Skip to content

OpenAI's Codex Security CLI and TypeScript SDK for finding, validating, and fixing security vulnerabilities. npm: https://www.npmjs.com/package/@openai/codex-security

fix(plugin): read source excerpts from the scanned target (#1063)

mldangelo-oai0c4bf047aeca
diff --git a/plugins/codex-security/scripts/workbench_source_excerpt.py b/plugins/codex-security/scripts/workbench_source_excerpt.py
index 63e9fbe25..87d4199c8 100644
--- a/plugins/codex-security/scripts/workbench_source_excerpt.py
+++ b/plugins/codex-security/scripts/workbench_source_excerpt.py
@@ -71,7 +71,12 @@ def scanned_source_text(scan: sqlite3.Row, target: Path, path: str) -> str | Non
snapshot_digest = scan["target_snapshot_digest"]
if snapshot_digest is not None and snapshot_digest != clean_worktree_content_digest():
return None
- object_name = f"{revision}:{path}"
+ if (
+ scan["diff_target_kind"] == "working_tree"
+ and scan["diff_content_digest"] != clean_worktree_content_digest()
+ ):
+ return None
+ object_name = f"{revision}:./{path}"
content = git_bytes(target, "cat-file", "blob", object_name)
return content.decode("utf-8", errors="replace") if content is not None else None
diff --git a/plugins/codex-security/tests/test_workbench_db.py b/plugins/codex-security/tests/test_workbench_db.py
index b21d854e2..cd0cfacbf 100644
--- a/plugins/codex-security/tests/test_workbench_db.py
+++ b/plugins/codex-security/tests/test_workbench_db.py
@@ -2248,7 +2248,14 @@ def test_workbench_rejects_working_tree_target_after_contents_change(tmp_path: P
def test_workbench_warns_after_working_tree_changes(tmp_path: Path) -> None:
state_dir = tmp_path / "state"
target = tmp_path / "target"
- revision = initialize_git_repository(target)
+ initialize_git_repository(target)
+ (target / "README.md").write_text("committed source\n" * 50)
+ subprocess.run(["git", "add", "README.md"], cwd=target, check=True)
+ subprocess.run(["git", "commit", "-qm", "Add source fixture"], cwd=target, check=True)
+ revision = subprocess.run(
+ ["git", "rev-parse", "HEAD"], cwd=target, check=True, capture_output=True, text=True
+ ).stdout.strip()
+ (target / "README.md").write_text("scanned working-tree source\n" * 50)
(target / "new-file.txt").write_text("selected content\n")
workspace_id = str(uuid.uuid4())
create_saved_working_tree_workspace(state_dir, target, workspace_id, revision)
@@ -2266,6 +2273,7 @@ def test_workbench_warns_after_working_tree_changes(tmp_path: Path) -> None:
scan_dir,
scan_id,
target,
+ relative_path="README.md",
target_kind="git_diff",
diff_base_revision=revision,
diff_head_revision=revision,
@@ -2276,6 +2284,9 @@ def test_workbench_warns_after_working_tree_changes(tmp_path: Path) -> None:
completed = scan_command(state_dir, "complete-scan", scan_id)
assert completed["scan"]["progress"]["status"] == "complete"
assert completed["scan"]["warnings"] == [WORKTREE_CHANGED_WARNING]
+ finding = completed["scan"]["findings"][0]
+ assert finding["locations"][0]["path"] == "README.md"
+ assert "sourceExcerpt" not in finding
manifest = json.loads((scan_dir / "scan-manifest.json").read_text())
assert manifest["scan"]["target"]["snapshotDigest"] == snapshot_digest
assert json.loads((scan_dir / "coverage.json").read_text())["completeness"] == "complete"
@@ -2938,19 +2949,26 @@ def test_completed_finding_projects_writeup_and_poc_artifact_paths(
]
[email protected]("nested", [False, True])
def test_workbench_populates_clean_git_scan_revision_with_large_source_excerpt(
tmp_path: Path,
+ nested: bool,
) -> None:
state_dir = tmp_path / "state"
- target = tmp_path / "target"
- initialize_git_repository(target)
+ repository = tmp_path / "target"
+ initialize_git_repository(repository)
+ target = repository
+ if nested:
+ (repository / "README.md").write_text("different root source\n" * 50)
+ target = repository / "nested"
+ target.mkdir()
(target / "README.md").write_text(
"\n".join(f"source line {line_number}" for line_number in range(1, 51))
+ "\n"
+ "x" * (1024 * 1024)
)
- subprocess.run(["git", "add", "README.md"], cwd=target, check=True)
- subprocess.run(["git", "commit", "-qm", "Add source fixture"], cwd=target, check=True)
+ subprocess.run(["git", "add", "."], cwd=repository, check=True)
+ subprocess.run(["git", "commit", "-qm", "Add source fixture"], cwd=repository, check=True)
revision = subprocess.run(
["git", "rev-parse", "HEAD"],
cwd=target,
@@ -3021,7 +3039,7 @@ def test_source_excerpt_breaks_lines_only_at_newlines(tmp_path: Path, separator:
)
excerpt = finding_source_excerpt(
- {"target_revision": revision, "target_snapshot_digest": None},
+ {"target_revision": revision, "target_snapshot_digest": None, "diff_target_kind": None},
target,
[{"path": "README.md", "startLine": 5, "endLine": 5}],
)
@@ -3043,7 +3061,7 @@ def test_source_excerpt_numbers_standard_line_endings(tmp_path: Path, line_endin
)
excerpt = finding_source_excerpt(
- {"target_revision": revision, "target_snapshot_digest": None},
+ {"target_revision": revision, "target_snapshot_digest": None, "diff_target_kind": None},
target,
[{"path": "README.md", "startLine": 5, "endLine": 5}],
)
@@ -3073,7 +3091,7 @@ def test_source_excerpt_preserves_final_lines(
finding_source_excerpt = namespace["finding_source_excerpt"]
target = tmp_path / "target"
revision = commit_source_fixture(target, source)
- scan = {"target_revision": revision, "target_snapshot_digest": None}
+ scan = {"target_revision": revision, "target_snapshot_digest": None, "diff_target_kind": None}
excerpt = finding_source_excerpt(
scan,
diff --git a/sdk/typescript/tests-ts/plugin-report-limits.test.ts b/sdk/typescript/tests-ts/plugin-report-limits.test.ts
index 1356466bb..7f1b044d5 100644
--- a/sdk/typescript/tests-ts/plugin-report-limits.test.ts
+++ b/sdk/typescript/tests-ts/plugin-report-limits.test.ts
@@ -22,7 +22,7 @@ describe("bundled scan report and source limits", () => {
" source = b'x' * (1024 * 1024 + 1)",
" excerpts.git_bytes = lambda *args: source",
" target = pathlib.Path(directory).resolve()",
- " excerpt = excerpts.scanned_source_text({'target_revision': 'deadbeef', 'target_snapshot_digest': None}, target, 'large.py')",
+ " excerpt = excerpts.scanned_source_text({'target_revision': 'deadbeef', 'target_snapshot_digest': None, 'diff_target_kind': None}, target, 'large.py')",
" hashes = finalizer._github_line_hashes(io.StringIO('line\\n' * 100001), {100001})",
" print(json.dumps({'documentBytes': len(document), 'sourceBytes': len(excerpt), 'lateSourceLine': 100001 in hashes, 'unsafePathRejected': excerpts.safe_source_path(target, '../outside') is None}))",
].join("\n");