space/gitea-codex
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

[fix]. Restore PR-scoped review + remove fix cmd
All checks were successful
ci / test (pull_request) Successful in 32s
Details
ci / publish (pull_request) Has been skipped
Details

Browse Source
This commit is contained in:
Space-Banane
2026-05-23 14:15:00 +02:00
parent 9392591429
commit 08075cb3c4
14 changed files with 155 additions and 131 deletions
Download Patch File Download Diff File Expand all files Collapse all files

86
tests/test_container_runner.py
View File

@@ -6,11 +6,13 @@ import pytest
from gitea_codex_bot.config import get_settings
from gitea_codex_bot.services.gitea import PullRequestContext
from gitea_codex_bot.services.repo_config import RepoReviewConfig
from gitea_codex_bot.types import ParsedCommand
from gitea_codex_bot.workers.container_runner import (
CONTAINER_CODEX_HOME,
RESULT_END_MARKER,
RESULT_START_MARKER,
_apply_repo_default_review_mode,
_build_docker_command,
_build_exec_review_prompt,
_build_install_and_run_command,
@@ -96,7 +98,13 @@ def test_build_install_command_chatgpt_mode_sets_git_checkout_and_review(monkeyp
settings = get_settings()
pr = _sample_pr()
command = _build_install_and_run_command(settings, pr=pr, review_prompt="review: security --full")
command = _build_install_and_run_command(
settings,
pr=pr,
review_prompt="review: security --full",
result_start_marker=f"{RESULT_START_MARKER}_x",
result_end_marker=f"{RESULT_END_MARKER}_x",
)
assert 'printf "%s" "$CODEX_AUTH_JSON_B64" | base64 -d > /root/.codex/auth.json' in command
assert "git -c http.extraHeader=" in command
@@ -115,15 +123,21 @@ def test_build_install_command_chatgpt_mode_sets_git_checkout_and_review(monkeyp
assert "npm install -g @openai/codex@latest" in command
assert "codex --version >/tmp/codex-version.log" in command
assert " - " not in command
assert f'echo "{RESULT_START_MARKER}"' in command
assert f'echo "{RESULT_END_MARKER}"' in command
assert f'echo "{RESULT_START_MARKER}_x"' in command
assert f'echo "{RESULT_END_MARKER}_x"' in command
def test_build_install_command_does_not_include_reasoning_effort_flag() -> None:
settings = get_settings()
pr = _sample_pr()
command = _build_install_and_run_command(settings, pr=pr, review_prompt="review: tests")
command = _build_install_and_run_command(
settings,
pr=pr,
review_prompt="review: tests",
result_start_marker=f"{RESULT_START_MARKER}_x",
result_end_marker=f"{RESULT_END_MARKER}_x",
)
assert "--reasoning-effort" not in command
@@ -132,7 +146,13 @@ def test_build_install_command_uses_upstream_remote_for_fork_pr_base_fetch() ->
settings = get_settings()
pr = _sample_fork_pr()
command = _build_install_and_run_command(settings, pr=pr, review_prompt="review: tests")
command = _build_install_and_run_command(
settings,
pr=pr,
review_prompt="review: tests",
result_start_marker=f"{RESULT_START_MARKER}_x",
result_end_marker=f"{RESULT_END_MARKER}_x",
)
assert "base_remote=upstream" in command
assert f"git remote add upstream {pr.base_clone_url}" in command
@@ -271,6 +291,10 @@ def test_run_review_ephemeral_single_attempt_success(monkeypatch: pytest.MonkeyP
return None
monkeypatch.setattr("gitea_codex_bot.workers.container_runner.GiteaClient", _FakeGiteaClient)
monkeypatch.setattr(
"gitea_codex_bot.workers.container_runner.uuid.uuid4",
lambda: type("U", (), {"hex": "abc123def4567890abc123def4567890"})(),
