20569549039

Committed 29 Dec 2025 09:30AM UTC coverage: 65.82% (+0.008%) from 65.812%

Build # 20569549039

Build Type

push

github

Committed by

kimata

Commit Message

fix: プライベート関数の呼び出しを修正

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

Coverage Stats

4 of 4 new or added lines in 1 file covered. (100.0%)

160 existing lines in 1 file now uncovered.

2902 of 4409 relevant lines covered (65.82%)

0.66 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

42.64

/src/my_lib/selenium_util.py

#!/usr/bin/env python3
"""
Selenium を Chrome Driver を使って動かします。

Usage:
  selenium_util.py [-c CONFIG] [-D]

Options:
  -c CONFIG         : CONFIG を設定ファイルとして読み込んで実行します。[default: tests/data/config.example.yaml]
  -D                : デバッグモードで動作します。
"""

from __future__ import annotations

import datetime
import inspect
import json
import logging
import os
import pathlib
import random
import re
import shutil
import signal
import sqlite3
import subprocess
import time
import io
from collections.abc import Callable
from dataclasses import dataclass
from typing import TYPE_CHECKING, Any

import PIL.Image
import psutil
import selenium
import selenium.common.exceptions
import selenium.webdriver.chrome.options
import selenium.webdriver.chrome.service
import selenium.webdriver.common.action_chains
import selenium.webdriver.common.by
import selenium.webdriver.common.keys
import selenium.webdriver.support.expected_conditions
import undetected_chromedriver

if TYPE_CHECKING:
    from selenium.webdriver.remote.webdriver import WebDriver
    from selenium.webdriver.support.wait import WebDriverWait

WAIT_RETRY_COUNT: int = 1


class SeleniumError(Exception):
    """Selenium 関連エラーの基底クラス"""


def _get_chrome_version() -> int | None:
    try:
        result = subprocess.run(
            ["google-chrome", "--version"],
            capture_output=True,
            text=True,
            timeout=10,
        )
        match = re.search(r"(\d+)\.", result.stdout)
        if match:
            return int(match.group(1))
    except Exception:
        logging.warning("Failed to detect Chrome version")
    return None


def _create_driver_impl(
    profile_name: str,
    data_path: pathlib.Path,
    is_headless: bool,
    use_subprocess: bool = True,
) -> WebDriver:  # noqa: ARG001
    chrome_data_path = data_path / "chrome"
    log_path = data_path / "log"

    # NOTE: Pytest を並列実行できるようにする
    suffix = os.environ.get("PYTEST_XDIST_WORKER", None)
    if suffix is not None:
        profile_name += "." + suffix

    chrome_data_path.mkdir(parents=True, exist_ok=True)
    log_path.mkdir(parents=True, exist_ok=True)

    options = selenium.webdriver.chrome.options.Options()

    if is_headless:
        options.add_argument("--headless=new")

    options.add_argument("--no-sandbox")  # for Docker
    options.add_argument("--disable-dev-shm-usage")  # for Docker
    options.add_argument("--disable-gpu")

    options.add_argument("--disable-popup-blocking")
    options.add_argument("--disable-plugins")

    options.add_argument("--no-first-run")

    options.add_argument("--lang=ja-JP")
    options.add_argument("--window-size=1920,1080")

    # NOTE: Accept-Language ヘッダーを日本語優先に設定
    options.add_experimental_option("prefs", {"intl.accept_languages": "ja-JP,ja,en-US,en"})

    options.add_argument("--user-data-dir=" + str(chrome_data_path / profile_name))

    options.add_argument("--enable-logging")
    options.add_argument("--v=1")

    chrome_log_file = log_path / f"chrome_{profile_name}.log"
    options.add_argument(f"--log-file={chrome_log_file!s}")

    if not is_headless:
        options.add_argument("--auto-open-devtools-for-tabs")

    service = selenium.webdriver.chrome.service.Service(
        service_args=["--verbose", f"--log-path={str(log_path / 'webdriver.log')!s}"],
    )

    chrome_version = _get_chrome_version()

    # NOTE: user_multi_procs=True は既存の chromedriver ファイルが存在することを前提としているため、
    # ファイルが存在しない場合（CI環境の初回実行など）は False にする
    uc_data_path = pathlib.Path("~/.local/share/undetected_chromedriver").expanduser()
    use_multi_procs = uc_data_path.exists() and any(uc_data_path.glob("*chromedriver*"))

    driver = undetected_chromedriver.Chrome(
        service=service,
        options=options,
        use_subprocess=use_subprocess,
        version_main=chrome_version,
        user_multi_procs=use_multi_procs,
    )

    driver.set_page_load_timeout(30)

    return driver


@dataclass
class _ProfileHealthResult:
    """プロファイル健全性チェックの結果"""

    is_healthy: bool
    errors: list[str]
    has_lock_files: bool = False
    has_corrupted_json: bool = False
    has_corrupted_db: bool = False


def _check_json_file(file_path: pathlib.Path) -> str | None:
    """JSON ファイルの整合性をチェック

