SED command examples

PURPOSE
Learn SED command by examples

ENVIRONMENT
LINUX: shell/bash
WINDOWS:cygwin

EXAMPLES

• we are going to work with the following file

$ cat file2
1. one
2. two
3. three
4. four
5. five

• replacing "one" by "OOO"...instead of "OOO" if you put "" it deletes "one"

$ sed s/one/OOO/ file2
1. OOO
2. two
3. three
4. four
5. five

• same as above multiple replace ( "one" by "OOONE" and "two" by "TWOOO")

$ sed -e s/one/OOONE/ -e s/two/TWOOO/ file2
1. OOONE
2. TWOOO
3. three
4. four
5. five

• replace character by character(here "o" by "O" and "t" by "T")

$ sed 'y/ot/OT/' file2
1. One
2. TwO
3. Three
4. fOur
5. five

• putting matches in a file and then running sed

$ cat matches

s/one/1111/

s/two/TTTT/

$ sed -f matches file2

1. 1111

2. TTTT

3. three

4. four

5. five

Finding pattern by line number or word

• syntax: （in the following syntaxes X、Y are numbers and <word> is a word)



①sed -n Xp filename  : prints line X of file
②sed -n /<word>/p filename: prints line where <word>
③sed -n 'X , Yp' filename : prints file from line N to line Y 
④sed -n 'X ~ Y'p filename  : prints every Y number of line after line X
⑤sed -n 'X , /<word>/p'  filename : prints file from line X to <word>
⑥sed -n '/<word>/, Yp' filename : prints file from <word> to line Y
⑦sed -n '/<word>/, +Yp' filename : prints from <word> and adds Y lines number
⑧sed -n '/<wordA>/ , /<wordB>/p' filename : prints file from <wordA> to <wordB>

※ without the single quotes it mostly works,just try it yourself

• ①  prints line 3

$ sed -n 3p file2

3. three

• ② prints line containing the word "two"

$ sed -n /two/p file2

2. two

• ③ prints line 3 to 4

$ sed -n 3,4p file2

3. three

4. four

• ④ prints every 2 lines after the 1st line

$ sed -n '1~2'p file2

1. one

3. three

5. five

• ⑤ prints line 3 to the line where "four" is found

$ sed -n '3,/four/p' file2

3. three

4. four

• ⑥ prints the line where "two" is found up until the 4th line

$ sed -n '/two/,4p' file2

2. two

3. three

4. four

• ⑦ prints 3lines after the matching the searched word "one"

$ sed -n '/one/,+3p' file2

1. one

2. two

3. three

4. four

• ⑧ prints everything between line where "two" is found to line where "four" is found

$ sed -n /two/,/four/p file2

2. two

3. three

4. four

WE can use regex with SED too

• erase "four" and delete line containing the word "six"

$ sed -e 's/four//;/six/d' file2 ( same as sed -e s/four// -e s/six// file2)

1. one

2. two

3. three

4.

5. five

• find a word which start with "o" and ends with "e" and replace by 3 times the match

$ sed 's/o\(.*\)e/\1\1\1/g' file2

1. nnn

2. two

3. three

4. four

5. five

• & represents the match of sed so lets replace "one" by "<one-one-one>"

$ sed 's/one/<&-&-&>/' file2

1. <one-one-one>

2. two

3. three

4. four

5. five

/div>

• using other delimes than "/"

$ echo "http://www.google.com/" | sed @http://www.google.com/@http://www.youtube.com/@

http://www.youtube.comsomelinks/

• writing changes directly to file(replacing "one" by "1111" directly in the target file)

$cp file2 file3    ( lets copy file2 into another file to not loose the original)
$sed -i s/one/1111/ file3
$ cat file3
1. 1111
2. two
3. three
4. four

5. five

• replacing options "g" and "i"
• 
  
• the following replaces only the 1st match

$ echo "oneONEone" | sed s/one/111/

111ONEone

• by adding the "g" option it replaces the match "Globally"

$ echo "oneONEone" | sed s/one/111/g

111ONE111

• by adding the "i" option it replaces the match "Ignoring " the case

$ echo "oneONEone" | sed s/one/111/gi

111111111