Using sed in scripts

Replace all occurences of ‘someText’ with ‘someOtherText’ in someFile

sed "s/someText/someOtherText/g" -i someFile

Explanation:

  • s/someText/someOtherText replaces ‘someText’ by ‘someOtherText’
  • ‘g’ after ‘someOtherText’ stands for ‘globally’, so all occurences of ‘someText’ will be replaced by ‘someOtherText’
  • with -i flag indicated, the replacements will be made in the file. There is no need to pipe the output to any other binary.

Delete all occurences of ‘someText’ from someFile

sed "s/someText//g" -i someFile

So, technically, we replace ‘someText’ with an empty string.

Replace only the first occurence of ‘someText’ variable value with ‘someOtherText’ variable value in someFile

sed -e ':a' -e '$!{N;ba' -e "}; s|${someText}|${someOtherText}"

Rearrange blocks of text in lines

Let’s suppose that input contains some lines which match the pattern:

someText...(any characters)...twoDigits (any digits)

We want to rearrange these lines in the following way:

twoDigits-someText

Solution:

sed "s|^\(someText\)\(.*\)\([0-9]\{2\}\)$|\3-\1|g"

Explanation:

  • in a regular expression (left part), a line is represented in the following way:
    • ‘^’ means the begginning of the line
    • \(someText\) - ‘someText’, represented as a group (the first one). We will need this group later, at the replacemet step
    • \(.*\) - any number ('*') of any characters ('.'), represented as another group (the second one)
    • \([0-9]\{2\}\) - any two ({2}) digits ([0-9]), represented as another group (the third one)
    • ‘$’ means the end of the line
  • the replacement (right part) consists of:
    • the third group (two digits)
    • ‘-’
    • the first group (someText)

Use sub-shells in sed expressions

Replace all current dd-mm-yyyy by yyyy-mm-dd:

sed "s|$(date +%d-%m-%Y)|$(date +%Y-%m-%d)|g"

We get and format current date using date command.