Apostrophes in legal abbreviations truncate case name extraction (Dep't, Nat'l, Comm'n, etc.)

[rlfordon]

Summary

eyecite's case name extraction regexes use character classes like [\w\-.]+ that don't include apostrophes. Legal abbreviations with internal apostrophes — Dep't, Nat'l, Comm'n, Att'y, P'ship, etc. — are truncated at the apostrophe, losing the suffix.

Reproduction

from eyecite import get_citations
from eyecite.models import FullCaseCitation

text = "Att'y Grievance Comm'n v. Glenn, 341 Md. 448 (2003)."
c = [c for c in get_citations(text) if isinstance(c, FullCaseCitation)][0]
print(c.metadata.plaintiff)
# Expected: "Att'y Grievance Comm'n"
# Actual:   "Att' Grievance Comm'"   (suffixes after apostrophes dropped)

text2 = "Keaau Dev. P'ship LLC v. Lawrence, 571 P.3d 958 (2025)."
c2 = [c for c in get_citations(text2) if isinstance(c, FullCaseCitation)][0]
print(c2.metadata.plaintiff)
# Expected: "Keaau Dev. P'ship LLC"
# Actual:   "Keaau Dev. P' LLC"

Affected abbreviations

These are all standard Indigo Book legal abbreviations — very common in case citations:

Cited as	Extracted as	Full form
Att'y	Att'	Attorney
Dep't	Dep'	Department
Gov't	Gov'	Government
Comm'n	Comm'	Commission
Comm'r	Comm'	Commissioner
P'ship	P'	Partnership
Nat'l	Nat'	National
Int'l	Int'	International
Sec'y	Sec'	Secretary
Ass'n	Ass'	Association
Adm'r	Adm'	Administrator

Root cause

Three regexes in eyecite/regexes.py use character classes that exclude apostrophes:

Regex	Character class	Purpose
SHORT_CITE_ANTECEDENT_REGEX	[\w\-.]+	Short cite antecedents
SUPRA_ANTECEDENT_REGEX	[\w\-.]+	Supra cite antecedents
PRE_FULL_CITATION_REGEX	[a-z\-.]+	Backward scan for case names

Suggested fix

Add straight and curly apostrophes to the character classes:

• [\w\-.]+ → [\w\-.'\u2019]+
• [a-z\-.]+ → [a-z\-.'\u2019]+

Including \u2019 (right single quotation mark) handles PDFs that use Unicode smart quotes.

This should be low risk — the regexes already allow periods and hyphens, so apostrophes are consistent. The change is additive: existing matches are unaffected, only previously-truncated words are now captured fully.

Impact

I found 8 case names directly affected in my corpus of 19 court opinion PDFs that I've been running through eyecite. These abbreviations are extremely common in legal citations generally — this likely affects a significant number of extractions.

Happy to submit a PR for this if you'd like.