Use Selenium for generating PDFs using chromium

Dockerfile

@@ -30,6 +30,7 @@ RUN apt-get -y update && \
     eatmydata apt-get install -y --no-install-recommends \
         build-essential \
         chromium \
+        chromium-driver \
         curl \
 	dumb-init \
 	gettext \

aleksis/core/util/pdf.py

+import base64
 import os
 import subprocess  # noqa
 from datetime import timedelta
 from tempfile import TemporaryDirectory
-from typing import Optional, Tuple, Union, Callable
+from typing import Callable, Optional, Tuple, Union
 from urllib.parse import urljoin
 
 from django.conf import settings
@@ -19,6 +20,7 @@ from django.utils.translation import gettext as _
 
 from celery.result import AsyncResult
 from celery_progress.backend import ProgressRecorder
+from selenium import webdriver
 
 from aleksis.core.celery import app
 from aleksis.core.models import PDFFile
@@ -26,6 +28,29 @@ from aleksis.core.util.celery_progress import recorded_task, render_progress_pag
 from aleksis.core.util.core_helpers import process_custom_context_processors
 
 
+def _generate_pdf_with_chromium(temp_dir, pdf_path, html_url, lang):
+    """Generate a PDF file from a HTML file."""
+    chrome_options = webdriver.ChromeOptions()
+    chrome_options.add_argument("--kiosk-printing")
+    chrome_options.add_argument("--headless")
+    chrome_options.add_argument("--no-sandbox")
+    chrome_options.add_argument("--disable-gpu")
+    chrome_options.add_argument("--disable-dev-shm-usage")
+    chrome_options.add_argument("--disable-setuid-sandbox")
+    chrome_options.add_argument("--dbus-stub")
+    chrome_options.add_argument("--temp-profile")
+    chrome_options.add_argument(f"--lang={lang}")
+
+    driver = webdriver.Chrome(options=chrome_options)
+    driver.get(html_url)
+    pdf = driver.execute_cdp_cmd(
+        "Page.printToPDF", {"printBackground": True, "preferCSSPageSize": True}
+    )
+    driver.close()
+    with open(pdf_path, "wb") as f:
+        f.write(base64.b64decode(pdf["data"]))
+
+
 @recorded_task
 def generate_pdf(
     file_pk: int, html_url: str, recorder: ProgressRecorder, lang: Optional[str] = None
@@ -40,26 +65,7 @@ def generate_pdf(
         pdf_path = os.path.join(temp_dir, "print.pdf")
         lang = lang or get_language()
 
-        # Run PDF generation using a headless Chromium
-        cmd = [
-            "chromium",
-            "--headless",
-            "--no-sandbox",
-            "--run-all-compositor-stages-before-draw",
-            "--temp-profile",
-            "--disable-dev-shm-usage",
-            "--disable-gpu",
-            "--disable-setuid-sandbox",
-            "--dbus-stub",
-            f"--home-dir={temp_dir}",
-            f"--lang={lang}",
-            f"--print-to-pdf={pdf_path}",
-            html_url,
-        ]
-        res = subprocess.run(cmd)  # noqa
-
-        # Let the task fail on a non-success return code
-        res.check_returncode()
+        _generate_pdf_with_chromium(temp_dir, pdf_path, html_url, lang)
 
         # Upload PDF file to media storage
         with open(pdf_path, "rb") as f:
@@ -80,13 +86,12 @@ def process_context_for_pdf(context: Optional[dict] = None, request: Optional[Ht
         processed_context = context
     return processed_context
 
+
 def generate_pdf_from_html(
-    html: str, request: Optional[HttpRequest] = None) -> Tuple[PDFFile, AsyncResult]:
+    html: str, request: Optional[HttpRequest] = None
+) -> Tuple[PDFFile, AsyncResult]:
     """Start a PDF generation task and return the matching file object and Celery result."""
-
-    file_object = PDFFile.objects.create(
-        html_file=ContentFile(html.encode(), name="source.html")
-    )
+    file_object = PDFFile.objects.create(html_file=ContentFile(html.encode(), name="source.html"))
 
     # As this method may be run in background and there is no request available,
     # we have to use a predefined URL from settings then
@@ -99,8 +104,12 @@ def generate_pdf_from_html(
 
     return file_object, result
 
+
 def generate_pdf_from_template(
-    template_name: str, context: Optional[dict] = None, request: Optional[HttpRequest] = None, render_method: Optional[Callable] = None
+    template_name: str,
+    context: Optional[dict] = None,
+    request: Optional[HttpRequest] = None,
+    render_method: Optional[Callable] = None,
 ) -> Tuple[PDFFile, AsyncResult]:
     """Start a PDF generation task and return the matching file object and Celery result."""
     processed_context = process_context_for_pdf(context, request)

docs/admin/10_install.rst

@@ -53,6 +53,7 @@ Install some packages from the Debian package system.
                yarnpkg \
                python3-virtualenv \
                chromium \
+               chromium-driver \
                redis-server \
                postgresql \
                locales-all \

docs/dev/01_setup.rst

@@ -47,7 +47,7 @@ Install native dependencies
 
 Some system libraries are required to install AlekSIS. On Debian, for example, this would be done with::
 
-  sudo apt install build-essential libpq-dev libpq5 libssl-dev python3-dev python3-pip python3-venv yarnpkg gettext chromium
+  sudo apt install build-essential libpq-dev libpq5 libssl-dev python3-dev python3-pip python3-venv yarnpkg gettext chromium chromium-driver
 
 Get Poetry
 ----------

pyproject.toml

@@ -130,6 +130,7 @@ django-iconify = "^0.3"
 customidenticon = "^0.1.5"
 graphene-django = "^2.15.0"
 django-webpack-loader = "^1.6.0"
+selenium = "^4.4.3"
 
 [tool.poetry.extras]
 ldap = ["django-auth-ldap"]