)
calls: list[list[str]] = []
def _fake_run(cmd, *args, **kwargs):
@@ -282,9 +306,9 @@ def test_run_review_ephemeral_single_attempt_success(monkeypatch: pytest.MonkeyP
"returncode": 0,
"stdout": (
'{"type":"response.started","model":"gpt-5.3-codex"}\n'
f"{RESULT_START_MARKER}\n"
f"{RESULT_START_MARKER}_abc123def4567890abc123def4567890\n"
'{"verdict":"correct","confidence":0.9,"summary":"ok","findings":[],"markdown_comment":"ok"}\n'
f"{RESULT_END_MARKER}\n"
f"{RESULT_END_MARKER}_abc123def4567890abc123def4567890\n"
),
"stderr": "",
},
@@ -308,30 +332,62 @@ def test_run_review_ephemeral_single_attempt_success(monkeypatch: pytest.MonkeyP
def test_parse_review_result_from_stdout_artifact() -> None:
stdout = (
"noise\n"
f"{RESULT_START_MARKER}\n"
f"{RESULT_START_MARKER}_test\n"
'{"verdict":"correct","confidence":0.9,"summary":"ok","findings":[],"markdown_comment":"ok"}\n'
f"{RESULT_END_MARKER}\n"
f"{RESULT_END_MARKER}_test\n"
)
parsed = _parse_review_result_from_stdout_artifact(
stdout,
result_start_marker=f"{RESULT_START_MARKER}_test",
result_end_marker=f"{RESULT_END_MARKER}_test",
)
parsed = _parse_review_result_from_stdout_artifact(stdout)
assert parsed["verdict"] == "correct"
assert parsed["summary"] == "ok"
def test_parse_review_result_from_stdout_artifact_fails_without_markers() -> None:
with pytest.raises(RuntimeError):
_parse_review_result_from_stdout_artifact("no markers here")
_parse_review_result_from_stdout_artifact(
"no markers here",
result_start_marker=f"{RESULT_START_MARKER}_x",
result_end_marker=f"{RESULT_END_MARKER}_x",
)
def test_build_exec_review_prompt_strips_mention_and_command() -> None:
prompt = _build_exec_review_prompt(
ParsedCommand(name="review", raw="@codex review security --full\nfocus session handling")
ParsedCommand(name="review", raw="@codex review security --full\nfocus session handling"),
RepoReviewConfig(),
_sample_pr(),
)
assert prompt == "review: security --full\nfocus session handling"
assert prompt.startswith("review: security --full\nfocus session handling")
assert "Compare exactly these commits:" in prompt
def test_build_exec_review_prompt_falls_back_when_no_extra_text() -> None:
prompt = _build_exec_review_prompt(ParsedCommand(name="rerun", raw="@codex rerun"))
assert prompt == "review: review this pull request and report introduced issues."
prompt = _build_exec_review_prompt(ParsedCommand(name="rerun", raw="@codex rerun"), RepoReviewConfig(), _sample_pr())
assert prompt.startswith("review: review this pull request and report introduced issues.")
def test_apply_repo_default_review_mode_for_review_command() -> None:
command = ParsedCommand(name="review", raw="@codex review")
cfg = RepoReviewConfig(default_mode="tests")
_apply_repo_default_review_mode(command, cfg)
assert command.mode == "tests"
def test_parse_review_result_from_stdout_artifact_uses_end_marker_after_start() -> None:
stdout = (
f"{RESULT_START_MARKER}_a\n"
'{"verdict":"correct","confidence":0.9,"summary":"contains marker text __CODEX_REVIEW_RESULT_END___a","findings":[],"markdown_comment":"ok"}\n'
f"{RESULT_END_MARKER}_a\n"
)
parsed = _parse_review_result_from_stdout_artifact(
stdout,
result_start_marker=f"{RESULT_START_MARKER}_a",
result_end_marker=f"{RESULT_END_MARKER}_a",
)
assert parsed["verdict"] == "correct"
def test_extract_result_meta_from_codex_stdout_collects_model_and_usage() -> None: