Upload public_routes.py

backend/media_analysis/public_routes.py

@@ -31,6 +31,10 @@ def _public_answer_text(answer):
         "", answer, flags=re.I)
     # Other formats of the same internal metadata are not useful to a reader.
     answer = re.sub(r"\bpaper\s+\d+\s*[,;]\s*label\s+(?:table|figure|fig)[_.-]0*\d+\b", "", answer, flags=re.I)
+    # Some models echo structured citation metadata in prose instead of using
+    # the separately returned provenance record (e.g. Source 1; paper_id=...).
+    answer = re.sub(r"\(\s*Source\s*\d+\s*;?\s*paper_id\s*=\s*[^);\s]+\s*;?\s*label\s*=\s*[^)]+\)", "", answer, flags=re.I)
+    answer = re.sub(r"\bSource\s*\d+\s*;\s*paper_id\s*=\s*[^;\s]+\s*;\s*label\s*=\s*(?:table|figure|fig)[_.-]0*\d+\b", "", answer, flags=re.I)
     return answer
 
 def _enabled():