[SOLVED] string search in .odt document directories

Alok Rai · 03-02-2021, 04:57 AM

Thanks, shruggy!

So, at the terminal prompt, I would run this command -

zipgrep pattern "$f" documents/*.odt

And then I would get a list that I could pipe to "more"????

Is that what "invoke it in a loop" means? Sorry if I am being dense, but please bear with me..

Thanks.

hish2021 · 03-02-2021, 05:13 AM

Would zipgrep be more convenient than the script mentioned in posts #5 and #6 if your usage is what was described in the original post?

shruggy · 03-02-2021, 05:13 AM

Quote:

Originally Posted by Alok Rai

So, at the terminal prompt, I would run this command

No. At the terminal prompt you can either use the for loop as shown above

Code:

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

or find

Code:

find documents/ -type f -name \*.odt -exec zipgrep pattern {} \;

Another possibility are grep-like tools that allow both searching inside archives and specifying multiple input files/directories, e.g. ugrep:

Code:

ug -z pattern documents/*.odt

chrism01 · 03-02-2021, 10:42 PM

According to the manpage https://linux.die.net/man/1/zipgrep, it offers all the usual egrep options https://linux.die.net/man/1/egrep, which of course include -r (recursive

) ....

Alok Rai · 03-03-2021, 02:05 AM

Thanks, Chris! That should do me fine!

Alok Rai · 03-03-2021, 04:50 AM

Ah well - spoke too soon!

I tried using zipgrep. Here is the command - and the output -

$ zipgrep pattern "Hindutva" *.odt
zipinfo: cannot find or open Hindutva, Hindutva.zip or Hindutva.ZIP.
/usr/bin/zipgrep: 97: test: -eq: unexpected operator
/usr/bin/zipgrep: 100: test: Illegal number:

I am in the directory Documents - and wish to search therein.

Help!

shruggy · 03-03-2021, 04:54 AM

pattern is not a command word, it stays for the pattern itself

Code:

zipgrep "Hindutva" file.odt

Alok Rai · 03-03-2021, 05:01 AM

Now, this:

$ zipgrep "Hindutva" Documents/*.odt
zipinfo: cannot find or open Documents/*.odt, Documents/*.odt.zip or Documents/*.odt.ZIP.

No zipfiles found.
/usr/bin/zipgrep: 97: test: -eq: unexpected operator
/usr/bin/zipgrep: 100: test: Illegal number:

shruggy · 03-03-2021, 05:03 AM

Sure, because you can specify only one file as an argument to zipgrep. Besides, if your current directory is already ~/Documents then specifying Documents/*.odt is wrong.

I agree with what hish2021 said above, I also think a Bash script as suggested in #8 would serve you better.

Alok Rai · 03-03-2021, 05:19 AM

Tried the command on a different directory - and managed to get some results. But am now even more puzzled - because it threw up filenames where the body of the files contains nary a mention of "Hindutva" - for instance, this - from a file named alice_impossible -

Alice laughed. “There’s no use trying,” she said: “one can’t believe impossible things.”
“I daresay you haven’t had much practice,” said the Queen. “When I was your age, I always did it for half-an-hour a day. Why, sometimes I’ve believed as many as six impossible things before breakfast.”
[Carroll, Through the Looking Glass, chapter 5.]

Weird, isn't it?

shruggy · 03-03-2021, 06:05 AM

zipgrep treats an .odt file as a zip archive. It will search not only in the document text, but in the metadata, too. E.g. if the document was created by user Hindutva it will be matched.

Alok Rai · 03-03-2021, 07:07 AM

Brilliant, thank you! So, now if I can find a way of accessing the metadata, and manipulating that, I can, in effect, attach keywords to documents!

So much to learn, thank you!

shruggy · 03-03-2021, 07:27 AM

You can manipulate LibreOffice document metadata with unoconv

Code:

unoconv -f odt -o output.odt -M Author="Alok Rai" input.odt

hish2021 · 03-03-2021, 08:04 AM

I tried to understand the code posted in #6

Code:

#!/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

In the process, I changed things to this:

Code:

#!/bin/bash

[ "$1" ] || { echo "You forgot the search string!" & exit; }
find ~/Documents -type f -name "*.od*" | while read -r i ; do
  unzip -c "$i" | grep -iq "$1" && echo "string found in $i"
done

I first had just "while read i" but https://www.shellcheck.net/ suggests using "read -r" (https://github.com/koalaman/shellcheck/wiki/SC2162).

Alok Rai · 03-04-2021, 12:10 AM

This is waaaaaaay beyond my range, but thanks a ton! I'll come back when I am big and strong....