Re: [pdf-devel] Text module unit tests failing

[jemarch]

Hi Aleks.

   Seems that a recent change in the code made all the Text module
   unit tests fail when calling pdf_text_init().

   http://gnupdf.org/Lib:Unit_Testing_Report

   It's related to the Locale:
   libgnupdf: ***DEBUG BASE***:base/pdf-text-context.c:138: Invalid locale
   info detected! 'C'

   I can reproduce it in my local copy. Any hint on what could be
   happening?

I can reproduce it in two systems.

A eeepc running Debian GNU/Linux:

  $ uname -a
  Linux termi 2.6.25-2-686-bigmem #1 SMP Thu Jun 12 17:11:59 UTC 2008 i686 
GNU/Linux
  $ locale
  LANG=
  LC_CTYPE="POSIX"
  LC_NUMERIC="POSIX"
  LC_TIME="POSIX"
  LC_COLLATE="POSIX"
  LC_MONETARY="POSIX"
  LC_MESSAGES="POSIX"
  LC_PAPER="POSIX"
  LC_NAME="POSIX"
  LC_ADDRESS="POSIX"
  LC_TELEPHONE="POSIX"
  LC_MEASUREMENT="POSIX"
  LC_IDENTIFICATION="POSIX"
  LC_ALL=

and in chapters.es.gnu.org:

  $ uname -a
  Linux chapters.gnu.org 2.6.21.5-grsec-xxxx-grs-ipv4-64 #6 SMP Mon Sep 17 
18:26:16 CEST 2007 x86_64 GNU/Linux
  $ locale
  LANG=en_GB.UTF-8
  LC_CTYPE="en_GB.UTF-8"
  LC_NUMERIC="en_GB.UTF-8"
  LC_TIME="en_GB.UTF-8"
  LC_COLLATE="en_GB.UTF-8"
  LC_MONETARY="en_GB.UTF-8"
  LC_MESSAGES="en_GB.UTF-8"
  LC_PAPER="en_GB.UTF-8"
  LC_NAME="en_GB.UTF-8"
  LC_ADDRESS="en_GB.UTF-8"
  LC_TELEPHONE="en_GB.UTF-8"
  LC_MEASUREMENT="en_GB.UTF-8"
  LC_IDENTIFICATION="en_GB.UTF-8"
  LC_ALL=

Running gdb in the eeepc I got the following sigfault:

  Program received signal SIGSEGV, Segmentation fault.
  0xb7d803a3 in strlen () from /lib/i686/cmov/libc.so.6
  (gdb) bt
  #0  0xb7d803a3 in strlen () from /lib/i686/cmov/libc.so.6
  #1  0xb7ec180d in pdf_text_substitute_line_ending (text=0xa024b20, 
      new_eol=0x0) at base/pdf-text-filter.c:251
  #2  0xb7ec34b3 in pdf_text_filter (text=0xa024b20, filter=1)
      at base/pdf-text.c:1284
  #3  0x08075c55 in pdf_text_filter_001 (_i=0) at base/text/pdf-text-filter.c:91
  #4  0x0807e421 in srunner_run_all ()
  #5  0x0807b767 in main () at runtests.c:43

That is not strange, since it seems that pdf_text_context_init fails
and thus the global text data structure is uninitialized.

I will debug it deeper tomorrow.