html2print: minor cleanups on windows

- get chrome binary_location from pybrowsers on windows
- --headless=new seems more stable on Windows, and seems to works in Linux too
- add entrypoint.sh to .gitattributes so that it keeps the line-endings (windows/docker)
- some added typing to make mypy happy

Author: thseiler

.gitattributes

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

html2print/html2print.py

@@ -13,6 +13,7 @@
 from time import sleep
 from typing import Dict, List, Optional
 
+import browsers
 import requests
 from requests import Response
 from selenium import webdriver
@@ -39,7 +40,8 @@
 
 
 class ChromeDriverManager:
-    def get_chrome_driver(self, path_to_cache_dir: str):
+    def get_chrome_driver(self, path_to_cache_dir: str) -> str:
+        """Return path to downloaded chromedriver."""
         chrome_version = self.get_chrome_version()
 
         # If Web Driver Manager cannot detect Chrome, it returns None.
@@ -104,13 +106,24 @@ 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 +156,11 @@ 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
@@ -160,7 +178,7 @@ def _download_chromedriver(
         return path_to_cached_chrome_driver
 
     @staticmethod
-    def send_http_get_request(url, params=None, **kwargs) -> Response:
+    def send_http_get_request(url, params=None, **kwargs) -> Optional[Response]:
         last_error: Optional[Exception] = None
         for attempt in range(1, 4):
             print(  # noqa: T201
@@ -180,6 +198,7 @@ def send_http_get_request(url, params=None, **kwargs) -> Response:
             f"html2print: "
             f"failed to get response for URL: {url} with error: {last_error}"
         )
+        return None
 
     @staticmethod
     def get_chrome_version():
@@ -312,30 +331,35 @@ def create_webdriver(
 ) -> webdriver.Chrome:
     print("html2print: creating ChromeDriver service.", flush=True)  # noqa: T201
     if chromedriver is None:
-        path_to_chrome = ChromeDriverManager().get_chrome_driver(
+        path_to_chrome_driver = ChromeDriverManager().get_chrome_driver(
             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(f"html2print: ChromeDriver available at path: {path_to_chrome_driver}")  # noqa: T201
 
     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)
+
+    path_to_chrome = ""
+    if platform.system() == "Windows":
+        path_to_chrome = browsers.get("chrome")["path"]
+    print(f"html2print: Chrome available at path: {path_to_chrome}")  # noqa: T201
 
     webdriver_options = Options()
+    webdriver_options.binary_location = path_to_chrome
     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")
-    # 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")
+    # 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")
 
     # The Chrome option --disable-dev-shm-usage disables the use of /dev/shm
     # (shared memory) for temporary storage in Chrome.
@@ -447,10 +471,10 @@ 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(f"html2print: ChromeDriver available at path: {path_to_chrome_driver}")  # noqa: T201
         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]