what is sed?

sed is a “stream editor” in the commandline. It can do a lot of things and I wrote down some basics examples here (to avoid re-googling them in a few months). A great and very thorough guide can be found elsewhere [0]. I only post the most useful examples that I needed so far.

sed typically works by reading an input string, looks for patterns and does something with these patterns. See this basic example:

# replace "a " with "not a "
$ echo "this is a sentence." | sed 's/a /not a /'

The echo command prints “this is a sentence.” and pipes the result (with the “|"-character) to sed which replaces “a " with “not a “. Usually the sed command is followed by ’s/PATTERN/REPLACEMENT/'.

typical usecases

numbers and letters

# replace a single instance of "number" with the letter "a_":
$ echo "012345679" | sed 's/[0-9]/a_/'
a_123456789

# replace a pattern that consists of 0-n instances of "number" with the letter "a_":
$ echo "012345679" | sed 's/[0-9]*/a_/'
a_

the same works for letters:

$ echo "abcdefghijklmnopqrstuvwxyz" | sed 's/[a-z]/A_/'
A_bcdefghijklmnopqrstuvwxyz

$ echo "012345679" | sed 's/[a-z]*/A_/'
A_

matching and replacing a pattern multiple times

# replace each occurrence of "b" with "B" 
$ echo "abababa" | sed 's/b/B/g'
aBaBaBa

using the matched pattern again

The character “&” represents the matched pattern.

# find "b" and replace it with " bb " by doubling the pattern
$ echo "abababa" | sed 's/b*/ && /'
a bb ababa

# do that for each ocurrence of "b"
$ echo "abababa" | sed 's/b*/ && /g'
a bb a bb a bb a

using multiple sed-commands after each other

instead of piping one results from sed to another instance of sed (and so on) sed can match multiple patterns.

# first: replace first instance of a letter with "a_"
# then: replace first instance of a letter with "firstletter"
echo "abcdefghijklmnopqrstuvwxyz" | sed 's/[a-z]/a_/ ; s/[a-z]/firstletter/'

Links:

[0] sed tutorial