    Returns:
        エラーメッセージ（正常な場合は None）
    """
    if not file_path.exists():
        return None

    try:
        content = file_path.read_text(encoding="utf-8")
        json.loads(content)
        return None
    except json.JSONDecodeError as e:
        return f"{file_path.name} is corrupted: {e}"
    except Exception as e:
        return f"{file_path.name} read error: {e}"


def _check_sqlite_db(db_path: pathlib.Path) -> str | None:
    """SQLite データベースの整合性をチェック

    Returns:
        エラーメッセージ（正常な場合は None）
    """
    if not db_path.exists():
        return None

    try:
        conn = sqlite3.connect(str(db_path), timeout=5)
        result = conn.execute("PRAGMA integrity_check").fetchone()
        conn.close()
        if result[0] != "ok":
            return f"{db_path.name} database is corrupted: {result[0]}"
        return None
    except sqlite3.DatabaseError as e:
        return f"{db_path.name} database error: {e}"
    except Exception as e:
        return f"{db_path.name} check error: {e}"


def _check_profile_health(profile_path: pathlib.Path) -> _ProfileHealthResult:
    """Chrome プロファイルの健全性をチェック

    Args:
        profile_path: Chrome プロファイルのディレクトリパス

    Returns:
        ProfileHealthResult: チェック結果
    """
    errors: list[str] = []
    has_lock_files = False
    has_corrupted_json = False
    has_corrupted_db = False

    if not profile_path.exists():
        # プロファイルが存在しない場合は健全（新規作成される）
        return _ProfileHealthResult(is_healthy=True, errors=[])

    default_path = profile_path / "Default"

    # 1. ロックファイルのチェック
    lock_files = ["SingletonLock", "SingletonSocket", "SingletonCookie"]
    existing_locks = []
    for lock_file in lock_files:
        lock_path = profile_path / lock_file
        if lock_path.exists() or lock_path.is_symlink():
            existing_locks.append(lock_file)
            has_lock_files = True
    if existing_locks:
        errors.append(f"Lock files exist: {', '.join(existing_locks)}")

    # 2. Local State の JSON チェック
    local_state_error = _check_json_file(profile_path / "Local State")
    if local_state_error:
        errors.append(local_state_error)
        has_corrupted_json = True

    # 3. Preferences の JSON チェック
    if default_path.exists():
        prefs_error = _check_json_file(default_path / "Preferences")
        if prefs_error:
            errors.append(prefs_error)
            has_corrupted_json = True

        # 4. SQLite データベースの整合性チェック
        for db_name in ["Cookies", "History", "Web Data"]:
            db_error = _check_sqlite_db(default_path / db_name)
            if db_error:
                errors.append(db_error)
                has_corrupted_db = True

    is_healthy = len(errors) == 0

    return _ProfileHealthResult(
        is_healthy=is_healthy,
        errors=errors,
        has_lock_files=has_lock_files,
        has_corrupted_json=has_corrupted_json,
        has_corrupted_db=has_corrupted_db,
    )


def _recover_corrupted_profile(profile_path: pathlib.Path) -> bool:
    """破損したプロファイルをバックアップして新規作成を可能にする

    Args:
        profile_path: Chrome プロファイルのディレクトリパス

    Returns:
        bool: リカバリが成功したかどうか
    """
    if not profile_path.exists():
        return True

    # バックアップ先を決定（タイムスタンプ付き）
    timestamp = datetime.datetime.now().strftime("%Y%m%d_%H%M%S")
    backup_path = profile_path.parent / f"{profile_path.name}.corrupted.{timestamp}"

    try:
        shutil.move(str(profile_path), str(backup_path))
        logging.warning(
            "Corrupted profile moved to backup: %s -> %s",
            profile_path,
            backup_path,
        )
        return True
    except Exception as e:
        logging.exception("Failed to backup corrupted profile: %s", e)
        return False


def _cleanup_profile_lock(profile_path: pathlib.Path) -> None:
    """プロファイルのロックファイルを削除する"""
    lock_files = ["SingletonLock", "SingletonSocket", "SingletonCookie"]
    found_locks = []
    for lock_file in lock_files:
        lock_path = profile_path / lock_file
        if lock_path.exists() or lock_path.is_symlink():
            found_locks.append(lock_path)

    if found_locks:
        logging.warning("Profile lock files found: %s", ", ".join(str(p.name) for p in found_locks))
        for lock_path in found_locks:
            try:
                lock_path.unlink()
            except OSError as e:
                logging.warning("Failed to remove lock file %s: %s", lock_path, e)


def _is_running_in_container() -> bool:
    """コンテナ内で実行中かどうかを判定"""
    return os.path.exists("/.dockerenv")


def _cleanup_orphaned_chrome_processes_in_container() -> None:
    """コンテナ内で実行中の場合のみ、残った Chrome プロセスをクリーンアップ

