000000000, - "trilhões": 1000000000000, - "quatrilhões": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/qu.py b/number_parser/data/qu.py index b93b645..e954fa8 100644 --- a/number_parser/data/qu.py +++ b/number_parser/data/qu.py @@ -1,26 +1,36 @@ info = { - "UNIT_NUMBERS": { - "chusaq": 0, - "huk": 1, - "iskay": 2, - "kinsa": 3, - "tawa": 4, - "phisqa": 5, - "suqta": 6, - "qanchis": 7, - "pusaq": 8, - "isqun": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "chusaq": 0, + "huk": 1, + "iskay": 2, + "kinsa": 3, + "tawa": 4, + "phisqa": 5, + "suqta": 6, + "qanchis": 7, + "pusaq": 8, + "isqun": 9 + }, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "pachak": 100, + "waranqa": 1000, + "hunu": 1000000, + "lluna": 1000000000, + "trilionu": 1000000000000, + "kvadrilionu": 1000000000000000 + } }, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "pachak": 100, - "waranqa": 1000, - "hunu": 1000000, - "lluna": 1000000000, - "trilionu": 1000000000000, - "kvadrilionu": 1000000000000000 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/ro.py b/number_parser/data/ro.py index 1495b6b..4a9ad09 100644 --- a/number_parser/data/ro.py +++ b/number_parser/data/ro.py @@ -1,35 +1,45 @@ info = { - "UNIT_NUMBERS": { - "zero": 0, - "una": 1, - "unu": 1, - "doi": 2, - "două": 2, - "trei": 3, - "patru": 4, - "cinci": 5, - "şase": 6, - "şapte": 7, - "opt": 8, - "nouă": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "zero": 0, + "una": 1, + "unu": 1, + "doi": 2, + "două": 2, + "trei": 3, + "patru": 4, + "cinci": 5, + "şase": 6, + "şapte": 7, + "opt": 8, + "nouă": 9 + }, + "DIRECT_NUMBERS": { + "zece": 10, + "unsprezece": 11 + }, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "sute": 100, + "mii": 1000, + "milioane": 1000000, + "milion": 1000000, + "miliard": 1000000000, + "miliarde": 1000000000, + "bilioane": 1000000000000, + "bilion": 1000000000000, + "biliard": 1000000000000000, + "biliarde": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "zece": 10, - "unsprezece": 11 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "sute": 100, - "mii": 1000, - "milioane": 1000000, - "milion": 1000000, - "miliard": 1000000000, - "miliarde": 1000000000, - "bilioane": 1000000000000, - "bilion": 1000000000000, - "biliard": 1000000000000000, - "biliarde": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/root.py b/number_parser/data/root.py index 0652d3f..6f6d231 100644 --- a/number_parser/data/root.py +++ b/number_parser/data/root.py @@ -1,8 +1,18 @@ info = { - "UNIT_NUMBERS": {}, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/ru.py b/number_parser/data/ru.py index 770ad4c..d36df25 100644 --- a/number_parser/data/ru.py +++ b/number_parser/data/ru.py @@ -1,216 +1,226 @@ info = { - "UNIT_NUMBERS": { - "ноль": 0, - "нолю": 0, - "ноля": 0, - "нуле": 0, - "нулем": 0, - "один": 1, - "одна": 1, - "одни": 1, - "одним": 1, - "одними": 1, - "одних": 1, - "одно": 1, - "одного": 1, - "одной": 1, - "одном": 1, - "одному": 1, - "одну": 1, - "два": 2, - "две": 2, - "двум": 2, - "двумя": 2, - "двух": 2, - "трем": 3, - "тремя": 3, - "трех": 3, - "три": 3, - "четыре": 4, - "четырем": 4, - "четырех": 4, - "четырьмя": 4, - "пяти": 5, - "пять": 5, - "пятью": 5, - "шести": 6, - "шесть": 6, - "шестью": 6, - "семи": 7, - "семь": 7, - "семью": 7, - "восемь": 8, - "восемью": 8, - "восьми": 8, - "девяти": 9, - "девять": 9, - "девятью": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "ноль": 0, + "нолю": 0, + "ноля": 0, + "нуле": 0, + "нулем": 0, + "один": 1, + "одна": 1, + "одни": 1, + "одним": 1, + "одними": 1, + "одних": 1, + "одно": 1, + "одного": 1, + "одной": 1, + "одном": 1, + "одному": 1, + "одну": 1, + "два": 2, + "две": 2, + "двум": 2, + "двумя": 2, + "двух": 2, + "трем": 3, + "тремя": 3, + "трех": 3, + "три": 3, + "четыре": 4, + "четырем": 4, + "четырех": 4, + "четырьмя": 4, + "пяти": 5, + "пять": 5, + "пятью": 5, + "шести": 6, + "шесть": 6, + "шестью": 6, + "семи": 7, + "семь": 7, + "семью": 7, + "восемь": 8, + "восемью": 8, + "восьми": 8, + "девяти": 9, + "девять": 9, + "девятью": 9 + }, + "DIRECT_NUMBERS": { + "десяти": 10, + "десять": 10, + "десятью": 10, + "одиннадцати": 11, + "одиннадцать": 11, + "одиннадцатью": 11, + "двенадцати": 12, + "двенадцать": 12, + "двенадцатью": 12, + "тринадцати": 13, + "тринадцать": 13, + "тринадцатью": 13, + "четырнадцати": 14, + "четырнадцать": 14, + "четырнадцатью": 14, + "пятнадцати": 15, + "пятнадцать": 15, + "пятнадцатью": 15, + "шестнадцати": 16, + "шестнадцать": 16, + "шестнадцатью": 16, + "семнадцати": 17, + "семнадцать": 17, + "семнадцатью": 17, + "восемнадцати": 18, + "восемнадцать": 18, + "восемнадцатью": 18, + "девятнадцати": 19, + "девятнадцать": 19, + "девятнадцатью": 19 + }, + "TENS": { + "двадцати": 20, + "двадцать": 20, + "двадцатью": 20, + "тридцати": 30, + "тридцать": 30, + "тридцатью": 30, + "сорок": 40, + "сорока": 40, + "пятидесяти": 50, + "пятьдесят": 50, + "пятьюдесятью": 50, + "шестидесяти": 60, + "шестьдесят": 60, + "шестьюдесятью": 60, + "семидесяти": 70, + "семьдесят": 70, + "семьюдесятью": 70, + "восемьдесят": 80, + "восемьюдесятью": 80, + "восьмидесяти": 80, + "девяноста": 90, + "девяносто": 90 + }, + "HUNDREDS": { + "ста": 100, + "сто": 100, + "двасти": 200, + "двести": 200, + "двумстам": 200, + "двумястами": 200, + "двухсот": 200, + "двухстах": 200, + "тремстам": 