html2print: minor cleanups on windows

Author: thseiler

.gitattributes

@@ -0,0 +1 @@
+entrypoint.sh -text

.github/workflows/ci-windows.yml

@@ -28,18 +28,7 @@ jobs:
       shell: powershell
 
     - name: Check Chrome Version
-      run: '& "C:\Program Files\Google\Chrome\Application\chrome.exe" --version'
-      shell: powershell
-
-    - name: Add Chrome to PATH
-      run: |
-        $chromePath = "C:\Program Files\Google\Chrome\Application"
-        echo "Adding $chromePath to PATH"
-        echo "$chromePath" | Out-File -Append -Encoding utf8 $env:GITHUB_PATH
-      shell: powershell
-
-    - name: Verify Chrome Installation
-      run: chrome --version
+      run: '(Get-Item "C:\Program Files\Google\Chrome\Application\chrome.exe").VersionInfo'
       shell: powershell
 
     - name: Upgrade pip

html2print/html2print.py

@@ -13,12 +13,12 @@
 from time import sleep
 from typing import Dict, List, Optional
 
+import browsers
 import requests
-from requests import Response
+from requests import RequestException, Response
 from selenium import webdriver
 from selenium.webdriver.chrome.options import Options
 from selenium.webdriver.chrome.service import Service
-from webdriver_manager.core.os_manager import ChromeType, OperationSystemManager
 
 __version__ = "0.0.15"
 
@@ -39,9 +39,7 @@
 
 
 class ChromeDriverManager:
-    def get_chrome_driver(self, path_to_cache_dir: str):
-        chrome_version = self.get_chrome_version()
-
+    def get_chrome_driver(self, chrome_version: str, path_to_cache_dir: str):
         # If Web Driver Manager cannot detect Chrome, it returns None.
         if chrome_version is None:
             raise RuntimeError(
@@ -104,13 +102,24 @@ def get_chrome_driver(self, path_to_cache_dir: str):
 
     @staticmethod
     def _download_chromedriver(
-        chrome_major_version,
+        chrome_major_version: str,
         os_type: str,
-        path_to_driver_cache_dir,
-        path_to_cached_chrome_driver,
-    ):
+        path_to_driver_cache_dir: str,
+        path_to_cached_chrome_driver: str,
+    ) -> str:
         url = "https://googlechromelabs.github.io/chrome-for-testing/known-good-versions-with-downloads.json"
-        response = ChromeDriverManager.send_http_get_request(url).json()
+        response = ChromeDriverManager.send_http_get_request(url)
+        if response is None:
+            raise RuntimeError(
+                "Could not download known-good-versions-with-downloads.json"
+            )
+
+        response = response.json()
+        if response is None:
+            raise RuntimeError(
+                "Could not parse known-good-versions-with-downloads.json"
+            )
+        assert isinstance(response, dict)
 
         matching_versions = [
             item
@@ -143,6 +152,11 @@ def _download_chromedriver(
         )
         response = ChromeDriverManager.send_http_get_request(driver_url)
 
+        if response is None:
+            raise Exception(
+                f"Could not download ChromeDriver from {driver_url}"
+            )
+
         Path(path_to_driver_cache_dir).mkdir(parents=True, exist_ok=True)
         zip_path = os.path.join(path_to_driver_cache_dir, "chromedriver.zip")
         print(  # noqa: T201
@@ -180,9 +194,12 @@ def send_http_get_request(url, params=None, **kwargs) -> Response:
             f"html2print: "
             f"failed to get response for URL: {url} with error: {last_error}"
         )
+        raise RequestException(
+            f"GET request failed after 3 attempts: {url}"
+        ) from last_error
 
     @staticmethod
-    def get_chrome_version():
+    def get_browser_info() -> browsers.Browser:
         # Special case: GitHub Actions macOS CI machines have both
         # Google Chrome for Testing and normal Google Chrome installed, and
         # sometimes their versions are of different major version families.
@@ -216,17 +233,24 @@ def get_chrome_version():
                     f"html2print: Google Chrome for Testing Version: {chrome_version}"
                 )
 
-                return chrome_version
+                return browsers.Browser(
+                    browser_type="chrome-for-testing",
+                    path=chrome_path,
+                    version=chrome_version,
+                    display_name="Google Chrome for Testing",
+                )
             except FileNotFoundError:
                 print("html2print: Chrome for Testing not available.")  # noqa: T201
             except Exception as e:
                 print(  # noqa: T201
                     f"html2print: Error getting Google Chrome for Testing version: {e}"
                 )
+        else:
+            chrome_browser_info = browsers.get("chrome") or browsers.get(
+                "chromium"
+            )
 