    NOTE: プロセスツリーに関係なくプロセス名で一律終了するのはコンテナ内限定
    """
    if not _is_running_in_container():
        return

    for proc in psutil.process_iter(["pid", "name"]):
        try:
            proc_name = proc.info["name"].lower() if proc.info["name"] else ""
            if "chrome" in proc_name:
                logging.info("Terminating orphaned Chrome process: PID %d", proc.info["pid"])
                os.kill(proc.info["pid"], signal.SIGTERM)
        except (psutil.NoSuchProcess, psutil.AccessDenied, ProcessLookupError, OSError):
            pass
    time.sleep(1)


def create_driver(
    profile_name: str,
    data_path: pathlib.Path,
    is_headless: bool = True,
    clean_profile: bool = False,
    auto_recover: bool = True,
    use_subprocess: bool = True,
) -> WebDriver:
    """Chrome WebDriver を作成する

    Args:
        profile_name: プロファイル名
        data_path: データディレクトリのパス
        is_headless: ヘッドレスモードで起動するか
        clean_profile: 起動前にロックファイルを削除するか
        auto_recover: プロファイル破損時に自動リカバリするか
        use_subprocess: サブプロセスで Chrome を起動するか
    """
    # NOTE: ルートロガーの出力レベルを変更した場合でも Selenium 関係は抑制する
    logging.getLogger("urllib3.connectionpool").setLevel(logging.WARNING)
    logging.getLogger("selenium.webdriver.common.selenium_manager").setLevel(logging.WARNING)
    logging.getLogger("selenium.webdriver.remote.remote_connection").setLevel(logging.WARNING)

    # NOTE: Pytest を並列実行できるようにする
    suffix = os.environ.get("PYTEST_XDIST_WORKER", None)
    actual_profile_name = profile_name + ("." + suffix if suffix is not None else "")
    profile_path = data_path / "chrome" / actual_profile_name

    # プロファイル健全性チェック
    health = _check_profile_health(profile_path)
    if not health.is_healthy:
        logging.warning("Profile health check failed: %s", ", ".join(health.errors))

        if health.has_lock_files and not (health.has_corrupted_json or health.has_corrupted_db):
            # ロックファイルのみの問題なら削除して続行
            logging.info("Cleaning up lock files only")
            _cleanup_profile_lock(profile_path)
        elif auto_recover and (health.has_corrupted_json or health.has_corrupted_db):
            # JSON または DB が破損している場合はプロファイルをリカバリ
            logging.warning("Profile is corrupted, attempting recovery")
            if _recover_corrupted_profile(profile_path):
                logging.info("Profile recovery successful, will create new profile")
            else:
                logging.error("Profile recovery failed")

    if clean_profile:
        _cleanup_profile_lock(profile_path)

    # NOTE: 1回だけ自動リトライ
    try:
        return _create_driver_impl(profile_name, data_path, is_headless, use_subprocess)
    except Exception as e:
        logging.warning("First attempt to create driver failed: %s", e)

        # コンテナ内で実行中の場合のみ、残った Chrome プロセスをクリーンアップ
        _cleanup_orphaned_chrome_processes_in_container()

        # プロファイルのロックファイルを削除
        _cleanup_profile_lock(profile_path)

        # 再度健全性チェック
        health = _check_profile_health(profile_path)
        if not health.is_healthy and auto_recover and (health.has_corrupted_json or health.has_corrupted_db):
            logging.warning("Profile still corrupted after first attempt, recovering")
            _recover_corrupted_profile(profile_path)

        return _create_driver_impl(profile_name, data_path, is_headless, use_subprocess)


def xpath_exists(driver: WebDriver, xpath: str) -> bool:
    return len(driver.find_elements(selenium.webdriver.common.by.By.XPATH, xpath)) != 0


def get_text(
    driver: WebDriver,
    xpath: str,
    safe_text: str,
    wait: WebDriverWait[WebDriver] | None = None,
) -> str:
    if wait is not None:
        wait.until(
            selenium.webdriver.support.expected_conditions.presence_of_all_elements_located(
                (selenium.webdriver.common.by.By.XPATH, xpath)
            )
        )

    if len(driver.find_elements(selenium.webdriver.common.by.By.XPATH, xpath)) != 0:
        return driver.find_element(selenium.webdriver.common.by.By.XPATH, xpath).text.strip()
    else:
        return safe_text


