refactor(src): コードベースの包括的リファクタリング（7フェーズ）#79
Merged
HappyOnigiri merged 18 commits intomainfrom Mar 13, 2026
Merged
Conversation
Contributor
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: Path: .coderabbit.yaml Review profile: CHILL Plan: Pro Run ID: 📒 Files selected for processing (2)
📝 WalkthroughWalkthroughPRContextデータクラスを導入し、PR処理をCI-fix／merge／review-fixのフェーズへ分割。サブプロセス呼び出しを Changes
Estimated code review effort🎯 4 (Complex) | ⏱️ ~60 minutes Possibly related PRs
Suggested labels
🚥 Pre-merge checks | ✅ 2 | ❌ 1❌ Failed checks (1 warning)
✅ Passed checks (2 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches
🧪 Generate unit tests (beta)
📝 Coding Plan
Comment |
…e-codebase-refactoring
HappyOnigiri
added
the
refix: running
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Actionable comments posted: 5
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (2)
src/pr_label.py (1)
158-175:⚠️ Potential issue | 🟠 Major
edit_pr_label()が一時的なgh障害で例外終了します。この関数は失敗時に warning を出して
Falseを返す設計ですが、run_command(cmd, check=False)への置き換えで timeout / 起動失敗がそのまま例外で抜けるようになっています。ラベル操作の一時障害で PR 処理全体が中断するので、ここでも例外を warning +Falseに畳んだほうが安全です。🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@src/pr_label.py` around lines 158 - 175, The edit_pr_label function currently calls run_command(cmd, check=False) but doesn't catch exceptions (e.g., timeout, OSError), so transient gh failures can bubble up; wrap the run_command call in a try/except around run_command(cmd, check=False) inside edit_pr_label, catch broad exceptions (at least Exception) and on exception log a warning similar to the existing print(message, file=sys.stderr) including the exception text, then return False; keep the existing success/stderr handling unchanged.src/coderabbit.py (1)
278-298:⚠️ Potential issue | 🟠 Major
_post_issue_comment()がFalseではなく例外で落ちる経路を持っています。この関数は投稿失敗を warning +
Falseで返す前提ですが、run_command(..., check=False)は timeout 時に例外を投げます。auto-resume コメント投稿の一時障害で全体処理まで落ちるので、ここで例外を握ってFalseに落とす必要があります。🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@src/coderabbit.py` around lines 278 - 298, The _post_issue_comment() path currently calls run_command(..., check=False) but doesn't catch exceptions (e.g., timeouts) so failures can raise instead of returning False; wrap the run_command call in a try/except that catches the exception types run_command may raise (broad Exception if specific types are not available), log the exception detail to sys.stderr or process logger just like the existing warning, and return False on exception; ensure you reference the _post_issue_comment function and the run_command invocation so reviewers can find and update that block to swallow errors and preserve the current False-return behavior.
