Bug#979011: libvigraimpex: FTBFS with doxygen 1.9.0 from experimental

[Paolo Greppi]

Source: libvigraimpex
Severity: important

Dear Maintainer,

in a partial rebuild of the doxygen build dependencies against 1.9.0 from experimental:
https://salsa.debian.org/debian/doxygen/-/wikis/ratt_doxygen_1.9.0-1_amd64-partial
this package failed to build with this error:

   Postprocessing html files
   cd /<<PKGBUILDDIR>>/obj.x86_64-linux-gnu/docsrc && /usr/bin/python3 /<<PKGBUILDDIR>>/docsrc/makeFunctionIndex.py /<<PKGBUILDDIR>>/doc/vigra
   Traceback (most recent call last):
     File "/<<PKGBUILDDIR>>/docsrc/makeFunctionIndex.py", line 141, in <module>
       generateFunctionIndex(functionList)
     File "/<<PKGBUILDDIR>>/docsrc/makeFunctionIndex.py", line 124, in generateFunctionIndex
       text = open(path + "/namespaces.html").read()
     File "/usr/lib/python3.9/codecs.py", line 322, in decode
       (result, consumed) = self._buffer_decode(data, self.errors, final)
   UnicodeDecodeError: 'utf-8' codec can't decode byte 0xb3 in position 138922: invalid start byte

I attach the full log.

The namespaces.html file found in doc/vigra is 142804 bytes long (for reference, the one installed in /usr/share/doc/libvigraimpex-dev/html/namespaces.html by libvigraimpex-doc is 7535 bytes).

I cheched with hexdump around position 138922 (hex 21EAA):

...
00021e60  3d 22 64 65 73 63 22 3e  43 6f 6d 70 75 74 61 74  |="desc">Computat|
00021e70  69 6f 6e 20 6f 66 20 57  69 67 6e 65 72 20 44 20  |ion of Wigner D |
00021e80  6d 61 74 72 69 78 20 2b  20 72 6f 74 61 74 69 6f  |matrix + rotatio|
00021e90  6e 20 66 75 6e 63 74 69  6f 6e 73 20 69 6e 20 53  |n functions in S|
00021ea0  48 2c 56 48 20 61 6e 64  20 52 b3 20 3c 2f 74 64  |H,VH and R. </td|
00021eb0  3e 3c 2f 74 72 3e 0a 3c  74 72 20 69 64 3d 22 72  |></tr>.<tr id="r|
...

It corresponds to this line:

tr id="row_1_250_" class="even" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classvigra_1_1WignerMatrix.html" target="_self">WignerMatrix</a></td><td class="desc">Computation of Wigner D matrix + rotation functions in SH,VH and R³ </td></tr>

This text is extracted from include/vigra/wigner-matrix.hxx:

...
00000b90  78 20 2b 20 72 6f 74 61  74 69 6f 6e 20 66 75 6e  |x + rotation fun|
00000ba0  63 74 69 6f 6e 73 20 0a  20 20 20 20 20 2a 20 20  |ctions .     *  |
00000bb0  20 20 20 20 20 20 20 69  6e 20 53 48 2c 56 48 20  |       in SH,VH |
00000bc0  61 6e 64 20 52 b3 0a 20  20 20 20 20 2a 0a 20 20  |and R..     *.  |
00000bd0  20 20 20 2a 20 41 6c 6c  20 72 6f 74 61 74 69 6f  |   * All rotatio|
...

The 0xb3 byte is present in your source, and doxygen merely moves it from one place to the other.
You can either fix the byte in the source or file or fix the Python script to be more lax with Unicode errors.

I am not yet 100% sure that it makes sense to rush doxygen 1.9.0 to bullseye, but I might try, and if so, I'll increase the severity of this bug to serious.

Paolo


-- System Information:
Debian Release: bullseye/sid
   APT prefers unstable-debug
   APT policy: (500, 'unstable-debug'), (500, 'unstable'), (1, 'experimental')
Architecture: amd64 (x86_64)

Kernel: Linux 5.9.0-5-amd64 (SMP w/4 CPU threads)
Locale: LANG=en_GB.UTF-8, LC_CTYPE=en_GB.UTF-8 (charmap=UTF-8), LANGUAGE=en_GB:en
Shell: /bin/sh linked to /bin/dash
Init: systemd (via /run/systemd/system)
LSM: AppArmor: enabled