[SOLVED] string search in .odt document directories

Alok Rai · 03-01-2021, 02:08 AM

Is it possible to use some grep-like command to search for a particular string in a whole directory of .odt documents?

I imagine that one could search individual documents using "search", one by painful one - but zapping a whole directory?

I am using LM 19 , XFCE variant.

lvm_ · 03-01-2021, 02:24 AM

File indexers which come with major desktop environments - baloo for KDE and tracker for Gnome can do it, but they are CPU and disk hogs. Not sure if XFCE has a similar tool. Actually odt file is just a buch of zipped xml files, so quick and dirty way would be to write a script unzipping them and grepping inside xml starting from /content.xml or maybe using a more advanced xml grep tool like xml_grep

syg00 · 03-01-2021, 02:43 AM

When looking for text I don't care about the xml - just unzip and pipe to normal grep. Stick it in a loop of choice.

Alok Rai · 03-01-2021, 03:22 AM

I fear I might have sounded more capable than I am! But - how would I unzip all the .odt files in a particular directory (and sub-directories) and then pipe them to grep?

And yes, thanks for the warning - I will take a careful backup of everything before I try any operation on them.

syg00 · 03-01-2021, 03:26 AM

Quick search found this - have a look at the script. Shouldn't affect any of your current files adversely.

Alok Rai · 03-01-2021, 03:38 AM

I probably need a Bash tutorial in order to make sense of this -

#!/bin/bash

find . -type f -name "*.od*" | while read i ; do
[ "$1" ] || { echo "You forgot search string!" ; exit 1 ; }
unzip -ca "$i" 2>/dev/null | grep -iq "$*"
if [ $? -eq 0 ] ; then
echo "string found in $i" | nl
fi
done

- but thanks! Will work at it.

hish2021 · 03-01-2021, 06:11 AM

Another tool is recoll. It's also an indexer. You can set it to run when you want it to and it's very customizable. It should be in your distro's repos.

allend · 03-01-2021, 10:33 AM

As an alternative, using libreoffice to convert .odt files to ,txt files

Code:

#!/bin/bash

# Script to search all .odt files in top directory specified in $1 for string specified in $2

topdir=$1
shopt -s globstar

for f in "$topdir"/**/*.odt; do
  libreoffice --convert-to "txt:Text (encoded):UTF8" --outdir "/tmp" "$f" 1>/dev/null
  tmpfile=${f/%odt/txt}
  tmpfile=/tmp/${tmpfile##*/}
  if $(grep -iq "$2" "$tmpfile"); then
    echo "Found $2 in $f"
  fi
  rm "$tmpfile"
done

shopt -u globstar

shruggy · 03-01-2021, 10:44 AM

Actually, the package unzip also provides the zipgrep command.

There's also odt2txt which is much faster than converting with the libreoffice command line (I mean the standalone package of that name, not the symbolic link /usr/bin/odt2txt provided by unoconv which is more or less the same as the headless libreoffice).

Alok Rai · 03-01-2021, 11:14 AM

Thank you, wonderful community! I have several options now: I can convert .odt files to .txt files, and then just run a search using grep. Or, I could experiment with using recoll. Or ack.

Thanks once again.

shruggy · 03-01-2021, 11:36 AM

ack? I don't think ack can search inside compressed files.

Alok Rai · 03-02-2021, 12:18 AM

Thanks, shruggy. ack can't, can it. I believe recoll can.

syg00 · 03-02-2021, 12:21 AM

I reckon shruggys suggestion of zipgrep is a better option. I have some very bad experiences of indexers - to the extent I disable them at every system install.

Alok Rai · 03-02-2021, 01:53 AM

I just looked up zipgrep, and it sounds like the answer to my prayers! One further question, though - is it possible to use zipgrep to search through a whole directory containing .odt files? Or would I first have to zip up the directory, so that I have a file called, say, documents.zip - and then run zipgrep on it?

shruggy · 03-02-2021, 04:49 AM

/usr/bin/zipgrep is just a shell script, and AFAICS, it only works with one zip file. But you can invoke it in a loop:

Code:

for f in documents/*.odt
do zipgrep pattern "$f"
done