300, + "тремястами": 300, + "трехсот": 300, + "трехстах": 300, + "триста": 300, + "четыремстам": 400, + "четыреста": 400, + "четырехсот": 400, + "четырехстах": 400, + "четырьмястами": 400, + "пятисот": 500, + "пятистам": 500, + "пятистах": 500, + "пятьсот": 500, + "пятьюстами": 500, + "шестисот": 600, + "шестистам": 600, + "шестистах": 600, + "шестьсот": 600, + "шестьюстами": 600, + "семисот": 700, + "семистам": 700, + "семистах": 700, + "семьсот": 700, + "семьюстами": 700, + "восемьсот": 800, + "восемьюстами": 800, + "восьмисот": 800, + "восьмистам": 800, + "восьмистах": 800, + "девятисот": 900, + "девятистам": 900, + "девятистах": 900, + "девятьсот": 900, + "девятьюстами": 900 + }, + "BIG_POWERS_OF_TEN": { + "тысяч": 1000, + "тысяча": 1000, + "тысячам": 1000, + "тысячами": 1000, + "тысячах": 1000, + "тысяче": 1000, + "тысячей": 1000, + "тысячи": 1000, + "тысячу": 1000, + "миллион": 1000000, + "миллиона": 1000000, + "миллионам": 1000000, + "миллионами": 1000000, + "миллионах": 1000000, + "миллионе": 1000000, + "миллионов": 1000000, + "миллионом": 1000000, + "миллиону": 1000000, + "миллиард": 1000000000, + "миллиарда": 1000000000, + "миллиардам": 1000000000, + "миллиардами": 1000000000, + "миллиардах": 1000000000, + "миллиарде": 1000000000, + "миллиардов": 1000000000, + "миллиардом": 1000000000, + "миллиарду": 1000000000, + "триллион": 1000000000000, + "триллиона": 1000000000000, + "триллионам": 1000000000000, + "триллионами": 1000000000000, + "триллионах": 1000000000000, + "триллионе": 1000000000000, + "триллионов": 1000000000000, + "триллионом": 1000000000000, + "триллиону": 1000000000000, + "квадриллион": 1000000000000000, + "квадриллиона": 1000000000000000, + "квадриллионам": 1000000000000000, + "квадриллионами": 1000000000000000, + "квадриллионах": 1000000000000000, + "квадриллионе": 1000000000000000, + "квадриллионов": 1000000000000000, + "квадриллионом": 1000000000000000, + "квадриллиону": 1000000000000000, + "квинтиллион": 1000000000000000000, + "квинтиллиона": 1000000000000000000, + "квинтиллионам": 1000000000000000000, + "квинтиллионами": 1000000000000000000, + "квинтиллионах": 1000000000000000000, + "квинтиллионе": 1000000000000000000, + "квинтиллионов": 1000000000000000000, + "квинтиллионом": 1000000000000000000, + "квинтиллиону": 1000000000000000000, + "секстиллион": 1000000000000000000000, + "секстиллиона": 1000000000000000000000, + "секстиллионам": 1000000000000000000000, + "секстиллионами": 1000000000000000000000, + "секстиллионах": 1000000000000000000000, + "секстиллионе": 1000000000000000000000, + "секстиллионов": 1000000000000000000000, + "секстиллионом": 1000000000000000000000, + "секстиллиону": 1000000000000000000000 + } }, - "DIRECT_NUMBERS": { - "десяти": 10, - "десять": 10, - "десятью": 10, - "одиннадцати": 11, - "одиннадцать": 11, - "одиннадцатью": 11, - "двенадцати": 12, - "двенадцать": 12, - "двенадцатью": 12, - "тринадцати": 13, - "тринадцать": 13, - "тринадцатью": 13, - "четырнадцати": 14, - "четырнадцать": 14, - "четырнадцатью": 14, - "пятнадцати": 15, - "пятнадцать": 15, - "пятнадцатью": 15, - "шестнадцати": 16, - "шестнадцать": 16, - "шестнадцатью": 16, - "семнадцати": 17, - "семнадцать": 17, - "семнадцатью": 17, - "восемнадцати": 18, - "восемнадцать": 18, - "восемнадцатью": 18, - "девятнадцати": 19, - "девятнадцать": 19, - "девятнадцатью": 19 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": { - "двадцати": 20, - "двадцать": 20, - "двадцатью": 20, - "тридцати": 30, - "тридцать": 30, - "тридцатью": 30, - "сорок": 40, - "сорока": 40, - "пятидесяти": 50, - "пятьдесят": 50, - "пятьюдесятью": 50, - "шестидесяти": 60, - "шестьдесят": 60, - "шестьюдесятью": 60, - "семидесяти": 70, - "семьдесят": 70, - "семьюдесятью": 70, - "восемьдесят": 80, - "восемьюдесятью": 80, - "восьмидесяти": 80, - "девяноста": 90, - "девяносто": 90 - }, - "HUNDREDS": { - "ста": 100, - "сто": 100, - "двасти": 200, - "двести": 200, - "двумстам": 200, - "двумястами": 200, - "двухсот": 200, - "двухстах": 200, - "тремстам": 300, - "тремястами": 300, - "трехсот": 300, - "трехстах": 300, - "триста": 300, - "четыремстам": 400, - "четыреста": 400, - "четырехсот": 400, - "четырехстах": 400, - "четырьмястами": 400, - "пятисот": 500, - "пятистам": 500, - "пятистах": 500, - "пятьсот": 500, - "пятьюстами": 500, - "шестисот": 600, - "шестистам": 600, - "шестистах": 600, - "шестьсот": 600, - "шестьюстами": 600, - "семисот": 700, - "семистам": 700, - "семистах": 700, - "семьсот": 700, - "семьюстами": 700, - "восемьсот": 800, - "восемьюстами": 800, - "восьмисот": 800, - "восьмистам": 800, - "восьмистах": 800, - "девятисот": 900, - "девятистам": 900, - "девятистах": 900, - "девятьсот": 900, - "девятьюстами": 900 - }, - "BIG_POWERS_OF_TEN": { - "тысяч": 1000, - "тысяча": 1000, - "тысячам": 1000, - "тысячами": 1000, - "тысячах": 1000, - "тысяче": 1000, - "тысячей": 1000, - "тысячи": 1000, - "тысячу": 1000, - "миллион": 1000000, - "миллиона": 1000000, - "миллионам": 1000000, - "миллионами": 1000000, - "миллионах": 1000000, - "миллионе": 1000000, - "миллионов": 1000000, - "миллионом": 1000000, - "миллиону": 1000000, - "миллиард": 1000000000, - "миллиарда": 1000000000, - "миллиардам": 1000000000, - "миллиардами": 1000000000, - "миллиардах": 1000000000, - "миллиарде": 1000000000, - "миллиардов": 1000000000, - "миллиардом": 1000000000, - "миллиарду": 