diagnostic output interferes with stdout redirection

[drew-parsons]

xhtml2pdf allows for stdout redirection, so xhtml2pdf - - receives input from stdout and sends output to stdout rather than to file.

But it also generates diagnostic messages during processing so these also get sent to stdout, putting meaningless data into the pdf data stream.

e.g.

$ echo "test" | xhtml2pdf  - - >/tmp/test.pdf 2>/dev/null ; head -2 /tmp/test.pdf 
Converting - to -...
%PDF-1.3

So the generated pdf is prefixed with a line "Converting - to -...".
This line should probably be sent to stderr instead of stdout.

This bug was reported at Debian Bug#592500, https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=592500

[luisza]

Could you test using quiet mode ?

echo "test" | xhtml2pdf -q - - >/tmp/test.pdf 2>/dev/null ; head -2 /tmp/test.pdf

[drew-parsons]

-q eliminates all output to stdout, so the generated file (/tmp/test.pdf) is empty

$ ls -l /tmp/test.pdf
-rw-r--r-- 1 user user 0 Jan 21 08:53 /tmp/test.pdf

[drew-parsons]

Hmm, perversely however, it's now also generating an empty file even without -q. Not sure what's happened overnight (apart from a reboot). Will report back when I can reproduce the original problem again.

[drew-parsons]

I see what happened now. xhtml2pdf was the python2 version, but python-xhtml2pdf has just been removed from Debian, leaving only the python3-xhtml2pdf package with xhtml2pdf3 as the executable file.

The behaviour between the python3 and python3 versions seems quite different.
echo "test" | xhtml2pdf3 - - >/tmp/test.pdf 2>/dev/null ; head -2 /tmp/test.pdf does not create a pdf file at all in /tmp/test.pdf, the file simply contains the plain text "Converting - to -..." with no pdf data.

But the -q option for xhtml2pdf3 still generates an empty file. This must be because xhtml2pdf3 - - is no longer well-defined: without redirecting stderr to /dev/null, I get

$ echo "test" | xhtml2pdf3 - -
Converting - to -...
Traceback (most recent call last):
  File "/usr/bin/xhtml2pdf3", line 11, in <module>
    load_entry_point('xhtml2pdf==0.2.2', 'console_scripts', 'xhtml2pdf')()
  File "/usr/lib/python3/dist-packages/xhtml2pdf/pisa.py", line 167, in command
    execute()
  File "/usr/lib/python3/dist-packages/xhtml2pdf/pisa.py", line 392, in execute
    xml_output=xml_output
  File "/usr/lib/python3/dist-packages/xhtml2pdf/document.py", line 183, in pisaDocument
    context.dest.write(data)  # TODO: context.dest is a tempfile as well...
TypeError: write() argument must be str, not bytes