-        os_manager = OperationSystemManager(os_type=None)
-        version = os_manager.get_browser_version_from_os(ChromeType.GOOGLE)
-        return version
+        return chrome_browser_info
 
 
 def get_inches_from_millimeters(mm: float) -> float:
@@ -305,34 +329,44 @@ class Done(Exception):
 
 
 def create_webdriver(
+    browser_info: browsers.Browser,
     chromedriver: Optional[str],
     path_to_cache_dir: str,
     page_load_timeout: int,
     debug: bool = False,
 ) -> webdriver.Chrome:
     print("html2print: creating ChromeDriver service.", flush=True)  # noqa: T201
+
     if chromedriver is None:
-        path_to_chrome = ChromeDriverManager().get_chrome_driver(
-            path_to_cache_dir
+        path_to_chrome_driver = ChromeDriverManager().get_chrome_driver(
+            chrome_version=browser_info["version"],
+            path_to_cache_dir=path_to_cache_dir,
         )
     else:
-        path_to_chrome = chromedriver
-    print(f"html2print: ChromeDriver available at path: {path_to_chrome}")  # noqa: T201
+        path_to_chrome_driver = chromedriver
+    print(  # noqa: T201
+        f"html2print: ChromeDriver available at path: {path_to_chrome_driver}"
+    )
 
     if debug:
         service = Service(
-            path_to_chrome, log_output=PATH_TO_CHROME_DRIVER_DEBUG_LOG
+            path_to_chrome_driver, log_output=PATH_TO_CHROME_DRIVER_DEBUG_LOG
         )
     else:
-        service = Service(path_to_chrome)
+        service = Service(path_to_chrome_driver)
 
     webdriver_options = Options()
+    # Workaround for Windows: chrome is not typically found in the PATH, so need to supply the exact binary location manually
+    if platform.system() == "Windows":
+        webdriver_options.binary_location = browser_info["path"]
     webdriver_options.add_argument("start-maximized")
     webdriver_options.add_argument("disable-infobars")
     # Doesn't seem to be needed.
     # webdriver_options.add_argument('--disable-gpu')  # noqa: ERA001
     webdriver_options.add_argument("--disable-extensions")
-    webdriver_options.add_argument("--headless=chrome")
+    # Use --headless=new, as it seems to be more stable on Windows (available since Chrome 109).
+    # see https://www.selenium.dev/blog/2023/headless-is-going-away/
+    webdriver_options.add_argument("--headless=new")
     # FIXME: This is not nice but otherwise it does not work in Ubuntu 24-based Docker image.
     # https://github.com/SeleniumHQ/selenium/issues/15327#issuecomment-2689287561
     webdriver_options.add_argument("--no-sandbox")
@@ -441,16 +475,31 @@ def main():
 
     args = parser.parse_args()
 
+    # Look for Chrome on the System...
+    browser_info = ChromeDriverManager.get_browser_info()
+
     path_to_cache_dir: str
     if args.command == "get_driver":
         path_to_cache_dir = (
             args.cache_dir if args.cache_dir is not None else DEFAULT_CACHE_DIR
         )
 
-        path_to_chrome = ChromeDriverManager().get_chrome_driver(
-            path_to_cache_dir
+        path_to_chrome_browser = browser_info["path"]
+        print(  # noqa: T201
+            f"html2print: Chrome available at path: {path_to_chrome_browser}"
+        )
+        chrome_version = browser_info["version"]
+        print(  # noqa: T201
+            f"html2print: Chrome version : {chrome_version}"
+        )
+
+        path_to_chrome_driver = ChromeDriverManager().get_chrome_driver(
+            chrome_version=browser_info["version"],
+            path_to_cache_dir=path_to_cache_dir,
+        )
+        print(  # noqa: T201
+            f"html2print: ChromeDriver available at path: {path_to_chrome_driver}"
         )
-        print(f"html2print: ChromeDriver available at path: {path_to_chrome}")  # noqa: T201
         sys.exit(0)
 
     elif args.command == "print":
@@ -462,6 +511,7 @@ def main():
             args.cache_dir if args.cache_dir is not None else DEFAULT_CACHE_DIR
         )
         driver = create_webdriver(
+            browser_info,
             args.chromedriver,
             path_to_cache_dir,
             page_load_timeout,

pyproject.toml

@@ -50,10 +50,12 @@ classifiers = [
 dependencies = [
     # HTML2PDF dependencies
     "selenium",
-    "webdriver-manager",
 
     # requests is used by HTML2PDF_HTTPClient.
     "requests",
+
+    # to detect the browers path (needed on windows)
+    "pybrowsers",
 ]
 
 [project.optional-dependencies]

tests/integration/lit.cfg

@@ -23,3 +23,6 @@ config.suffixes = ['.itest', '.c']
 config.is_windows = lit_config.isWindows
 if not lit_config.isWindows:
     config.available_features.add('PLATFORM_IS_NOT_WINDOWS')
+
+# In Linux CI, $HOME is required for chrome as it needs to access things in ~/.local
+config.environment['HOME'] = os.environ.get('HOME', '/tmp')