1000000000, - "триллион": 1000000000000, - "триллиона": 1000000000000, - "триллионам": 1000000000000, - "триллионами": 1000000000000, - "триллионах": 1000000000000, - "триллионе": 1000000000000, - "триллионов": 1000000000000, - "триллионом": 1000000000000, - "триллиону": 1000000000000, - "квадриллион": 1000000000000000, - "квадриллиона": 1000000000000000, - "квадриллионам": 1000000000000000, - "квадриллионами": 1000000000000000, - "квадриллионах": 1000000000000000, - "квадриллионе": 1000000000000000, - "квадриллионов": 1000000000000000, - "квадриллионом": 1000000000000000, - "квадриллиону": 1000000000000000, - "квинтиллион": 1000000000000000000, - "квинтиллиона": 1000000000000000000, - "квинтиллионам": 1000000000000000000, - "квинтиллионами": 1000000000000000000, - "квинтиллионах": 1000000000000000000, - "квинтиллионе": 1000000000000000000, - "квинтиллионов": 1000000000000000000, - "квинтиллионом": 1000000000000000000, - "квинтиллиону": 1000000000000000000, - "секстиллион": 1000000000000000000000, - "секстиллиона": 1000000000000000000000, - "секстиллионам": 1000000000000000000000, - "секстиллионами": 1000000000000000000000, - "секстиллионах": 1000000000000000000000, - "секстиллионе": 1000000000000000000000, - "секстиллионов": 1000000000000000000000, - "секстиллионом": 1000000000000000000000, - "секстиллиону": 1000000000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/se.py b/number_parser/data/se.py index 7a66e6d..3526a55 100644 --- a/number_parser/data/se.py +++ b/number_parser/data/se.py @@ -1,26 +1,36 @@ info = { - "UNIT_NUMBERS": { - "nolla": 0, - "okta": 1, - "guokte": 2, - "golbma": 3, - "njeallje": 4, - "vihtta": 5, - "guhtta": 6, - "čieža": 7, - "gávcci": 8, - "ovcci": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "nolla": 0, + "okta": 1, + "guokte": 2, + "golbma": 3, + "njeallje": 4, + "vihtta": 5, + "guhtta": 6, + "čieža": 7, + "gávcci": 8, + "ovcci": 9 + }, + "DIRECT_NUMBERS": { + "logi": 10 + }, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "miljon": 1000000, + "miljard": 1000000000, + "biljon": 1000000000000, + "biljard": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "logi": 10 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "miljon": 1000000, - "miljard": 1000000000, - "biljon": 1000000000000, - "biljard": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/sk.py b/number_parser/data/sk.py index 44b8aa1..0a30c13 100644 --- a/number_parser/data/sk.py +++ b/number_parser/data/sk.py @@ -1,51 +1,61 @@ info = { - "UNIT_NUMBERS": { - "nula": 0, - "jeden": 1, - "jedna": 1, - "jedno": 1, - "dva": 2, - "dve": 2, - "tri": 3, - "štyri": 4, - "päť": 5, - "šesť": 6, - "sedem": 7, - "osem": 8, - "deväť": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "nula": 0, + "jeden": 1, + "jedna": 1, + "jedno": 1, + "dva": 2, + "dve": 2, + "tri": 3, + "štyri": 4, + "päť": 5, + "šesť": 6, + "sedem": 7, + "osem": 8, + "deväť": 9 + }, + "DIRECT_NUMBERS": { + "desať": 10, + "jedenásť": 11, + "dvanásť": 12, + "trinásť": 13, + "štrnásť": 14, + "pätnásť": 15, + "šestnásť": 16, + "sedemnásť": 17, + "osemnásť": 18, + "devätnásť": 19 + }, + "TENS": { + "dvadsať": 20, + "tridsať": 30, + "štyridsať": 40 + }, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "tisíc": 1000, + "milión": 1000000, + "miliónov": 1000000, + "milióny": 1000000, + "miliarda": 1000000000, + "miliardov": 1000000000, + "miliardy": 1000000000, + "bilión": 1000000000000, + "biliónov": 1000000000000, + "bilióny": 1000000000000, + "biliarda": 1000000000000000, + "biliardov": 1000000000000000, + "biliardy": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "desať": 10, - "jedenásť": 11, - "dvanásť": 12, - "trinásť": 13, - "štrnásť": 14, - "pätnásť": 15, - "šestnásť": 16, - "sedemnásť": 17, - "osemnásť": 18, - "devätnásť": 19 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": { - "dvadsať": 20, - "tridsať": 30, - "štyridsať": 40 - }, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "tisíc": 1000, - "milión": 1000000, - "miliónov": 1000000, - "milióny": 1000000, - "miliarda": 1000000000, - "miliardov": 1000000000, - "miliardy": 1000000000, - "bilión": 1000000000000, - "biliónov": 1000000000000, - "bilióny": 1000000000000, - "biliarda": 1000000000000000, - "biliardov": 1000000000000000, - "biliardy": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/sl.py b/number_parser/data/sl.py index 867fbe6..4c8bb55 100644 --- a/number_parser/data/sl.py +++ b/number_parser/data/sl.py @@ -1,54 +1,64 @@ info = { - "UNIT_NUMBERS": { - "nič": 0, - "ena": 1, - "dva": 2, - "dve": 2, - "dvije": 2, - "tri": 3, - "štiri": 4, - "pet": 5, - "šest": 6, - "sedem": 7, - "osem": 8, - "devet": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "nič": 0, + "ena": 1, + "dva": 2, + "dve": 2, + "dvije": 2, + "tri": 3, + "štiri": 4, + "pet": 5, + "šest": 6, + "sedem": 7, + "osem": 8, + "devet": 9 + }, + "DIRECT_NUMBERS": { + "deset": 10, + "enajst": 11, + "dvanajst": 12, + "trinajst": 13, + "štrinajst": 14, + "petnajst": 15, + "šestnajst": 16, + "sedemnajst": 17, + "asemnajst": 18, + "devetnajst": 19 + }, + "TENS": { + "dvaset": 20 + }, + "HUNDREDS": { + "dvjesto": 200, + "tristo": 300, + "štiristo": 400, + "petsto": 500, + "šesto": 600, + "sedemsto": 700, + "osemsto": 800, + "devetsto": 900 + }, + "BIG_POWERS_OF_TEN": { + "tisuće": 1000, + "tisuću": 1000, + "milijun": 