🧹 Nitpick comments (3)
src/ci_check.py (1)
148-153:Exceptionのキャッチが広すぎる可能性
run_commandはタイムアウト時にSubprocessErrorを送出します（subprocess_helpers.py参照）。ここでExceptionをキャッチすると、予期しないエラー（例:KeyboardInterrupt以外のプログラミングエラー）も静かに握りつぶしてしまう可能性があります。
SubprocessErrorを明示的にキャッチすることを検討してください。♻️ 提案する修正
+from errors import SubprocessError + # ... try: run_view_result = run_command( ["gh", "run", "view", run_id, "--repo", repo, "--log-failed"], check=False, timeout=60, ) - except Exception: + except SubprocessError: print( f"Warning: timed out fetching CI logs for run {run_id}; skipping", file=sys.stderr, ) continue🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@src/ci_check.py` around lines 148 - 153, Replace the overly broad "except Exception:" that swallows all errors with an explicit "except SubprocessError:" to only handle the timeout/ subprocess failure raised by run_command; ensure SubprocessError is imported or referenced correctly (it's the builtin SubprocessError) and let all other exceptions propagate (so KeyboardInterrupt and programming errors aren't silently ignored) — update the except block around the run_command/CI-log fetch logic in ci_check.py accordingly.tests/test_ci_check.py (1)
188-203:pr_reviewer.run_commandのパッチは不要に見えます。
are_all_ci_checks_successfulはci_checkモジュール内のrun_commandのみを使用しており、pr_reviewer.run_commandのパッチは効果がありません。テストは正常に動作しますが、不要なモックは削除を検討してください。♻️ 不要なパッチの削除
def test_empty_checks_ci_empty_as_success_false_returns_false(self): - with ( - patch("ci_check.run_command") as mock_run, - patch( - "pr_reviewer.run_command", - return_value=Mock(returncode=0, stdout="{}", stderr=""), - ), - ): + with patch("ci_check.run_command") as mock_run: mock_run.side_effect = [同様のパターンが
test_empty_checks_commit_old_treats_as_success、test_empty_checks_commit_recent_returns_false、test_non_empty_checks_all_success_returns_trueにも存在します。🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@tests/test_ci_check.py` around lines 188 - 203, Remove the unnecessary patching of pr_reviewer.run_command in the tests that target ci_check.are_all_ci_checks_successful: keep only the patch for ci_check.run_command (mock_run) since are_all_ci_checks_successful calls run_command inside the ci_check module; update the three affected tests (the one shown plus test_empty_checks_commit_old_treats_as_success and test_empty_checks_commit_recent_returns_false and test_non_empty_checks_all_success_returns_true) to delete the patch(...) block for "pr_reviewer.run_command" and any related return_value setup so the tests rely solely on mock_run.side_effect and mock_run.call_count assertions.src/auto_fixer.py (1)
115-136: 型アノテーションの改善を検討してください。
works_dirとreports_dirにAnyを使用していますが、実際の型はPath | Noneです。より正確な型アノテーションにすることで、静的解析ツールの恩恵を受けられます。♻️ 型アノテーションの改善
- works_dir: Any # Path + works_dir: Path | None labels: list[dict] ... - reports_dir: Any # Path | None + reports_dir: Path | None🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@src/auto_fixer.py` around lines 115 - 136, Update the type annotations for works_dir and reports_dir from Any to the concrete Path | None type: import Path from pathlib if not already imported and change the annotations for the variables named works_dir and reports_dir to Path | None (or Optional[Path] if your codebase prefers typing.Optional). Ensure the new import doesn't conflict with existing names and run static type checks to confirm no further adjustments are needed for usages of works_dir and reports_dir.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@src/config.py`:
- Around line 375-379: The call to run_command inside expand_repositories can
raise exceptions (e.g., TimeoutError, SubprocessError, OSError) which are not
being converted into ConfigError; wrap the run_command(cmd, check=False) call in
a try/except that catches those exceptions and re-raises a ConfigError including
the exception text and repo_name, and still handle the non-zero return case as
you do now (inspect result.returncode and raise ConfigError with stderr);
reference the expand_repositories function, the run_command call, and the
ConfigError class when making this change.
- Around line 91-105: The int parsing currently uses int(value) which silently
accepts floats and some booleans; fix by explicitly rejecting bools and floats
before conversion: if isinstance(value, bool): if spec.reject_bool: raise
ConfigError(...) else proceed with the existing bool-handling path (or treat as
invalid per spec) and if isinstance(value, float): raise ConfigError(f"{key}
must be an integer, not a float."); perform these checks (referencing value,
spec.reject_bool, spec.min_value, int_value, clamp, ConfigError, key)
immediately before the try/except that calls int(value) so only truly
integer-compatible inputs reach int().