def input_xpath(
    driver: WebDriver,
    xpath: str,
    text: str,
    wait: WebDriverWait[WebDriver] | None = None,
    is_warn: bool = True,
) -> bool:
    if wait is not None:
        wait.until(
            selenium.webdriver.support.expected_conditions.element_to_be_clickable(
                (selenium.webdriver.common.by.By.XPATH, xpath)
            )
        )
        time.sleep(0.05)

    if xpath_exists(driver, xpath):
        driver.find_element(selenium.webdriver.common.by.By.XPATH, xpath).send_keys(text)
        return True
    else:
        if is_warn:
            logging.warning("Element is not found: %s", xpath)
        return False


def click_xpath(
    driver: WebDriver,
    xpath: str,
    wait: WebDriverWait[WebDriver] | None = None,
    is_warn: bool = True,
    move: bool = False,
) -> bool:
    if wait is not None:
        wait.until(
            selenium.webdriver.support.expected_conditions.element_to_be_clickable(
                (selenium.webdriver.common.by.By.XPATH, xpath)
            )
        )
        time.sleep(0.05)

    if xpath_exists(driver, xpath):
        elem = driver.find_element(selenium.webdriver.common.by.By.XPATH, xpath)
        if move:
            action = selenium.webdriver.common.action_chains.ActionChains(driver)
            action.move_to_element(elem)
            action.perform()

        elem.click()
        return True
    else:
        if is_warn:
            logging.warning("Element is not found: %s", xpath)
        return False


def is_display(driver: WebDriver, xpath: str) -> bool:
    return (len(driver.find_elements(selenium.webdriver.common.by.By.XPATH, xpath)) != 0) and (
        driver.find_element(selenium.webdriver.common.by.By.XPATH, xpath).is_displayed()
    )


def random_sleep(sec: float) -> None:
    RATIO = 0.8

    time.sleep((sec * RATIO) + (sec * (1 - RATIO) * 2) * random.random())  # noqa: S311


def wait_patiently(
    driver: WebDriver,
    wait: WebDriverWait[WebDriver],
    target: Any,
) -> None:
    error: selenium.common.exceptions.TimeoutException | None = None
    for i in range(WAIT_RETRY_COUNT + 1):
        try:
            wait.until(target)
            return
        except selenium.common.exceptions.TimeoutException as e:  # noqa: PERF203
            logging.warning(
                "タイムアウトが発生しました。(%s in %s line %d)",
                inspect.stack()[1].function,
                inspect.stack()[1].filename,
                inspect.stack()[1].lineno,
            )
            error = e

            logging.info(i)
            if i != WAIT_RETRY_COUNT:
                logging.info("refresh")
                driver.refresh()

    if error is not None:
        raise error


def dump_page(
    driver: WebDriver,
    index: int,
    dump_path: pathlib.Path,
    stack_index: int = 1,
) -> None:
    name = inspect.stack()[stack_index].function.replace("<", "").replace(">", "")

    dump_path.mkdir(parents=True, exist_ok=True)

    png_path = dump_path / f"{name}_{index:02d}.png"
    htm_path = dump_path / f"{name}_{index:02d}.htm"

    driver.save_screenshot(str(png_path))

    with htm_path.open("w", encoding="utf-8") as f:
        f.write(driver.page_source)

    logging.info(
        "page dump: %02d from %s in %s line %d",
        index,
        inspect.stack()[stack_index].function,
        inspect.stack()[stack_index].filename,
        inspect.stack()[stack_index].lineno,
    )


def clear_cache(driver: WebDriver) -> None:
    driver.execute_cdp_cmd("Network.clearBrowserCache", {})


def clean_dump(dump_path: pathlib.Path, keep_days: int = 1) -> None:
    if not dump_path.exists():
        return

    time_threshold = datetime.timedelta(keep_days)

    for item in dump_path.iterdir():
        if not item.is_file():
            continue
        try:
            time_diff = datetime.datetime.now(datetime.timezone.utc) - datetime.datetime.fromtimestamp(
                item.stat().st_mtime, datetime.timezone.utc
            )
        except FileNotFoundError:
            # ファイルが別プロセスにより削除された場合（SQLiteの一時ファイルなど）
            continue
        if time_diff > time_threshold:
            logging.info("remove %s [%s day(s) old].", item.absolute(), f"{time_diff.days:,}")

            item.unlink(missing_ok=True)


def get_memory_info(driver: WebDriver) -> dict[str, Any]:
    """ブラウザのメモリ使用量を取得（単位: KB）"""
    total_bytes = subprocess.Popen(  # noqa: S602
        "smem -t -c pss -P chrome | tail -n 1",  # noqa: S607
        shell=True,
        stdout=subprocess.PIPE,
    ).communicate()[0]
    total = int(str(total_bytes, "utf-8").strip())  # smem の出力は KB 単位

