Tesseract doesn't accept process substitution

Question

I'm making a quick script that is supposed to use OCR tool (tesseract) on image in clipboard to convert it to text and output it. It looks like this:

#!/bin/sh

temp="$(mktemp tmpXXX.png)"
xclip -selection clipboard -t image/png -o > $temp
tesseract $temp stdout 2>/dev/null
rm $temp

What I'm wondering is why doesn't this one-liner tesseract <(xclip -selection clipboard -t image/png -o) stdout work? From what I know, process substitution is supposed to make temporary file (similar to my full script) that tesseract uses as input file. Alas, this leads to an error:

Error in pixReadStream: Unknown format: no pix returned
Error in pixRead: pix not read
Error during processing.

Does anybody have an idea why this happens?

Thanks in advance.

It depends on the OS, but generally process substitution creates a pipe rather than a temporary file. Pipes don't support all the operations that regular files do (e.g. you can't read them out of order), so my guess is `tesseract` is trying to do something that's not compatible with a pipe. — Gordon Davisson, Apr 04 '22 at 17:16

score 2 · Accepted Answer · answered Apr 04 '22 at 18:10

Process substitution, a ksh feature by the way (also available in zsh and bash), not a sh one uses pipes (unnamed on systems that support /dev/fd/n, named otherwise), not temporary files.

Here, tesseract may need to be able to seek within the file or know its size in advance, which is not possible with pipes. It could also expect the file name to have certain extensions.

For a process substitution that uses temporary files, you need the =(...) form which is only available in the zsh shell.

tesseract =(xclip -selection clipboard -t image/png -o) stdout

You can set the $TMPPREFIX (defaults to /tmp/zsh) and/or $TMPSUFFIX (empty by default) to have a control on how the temporary files created by =(...) are named.

TMPSUFFIX=.png
tesseract =(xclip -selection clipboard -t image/png -o) stdout

For instance.

In zsh and bash versions prior to 5.0, and on Linux or Cygwin, you can also do:

{
  xclip -selection clipboard -t image/png -o > /dev/fd/3 &&
    tesseract /dev/fd/3 stdout
} 3<<< ''

As heredocs and herestrings are implemented (used to be in bash) as deleted temporary files and on Linux and Cygwin, opening /dev/fd/n where n is a file descriptor opened on a file opens that file instead of duplicating the fd like it does on most if not all other systems.

Tesseract doesn't accept process substitution

1 Answers1