In `@src/pr_reviewer.py`:
- Around line 46-48: Calls to run_command (e.g., in _fetch_check_runs_via_rest
and other functions that currently check result.returncode != 0 and return
fallback values) can raise SubprocessError on timeout and bypass the intended
fallback; wrap each run_command invocation in a try/cept SubprocessError block,
log or ignore the exception as appropriate, and return the existing contract
value (empty list, empty dict, or False) on exception instead of letting it
propagate; specifically update the call sites that currently do "result =
run_command(...); if result.returncode != 0: return X" so they become a try:
result = run_command(...) ... except SubprocessError: return X, preserving the
same return value X used for non-zero return codes.
In `@src/prompt_builder.py`:
- Around line 92-95: 現在のループ only appends when comment.get("id") is truthy, so
comments that only have "_state_comment_id" are skipped; change the logic in the
loop over comments to first obtain the resolved id via
inline_comment_state_id(comment) (which accepts "_state_comment_id"), check that
the returned id is non-empty, and append that to target_ids; refer to the
variables/comments and the inline_comment_state_id function to implement this
replacement so restored inline comments are included for summarization instead
of falling back to raw body.
In `@src/subprocess_helpers.py`:
- Around line 29-45: The run_command function currently only wraps
subprocess.TimeoutExpired into SubprocessError, leaving
FileNotFoundError/OSError from subprocess.run unhandled; update the try/except
around subprocess.run in run_command to also catch FileNotFoundError and OSError
and re-raise them as SubprocessError (same shape as the TimeoutExpired branch:
provide a clear message including the command name, set returncode=-1 and
stderr="" and chain the original exception), ensuring existing TimeoutExpired
handling remains and that check/result logic after the call is unchanged.
---
Outside diff comments:
In `@src/coderabbit.py`:
- Around line 278-298: The _post_issue_comment() path currently calls
run_command(..., check=False) but doesn't catch exceptions (e.g., timeouts) so
failures can raise instead of returning False; wrap the run_command call in a
try/except that catches the exception types run_command may raise (broad
Exception if specific types are not available), log the exception detail to
sys.stderr or process logger just like the existing warning, and return False on
exception; ensure you reference the _post_issue_comment function and the
run_command invocation so reviewers can find and update that block to swallow
errors and preserve the current False-return behavior.
In `@src/pr_label.py`:
- Around line 158-175: The edit_pr_label function currently calls
run_command(cmd, check=False) but doesn't catch exceptions (e.g., timeout,
OSError), so transient gh failures can bubble up; wrap the run_command call in a
try/except around run_command(cmd, check=False) inside edit_pr_label, catch
broad exceptions (at least Exception) and on exception log a warning similar to
the existing print(message, file=sys.stderr) including the exception text, then
return False; keep the existing success/stderr handling unchanged.
---
Nitpick comments:
In `@src/auto_fixer.py`:
- Around line 115-136: Update the type annotations for works_dir and reports_dir
from Any to the concrete Path | None type: import Path from pathlib if not
already imported and change the annotations for the variables named works_dir
and reports_dir to Path | None (or Optional[Path] if your codebase prefers
typing.Optional). Ensure the new import doesn't conflict with existing names and
run static type checks to confirm no further adjustments are needed for usages
of works_dir and reports_dir.
In `@src/ci_check.py`:
- Around line 148-153: Replace the overly broad "except Exception:" that
swallows all errors with an explicit "except SubprocessError:" to only handle
the timeout/ subprocess failure raised by run_command; ensure SubprocessError is
imported or referenced correctly (it's the builtin SubprocessError) and let all
other exceptions propagate (so KeyboardInterrupt and programming errors aren't
silently ignored) — update the except block around the run_command/CI-log fetch
logic in ci_check.py accordingly.
In `@tests/test_ci_check.py`:
- Around line 188-203: Remove the unnecessary patching of
pr_reviewer.run_command in the tests that target
ci_check.are_all_ci_checks_successful: keep only the patch for
ci_check.run_command (mock_run) since are_all_ci_checks_successful calls
run_command inside the ci_check module; update the three affected tests (the one
shown plus test_empty_checks_commit_old_treats_as_success and
test_empty_checks_commit_recent_returns_false and
test_non_empty_checks_all_success_returns_true) to delete the patch(...) block
for "pr_reviewer.run_command" and any related return_value setup so the tests
rely solely on mock_run.side_effect and mock_run.call_count assertions.