    try:
        memory_info = driver.execute_cdp_cmd("Memory.getAllTimeSamplingProfile", {})
        heap_usage = driver.execute_cdp_cmd("Runtime.getHeapUsage", {})

        heap_used = heap_usage.get("usedSize", 0) // 1024  # bytes → KB
        heap_total = heap_usage.get("totalSize", 0) // 1024  # bytes → KB
    except Exception as e:
        logging.debug("Failed to get memory usage: %s", e)

        memory_info = None
        heap_used = 0
        heap_total = 0

    return {
        "total": total,
        "heap_used": heap_used,
        "heap_total": heap_total,
        "memory_info": memory_info,
    }


def log_memory_usage(driver: WebDriver) -> None:
    mem_info = get_memory_info(driver)
    logging.info(
        "Chrome memory: %s MB (JS heap: %s MB)",
        f"""{mem_info["total"] // 1024:,}""",
        f"""{mem_info["heap_used"] // 1024:,}""",
    )


def _warmup(
    driver: WebDriver,
    keyword: str,
    url_pattern: str,
    sleep_sec: int = 3,
) -> None:
    # NOTE: ダミーアクセスを行って BOT ではないと思わせる。(効果なさそう...)
    driver.get("https://www.yahoo.co.jp/")
    time.sleep(sleep_sec)

    driver.find_element(selenium.webdriver.common.by.By.XPATH, '//input[@name="p"]').send_keys(keyword)
    driver.find_element(selenium.webdriver.common.by.By.XPATH, '//input[@name="p"]').send_keys(
        selenium.webdriver.common.keys.Keys.ENTER
    )

    time.sleep(sleep_sec)

    driver.find_element(
        selenium.webdriver.common.by.By.XPATH, f'//a[contains(@href, "{url_pattern}")]'
    ).click()

    time.sleep(sleep_sec)


class browser_tab:  # noqa: N801
    def __init__(self, driver: WebDriver, url: str) -> None:  # noqa: D107
        self.driver = driver
        self.url = url
        self.original_window: str | None = None

    def __enter__(self) -> None:  # noqa: D105
        self.original_window = self.driver.current_window_handle
        self.driver.execute_script("window.open('');")
        self.driver.switch_to.window(self.driver.window_handles[-1])
        self.driver.get(self.url)

    def __exit__(
        self,
        exception_type: type[BaseException] | None,
        exception_value: BaseException | None,
        traceback: Any,
    ) -> None:  # noqa: D105
        try:
            self.driver.close()
            if self.original_window is not None:
                self.driver.switch_to.window(self.original_window)
            time.sleep(0.1)
        except Exception:
            # NOTE: Chromeがクラッシュした場合は無視（既に終了しているため操作不可）
            logging.exception("Failed to close browser tab (Chrome may have crashed)")


class error_handler:  # noqa: N801
    """Selenium操作時のエラーハンドリング用コンテキストマネージャ

    エラー発生時に自動でログ出力、スクリーンショット取得、コールバック呼び出しを行う。

    Args:
        driver: WebDriver インスタンス
        message: ログに出力するエラーメッセージ
        on_error: エラー時に呼ばれるコールバック関数 (exception, screenshot: PIL.Image.Image | None) -> None
        capture_screenshot: スクリーンショットを自動取得するか（デフォルト: True）
        reraise: 例外を再送出するか（デフォルト: True）

    Attributes:
        exception: 発生した例外（エラーがなければ None）
        screenshot: 取得したスクリーンショット（PIL.Image.Image、取得失敗時は None）

    Examples:
        基本的な使用方法::

            with my_lib.selenium_util.error_handler(driver, message="ログイン処理に失敗") as handler:
                driver.get(login_url)
                driver.find_element(...).click()

        コールバック付き（Slack通知など）::

            def notify(exc, screenshot):
                slack.error("エラー発生", str(exc), screenshot)

            with my_lib.selenium_util.error_handler(
                driver,
                message="クロール処理に失敗",
                on_error=notify,
            ):
                crawl_page(driver)

        例外を抑制して続行::

            with my_lib.selenium_util.error_handler(driver, reraise=False) as handler:
                risky_operation()

            if handler.exception:
                logging.warning("処理をスキップしました")
    """

    def __init__(
        self,
        driver: WebDriver,
        message: str = "Selenium operation failed",
        on_error: Callable[[Exception, PIL.Image.Image | None], None] | None = None,
        capture_screenshot: bool = True,
        reraise: bool = True,
    ) -> None:
        self.driver = driver
        self.message = message
        self.on_error = on_error
        self.capture_screenshot = capture_screenshot
        self.reraise = reraise
        self.exception: Exception | None = None
        self.screenshot: PIL.Image.Image | None = None

    def __enter__(self) -> "error_handler":
        return self

    def __exit__(
        self,
        exception_type: type[BaseException] | None,
        exception_value: BaseException | None,
        traceback: Any,
    ) -> bool:
        if exception_value is None:
            return False

