Read lines and match against pattern

Question

I can't figure this out. I need to look at every line in a file and check whether it matches a word that is given in a variable.

I started with command read, but I don't know what I am supposed to use after that. I tried grep, but I probably used it wrongly.

while read line; do 
  if [ $condition  ] ;then echo "ok" fi
done < file.txt

Answer 1

Here's a quickie for you, simply what we're doing is

Line 1: While reading file into variable line

Line 2: Match a regex, echo the $line if matching the word "bird" echo that line. Do whatever actions you need here, in this if statement.

Line 3: End of while loop, which pipes in the file foo.text

#!/bin/bash
while read line; do
  if [[ $line =~ bird ]] ; then echo $line; fi
done <foo.text

Note that "bird" is a regex. So that you could replace it with for example: bird.*word to match the same line with a regular expression.

Try it with a file like so, called foo.text with the contents:

my dog is brown
her cat is white
the bird is the word

Answer 2

The easier way to do this, is using grep (or egrep).

grep bird file.txt

Answer 3

Here's a simple script that passes in two arguments (filename and variable)... There's also a hardcoded version beneath that.

#!/bin/bash

[ $# -lt 2 ] && echo "wrong number of args" && exit 1
file=$1
var=$2
while read line; do
    if  echo "$line" | grep -q "$var"; then echo "ok"; fi
done < "$file"
exit 0

[user@myserver ~]$ myscript file.txt some_text

Alternatively, if you prefer to hardcode the file (and variable) rather than passing in as an arguments...

while read line; do
    if  echo "$line" | grep -q "some_text"; then echo "ok"; fi
done < file.txt

Answer 4

Depending on what processing you want to do for each line, sed is another option. I'd only recommend it if you already intend to use sed on each line, though.

sed -n '/regex/p' foo.text

I'm using sed's -n option to suppress normal output. And I'm using the p command to print only the matching lines.

You could combine this with editing commands, too (the processing I was referring to).

sed -n '/regex/{s/pattern/replacement/;p}' foo.text

If you don't want to do any processing on each line, then I agree with steviethecat's answer that grep is the best solution, being the cleanest and fastest.

grep regex foo.text

If you want to do processing that requires additional Unix commands, then (and only then) do I recommend using read.