html2print: minor cleanups on windows

- add entrypoint.sh to .gitattributes so that it keeps the line-endings (windows/docker)

Author: thseiler

.gitattributes

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

.github/workflows/ci-windows.yml

@@ -28,7 +28,7 @@ jobs:
       shell: powershell
 
     - name: Check Chrome Version
-      run: '& "C:\Program Files\Google\Chrome\Application\chrome.exe" --version'
+      run: '(Get-Item "C:\Program Files\Google\Chrome\Application\chrome.exe").VersionInfo'
       shell: powershell
 
     - name: Add Chrome to PATH
@@ -38,10 +38,6 @@ jobs:
         echo "$chromePath" | Out-File -Append -Encoding utf8 $env:GITHUB_PATH
       shell: powershell
 
-    - name: Verify Chrome Installation
-      run: chrome --version
-      shell: powershell
-
     - name: Upgrade pip
       run: |
         python -m pip install --upgrade pip

html2print/html2print.py

@@ -13,8 +13,9 @@
 from time import sleep
 from typing import Dict, List, Optional
 
+import browsers
 import requests
-from requests import Response
+from requests import Response, RequestException
 from selenium import webdriver
 from selenium.webdriver.chrome.options import Options
 from selenium.webdriver.chrome.service import Service
@@ -180,6 +181,8 @@ 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():
@@ -332,7 +335,9 @@ def create_webdriver(
     # 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")
@@ -447,10 +452,18 @@ def main():
             args.cache_dir if args.cache_dir is not None else DEFAULT_CACHE_DIR
         )
 
-        path_to_chrome = ChromeDriverManager().get_chrome_driver(
+        path_to_chrome_driver = ChromeDriverManager().get_chrome_driver(
             path_to_cache_dir
         )
-        print(f"html2print: ChromeDriver available at path: {path_to_chrome}")  # noqa: T201
+        print(  # noqa: T201
+            f"html2print: ChromeDriver available at path: {path_to_chrome_driver}"
+        )
+
+        if browsers.get("chrome") is not None:
+            path_to_chrome_browser = browsers.get("chrome")["path"]
+            print(  # noqa: T201
+                f"html2print: Chrome available at path:{path_to_chrome_browser}"
+            )
         sys.exit(0)
 
     elif args.command == "print":

pyproject.toml

@@ -54,6 +54,9 @@ dependencies = [
 
     # requests is used by HTML2PDF_HTTPClient.
     "requests",
+
+    # to detect the browers path (needed on windows)
+    "pybrowsers",
 ]
 
 [project.optional-dependencies]