fix(codex-cli): improve ripgrep download diagnostics (#8486)

When rg download fails during npm package staging, log the
target/platform/url and preserve the original exception as the cause.
Emit GitHub Actions log groups and error annotations so the failure is
easier to spot.

Document why a urlopen timeout is set (the default can hang
indefinitely).

This is to make failures in the specific build step easier to understand
/ work out what's failing rather than having a big wall of text (or at
least having an obvious part of it that helps narrow that wall)

---------

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
This commit is contained in:
Josh McKinney
2025-12-23 13:18:35 -08:00
committed by GitHub
parent 0130a2fa40
commit f2b740c95d

View File

@@ -2,6 +2,7 @@
"""Install Codex native binaries (Rust CLI plus ripgrep helpers)."""
import argparse
from contextlib import contextmanager
import json
import os
import shutil
@@ -12,6 +13,7 @@ import zipfile
from dataclasses import dataclass
from concurrent.futures import ThreadPoolExecutor, as_completed
from pathlib import Path
import sys
from typing import Iterable, Sequence
from urllib.parse import urlparse
from urllib.request import urlopen
@@ -77,6 +79,45 @@ RG_TARGET_PLATFORM_PAIRS: list[tuple[str, str]] = [
RG_TARGET_TO_PLATFORM = {target: platform for target, platform in RG_TARGET_PLATFORM_PAIRS}
DEFAULT_RG_TARGETS = [target for target, _ in RG_TARGET_PLATFORM_PAIRS]
# urllib.request.urlopen() defaults to no timeout (can hang indefinitely), which is painful in CI.
DOWNLOAD_TIMEOUT_SECS = 60
def _gha_enabled() -> bool:
# GitHub Actions supports "workflow commands" (e.g. ::group:: / ::error::) that make logs
# much easier to scan: groups collapse noisy sections and error annotations surface the
# failure in the UI without changing the actual exception/traceback output.
return os.environ.get("GITHUB_ACTIONS") == "true"
def _gha_escape(value: str) -> str:
# Workflow commands require percent/newline escaping.
return value.replace("%", "%25").replace("\r", "%0D").replace("\n", "%0A")
def _gha_error(*, title: str, message: str) -> None:
# Emit a GitHub Actions error annotation. This does not replace stdout/stderr logs; it just
# adds a prominent summary line to the job UI so the root cause is easier to spot.
if not _gha_enabled():
return
print(
f"::error title={_gha_escape(title)}::{_gha_escape(message)}",
flush=True,
)
@contextmanager
def _gha_group(title: str):
# Wrap a block in a collapsible log group on GitHub Actions. Outside of GHA this is a no-op
# so local output remains unchanged.
if _gha_enabled():
print(f"::group::{_gha_escape(title)}", flush=True)
try:
yield
finally:
if _gha_enabled():
print("::endgroup::", flush=True)
def parse_args() -> argparse.Namespace:
parser = argparse.ArgumentParser(description="Install native Codex binaries.")
@@ -131,18 +172,20 @@ def main() -> int:
workflow_id = workflow_url.rstrip("/").split("/")[-1]
print(f"Downloading native artifacts from workflow {workflow_id}...")
with tempfile.TemporaryDirectory(prefix="codex-native-artifacts-") as artifacts_dir_str:
artifacts_dir = Path(artifacts_dir_str)
_download_artifacts(workflow_id, artifacts_dir)
install_binary_components(
artifacts_dir,
vendor_dir,
[BINARY_COMPONENTS[name] for name in components if name in BINARY_COMPONENTS],
)
with _gha_group(f"Download native artifacts from workflow {workflow_id}"):
with tempfile.TemporaryDirectory(prefix="codex-native-artifacts-") as artifacts_dir_str:
artifacts_dir = Path(artifacts_dir_str)
_download_artifacts(workflow_id, artifacts_dir)
install_binary_components(
artifacts_dir,
vendor_dir,
[BINARY_COMPONENTS[name] for name in components if name in BINARY_COMPONENTS],
)
if "rg" in components:
print("Fetching ripgrep binaries...")
fetch_rg(vendor_dir, DEFAULT_RG_TARGETS, manifest_path=RG_MANIFEST)
with _gha_group("Fetch ripgrep binaries"):
print("Fetching ripgrep binaries...")
fetch_rg(vendor_dir, DEFAULT_RG_TARGETS, manifest_path=RG_MANIFEST)
print(f"Installed native dependencies into {vendor_dir}")
return 0
@@ -203,7 +246,14 @@ def fetch_rg(
for future in as_completed(future_map):
target = future_map[future]
results[target] = future.result()
try:
results[target] = future.result()
except Exception as exc:
_gha_error(
title="ripgrep install failed",
message=f"target={target} error={exc!r}",
)
raise RuntimeError(f"Failed to install ripgrep for target {target}.") from exc
print(f" installed ripgrep for {target}")
return [results[target] for target in targets]
@@ -301,6 +351,8 @@ def _fetch_single_rg(
url = providers[0]["url"]
archive_format = platform_info.get("format", "zst")
archive_member = platform_info.get("path")
digest = platform_info.get("digest")
expected_size = platform_info.get("size")
dest_dir = vendor_dir / target / "path"
dest_dir.mkdir(parents=True, exist_ok=True)
@@ -313,10 +365,32 @@ def _fetch_single_rg(
tmp_dir = Path(tmp_dir_str)
archive_filename = os.path.basename(urlparse(url).path)
download_path = tmp_dir / archive_filename
_download_file(url, download_path)
print(
f" downloading ripgrep for {target} ({platform_key}) from {url}",
flush=True,
)
try:
_download_file(url, download_path)
except Exception as exc:
_gha_error(
title="ripgrep download failed",
message=f"target={target} platform={platform_key} url={url} error={exc!r}",
)
raise RuntimeError(
"Failed to download ripgrep "
f"(target={target}, platform={platform_key}, format={archive_format}, "
f"expected_size={expected_size!r}, digest={digest!r}, url={url}, dest={download_path})."
) from exc
dest.unlink(missing_ok=True)
extract_archive(download_path, archive_format, archive_member, dest)
try:
extract_archive(download_path, archive_format, archive_member, dest)
except Exception as exc:
raise RuntimeError(
"Failed to extract ripgrep "
f"(target={target}, platform={platform_key}, format={archive_format}, "
f"member={archive_member!r}, url={url}, archive={download_path})."
) from exc
if not is_windows:
dest.chmod(0o755)
@@ -326,7 +400,9 @@ def _fetch_single_rg(
def _download_file(url: str, dest: Path) -> None:
dest.parent.mkdir(parents=True, exist_ok=True)
with urlopen(url) as response, open(dest, "wb") as out:
dest.unlink(missing_ok=True)
with urlopen(url, timeout=DOWNLOAD_TIMEOUT_SECS) as response, open(dest, "wb") as out:
shutil.copyfileobj(response, out)