[CI] Harden GitHub automation inputs (#1761)

Author: peterdsharpe

.github/scripts/analyze_flaky_tests.py

@@ -106,17 +106,28 @@ def list_workflow_runs(
     while len(runs) < num_runs:
         endpoint = (
             f"/repos/{repo}/actions/workflows/{workflow_name}/runs"
-            f"?branch={branch}&status=completed&per_page={per_page}&page={page}"
+            f"?branch={branch}&event=push&status=completed"
+            f"&per_page={per_page}&page={page}"
         )
         data = gh_api(endpoint)
         if not data or "workflow_runs" not in data:
             break
-        batch = data["workflow_runs"]
-        if not batch:
+        raw_batch = data["workflow_runs"]
+        if not raw_batch:
             break
+        # The API filters above are the primary boundary. Keep this fail-closed
+        # check as defense in depth before downloading executable-repository
+        # artifacts such as JUnit XML.
+        batch = [
+            run
+            for run in raw_batch
+            if run.get("event") == "push"
+            and run.get("head_branch") == branch
+            and (run.get("head_repository") or {}).get("full_name") == repo
+        ]
         runs.extend(batch)
         page += 1
-        if len(batch) < per_page:
+        if len(raw_batch) < per_page:
             break
 
     return runs[:num_runs]

.github/tdbot/tdbot.py

@@ -80,7 +80,8 @@ def get_pr_info(repo: str, pr_number: int) -> dict:
         "--repo",
         repo,
         "--json",
-        "headRefName,baseRefName,author,title,url,labels,mergeable,reviewDecision",
+        "headRefName,baseRefName,author,title,url,labels,mergeable,reviewDecision,"
+        "isCrossRepository",
     )
     return json.loads(result.stdout)
 
@@ -113,6 +114,20 @@ def is_ghstack_pr(head_branch: str) -> bool:
     return bool(re.match(r"^gh/[^/]+/\d+/head$", head_branch))
 
 
+def reject_cross_repository_write(ctx: CommandContext, command: str) -> bool:
+    """Reject commands that would push through the base repository remote."""
+    if ctx.pr_info.get("isCrossRepository") is False:
+        return False
+    post_comment(
+        ctx.repo,
+        ctx.pr_number,
+        f"`{command}` cannot safely update a branch from a fork. "
+        "Please perform the update in the fork and push it from an account "
+        "that controls that repository.",
+    )
+    return True
+
+
 def find_ghstack_stack_top(repo: str, head_branch: str) -> int:
     """Return the PR number at the top of the ghstack stack containing *head_branch*.
 
@@ -279,6 +294,9 @@ def cmd_rebase(ctx: CommandContext, args: argparse.Namespace) -> None:
         )
         return
 
+    if reject_cross_repository_write(ctx, "rebase"):
+        return
+
     if is_ghstack_pr(head):
         _rebase_ghstack(ctx, args)
     else:
@@ -380,6 +398,9 @@ def cmd_lint(ctx: CommandContext, _args: argparse.Namespace) -> None:
         )
         return
 
+    if reject_cross_repository_write(ctx, "lint"):
+        return
+
     post_comment(
         ctx.repo,
         ctx.pr_number,

.github/workflows/pr-label.yml

@@ -91,8 +91,16 @@ jobs:
           **Note:** Matching is case-insensitive. Common variations (singular/plural) are supported.
           COMMENT_EOF
             sed -i 's/^ *//' /tmp/pr_comment.md
-            sed -i "s/REASON_PLACEHOLDER/$reason/" /tmp/pr_comment.md
-            sed -i "s|TITLE_PLACEHOLDER|$current_title|" /tmp/pr_comment.md
+            COMMENT_REASON="$reason" COMMENT_TITLE="$current_title" python3 - <<'PY'
+          import os
+          from pathlib import Path
+
+          path = Path("/tmp/pr_comment.md")
+          comment = path.read_text()
+          comment = comment.replace("REASON_PLACEHOLDER", os.environ["COMMENT_REASON"])
+          comment = comment.replace("TITLE_PLACEHOLDER", os.environ["COMMENT_TITLE"])
+          path.write_text(comment)
+          PY
             gh pr comment "$PR_NUMBER" --repo "$REPO" --body-file /tmp/pr_comment.md
           }