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
Run ID: 6da8a085-1128-4e87-b38a-35a90ced5ee7
📒 Files selected for processing (30)
src/auto_fixer.pysrc/ci_check.pysrc/ci_log.pysrc/claude_runner.pysrc/coderabbit.pysrc/config.pysrc/errors.pysrc/git_ops.pysrc/github_pr_fetcher.pysrc/pr_label.pysrc/pr_reviewer.pysrc/prompt_builder.pysrc/report.pysrc/state_manager.pysrc/subprocess_helpers.pysrc/summarizer.pytests/test_auto_fixer.pytests/test_ci_check.pytests/test_ci_log.pytests/test_claude_runner.pytests/test_coderabbit.pytests/test_config.pytests/test_git_ops.pytests/test_github_pr_fetcher.pytests/test_pr_label.pytests/test_pr_reviewer.pytests/test_prompt_builder.pytests/test_report.pytests/test_state_manager.pytests/test_summarizer.py
Owner
Author
🤖 Refix Status対応済みレビュー一覧
実行レポートCI 修正2026-03-13 07:10:00 UTC src/config.py,src/pr_reviewer.py CI formatter diff の修正開始
2026-03-13 07:10:30 UTC src/config.py,src/pr_reviewer.py 修正完了・push済み2ファイルの formatter 差分を修正し、commit ( |
- config.py: raise ConfigError を1行に折りたたむ - pr_reviewer.py: print() を複数行に展開する
Contributor
There was a problem hiding this comment.
Actionable comments posted: 1
♻️ Duplicate comments (1)
src/config.py (1)
92-100:⚠️ Potential issue | 🟠 Majorint フィールドで
boolが通ってしまいます。Line 94 は
spec.reject_boolがTrueの場合しかboolを拒否しないため、ci_log_max_lines: trueのような誤設定がint(True)経由で受理されます（結果は clamp で 20）。設定ミスを検知できないので、int フィールドでは常にboolを拒否すべきです。修正例
def _validate_scalar_field(key: str, value: Any, spec: FieldSpec) -> Any: @@ - if isinstance(value, float): + if isinstance(value, bool): + raise ConfigError(f"{key} must be an integer.") + if isinstance(value, float): raise ConfigError(f"{key} must be an integer.") - if spec.reject_bool and isinstance(value, bool): - min_str = f">= {spec.min_value}" if spec.min_value is not None else "" - raise ConfigError( - f"{key} must be a non-negative integer{(' ' + min_str) if min_str else ''}." - )🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@src/config.py` around lines 92 - 100, The int-field validation currently only rejects bool when spec.reject_bool is True, allowing values like ci_log_max_lines: true to be accepted via int(True); update the int validation in the block that handles numeric ints (the code around value, key, spec.reject_bool, int_value and clamp) to always treat booleans as invalid by checking isinstance(value, bool) first and raising a ConfigError with the same message used for non-negative integer rejection, before attempting int(value) conversion.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@src/pr_label.py`:
- Around line 56-57: run_command が timeout/起動失敗で SubprocessError
を投げるため、get_result = run_command(get_cmd, check=False)
を直接参照すると例外で処理が中断します。該当する呼び出し箇所（get_result/get_cmd を使う箇所、同パターンがある行群）を try/except
SubprocessError で包み、例外時は returncode を非ゼロにしたダミー結果オブジェクト（例:
SimpleNamespace(returncode=1, stdout="", stderr=str(err))）を代入して既存の returncode
ベースの分岐（if get_result.returncode == 0 / False 判定）にフォールバックするようにしてください。対象シンボル:
run_command, get_result, get_cmd（および同様に使われている変数名がある他の呼び出し箇所）。
---
Duplicate comments:
In `@src/config.py`:
- Around line 92-100: The int-field validation currently only rejects bool when
spec.reject_bool is True, allowing values like ci_log_max_lines: true to be
accepted via int(True); update the int validation in the block that handles
numeric ints (the code around value, key, spec.reject_bool, int_value and clamp)
to always treat booleans as invalid by checking isinstance(value, bool) first
and raising a ConfigError with the same message used for non-negative integer
rejection, before attempting int(value) conversion.
