add_page_numbers.py: modified reSENTTERMINAL for better prefix tagging

nert-nlp · Dec 31, 2023 · 613a8c2 · nschneid · Jan 1, 2024 · tomlup
1 parent 2a0f024
commit 613a8c2
Show file tree

Hide file tree

Showing 3 changed files with 176 additions and 176 deletions.
diff --git a/all-examples/add_page_numbers.py b/all-examples/add_page_numbers.py
@@ -14,7 +14,7 @@
 reFXNS = re.compile(r'^((Head|Mod(ifier)?|P(redicator)?|Predicate|Comp[12]?|PredComp|Nucleus|Prenucleus|Subj(ect)?|O(bj(ect)?)?[12]?|Det|Det-Head|Mod-Head|Subj-det|Subj-det-Head|PredicatorPredComp|Marker|Coordinate[12]|Supplement):)+$')
 reTreeHeader = re.compile(r'^\[\d+\]a\.(Clause|NP|NPinterrog|PP|VP)b\.(Clause|NP|NPinterrog|PP|VP)(c\.(Clause|NP|NPinterrog|PP|VP))?$')
 reNUMERICEX = re.compile(r'^\[\d+\]')
-reSENTTERMINAL = re.compile(r'[.!?](\t|$)')
+reSENTTERMINAL = re.compile(r'(((?<!etc)\.)|[!?])(\t|$|])')
 
 reLI = re.compile(r'(^\[[0-9A-Z]+\](?=\t))|\t[xvi]+(?=\t)|\t([a-z])\.(?=\t)')