1000000, + "milijuny": 1000000, + "miliarda": 1000000000, + "miliardy": 1000000000, + "bilijun": 1000000000000, + "bilijuny": 1000000000000, + "biliarda": 1000000000000000, + "biliardy": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "deset": 10, - "enajst": 11, - "dvanajst": 12, - "trinajst": 13, - "štrinajst": 14, - "petnajst": 15, - "šestnajst": 16, - "sedemnajst": 17, - "asemnajst": 18, - "devetnajst": 19 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": { - "dvaset": 20 - }, - "HUNDREDS": { - "dvjesto": 200, - "tristo": 300, - "štiristo": 400, - "petsto": 500, - "šesto": 600, - "sedemsto": 700, - "osemsto": 800, - "devetsto": 900 - }, - "BIG_POWERS_OF_TEN": { - "tisuće": 1000, - "tisuću": 1000, - "milijun": 1000000, - "milijuny": 1000000, - "miliarda": 1000000000, - "miliardy": 1000000000, - "bilijun": 1000000000000, - "bilijuny": 1000000000000, - "biliarda": 1000000000000000, - "biliardy": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/sq.py b/number_parser/data/sq.py index 86934a2..0b7a11f 100644 --- a/number_parser/data/sq.py +++ b/number_parser/data/sq.py @@ -1,32 +1,42 @@ info = { - "UNIT_NUMBERS": { - "zero": 0, - "një": 1, - "dy": 2, - "tre": 3, - "tri": 3, - "katër": 4, - "pesë": 5, - "gjashtë": 6, - "shtatë": 7, - "tetë": 8, - "nëntë": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "zero": 0, + "një": 1, + "dy": 2, + "tre": 3, + "tri": 3, + "katër": 4, + "pesë": 5, + "gjashtë": 6, + "shtatë": 7, + "tetë": 8, + "nëntë": 9 + }, + "DIRECT_NUMBERS": { + "dhjetë": 10 + }, + "TENS": { + "njëzet": 20, + "tridhjetë": 30, + "dyzet": 40 + }, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "mijë": 1000, + "milionë": 1000000, + "miliarë": 1000000000, + "bilionë": 1000000000000, + "biliarë": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "dhjetë": 10 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": { - "njëzet": 20, - "tridhjetë": 30, - "dyzet": 40 - }, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "mijë": 1000, - "milionë": 1000000, - "miliarë": 1000000000, - "bilionë": 1000000000000, - "biliarë": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/sr-Latn.py b/number_parser/data/sr-Latn.py index cb657d6..a3dc4e5 100644 --- a/number_parser/data/sr-Latn.py +++ b/number_parser/data/sr-Latn.py @@ -1,52 +1,62 @@ info = { - "UNIT_NUMBERS": { - "nula": 0, - "jedan": 1, - "jedna": 1, - "jedno": 1, - "dva": 2, - "dve": 2, - "tri": 3, - "četiri": 4, - "pet": 5, - "šest": 6, - "sedam": 7, - "osam": 8, - "devet": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "nula": 0, + "jedan": 1, + "jedna": 1, + "jedno": 1, + "dva": 2, + "dve": 2, + "tri": 3, + "četiri": 4, + "pet": 5, + "šest": 6, + "sedam": 7, + "osam": 8, + "devet": 9 + }, + "DIRECT_NUMBERS": { + "deset": 10, + "jedanaest": 11, + "dvanaest": 12, + "trinaest": 13, + "četrnaest": 14, + "petnaest": 15, + "šesnaest": 16, + "sedamnaest": 17, + "osamnaest": 18, + "devetnaest": 19 + }, + "TENS": { + "dvadeset": 20, + "trideset": 30, + "četrdeset": 40, + "pedeset": 50, + "šezdeset": 60, + "sedamdeset": 70, + "osamdeset": 80, + "devedeset": 90 + }, + "HUNDREDS": { + "dvesta": 200, + "trista": 300 + }, + "BIG_POWERS_OF_TEN": { + "hiljada": 1000, + "hiljadu": 1000, + "milion": 1000000, + "milijarda": 1000000000, + "bilion": 1000000000000, + "bilijarda": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "deset": 10, - "jedanaest": 11, - "dvanaest": 12, - "trinaest": 13, - "četrnaest": 14, - "petnaest": 15, - "šesnaest": 16, - "sedamnaest": 17, - "osamnaest": 18, - "devetnaest": 19 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": { - "dvadeset": 20, - "trideset": 30, - "četrdeset": 40, - "pedeset": 50, - "šezdeset": 60, - "sedamdeset": 70, - "osamdeset": 80, - "devedeset": 90 - }, - "HUNDREDS": { - "dvesta": 200, - "trista": 300 - }, - "BIG_POWERS_OF_TEN": { - "hiljada": 1000, - "hiljadu": 1000, - "milion": 1000000, - "milijarda": 1000000000, - "bilion": 1000000000000, - "bilijarda": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/sr.py b/number_parser/data/sr.py index b07e479..144e5ab 100644 --- a/number_parser/data/sr.py +++ b/number_parser/data/sr.py @@ -1,52 +1,62 @@ info = { - "UNIT_NUMBERS": { - "нула": 0, - "један": 1, - "једна": 1, - "једно": 1, - "два": 2, - "две": 2, - "три": 3, - "четири": 4, - "пет": 5, - "шест": 6, - "седам": 7, - "осам": 8, - "девет": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "нула": 0, + "један": 1, + "једна": 1, + "једно": 1, + "два": 2, + "две": 2, + "три": 3, + "четири": 4, + "пет": 5, + "шест": 6, + "седам": 7, + "осам": 8, + "девет": 9 + }, + "DIRECT_NUMBERS": { + "десет": 10, + "једанаест": 11, + "дванаест": 12, + "тринаест": 13, + "четрнаест": 14, + "петнаест": 15, + "шеснаест": 16, + "седамнаест": 17, + "осамнаест": 18, + "деветнаест": 19 + }, + "TENS": { + "двадесет": 20, + "тридесет": 30, + "четрдесет": 40, + "педесет": 50, + "шездесет": 60, + "седамдесет": 70, + "осамдесет": 80, + "деведесет": 90 + }, + "HUNDREDS": { + "двеста": 200, + "триста": 300 + }, + "BIG_POWERS_OF_TEN": { + "хиљада": 1000, + "хиљаду": 1000, + "милион": 1000000, + "милијарда": 1000000000, + "билион": 1000000000000, + "билијарда": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "десет": 10, - "једанаест": 11, - "дванаест": 12, - "тринаест": 13, - "четрнаест": 14, - "петнаест": 15, - "шеснаест": 16, - "седамнаест": 