From 6bfaf9ef42c932ce1e31b4670476384e1cbfe375 Mon Sep 17 00:00:00 2001
From: Andy Shultz <ashultz@edx.org>
Date: Thu, 16 Nov 2023 15:22:08 -0500
Subject: [PATCH] feat: add logging after ES to go with the full arg log

because hits can be quite fat we don't want to log everything,
but we can log summary fields and validate that the hit translation
doesn't lose them
---
 edxsearch/__init__.py |  2 +-
 search/elastic.py     | 16 +++++++++++-----
 2 files changed, 12 insertions(+), 6 deletions(-)

diff --git a/edxsearch/__init__.py b/edxsearch/__init__.py
index 0f6ce6a7..1eaf4653 100644
--- a/edxsearch/__init__.py
+++ b/edxsearch/__init__.py
@@ -1,3 +1,3 @@
 """ Container module for testing / demoing search """
 
-__version__ = '3.7.1'
+__version__ = '3.7.2'
diff --git a/search/elastic.py b/search/elastic.py
index 4e38093a..4ace3083 100644
--- a/search/elastic.py
+++ b/search/elastic.py
@@ -22,7 +22,7 @@
 RESERVED_CHARACTERS = "+=><!(){}[]^~*:\\/&|?"
 
 
-def _translate_hits(es_response):
+def _translate_hits(es_response, log_hit_data=False):
     """
     Provide result set in our desired format from elasticsearch results.
 
@@ -123,11 +123,14 @@ def translate_agg_bucket(bucket, agg_result):
             "other": agg_item["sum_other_doc_count"],
         }
 
-    results = list(map(translate_result, es_response["hits"]["hits"]))
+    hits_result = es_response["hits"]["hits"]
+    total = es_response["hits"]["total"]["value"]
+    max_score = es_response["hits"]["max_score"]
+    results = list(map(translate_result, hits_result))
     response = {
         "took": es_response["took"],
-        "total": es_response["hits"]["total"]["value"],
-        "max_score": es_response["hits"]["max_score"],
+        "total": total,
+        "max_score": max_score,
         "results": results,
     }
     if "aggregations" in es_response:
@@ -137,6 +140,9 @@ def translate_agg_bucket(bucket, agg_result):
             if "total_" not in bucket
         }
 
+    if log_hit_data:
+        log.info(f"elastic search results: total {total}, max_score {max_score}, raw hit count {len(hits_result)}")
+
     return response
 
 
@@ -663,4 +669,4 @@ def search(self,
             log.exception("error while searching index - %r", ex)
             raise
 
-        return _translate_hits(es_response)
+        return _translate_hits(es_response, log_hit_data=log_search_params)