Switch from pyppeteer to playwright.

Fixes #1262.

This doesn't really improve a whole lot. playwright notably disables the
Chromium sandbox by default, so we don't need to do that explicitly any more,
but other than that the code looks largely the same.

The generated PDFs look slightly different, I think because playwright
install-deps installs more fonts, and we then end up using Arial for the
receipts.

Because this is substantially invasive and in the critical path for generating
receipts I don't think this is worthwhile merging for 2024.

Dockerfile

@@ -3,7 +3,9 @@ FROM ghcr.io/emfcamp/website-base-dev:latest
 COPY pyproject.toml poetry.lock Makefile /app/
 WORKDIR /app
 
-RUN poetry install && poetry run pyppeteer-install
+RUN poetry install \
+	&& poetry run playwright install-deps \
+	&& poetry run playwright install chromium
 
 ENV SHELL=/bin/bash
 ENTRYPOINT ["./docker/dev_entrypoint.sh"]

Dockerfile.prod

@@ -8,7 +8,10 @@ FROM ghcr.io/emfcamp/website-base:latest
 COPY . /app/
 WORKDIR /app
 
-RUN poetry install --no-dev && poetry run pyppeteer-install && rm -Rf ./static
+RUN poetry install --no-dev \
+	&& poetry run playwright install-deps \
+	&& poetry run playwright install chromium \
+	&& rm -Rf ./static
 
 COPY --from=static /app/static /app/static
 

apps/common/receipt.py

@@ -2,9 +2,9 @@
 from lxml import etree
 import asyncio
 
-from flask import render_template, current_app as app
+from flask import render_template
 from markupsafe import Markup
-from pyppeteer.launcher import launch
+from playwright.async_api import async_playwright
 import barcode
 from barcode.writer import ImageWriter, SVGWriter
 import segno
@@ -63,29 +63,22 @@ def render_pdf(url, html):
     # you're running a dev server, use app.run(processes=2)
 
     async def to_pdf():
-        browser = await launch(
-            # Handlers don't work as we're not in the main thread.
-            handleSIGINT=False,
-            handleSIGTERM=False,
-            handleSIGHUP=False,
-            # --no-sandbox is necessary as we're running as root (in docker!)
-            args=["--no-sandbox"],
-        )
-        page = await browser.newPage()
-
-        async def request_intercepted(request):
-            app.logger.debug("Intercepted URL: %s", request.url)
-            if request.url == url:
-                await request.respond({"body": html})
-            else:
-                await request.continue_()
-
-        page.on("request", lambda r: asyncio.ensure_future(request_intercepted(r)))
-        await page.setRequestInterception(True)
-
-        await page.goto(url)
-        pdf = await page.pdf(format="A4")
-        await browser.close()
+        async with async_playwright() as p:
+            browser = await p.chromium.launch(
+                # Handlers don't work as we're not in the main thread.
+                handle_sigint=False,
+                handle_sigterm=False,
+                handle_sighup=False,
+            )
+            context = await browser.new_context()
+            page = await browser.new_page()
+
+            await page.route(url, lambda route: route.fulfill(body=html))
+            await page.goto(url)
+            pdf = await page.pdf(format="A4")
+            await page.close()
+            await context.close()
+            await browser.close()
         return pdf
 
     loop = asyncio.new_event_loop()

docker/Dockerfile.base

@@ -12,7 +12,7 @@ COPY pyproject.toml poetry.lock /app/
 # doesn't confuse poetry.
 ENV POETRY_VIRTUALENVS_IN_PROJECT=false
 
-# All the X11 packages are required for chromium/pyppetteer which we use for
+# All the X11 packages are required for chromium/playwright which we use for
 # generating PDFs.
 #
 # It would be nice to be able to build an image without build tools and dev
@@ -26,5 +26,7 @@ RUN apt-get update && \
       libxrandr2 libasound2 libatk1.0-0 libatk-bridge2.0-0 libgtk-3-0 libzbar0 \
       coinor-cbc curl && \
     pip3 install poetry==1.7.1 && \
-    poetry install && poetry run pyppeteer-install && \
+    poetry install && \
+    poetry run playwright install-deps && \
+    poetry run playwright install chromium && \
     rm -rf /var/lib/apt/lists/*

logging.yaml

@@ -31,7 +31,7 @@ loggers:
     propagate: false
   iso8601:
     level: WARN
-  pyppeteer:
+  playwright:
     level: WARN
   websockets.protocol:
     level: WARN