Hello,
Sebastian Humenda, le dim. 17 avril 2022 20:51:04 +0200, a ecrit:
> E [16/Apr/2022:15:11:58 +0200] [Job 12] Job submission timed out.
> E [16/Apr/2022:15:11:58 +0200] [Job 12] Aborting job because it has no files.
> E [16/Apr/2022:15:24:17 +0200] [CGI] ippfind (PID 19352) stopped with status
> 1!
> E [16/Apr/2022:15:24:17 +0200] [cups-deviced] PID 19344 (driverless) stopped
> with status 1!
>
> The embosser stopped after 2/3rd of the page
So it indeed started printing the image, but didn't manage to finish it?
> and refused to interact with to any key press. I had to force-reboot
> it.
Uh, perhaps there is a low-level communication problem, was that through
USB or Ethernet?
> Any idea how to get more helpful output?
No better idea than just raising the cups loglevel, but I doubt it'll
provide anything useful. Perhaps check dmesg?
> I read the cups-filters docs in /usr, but it looked as if all the conversion
> magic for the image would be done in the background.
Yes, it's done by a cups filter. Since apparently some of the picture
did show up, that did happen fine.
> E [16/Apr/2022:14:50:28 +0200] [Job 9] text conversion pipeline pdftotext
> -raw - - | file2brl -p -Chyphenate=yes
> -CliteraryTextTable=en-us-brf.dis,en_GB.tbl,de-g2.ctb,braille-patterns.cti
> -CinputTextEncoding=UTF8 -CbraillePages=yes -CbraillePageNumberAt=bottom
> -CpageNumberBottomSeparateLine=yes -CprintPages=yes -CprintPageNumberAt=top
> -CpageNumberTopSeparateLine=yes -CpageSeparator=yes -CpageSeparatorNumber=yes
> -CcontinuePages=yes -CcellsPerLine=29 -ClinesPerPage=27 | | addmargins
> failed
>
> The embosser just emitted an empty page.
> It looks a bit worrying that there are two pipes in the conversion command.
That's odd indeed. The script does check that $TRANSLATE is not empty
before putting it there...
> Dissecting this further, I figured out that some translation tables seem to be
> missing:
>
> $ echo 'Hello, world!' |strace file2brl -p -Chyphenate=yes
> -CliteraryTextTable=en-us-brf.dis,en_GB.tbl,de-g2.ctb,braille-patterns.cti
> -CinputTextEncoding=UTF8 -CbraillePages=yes -CbraillePageNumberAt=bottom
> -CpageNumberBottomSeparateLine=yes -CprintPages=yes -CprintPageNumberAt=top
> -CpageNumberTopSeparateLine=yes -CpageSeparator=yes -CpageSeparatorNumber=yes
> -CcontinuePages=yes -CcellsPerLine=29 -ClinesPerPage=27
> [output without strace, config cropped]
> :2: invalid literaryTextTable
[...]
> openat(AT_FDCWD, "/usr/share/liblouis/tables/braille-patterns.cti", O_RDONLY)
> = 6
> stat("/usr/share/liblouisutdml/lbu_files/en-us-brf.dis,en_GB.tbl,de-g2.ctb,braille-patterns.cti",
> 0x7fffd96dd9b0) = -1 ENOENT (No such file or directory)
>
> It seems as if the format to pass multiple braille tables is wrong. The list
> seems to be interpreted like a literal file name. Any suggestions on how to
> fix this?
Mmm, I'm not getting any issue with:
echo 'Hello, world!' | file2brl -p -Chyphenate=yes
-CliteraryTextTable=en-us-brf.dis,en_GB.tbl,de-g2.ctb,braille-patterns.cti
-CinputTextEncoding=UTF8 -CbraillePages=yes -CbraillePageNumberAt=bottom
-CpageNumberBottomSeparateLine=yes -CprintPages=yes -CprintPageNumberAt=top
-CpageNumberTopSeparateLine=yes -CpageSeparator=yes -CpageSeparatorNumber=yes
-CcontinuePages=yes -CcellsPerLine=29 -ClinesPerPage=27
It does print "HEQO1 _W6". In strace there is such bogus stat call, but
after that it tries to open files one by one too.
Which version of liblouis is that?
Samuel
