pre-commit: filter out large diffs

scripts/filter_pr_changes.py

@@ -0,0 +1,28 @@
+#!/usr/bin/python3
+
+import subprocess
+
+max_diff_per_file = 20000
+max_diff_total = 20000
+max_file_total = 300 - 1
+
+stats = subprocess.check_output(['git', 'diff', '--numstat']).decode().splitlines()
+diffs = []
+for line in stats:
+    add, sub, file = line.removesuffix('\n').split()
+    count = int(add)+int(sub)
+    if count > max_diff_per_file:
+        continue
+    diffs.append((file, count))
+diffs.sort(key=lambda x: x[1])
+
+file_count = 0
+diff_count = 0
+
+for file, count in diffs:
+    if file_count < max_file_total and diff_count + count <= max_diff_total:
+        file_count += 1
+        diff_count += count
+        subprocess.run(['git', 'add', file])
+    else:
+        break

scripts/update_optimized.sh

@@ -33,12 +33,15 @@ then
     echo "SHOULD_OPEN_ISSUE=0" >> $GITHUB_OUTPUT
   fi
 else
-  git commit -a -m "pre-commit: Update"
+  diff_stat=$(git diff --shortstat)
+  ./scripts/filter_pr_changes.py
+  git commit -m "pre-commit: Update"
   git push -f
   echo "baseline: https://github.com/llvm/llvm-project/commit/$LLVM_REVISION" > scripts/pr-comment.md
   echo "patch: $COMMIT_URL" >> scripts/pr-comment.md
   echo "sha256: $PATCH_SHA256" >> scripts/pr-comment.md
   echo "commit: $(git rev-parse HEAD)" >> scripts/pr-comment.md
+  echo "$diff_stat" >> scripts/pr-comment.md
   head -100 test.log >> scripts/pr-comment.md
   git show --name-only --oneline | head -100 >> scripts/pr-comment.md
 fi