17, - "осамнаест": 18, - "деветнаест": 19 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": { - "двадесет": 20, - "тридесет": 30, - "четрдесет": 40, - "педесет": 50, - "шездесет": 60, - "седамдесет": 70, - "осамдесет": 80, - "деведесет": 90 - }, - "HUNDREDS": { - "двеста": 200, - "триста": 300 - }, - "BIG_POWERS_OF_TEN": { - "хиљада": 1000, - "хиљаду": 1000, - "милион": 1000000, - "милијарда": 1000000000, - "билион": 1000000000000, - "билијарда": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/su.py b/number_parser/data/su.py index 11319bc..186d34b 100644 --- a/number_parser/data/su.py +++ b/number_parser/data/su.py @@ -1,23 +1,33 @@ info = { - "UNIT_NUMBERS": { - "nol": 0, - "hiji": 1, - "sa": 1, - "dua": 2, - "tilu": 3, - "opat": 4, - "lima": 5, - "genep": 6, - "tujuh": 7, - "dalapan": 8, - "salapan": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "nol": 0, + "hiji": 1, + "sa": 1, + "dua": 2, + "tilu": 3, + "opat": 4, + "lima": 5, + "genep": 6, + "tujuh": 7, + "dalapan": 8, + "salapan": 9 + }, + "DIRECT_NUMBERS": { + "sapuluh": 10, + "sabelas": 11 + }, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "DIRECT_NUMBERS": { - "sapuluh": 10, - "sabelas": 11 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/sv.py b/number_parser/data/sv.py index 68f433c..3bfa647 100644 --- a/number_parser/data/sv.py +++ b/number_parser/data/sv.py @@ -1,46 +1,56 @@ info = { - "UNIT_NUMBERS": { - "noll": 0, - "en": 1, - "et": 1, - "ett": 1, - "två": 2, - "tre": 3, - "fyra": 4, - "fem": 5, - "sex": 6, - "sju": 7, - "åtta": 8, - "nio": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "noll": 0, + "en": 1, + "et": 1, + "ett": 1, + "två": 2, + "tre": 3, + "fyra": 4, + "fem": 5, + "sex": 6, + "sju": 7, + "åtta": 8, + "nio": 9 + }, + "DIRECT_NUMBERS": { + "tio": 10, + "elva": 11, + "tolv": 12, + "tretton": 13, + "fjorton": 14, + "femton": 15, + "sexton": 16, + "sjutton": 17, + "arton": 18, + "nitton": 19 + }, + "TENS": { + "tjugo": 20, + "trettio": 30, + "fyrtio": 40, + "femtio": 50, + "sextio": 60, + "sjuttio": 70, + "åttio": 80, + "nittio": 90 + }, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "miljoner": 1000000, + "miljarder": 1000000000, + "biljoner": 1000000000000, + "biljarder": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "tio": 10, - "elva": 11, - "tolv": 12, - "tretton": 13, - "fjorton": 14, - "femton": 15, - "sexton": 16, - "sjutton": 17, - "arton": 18, - "nitton": 19 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": { - "tjugo": 20, - "trettio": 30, - "fyrtio": 40, - "femtio": 50, - "sextio": 60, - "sjuttio": 70, - "åttio": 80, - "nittio": 90 - }, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "miljoner": 1000000, - "miljarder": 1000000000, - "biljoner": 1000000000000, - "biljarder": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/sw.py b/number_parser/data/sw.py index f16a7d2..642463b 100644 --- a/number_parser/data/sw.py +++ b/number_parser/data/sw.py @@ -1,28 +1,38 @@ info = { - "UNIT_NUMBERS": { - "sifuri": 0, - "moja": 1, - "mbili": 2, - "tatu": 3, - "nne": 4, - "tano": 5, - "sita": 6, - "saba": 7, - "nane": 8, - "tisa": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "sifuri": 0, + "moja": 1, + "mbili": 2, + "tatu": 3, + "nne": 4, + "tano": 5, + "sita": 6, + "saba": 7, + "nane": 8, + "tisa": 9 + }, + "DIRECT_NUMBERS": {}, + "TENS": { + "ishirini": 20, + "thelathini": 30, + "arobaini": 40, + "hamsini": 50, + "sitini": 60, + "sabini": 70, + "themanini": 80, + "tisini": 90 + }, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "DIRECT_NUMBERS": {}, - "TENS": { - "ishirini": 20, - "thelathini": 30, - "arobaini": 40, - "hamsini": 50, - "sitini": 60, - "sabini": 70, - "themanini": 80, - "tisini": 90 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/ta.py b/number_parser/data/ta.py index aee6eb7..14e8ced 100644 --- a/number_parser/data/ta.py +++ b/number_parser/data/ta.py @@ -1,52 +1,62 @@ info = { - "UNIT_NUMBERS": { - "பூஜ்யம்": 0, - "ஒன்று": 1, - "இரண்டு": 2, - "மூன்று": 3, - "நான்கு": 4, - "ஐந்து": 5, - "ஆறு": 6, - "ஏழு": 7, - "எட்டு": 8, - "ஒன்பது": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "பூஜ்யம்": 0, + "ஒன்று": 1, + "இரண்டு": 2, + "மூன்று": 3, + "நான்கு": 4, + "ஐந்து": 5, + "ஆறு": 6, + "ஏழு": 7, + "எட்டு": 8, + "ஒன்பது": 9 + }, + "DIRECT_NUMBERS": { + "பத்து": 10, + "பதினொன்று": 11, + "பன்னிரண்டு": 12, + "பதின்மூன்று": 13, + "பதினான்கு": 14, + "பதினைந்து": 15, + "பதினாறு": 16, + "பதினேழு": 17, + "பதினெட்டு": 18, + "பத்தொன்பது": 19 + }, + "TENS": { + "இருபது": 20, + "முப்பது": 30, + "நாற்பது": 40, + "ஐம்பது": 50, + "அறுபது": 60, + "எழுபது": 70, + "எண்பது": 80, + "தொண்ணூறு": 90 + }, + "HUNDREDS": { + "இருநூறு": 200, + "முந்நூறு": 300, + "நாநூறூ": 400, + "ஐநூறு": 500, + "அறுநூறு": 600, + "எழுநூறு": 700, + "எண்நூறு": 800, + "தொள்ளாயிரம்": 900 + }, + "BIG_POWERS_OF_TEN": { + "ஆயிரம்": 1000, + "லட்சம்": 100000, + "கோடி": 10000000 + } }, - "DIRECT_NUMBERS": { - "பத்து": 10, - "பதினொன்று": 11, - "பன்னிரண்டு": 12, - "பதின்மூன்று": 13, - "பதினான்கு": 14, - "பதினைந்து": 15, - "பதினாறு": 16, - "பதினேழு": 17, - "பதினெட்டு": 18, - "பத்தொன்பது": 19 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": { - "இருபது": 20, - "முப்பது": 30, - "நாற்பது": 40, - "ஐம்பது": 50, - "அறுபது": 60, - "எழுபது": 70, - "எண்பது": 80, - "தொண்ணூறு": 90 - }, - "HUNDREDS": { - "இருநூறு": 