0

For a migration process, I need to do some replacement in my bash script.

So in my .txt files, I have these references for example:

[[File:My Image.png|400px|thumb|center|My Image]]
[[File:My Image.png|400px|thumb|center]]
[[File:My Image.png|400px|thumb]]
[[File:My Image.png|400px]]
[[File:My Image.png]]

What I need to do is to replace all these occurrences with this line (only - so no more size, description, or other informations):

[[File:My Image.png]]

So, I tried to do a is to build a PCRE regex to extract all images names with:

/File:(.*\..{3})/g

I've built this final command to find all occurrences in my .txt files and extract image name with my regex:

find . -name "*.txt" | xargs perl -i -p -e 's/File:(.*\..{3})/$1/g'

But, I've encountered some problems as:

  • An error:

xargs: unterminated quote

  • And finally I don't know how to use extracted image name to replace all occurrences (complete lines)

PS: I'm on MacOS system and use bash v4

0

Try

find . -name '*.txt' -exec perl -i -pe 's/File:[^|]+\K\|[^]]+//g' {} \;
  • File:[^|]+ match File: followed by non | characters
  • \K so that we don't have to capture the preceding string and put it back in replacement section
  • \|[^]]+ match | followed by non ] characters to be deleted
  • Can also use sed -i '' 's/\(File:[^|]*\)|[^]]*/\1/g'instead of perl

Further reading:

  • 1
    Hi, Thank you for your help and your very great explanation, amazing, it' working great as expected and it's very simple. Nice to see some further reading also, I will take a look as soon as possible ;) Have a nice day – Sébastien Robert Apr 4 '18 at 10:53
  • 1
    That regex will also match for example File:x|y (without the preceding [[). – haukex Apr 4 '18 at 11:00
  • @haukex true, and it is a good point :) I was modifying OP's attempt which didn't include the starting [[ or ending ]], so probably they do not make a difference for given input... – Sundeep Apr 4 '18 at 11:12
1

I've written a new regex which matches the whole [[...]], and replaces it by only the things you want to keep. It assumes that the filenames don't contain pipe | characters or the terminator ]]. I can't reproduce your issue with xargs, but I replaced it by find's -exec option anyway; the following works for me on Linux.

find . -name "*.txt" -exec perl -i -pe 's/(\[\[File:[^|]*).*?(\]\])/$1$2/g' '{}' +

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.