degbu

generic_ner.py

@@ -277,16 +277,20 @@ from stopwordsiso import stopwords
 stop_words = stopwords(["en", "fr", "de"])
 
 
-def remove_trailing_stopwords(entities):
-    # This function removes stopwords from the end of each entity's text
+def remove_trailing_stopwords(entities, stop_words):
+    # This function removes stopwords from both the beginning and end of each entity's text
     for entity in entities:
         words = entity["text"].split()
 
-        # Continue removing stopwords from the end of the text
+        # Remove stopwords from the beginning
+        while words and words[0].lower() in stop_words:
+            words.pop(0)  # Remove the first word if it's a stopword
+
+        # Remove stopwords from the end
         while words and words[-1].lower() in stop_words:
             words.pop()  # Remove the last word if it's a stopword
 
-        # Join the words back together and update the entity's text
+        # Join the remaining words back together and update the entity's text
         entity["text"] = " ".join(words)
 
     return entities