200, - "முந்நூறு": 300, - "நாநூறூ": 400, - "ஐநூறு": 500, - "அறுநூறு": 600, - "எழுநூறு": 700, - "எண்நூறு": 800, - "தொள்ளாயிரம்": 900 - }, - "BIG_POWERS_OF_TEN": { - "ஆயிரம்": 1000, - "லட்சம்": 100000, - "கோடி": 10000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/th.py b/number_parser/data/th.py index dade076..c781ee2 100644 --- a/number_parser/data/th.py +++ b/number_parser/data/th.py @@ -1,21 +1,31 @@ info = { - "UNIT_NUMBERS": { - "ศูนย์": 0, - "หนึ่ง": 1, - "สอง": 2, - "สาม": 3, - "สี่": 4, - "ห้า": 5, - "หก": 6, - "เจ็ด": 7, - "แปด": 8, - "เก้า": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "ศูนย์": 0, + "หนึ่ง": 1, + "สอง": 2, + "สาม": 3, + "สี่": 4, + "ห้า": 5, + "หก": 6, + "เจ็ด": 7, + "แปด": 8, + "เก้า": 9 + }, + "DIRECT_NUMBERS": {}, + "TENS": { + "ยี่​สิบ": 20 + }, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "DIRECT_NUMBERS": {}, - "TENS": { - "ยี่​สิบ": 20 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/tr.py b/number_parser/data/tr.py index d5e8a1e..64cfbbb 100644 --- a/number_parser/data/tr.py +++ b/number_parser/data/tr.py @@ -1,35 +1,45 @@ info = { - "UNIT_NUMBERS": { - "sıfır": 0, - "bir": 1, - "iki": 2, - "üç": 3, - "dört": 4, - "beş": 5, - "altı": 6, - "yedi": 7, - "sekiz": 8, - "dokuz": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "sıfır": 0, + "bir": 1, + "iki": 2, + "üç": 3, + "dört": 4, + "beş": 5, + "altı": 6, + "yedi": 7, + "sekiz": 8, + "dokuz": 9 + }, + "DIRECT_NUMBERS": {}, + "TENS": { + "yirmi": 20, + "otuz": 30, + "kırk": 40, + "elli": 50, + "altmış": 60, + "yetmiş": 70, + "seksen": 80, + "doksan": 90 + }, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "yüz": 100, + "bin": 1000, + "milyon": 1000000, + "milyar": 1000000000, + "trilyon": 1000000000000, + "katrilyon": 1000000000000000 + } }, - "DIRECT_NUMBERS": {}, - "TENS": { - "yirmi": 20, - "otuz": 30, - "kırk": 40, - "elli": 50, - "altmış": 60, - "yetmiş": 70, - "seksen": 80, - "doksan": 90 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "yüz": 100, - "bin": 1000, - "milyon": 1000000, - "milyar": 1000000000, - "trilyon": 1000000000000, - "katrilyon": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/uk.py b/number_parser/data/uk.py index 471dacb..b9af23e 100644 --- a/number_parser/data/uk.py +++ b/number_parser/data/uk.py @@ -1,67 +1,77 @@ info = { - "UNIT_NUMBERS": { - "нуль": 0, - "один": 1, - "одна": 1, - "одне": 1, - "два": 2, - "дві": 2, - "три": 3, - "чотири": 4, - "пʼять": 5, - "шість": 6, - "сім": 7, - "вісім": 8, - "девʼять": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "нуль": 0, + "один": 1, + "одна": 1, + "одне": 1, + "два": 2, + "дві": 2, + "три": 3, + "чотири": 4, + "пʼять": 5, + "шість": 6, + "сім": 7, + "вісім": 8, + "девʼять": 9 + }, + "DIRECT_NUMBERS": { + "десять": 10, + "одинадцять": 11, + "дванадцять": 12, + "тринадцять": 13, + "чотирнадцять": 14, + "пʼятнадцять": 15, + "шістнадцять": 16, + "сімнадцять": 17, + "вісімнадцять": 18, + "девʼятнадцять": 19 + }, + "TENS": { + "двадцять": 20, + "тридцять": 30, + "сорок": 40, + "пʼятдесят": 50, + "шістдесят": 60, + "сімдесят": 70, + "вісімдесят": 80, + "девʼяносто": 90 + }, + "HUNDREDS": { + "двісті": 200, + "триста": 300, + "чотириста": 400, + "пʼятсот": 500, + "шістсот": 600, + "сімсот": 700, + "вісімсот": 800, + "девʼятсот": 900 + }, + "BIG_POWERS_OF_TEN": { + "тисяч": 1000, + "тисяча": 1000, + "тисячі": 1000, + "мільйон": 1000000, + "мільйони": 1000000, + "мільйонів": 1000000, + "мільярд": 1000000000, + "мільярди": 1000000000, + "мільярдів": 1000000000, + "більйон": 1000000000000, + "більйони": 1000000000000, + "більйонів": 1000000000000, + "більярд": 1000000000000000, + "більярди": 1000000000000000, + "більярдів": 1000000000000000 + } }, - "DIRECT_NUMBERS": { - "десять": 10, - "одинадцять": 11, - "дванадцять": 12, - "тринадцять": 13, - "чотирнадцять": 14, - "пʼятнадцять": 15, - "шістнадцять": 16, - "сімнадцять": 17, - "вісімнадцять": 18, - "девʼятнадцять": 19 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "TENS": { - "двадцять": 20, - "тридцять": 30, - "сорок": 40, - "пʼятдесят": 50, - "шістдесят": 60, - "сімдесят": 70, - "вісімдесят": 80, - "девʼяносто": 90 - }, - "HUNDREDS": { - "двісті": 200, - "триста": 300, - "чотириста": 400, - "пʼятсот": 500, - "шістсот": 600, - "сімсот": 700, - "вісімсот": 800, - "девʼятсот": 900 - }, - "BIG_POWERS_OF_TEN": { - "тисяч": 1000, - "тисяча": 1000, - "тисячі": 1000, - "мільйон": 1000000, - "мільйони": 1000000, - "мільйонів": 1000000, - "мільярд": 1000000000, - "мільярди": 1000000000, - "мільярдів": 1000000000, - "більйон": 1000000000000, - "більйони": 1000000000000, - "більйонів": 1000000000000, - "більярд": 1000000000000000, - "більярди": 1000000000000000, - "більярдів": 1000000000000000 - }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/vi.py b/number_parser/data/vi.py index c0a6c90..c9a51e8 100644 --- a/number_parser/data/vi.py +++ b/number_parser/data/vi.py @@ -1,24 +1,34 @@ info = { - "UNIT_NUMBERS": { - "không": 0, - "một": 1, - "hai": 2, - "ba": 3, - "bốn": 4, - "năm": 5, - "sáu": 6, - "bảy": 7, - "tám": 8, - "chín": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "không": 0, + "một": 1, + "hai": 2, + "ba": 3, + "bốn": 4, + "năm": 5, + "sáu": 6, + "bảy": 7, + "tám": 8, + "chín": 9 + }, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": { + "trăm": 100, + "nghìn": 1000, + "triệu": 