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
Run ID: 5c5147e0-2083-43bb-b1e1-c42b3925ec42
📒 Files selected for processing (7)
src/ci_check.pysrc/coderabbit.pysrc/config.pysrc/pr_label.pysrc/pr_reviewer.pysrc/prompt_builder.pysrc/subprocess_helpers.py
🚧 Files skipped from review as they are similar to previous changes (1)
- src/prompt_builder.py
Contributor
There was a problem hiding this comment.
♻️ Duplicate comments (1)
src/config.py (1)
92-101:⚠️ Potential issue | 🟠 Majorbool 値が整数設定として受理される経路が残っています。
Line 94-101 では
spec.reject_bool=Falseの整数フィールドでTrue/Falseがint()により1/0として通過します（例:ci_log_max_lines: true）。設定ミスを静かに受理するため、整数フィールドでは bool を明示的に拒否してください。修正案
def _validate_scalar_field(key: str, value: Any, spec: FieldSpec) -> Any: """スカラーフィールドを FieldSpec に従って検証・変換する。""" if spec.type_ is bool: if not isinstance(value, bool): raise ConfigError(f"{key} must be a boolean.") return value # int field if isinstance(value, float): raise ConfigError(f"{key} must be an integer.") - if spec.reject_bool and isinstance(value, bool): - min_str = f">= {spec.min_value}" if spec.min_value is not None else "" - raise ConfigError( - f"{key} must be a non-negative integer{(' ' + min_str) if min_str else ''}." - ) + if isinstance(value, bool): + if spec.reject_bool: + min_str = f">= {spec.min_value}" if spec.min_value is not None else "" + raise ConfigError( + f"{key} must be a non-negative integer{(' ' + min_str) if min_str else ''}." + ) + raise ConfigError(f"{key} must be an integer.") try: int_value = int(value)#!/bin/bash # Verify current branch conditions and Python bool-to-int coercion. rg -n "ci_log_max_lines\": FieldSpec\\(int" src/config.py sed -n '85,105p' src/config.py python - <<'PY' v = True print("isinstance(v, float):", isinstance(v, float)) print("isinstance(v, bool):", isinstance(v, bool)) print("int(v):", int(v)) PY期待結果:
ci_log_max_linesが bool 拒否設定なしであること、およびint(True) == 1が確認され、現行ロジックで bool が整数化され得ることを再現できます。🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@src/config.py` around lines 92 - 101, The validation currently coerces bool to int via int(value); add an explicit check for isinstance(value, bool) before the int() conversion and raise ConfigError referencing the same message style (use key and spec.min_value if present) so boolean values are rejected for integer fields; modify the block that uses spec.reject_bool / int_value (symbols: spec, key, value, int_value) to raise when isinstance(value, bool) rather than allowing int(True)/int(False) to pass.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Duplicate comments:
In `@src/config.py`:
- Around line 92-101: The validation currently coerces bool to int via
int(value); add an explicit check for isinstance(value, bool) before the int()
conversion and raise ConfigError referencing the same message style (use key and
spec.min_value if present) so boolean values are rejected for integer fields;
modify the block that uses spec.reject_bool / int_value (symbols: spec, key,
value, int_value) to raise when isinstance(value, bool) rather than allowing
int(True)/int(False) to pass.
