Replace each line with different text using 'sed' (shell script)

Question

I have an XML file with the following structure:

    <root>
       <station name = "insert_text"/>
       <station name = "insert_text"/>
       <station name = "insert_text"/>
    </root>

I wanted to replace the text "insert_text" with the values from a text file which looks like:

    Station1
    Station2
    Station3

After reading the input file above, the generated XML file should look like:

    <root>
       <station name = "Station1"/>
       <station name = "Station2"/>
       <station name = "Station3"/>
    </root>

The script I now use is as follows and it replaces all the 3 lines only with 'Station1'.

    while read a
    do 
        sed -i -e "s/insert_text/$a/g" filename.xml
    done<inputfile.txt

What should I change in order to replace each of the lines in the XML file with different texts?

Every time you write a loop in shell to manipulate text you have the wrong approach. See for example unix.stackexchange.com/questions/169716/…. This is a job for awk. — Ed Morton
– Ed Morton, Commented Jan 29, 2016 at 17:06

nu11p01n73R · Accepted Answer · 2016-02-01 03:32:35Z

3

Problem

The issue with your code is that once it replaces insert_text in the file, there are no more insert_texts to be replaced by Station2 and Station3.

Solution

The solution is to quit the replacing once the first insert_text has be replaced, and read the next line from the file

You can do this by

while read a;     
do          
   sed  "1,/insert_text/ s/insert_text/$a/; " input;     
done < file

What it does?

1,/insert_text/ This is an address range, tells the sed to do the replacing from the line 1 till the first insert_text. No other insert_texts are replaced

OR

sed  "/insert_text/{s/insert_text/$a/;q;}" input;

edited Feb 1, 2016 at 3:32

answered Jan 29, 2016 at 14:07

nu11p01n73R

26.8k3 gold badges42 silver badges52 bronze badges

Sign up to request clarification or add additional context in comments.

7 Comments

William Pursell Over a year ago

Good approach, but I would implement this with sed "/insert_text/{s//$a/g; q;}"

nu11p01n73R Over a year ago

@WilliamPursell Yeah, that should work in gnu-sed, I was testing in BSD sed and was not able to make it work.

mklement0 Over a year ago

P.S.: BSD Sed requires the last command in a block to be ;-terminated, unlike GNU Sed (... q; } vs. { ... q }). That's why @WilliamPursell's command works even with BSD Sed, but the one you quoted in your answer does not. Also note his omission of the regex in the s// call, which conveniently reuses the one that was used to match the line.

ghoti Over a year ago

@mklement0 - I was just going to point that out; you beat me to it. Thanks for keeping the torch burning for us BSD users. ;-)

nu11p01n73R Over a year ago

@mklement0 The ; was something that I missed :(. Thanks for the tip that s// will replace the already matched pattern. That was new for me

|

Juan Diego Godoy Robles · Accepted Answer · 2016-01-29 19:21:24Z

3

I think awk is better suited for this task:

Demo

$ cat inputfile.dat 
Station1
Station2
Station3
$ cat filename.xml 
   <root>
       <station name = "insert_text"/>
       <station name = "insert_text"/>
       <station name = "insert_text"/>
    </root>


$ awk 'NR==FNR{a[++c]=$1;next}/station name/{sub("insert_text",a[++i])}1' inputfile.dat filename.xml 
   <root>
       <station name = "Station1"/>
       <station name = "Station2"/>
       <station name = "Station3"/>
    </root>

edited Jan 29, 2016 at 19:21

answered Jan 29, 2016 at 13:53

Juan Diego Godoy Robles

15k2 gold badges43 silver badges57 bronze badges

Comments

ghoti · Accepted Answer · 2016-02-01 03:38:43Z

Another awk take on this... Rather than reading one file into a an array in memory and then reading the second file for text to replace, you can stream through both files at once:

$ cat inp1
Station1
Station2
Station3
$ cat inp2
    <root>
       <station name = "insert_text"/>
       <station name = "insert_text"/>
       <station name = "insert_text"/>
    </root>
$ awk -vtext="insert_text" '$0~text{getline stn <"inp1"; sub(/insert_text/,stn)} 1' inp2
    <root>
       <station name = "Station1"/>
       <station name = "Station2"/>
       <station name = "Station3"/>
    </root>

This approach will be advantageous if your list of stations is larger than you can comfortably fit in memory.

Broken out for easier reading, here's what the script does:

$0 ~ text {                 # run these commands if we find your text
  getline stn <"inp1"       # read a new input line from the file "inp1"
  sub(/insert_text/,stn)    # do the substitution.
}
1                           # shorthand for "print the current line".

Collectives™ on Stack Overflow

Replace each line with different text using 'sed' (shell script)

3 Answers 3

7 Comments

Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

3 Answers 3

7 Comments

Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related