        # 例外を記録
        if isinstance(exception_value, Exception):
            self.exception = exception_value
        else:
            # BaseException（KeyboardInterrupt など）は処理せず再送出
            return False

        # ログ出力
        logging.exception(self.message)

        # スクリーンショット取得
        if self.capture_screenshot:
            try:
                screenshot_bytes = self.driver.get_screenshot_as_png()
                self.screenshot = PIL.Image.open(io.BytesIO(screenshot_bytes))
            except Exception:
                logging.debug("Failed to capture screenshot for error handling")

        # コールバック呼び出し
        if self.on_error is not None:
            try:
                self.on_error(self.exception, self.screenshot)
            except Exception:
                logging.exception("Error in on_error callback")

        # reraise=False なら例外を抑制
        return not self.reraise


def _is_chrome_related_process(process: psutil.Process) -> bool:
    """プロセスがChrome関連かどうかを判定"""
    try:
        process_name = process.name().lower()
        # Chrome関連のプロセス名パターン
        chrome_patterns = ["chrome", "chromium", "google-chrome", "undetected_chro"]
        # chromedriverは除外
        if "chromedriver" in process_name:
            return False
        return any(pattern in process_name for pattern in chrome_patterns)
    except (psutil.NoSuchProcess, psutil.AccessDenied):
        return False


def _get_chrome_processes_by_pgid(chromedriver_pid: int, existing_pids: set[int]) -> list[int]:
    """プロセスグループIDで追加のChrome関連プロセスを取得"""
    additional_pids = []
    try:
        pgid = os.getpgid(chromedriver_pid)
        for proc in psutil.process_iter(["pid", "name", "ppid"]):
            if proc.info["pid"] in existing_pids:
                continue
            try:
                if os.getpgid(proc.info["pid"]) == pgid:
                    proc_obj = psutil.Process(proc.info["pid"])
                    if _is_chrome_related_process(proc_obj):
                        additional_pids.append(proc.info["pid"])
                        logging.debug(
                            "Found Chrome-related process by pgid: PID %d, name: %s",
                            proc.info["pid"],
                            proc.info["name"],
                        )
            except (psutil.NoSuchProcess, psutil.AccessDenied, OSError):
                pass
    except (OSError, psutil.NoSuchProcess):
        logging.debug("Failed to get process group ID for chromedriver")
    return additional_pids


def _get_chrome_related_processes(driver: WebDriver) -> list[int]:
    """Chrome関連の全子プロセスを取得

    undetected_chromedriver 使用時、Chrome プロセスは chromedriver の子ではなく
    Python プロセスの直接の子として起動されることがあるため、両方を検索する。
    """
    chrome_pids = set()

    # 1. driver.service.process の子プロセスを検索
    try:
        if hasattr(driver, "service") and driver.service and hasattr(driver.service, "process"):  # type: ignore[attr-defined]
            process = driver.service.process  # type: ignore[attr-defined]
            if process and hasattr(process, "pid"):
                chromedriver_pid = process.pid

                # psutilでプロセス階層を取得
                parent_process = psutil.Process(chromedriver_pid)
                children = parent_process.children(recursive=True)

                for child in children:
                    chrome_pids.add(child.pid)
                    logging.debug(
                        "Found Chrome-related process (service child): PID %d, name: %s",
                        child.pid,
                        child.name(),
                    )
    except Exception:
        logging.exception("Failed to get Chrome-related processes from service")

    # 2. 現在の Python プロセスの全子孫から Chrome 関連プロセスを検索
    try:
        current_process = psutil.Process()
        all_children = current_process.children(recursive=True)

        for child in all_children:
            if child.pid in chrome_pids:
                continue
            try:
                if _is_chrome_related_process(child):
                    chrome_pids.add(child.pid)
                    logging.debug(
                        "Found Chrome-related process (python child): PID %d, name: %s",
                        child.pid,
                        child.name(),
                    )
            except (psutil.NoSuchProcess, psutil.AccessDenied):
                pass
    except Exception:
        logging.exception("Failed to get Chrome-related processes from python children")

    return list(chrome_pids)


def _send_signal_to_processes(pids: list[int], sig: signal.Signals, signal_name: str) -> None:
    """プロセスリストに指定されたシグナルを送信"""
    errors = []
    for pid in pids:
        try:
            # プロセス名を取得
            try:
                process = psutil.Process(pid)
                process_name = process.name()
            except (psutil.NoSuchProcess, psutil.AccessDenied):
                process_name = "unknown"

            if sig == signal.SIGKILL:
                # プロセスがまだ存在するかチェック
                os.kill(pid, 0)  # シグナル0は存在確認
            os.kill(pid, sig)
            logging.info("Sent %s to process: PID %d (%s)", signal_name, pid, process_name)
        except (ProcessLookupError, OSError) as e:  # noqa: PERF203
            # プロセスが既に終了している場合は無視
            errors.append((pid, e))