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
Run ID: cdd0223d-0f57-43b5-bbe9-b2bee9fc1ac3
📒 Files selected for processing (2)
src/config.pysrc/pr_reviewer.py
🚧 Files skipped from review as they are similar to previous changes (1)
- src/pr_reviewer.py
reject_bool フラグの有無に関わらず bool 値を拒否するよう修正。 これにより ci_log_max_lines: true のような値が int() 変換で 暗黙的に受理されなくなる。
タイムアウトや起動失敗時に SubprocessError が送出されると returncode ベースのフォールバック設計が崩れるため、 _ensure_repo_label_exists・_mark_pr_merged_label_if_needed・ backfill_merged_labels・_trigger_pr_auto_merge の各呼び出しを try/except SubprocessError で包んで安全に処理する。
Contributor
There was a problem hiding this comment.
Actionable comments posted: 2
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@src/config.py`:
- Around line 176-180: load_config currently only catches FileNotFoundError when
calling Path(filepath).read_text, so other I/O errors (e.g.,
PermissionError/OSError) leak; update the try/except around
Path(filepath).read_text(encoding="utf-8") to catch OSError (or a broader
exception covering I/O errors) and re-raise as ConfigError with the original
exception as the cause (i.e., raise ConfigError(f"config file not found:
{filepath}") from exc or a message adjusted for generic I/O errors), referencing
load_config, filepath, and Path.read_text to locate the change.
In `@src/pr_label.py`:
- Around line 56-59: The try/except blocks that call run_command(get_cmd,
check=False) are swallowing SubprocessError and returning False without logging;
change both occurrences (the block around run_command(get_cmd, check=False) and
the similar block at lines ~85-88) to catch the exception as e and write the
exception details to stderr (or use the module logger) before returning False so
the root cause is preserved (reference the run_command call and the
SubprocessError handler in pr_label.py).
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
Run ID: 92e87b71-6df0-4693-a4cc-f2add0a34be5
📒 Files selected for processing (2)
src/config.pysrc/pr_label.py
HappyOnigiri
added
refix: done
HappyOnigiri
added
the
refix: auto-merge-requested
HappyOnigiri
added
refix: merged
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
概要 / Summary
コードベースの包括的リファクタリング（7フェーズ）。subprocessヘルパーの統一・型付き例外の導入・テスト分割・config簡素化・公開API命名整理・
_process_single_pr()の分解・不足テストの追加を実施。変更内容 / Changes
src/errors.pyを新規追加：RefixError→ConfigError/SubprocessError→GitHubAPIError/GitErrorの例外階層を定義src/subprocess_helpers.pyを新規追加：run_command/run_git/run_gh/run_gh_apiをデフォルトタイムアウト付きで提供し、全9モジュールのボイラープレートを統合src/config.pyのsys.exit(1)をConfigError送出に変更し、auto_fixer.main()でキャッチsrc/pr_reviewer.pyのsys.exit(1)をSubprocessError送出に変更_merge_base_branch・_are_all_ci_checks_successful・_emit_runtime_pain_report等30件超）をアンダースコアなしにリネームsrc/config.pyにFieldSpecデータクラスと_SCALAR_FIELDSスキーマを導入し、スカラーフィールド検証を約150行から30行へ圧縮src/auto_fixer.pyの_process_single_pr()（約1,085行・29パラメーター）にPRContextデータクラスを導入し、_fetch_pr_context/_handle_coderabbit_status/_run_ci_fix_phase/_run_merge_phase/_run_review_fix_phaseの5フェーズ関数に分解tests/test_auto_fixer.pyからTestGeneratePrompt/TestLoadConfig/TestSetupClaudeSettings/TestCiFixHelpers/TestCodeRabbitRateLimitHelpers/TestRefixLabeling/TestAreAllCiChecksSuccessful/TestMergeStrategyHelpers/TestRunClaudePrompt/TestExpandRepositories/TestPerRunLimitsConfigを各モジュール専用テストファイルへ分割（11クラス → 新規7ファイル）tests/test_git_ops.py/tests/test_ci_log.py/tests/test_report.pyを新規追加（不足テストのカバレッジ補完）テスト / Testing
make ci（ruff format / ruff check / mypy / pyright / pytest）全チェック通過済み関連 issues / Related issues
なし
Summary by CodeRabbit
New Features
信頼性向上
設定検証
レポート改善
テスト