1000000, + "tỷ": 1000000000 + } }, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": { - "trăm": 100, - "nghìn": 1000, - "triệu": 1000000, - "tỷ": 1000000000 + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "SKIP_TOKENS": [] + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/yue-Hans.py b/number_parser/data/yue-Hans.py index 37557f5..189b868 100644 --- a/number_parser/data/yue-Hans.py +++ b/number_parser/data/yue-Hans.py @@ -1,29 +1,39 @@ info = { - "UNIT_NUMBERS": { - "零": 0, - "一": 1, - "壹": 1, - "两": 2, - "二": 2, - "贰": 2, - "三": 3, - "叁": 3, - "四": 4, - "肆": 4, - "五": 5, - "伍": 5, - "六": 6, - "陆": 6, - "七": 7, - "柒": 7, - "八": 8, - "捌": 8, - "九": 9, - "玖": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "零": 0, + "一": 1, + "壹": 1, + "两": 2, + "二": 2, + "贰": 2, + "三": 3, + "叁": 3, + "四": 4, + "肆": 4, + "五": 5, + "伍": 5, + "六": 6, + "陆": 6, + "七": 7, + "柒": 7, + "八": 8, + "捌": 8, + "九": 9, + "玖": 9 + }, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/yue.py b/number_parser/data/yue.py index 7ceab60..73b8be8 100644 --- a/number_parser/data/yue.py +++ b/number_parser/data/yue.py @@ -1,29 +1,39 @@ info = { - "UNIT_NUMBERS": { - "零": 0, - "一": 1, - "壹": 1, - "二": 2, - "兩": 2, - "貳": 2, - "三": 3, - "叁": 3, - "四": 4, - "肆": 4, - "五": 5, - "伍": 5, - "六": 6, - "陸": 6, - "七": 7, - "柒": 7, - "八": 8, - "捌": 8, - "九": 9, - "玖": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "零": 0, + "一": 1, + "壹": 1, + "二": 2, + "兩": 2, + "貳": 2, + "三": 3, + "叁": 3, + "四": 4, + "肆": 4, + "五": 5, + "伍": 5, + "六": 6, + "陸": 6, + "七": 7, + "柒": 7, + "八": 8, + "捌": 8, + "九": 9, + "玖": 9 + }, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/zh-Hant.py b/number_parser/data/zh-Hant.py index 26d0c8b..97069cd 100644 --- a/number_parser/data/zh-Hant.py +++ b/number_parser/data/zh-Hant.py @@ -1,30 +1,40 @@ info = { - "UNIT_NUMBERS": { - "〇": 0, - "零": 0, - "一": 1, - "壹": 1, - "二": 2, - "兩": 2, - "貳": 2, - "三": 3, - "參": 3, - "四": 4, - "肆": 4, - "五": 5, - "伍": 5, - "六": 6, - "陸": 6, - "七": 7, - "柒": 7, - "八": 8, - "捌": 8, - "九": 9, - "玖": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "〇": 0, + "零": 0, + "一": 1, + "壹": 1, + "二": 2, + "兩": 2, + "貳": 2, + "三": 3, + "參": 3, + "四": 4, + "肆": 4, + "五": 5, + "伍": 5, + "六": 6, + "陸": 6, + "七": 7, + "柒": 7, + "八": 8, + "捌": 8, + "九": 9, + "玖": 9 + }, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/number_parser/data/zh.py b/number_parser/data/zh.py index 2d95b3e..f3827e1 100644 --- a/number_parser/data/zh.py +++ b/number_parser/data/zh.py @@ -1,30 +1,40 @@ info = { - "UNIT_NUMBERS": { - "〇": 0, - "零": 0, - "一": 1, - "壹": 1, - "两": 2, - "二": 2, - "贰": 2, - "三": 3, - "叁": 3, - "四": 4, - "肆": 4, - "五": 5, - "伍": 5, - "六": 6, - "陆": 6, - "七": 7, - "柒": 7, - "八": 8, - "捌": 8, - "九": 9, - "玖": 9 + "NUMBERS": { + "UNIT_NUMBERS": { + "〇": 0, + "零": 0, + "一": 1, + "壹": 1, + "两": 2, + "二": 2, + "贰": 2, + "三": 3, + "叁": 3, + "四": 4, + "肆": 4, + "五": 5, + "伍": 5, + "六": 6, + "陆": 6, + "七": 7, + "柒": 7, + "八": 8, + "捌": 8, + "九": 9, + "玖": 9 + }, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} }, - "DIRECT_NUMBERS": {}, - "TENS": {}, - "HUNDREDS": {}, - "BIG_POWERS_OF_TEN": {}, - "SKIP_TOKENS": [] + "ORDINAL_NUMBERS": { + "UNIT_NUMBERS": {}, + "DIRECT_NUMBERS": {}, + "TENS": {}, + "HUNDREDS": {}, + "BIG_POWERS_OF_TEN": {} + }, + "SKIP_TOKENS": [], + "IS_LONG": 0 } diff --git a/scripts/write_complete_language_data.py b/scripts/write_complete_language_data.py index d8b1ef4..79e181a 100644 --- a/scripts/write_complete_language_data.py +++ b/scripts/write_complete_language_data.py @@ -12,19 +12,20 @@ SUPPLEMENTARY_PATH = "../number_parser_data/supplementary_translation_data/" TARGET_PATH = "../number_parser/data/" -VALID_KEYS = ["spellout-cardinal", "spellout-numbering"] -INVALID_KEYS = ["cents"] +VALID_CARDINAL_KEYS = ["spellout-cardinal", "spellout-numbering"] +INVALID_CARDINAL_KEYS = ["cents"] CAPTURE_BRACKET_CONTENT = r'\{(.*?)\}' REQUIRED_NUMBERS_DATA = ["UNIT_NUMBERS", "DIRECT_NUMBERS", "TENS", "HUNDREDS", "BIG_POWERS_OF_TEN"] +TYPES_OF_NUMBERS = ["NUMBERS", "ORDINAL_NUMBERS"] -def _is_valid(key): +def _is_valid_cardinal(key): """Identifying whether the given key of the source language file needs to be extracted.""" is_valid = False - for valid_key in VALID_KEYS: + for valid_key in VALID_CARDINAL_KEYS: if valid_key in key: is_valid = True - for invalid_key in INVALID_KEYS: + for invalid_key in INVALID_CARDINAL_KEYS: if invalid_key in key: is_valid = False return is_valid @@ -129,8 +130,13 @@ def write_complete_data(): full_target_path = os.path.join(TARGET_PATH, file_name.split(".")