    # エラーが発生した場合はまとめてログ出力
    if errors:
        logging.debug("Failed to send %s to some processes: %s", signal_name, errors)


def _terminate_chrome_processes(chrome_pids: list[int], timeout: float = 5.0) -> None:
    """Chrome関連プロセスを段階的に終了

    Args:
        chrome_pids: 終了対象のプロセスIDリスト
        timeout: SIGTERM後にプロセス終了を待機する最大時間（秒）
    """
    if not chrome_pids:
        return

    # 優雅な終了（SIGTERM）
    _send_signal_to_processes(chrome_pids, signal.SIGTERM, "SIGTERM")

    # プロセスの終了を待機（ポーリング）
    remaining_pids = list(chrome_pids)
    poll_interval = 0.2
    elapsed = 0.0

    while remaining_pids and elapsed < timeout:
        time.sleep(poll_interval)
        elapsed += poll_interval

        # まだ生存しているプロセスをチェック
        still_alive = []
        for pid in remaining_pids:
            try:
                if psutil.pid_exists(pid):
                    process = psutil.Process(pid)
                    if process.is_running() and process.status() != psutil.STATUS_ZOMBIE:
                        still_alive.append(pid)
            except (psutil.NoSuchProcess, psutil.AccessDenied):
                pass

        remaining_pids = still_alive

    # タイムアウト後もまだ残っているプロセスにのみ SIGKILL を送信
    if remaining_pids:
        logging.warning(
            "Chrome processes still alive after %.1fs, sending SIGKILL to %d processes",
            elapsed,
            len(remaining_pids),
        )
        _send_signal_to_processes(remaining_pids, signal.SIGKILL, "SIGKILL")


def _reap_single_process(pid: int) -> None:
    """単一プロセスをwaitpidで回収"""
    try:
        # ノンブロッキングでwaitpid
        result_pid, status = os.waitpid(pid, os.WNOHANG)
        if result_pid == pid:
            # プロセス名を取得
            try:
                process = psutil.Process(pid)
                process_name = process.name()
            except (psutil.NoSuchProcess, psutil.AccessDenied):
                process_name = "unknown"
            logging.debug("Reaped Chrome process: PID %d (%s)", pid, process_name)
    except (ChildProcessError, OSError):
        # 子プロセスでない場合や既に回収済みの場合は無視
        pass


def _reap_chrome_processes(chrome_pids: list[int]) -> None:
    """Chrome関連プロセスを明示的に回収してゾンビ化を防ぐ"""
    for pid in chrome_pids:
        _reap_single_process(pid)


def _get_remaining_chrome_pids(chrome_pids: list[int]) -> list[int]:
    """指定されたPIDリストから、まだ生存しているChrome関連プロセスを取得"""
    remaining = []
    for pid in chrome_pids:
        try:
            if psutil.pid_exists(pid):
                process = psutil.Process(pid)
                if process.is_running() and process.status() != psutil.STATUS_ZOMBIE:
                    if _is_chrome_related_process(process):
                        remaining.append(pid)
        except (psutil.NoSuchProcess, psutil.AccessDenied):
            pass
    return remaining


def _wait_for_processes_with_check(
    chrome_pids: list[int],
    timeout: float,
    poll_interval: float = 0.2,
    log_interval: float = 1.0,
) -> list[int]:
    """プロセスの終了を待機しつつ、残存プロセスをチェック

    Args:
        chrome_pids: 監視対象のプロセスIDリスト
        timeout: 最大待機時間（秒）
        poll_interval: チェック間隔（秒）
        log_interval: ログ出力間隔（秒）

    Returns:
        タイムアウト後も残存しているプロセスIDのリスト
    """
    elapsed = 0.0
    last_log_time = 0.0
    remaining_pids = list(chrome_pids)

    while remaining_pids and elapsed < timeout:
        time.sleep(poll_interval)
        elapsed += poll_interval
        remaining_pids = _get_remaining_chrome_pids(remaining_pids)

        if remaining_pids and (elapsed - last_log_time) >= log_interval:
            logging.info(
                "Found %d remaining Chrome processes after %.0fs",
                len(remaining_pids),
                elapsed,
            )
            last_log_time = elapsed

    return remaining_pids


def quit_driver_gracefully(
    driver: WebDriver | None,
    wait_sec: float = 5.0,
    sigterm_wait_sec: float = 5.0,
    sigkill_wait_sec: float = 5.0,
) -> None:  # noqa: C901, PLR0912
    """Chrome WebDriverを確実に終了する

