Bug 269167
| Summary: | converters/wkhtmltopdf crashes with "Bus error" | ||
|---|---|---|---|
| Product: | Ports & Packages | Reporter: | SolarCatcher <solarcatcher> |
| Component: | Individual Port(s) | Assignee: | Kurt Jaeger <pi> |
| Status: | Open --- | ||
| Severity: | Affects Many People | CC: | freebsd-bugzilla, grahamperrin, jan.catrysse, me, pat |
| Priority: | --- | Keywords: | crash, needs-qa |
| Version: | Latest | Flags: | bugzilla:
maintainer-feedback?
(pi) |
| Hardware: | amd64 | ||
| OS: | Any | ||
| See Also: | https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=269313 | ||
| Attachments: | |||
|
Description
SolarCatcher
2023-01-26 12:56:55 UTC
Thanks for the bug report. Can you name URLs where this crash can be reproduced with wkhtmltopdf ? My use case is rendering of static HTML to PDF, so I have not yet seen the problem. Thanks for answering so quickly. This is one example of a page that leads to the bus error: https://www.greenjobs.de/angebote/neueste.html?id=100117275&anz=html&frame=1 (In reply to SolarCatcher from comment #2) There are folks interested in deprecating wkhtmltopdf, and they suggest alternatives, please have a look here: https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=269313#c3 I have found print/py-weasyprint to be able to cover my use-case, can you investigate if one of the suggested tools covers your use-case ? (In reply to SolarCatcher from comment #2) I reproduced the core-dump on 14.0. (In reply to SolarCatcher from comment #2) I tested the page with print/py-weasyprint and it somewhat worked. It was version 57.2, and I still need to update the port to that version. Thank you for the additional info. I was not aware that wkhtmltopdf was EOL. In this case, I will need to look for a suitable alternative. I have not yet tried print/py-weasyprint but from what I read this will not work for us, because we cannot control the HTML/CSS of the pages we need to convert to PDF. And weasyprint seems to be a great tool if you can, and build relatively simple pages. As I have all my machines on quarterly packages, I cannot easily test it for now. For me the gold standard for HTML to PDF conversion is Prince, and it is even supported on FreeBSD. But, unfortunately, it is a bit too costly (prices for installation on a server start at 3,800 USD). So we will have to look a around for other options. If you like, you can close this bug report. I understand there will not be path forward with this software. Thanks for your work on this port. I have used it for a long time. While looking around for alternatives, I had the idea that Chromium in headless mode could be a solution for us. I found this little how-to, which looks promising: https://blog.grio.com/2020/08/understanding-pdf-generation-with-headless-chrome.html On the command-line it already does what it promises. However, the author points out how to create a more clever setup using Node and chromium-launcher. This would add quite a bit of overhead, of course (maybe I could shut this away in a separate jail). In any case, it could be a solution for us and others looking to replace wkhtmltopdf. That's why I wanted to mention it here. I have similar issues. Doing a simple: "wkhtmltopdf https://www.google.com google.pdf" results in a "Segmentation fault (core dumped)". The issue seems to be related to JavaScript. Adding the command line option: "--disable-javascript" resolves the issue. Of course, without the well-needed JavaScript… This worked fine with previous builds of the same version 0.12.6. I do think this is due to the way this package is built and not per se the source code of wkhtmltopdf. I do not well understand what would be the root cause... but as many many many people are using this package, a solution would greatly be appreciated. I tried reverting some changes in the ports tree… but the result is identical. It would be nice to find the culprit, but I am afraid I am unable to. Information for other users having the same issue: For the moment, I am using the Linux emulation FreeBSD provides: https://docs.freebsd.org/en/books/handbook/linuxemu/ And the CentOS 8 executable from https://rubygems.org/gems/wkhtmltopdf-binary/ Created attachment 252762 [details]
Script to instal CentOS 7 version of wkhtmltopdf with all requirements in linux mode
Since it took me some time to figure out how to install wkhtmltopdf in Linux mode, I created a script that does it all when I finally had everything right. What it does: - enable and start Linux compatibility mode - install CentOS 7 base system with X11 support and other requirements - install latest CentOS 7 version of wkhtmltopdf rpm - make a symlink in /usr/local/bin so other programs can find it Download the attached script, make it executable and run as root. Created attachment 252763 [details] Script to instal CentOS 7 version of wkhtmltopdf with all requirements in linux mode - corrected I had experimented with different versions of the installed files, and did not check with a completely clean install at the end. The script I posted in comment #11 did not work on a clean install; this one does. |