[0]+".py") full_supplementary_path = os.path.join(SUPPLEMENTARY_PATH, file_name) - language_data = {key: {} for key in REQUIRED_NUMBERS_DATA} - ordered_language_data = OrderedDict((key, {}) for key in REQUIRED_NUMBERS_DATA) + language_data = {} + ordered_language_data = OrderedDict() + + for number_types in TYPES_OF_NUMBERS: + language_data[number_types] = {key: {} for key in REQUIRED_NUMBERS_DATA} + ordered_language_data[number_types] = OrderedDict((key, {}) for key in REQUIRED_NUMBERS_DATA) + with open(full_source_path, 'r') as source: data = json.load(source) try: @@ -140,22 +146,29 @@ def write_complete_data(): continue for keys, vals in requisite_data.items(): - if _is_valid(keys): + if _is_valid_cardinal(keys): for key, val in vals.items(): # Removing soft-hyphens from the source file. val = val.replace('\xad', '') - _extract_information(key, val, language_data) + _extract_information(key, val, language_data["NUMBERS"]) with open(full_supplementary_path, 'r') as supplementary_data: data = json.load(supplementary_data) - for keys in REQUIRED_NUMBERS_DATA: - language_data[keys].update(data[keys]) - sorted_tuples = sorted(language_data[keys].items(), key=lambda x: (x[1], x[0])) - for items in sorted_tuples: - word, number = items[0], items[1] - ordered_language_data[keys][word] = int(number) + for number_types in TYPES_OF_NUMBERS: + for keys in REQUIRED_NUMBERS_DATA: + language_data[number_types][keys].update(data[number_types][keys]) + sorted_tuples = sorted(language_data[number_types][keys].items(), key=lambda x: (x[1], x[0])) + for items in sorted_tuples: + word, number = items[0], items[1] + ordered_language_data[number_types][keys][word] = int(number) skip_tokens = sorted(data["SKIP_TOKENS"]) + is_long = data["IS_LONG"] ordered_language_data["SKIP_TOKENS"] = skip_tokens + if is_long: + ordered_language_data["IS_LONG"] = 1 + else: + ordered_language_data["IS_LONG"] = 0 + translation_data = json.dumps(ordered_language_data, indent=4, ensure_ascii=False) out_text = ('info = ' + translation_data + '\n') with open(full_target_path, 'w+') as target_file: From ec1d0bc1e931a6b857c096702e6976d1dd6ce953 Mon Sep 17 00:00:00 2001 From: arnavkapoor Date: Tue, 4 Aug 2020 21:02:13 +0530 Subject: [PATCH 4/4] removing temporary file --- scripts/modify_supplementary_data_files.py | 124 --------------------- 1 file changed, 124 deletions(-) delete mode 100644 scripts/modify_supplementary_data_files.py diff --git a/scripts/modify_supplementary_data_files.py b/scripts/modify_supplementary_data_files.py deleted file mode 100644 index 8274e94..0000000 --- a/scripts/modify_supplementary_data_files.py +++ /dev/null @@ -1,124 +0,0 @@ -""" -The raw CLDR data was retreived on 25th Jun , 2020 from the following link -https://github.com/unicode-cldr/cldr-rbnf -""" - -import os -import json -import re -from collections import OrderedDict - -SUPPLEMENTARY_PATH = "../number_parser_data/supplementary_translation_data/" -f = ["UNIT_NUMBERS","DIRECT_NUMBERS","TENS","HUNDREDS","BIG_POWERS_OF_TEN"] -tokens = "SKIP_TOKENS" - -for file_name in os.listdir(SUPPLEMENTARY_PATH): - if file_name == "en.json": - continue - new_dic = {"NUMBERS": {} , "ORDINAL_NUMBERS": {}} - full_supplementary_path = os.path.join(SUPPLEMENTARY_PATH, file_name) - with open(full_supplementary_path, 'r') as source: - data = json.load(source) - for i in range(5): - new_dic["NUMBERS"][f[i]] = data[f[i]] - new_dic["ORDINAL_NUMBERS"][f[i]] = {} - - new_dic[tokens] = data[tokens] - new_dic["IS_LONG"] = False - # print(new_dic,file_name) - # break - mfinal = json.dumps(new_dic, indent=4, ensure_ascii=False) - with open(full_supplementary_path, 'w') as ff: - ff.write(mfinal) - # break - -# VALID_KEYS = ["spellout-cardinal", "spellout-numbering"] -# def _is_valid(key): -# """Identifying whether the given key of the source language file needs to be extracted.""" -# is_valid = False -# for valid_key in VALID_KEYS: -# if valid_key in key: -# is_valid = True -# return is_valid - -# def _count_zero(number): -# """Counting the number of zeroes in the given number.""" -# zero_count = 0 -# while number > 9: -# if number % 10 == 0: -# zero_count += 1 -# number /= 10 -# else: -# break -# return zero_count - - -# fset = {} - - -# PATH_TO_FILE = "/home/arnav/GSOC_2020/number-parser/number_parser_data/raw_cldr_translation_data/" -# with open(PATH_TO_FILE + "ru.json") as f: -# data = json.load(f) -# requisite_data = data['rbnf']['rbnf']['SpelloutRules'] -# for keys, vals in requisite_data.items(): -# if _is_valid(keys): -# for key, val in vals.items(): -# try: -# if _count_zero((int)(key)) == 2: -# if key[0] == "1": -# continue -# get_vals = val.split("<") -# get_val = get_vals[1] -# suffix = get_vals[-1].split("[")[0] -# rq1 = get_val[1:] -# rq2 = rq1.replace("feminine", "masculine") -# rq3 = rq1.replace("masculine", "feminine") - -# fd = (int)(key) // 100 -# d1 = requisite_data["%" + rq1] -# d2 = requisite_data["%" + rq2] -# d3 = requisite_data["%" + rq3] - -# if(get_vals[-1] == "сти[ >>];"): -# for i in range(fd,3): -# prefix = d3[str(i)] -# prefix = prefix[:-1] -# fset[prefix+suffix] = i*100 -# prefix = d2[str(i)] -# prefix = prefix[:-1] -# fset[prefix+suffix] = i*100 - - -# elif (get_vals[-1] == "ста[ >>];"): -# for i in range(fd,5): -# prefix = d2[str(i)] -# prefix = prefix[:-1] -# fset[prefix+suffix] = i*100 -# # print(i*100,prefix+suffix) -# else: -# for i in range(fd,10): -# prefix = d2[str(i)] -# prefix = prefix[:-1] -# fset[prefix+suffix] = i*100 -# # print(i*100,prefix+suffix) -# # print(d2) -# except: -# pass - -# fdict = OrderedDict() -# mvals = (sorted(fset.items(), key=lambda x: x[1])) - -# for each in mvals: -# fdict[each[0]] = each[1] - -# RT_PATH = "/home/arnav/GSOC_2020/number-parser/number_parser_data/supplementary_translation_data" -# mfinal = {} -# with open(RT_PATH + "/ru.json" ) as f: -# data = json.load(f) -# data["HUNDREDS"].update(fdict) -# mfinal = json.dumps(data, indent=4, ensure_ascii=False) -# print(mfinal) -# with open(RT_PATH + "/ru.json","w" ) as ff: -# ff.write(mfinal) - -