    終了フロー:
    1. driver.quit() を呼び出し
    2. wait_sec 秒待機しつつプロセス終了をチェック
    3. 残存プロセスがあれば SIGTERM を送信
    4. sigterm_wait_sec 秒待機しつつプロセス終了をチェック
    5. 残存プロセスがあれば SIGKILL を送信
    6. sigkill_wait_sec 秒待機

    Args:
        driver: 終了する WebDriver インスタンス
        wait_sec: quit 後にプロセス終了を待機する秒数（デフォルト: 5秒）
        sigterm_wait_sec: SIGTERM 後にプロセス終了を待機する秒数（デフォルト: 5秒）
        sigkill_wait_sec: SIGKILL 後にプロセス回収を待機する秒数（デフォルト: 5秒）
    """
    if driver is None:
        return

    # quit前にChrome関連プロセスを記録
    chrome_pids_before = _get_chrome_related_processes(driver)

    try:
        # WebDriverの正常終了を試行（これがタブのクローズも含む）
        driver.quit()
        logging.info("WebDriver quit successfully")
    except Exception:
        logging.warning("Failed to quit driver normally", exc_info=True)
    finally:
        # undetected_chromedriver の __del__ がシャットダウン時に再度呼ばれるのを防ぐ
        if hasattr(driver, "_has_quit"):
            driver._has_quit = True  # type: ignore[attr-defined]

    # ChromeDriverサービスの停止を試行
    try:
        if hasattr(driver, "service") and driver.service and hasattr(driver.service, "stop"):  # type: ignore[attr-defined]
            driver.service.stop()  # type: ignore[attr-defined]
    except (ConnectionResetError, OSError):
        # Chrome が既に終了している場合は無視
        logging.debug("Chrome service already stopped")
    except Exception:
        logging.warning("Failed to stop Chrome service", exc_info=True)

    # Step 1: quit 後に wait_sec 秒待機しつつプロセス終了をチェック
    remaining_pids = _wait_for_processes_with_check(chrome_pids_before, wait_sec)

    if not remaining_pids:
        logging.debug("All Chrome processes exited normally")
        return

    # Step 2: 残存プロセスに SIGTERM を送信
    logging.info(
        "Found %d remaining Chrome processes after %.0fs, sending SIGTERM",
        len(remaining_pids),
        wait_sec,
    )
    _send_signal_to_processes(remaining_pids, signal.SIGTERM, "SIGTERM")

    # Step 3: SIGTERM 後に sigterm_wait_sec 秒待機しつつプロセス終了をチェック
    remaining_pids = _wait_for_processes_with_check(remaining_pids, sigterm_wait_sec)

    if not remaining_pids:
        logging.info("All Chrome processes exited after SIGTERM")
        _reap_chrome_processes(chrome_pids_before)
        return

    # Step 4: 残存プロセスに SIGKILL を送信
    logging.warning(
        "Chrome processes still alive after SIGTERM + %.1fs, sending SIGKILL to %d processes",
        sigterm_wait_sec,
        len(remaining_pids),
    )
    _send_signal_to_processes(remaining_pids, signal.SIGKILL, "SIGKILL")

    # Step 5: SIGKILL 後に sigkill_wait_sec 秒待機してプロセス回収
    time.sleep(sigkill_wait_sec)
    _reap_chrome_processes(chrome_pids_before)

    # 最終チェック：まだ残っているプロセスがあるか確認
    still_remaining = _get_remaining_chrome_pids(remaining_pids)

    # 回収できなかったプロセスについて警告
    if still_remaining:
        for pid in still_remaining:
            try:
                process = psutil.Process(pid)
                logging.warning("Failed to collect Chrome-related process: PID %d (%s)", pid, process.name())
            except (psutil.NoSuchProcess, psutil.AccessDenied):
                pass


if __name__ == "__main__":
    import pathlib

    import docopt
    import selenium.webdriver.support.wait

    import my_lib.config
    import my_lib.logger

    assert __doc__ is not None
    args = docopt.docopt(__doc__)

    config_file = args["-c"]
    debug_mode = args["-D"]

    my_lib.logger.init("test", level=logging.DEBUG if debug_mode else logging.INFO)

    config = my_lib.config.load(config_file)

    driver = create_driver("test", pathlib.Path(config["data"]["selenium"]))
    wait = selenium.webdriver.support.wait.WebDriverWait(driver, 5)

    driver.get("https://www.google.com/")
    wait.until(
        selenium.webdriver.support.expected_conditions.presence_of_element_located(
            (selenium.webdriver.common.by.By.XPATH, '//input[contains(@value, "Google")]')
        )
    )

    quit_driver_gracefully(driver)

kimata / my-py-lib / 20569549039

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous