CODEWITHSUNDEEP
bashcommand-linesedxargs
Jesper
Jesper

Reputation: 810

Regexp lines from file and run command

I have a file with output from the identify command, looks like this (following format: FILENAME FORMAT SIZE METADATA)

/foo/bar.jpg JPEG 2055x1381 2055x1381+0+0 8-bit DirectClass 
/foo/ham spam.jpg JPEG 855x781 855x781+0+0 8-bit DirectClass
...

Note that the filenames can contain spaces! What I want to do is to basically run this on each of those lines:

convert -size <SIZE> -colors 1 xc:black <FILENAME>

In other words, creating blank images of existing ones. I've tried doing this with cat/sed/xargs but it's making my head explode. Any hints? And preferably a command-line solution..

Upvotes: 0

Views: 79

Answers (3)

jfg956
jfg956

Reputation: 16750

cat data_file | sed -e 's/\(.*\) JPEG \([^ ]*\) .*/convert -size \2 -colors 1 xc:black "\1"/' | bash

Upvotes: 1

Diego Sevilla
Diego Sevilla

Reputation: 29021

You can do what Michał suggests. Also, if the metadata has a fixed number of words, you could do this easily like the following (supposing you process every line):

FILENAME=`echo $LINE | rev | cut -d\  -f 6- | rev`

(that is, reverse the line, and take the name from the sixth parameter on, then you have to reverse to obtain the filename proper.)

If not, you can use the fact that all the images have an extension and that the extension itself doesn't have spaces, and search for the extension till the first space afterwards:

FILENAME=`echo $LINE | sed -e '/([^.]+) .*$/\1/'`

Upvotes: 0

Michał Šrajer
Michał Šrajer

Reputation: 31182

Assuming, that filename is the string before " JPEG":

LINE="/foo/ham spam.jpg JPEG 855x781 855x781+0+0 8-bit DirectClass"

You can get file name as:

FILENAME=$(echo "$LINE" | sed 's/\(.*\) JPEG.*/\1/')

Upvotes: 1

Related Questions