Merge pull request #57 from dannyl1u/dliu/possibility-threshold

Comment most similar issue even if similarity threshold not met

src/issue_handler.py

@@ -1,5 +1,4 @@
 import logging
-import re
 
 from config import SIMILARITY_THRESHOLD
 from src.github_api import close_issue, leave_comment
@@ -13,11 +12,7 @@ def handle_new_issue(
 ):
     logger.info(f"New issue opened: {issue_number} in {repo_full_name}")
     full_issue = f"{issue_title} {issue_body}"
-
-    stripped_title = re.sub(
-        r"^(bug:|feat:|enh:|chore:)\s*", "", issue_title, flags=re.IGNORECASE
-    )
-    similar_issue = query_similar_issue(stripped_title, repo_id)
+    similar_issue = query_similar_issue(full_issue, repo_id)
 
     if similar_issue and similar_issue["distance"] < 1 - SIMILARITY_THRESHOLD:
         comment_text = f"Closed due to high similarity with issue #{similar_issue['issue_number']} with title '{similar_issue['title']}'"
@@ -26,9 +21,17 @@ def handle_new_issue(
         logger.info(
             f"The new issue #{issue_number} with title '{issue_title}' is most similar to existing issue #{similar_issue['issue_number']} with title '{similar_issue['title']}', with a cosine similarity of {1 - similar_issue['distance']:.2f}."
         )
+    elif similar_issue and similar_issue["distance"] < 1 - (SIMILARITY_THRESHOLD * 0.5):
+        comment_text = f"Possibly related to issue #{similar_issue['issue_number']} with title '{similar_issue['title']}'"
+        leave_comment(installation_id, repo_full_name, issue_number, comment_text)
+        logger.info(
+            f"The new issue #{issue_number} with title '{issue_title}' is possibly similar to existing issue #{similar_issue['issue_number']} with title '{similar_issue['title']}', with a cosine similarity of {1 - similar_issue['distance']:.2f}."
+        )
     else:
+        comment_text = f"Most likely a new issue, most similar issue: #{similar_issue['issue_number']} with title '{similar_issue['title']}'"
+        leave_comment(installation_id, repo_full_name, issue_number, comment_text)
         logger.info(
-            f"The new issue #{issue_number} with title '{issue_title}' is not similar to existing issues or cannot be closed due to missing information."
+            f"The new issue #{issue_number} with title '{issue_title}' is not similar enough to close, most similar: #{similar_issue['issue_number']} with title '{similar_issue['title']}', with a cosine similarity of {1 - similar_issue['distance']:.2f}"
         )
 
     add_issue_to_chroma(full_issue, issue_number, issue_title, repo_id)