better file filtering (#3983)

better file fitlereing

Co-authored-by: Martin Ye <[email protected]>

sweepai/config/client.py

@@ -1,6 +1,7 @@
 from __future__ import annotations
 
 import os
+import re
 import traceback
 from functools import lru_cache
 
@@ -313,6 +314,15 @@ def is_file_suitable(self, file_contents: str) -> tuple[bool, str]:
         if len(file_contents)/line_count > 200:
             return False, "This file was determined to be non human readable due to the average line length."
         return True, ""
+
+    # returns if a file is likely autogenerated or not
+    def is_file_auto_generated(self, file_name: str) -> tuple[bool, str]:
+        # if there is a string of numbers in the file name that is more than 4 characters long, it is likely autogenerated
+        pattern = r'\d{4,}'
+        match = re.search(pattern, file_name)
+        if bool(match):
+            return True, "The filename means that this file is likely auto generated."
+        return False, ""
 
 
 

sweepai/core/review_utils.py

@@ -133,6 +133,13 @@ def get_pr_changes(repo: Repository, pr: PullRequest) -> tuple[dict[str, PRChang
                 errored = True
                 e = UnsuitableFileException(reason)
                 unsuitable_files.append((file_name, e))
+            else:
+                # drop likely autogenerated files based on file name
+                auto_generated, reason = sweep_config.is_file_auto_generated(file_name)
+                if auto_generated:
+                    errored = True
+                    e = UnsuitableFileException(reason)
+                    unsuitable_files.append((file_name, e))
 
         if errored:
             posthog.capture(