Sed

More Complex Sed
Compliments of Dr. Dailey

Command	Description
$ cat simple Here is a file containing some words. A total of 3 lines have been used. This line makes the last one right.	A file.
$ sed -n /A/,3p simple A total of 3 lines have been used. This line makes the last one right.	Printing from the first line that contains 'A' through line number 3 of the file.
$wc -l simple 3 simple ---------------------------------------- $c=3 $sed -n /A/,"$c"p simple A total of 3 lines have been used. This line makes the last one right.	Suppose we wished to print from /A/ through the end of the file, but didn't know how many lines there were. We can use wc -l to tell us how many lines, but then suppose we wish to use that output in our sed statement. The output of the wc -l command is non-numeric and cannot directly be used. So how might we use sed to transform the output of wc -l to just the numeric part?
Sample
$c=`wc -l simple\|sed s/.simple//` $echo $c 3 $ sed -n /A/,"$c"p simple A total of 3 lines have been used. This line makes the last one right.	We define $c as the number of lines in the file (by catching the output of wc and removing everything after the whitespace). Then we print every line of simple starting with the first line which contains uppercase 'A' until the end of the file.
$wc -l simple\|sed "s/$[^' ']$\ [^' ']/\1/" 3	If we wanted to generalize the above so it works with any file and not just 'simple', then we could look at any blanks which occur between strings of nonblanks and delete from there to the end of the line by remembering, and later recalling, the preceding non-blanks.
$ wc -l simple\|sed "s/\ \ [^' ']//2" 3 $c=`wc -l simple\|sed "s/\ \ [^' ']//2"` [ddailey]$sed -n /A/,"$c"p simple A total of 3 lines have been used. This line makes the last one right.	Another (perhaps simpler?) approach would be to second occurence of non-blanks following blanks and simply delete those.
$ sed y/aeiou/iouae/ simple Horo us i fulo cantiunung samo wards. A tatil af 3 lunos hivo boon esod. Thus luno mikos tho list ano rught.	Transform occurrences of a to i; e to o; i to u; o to a; and u to e throughout the file.
$ sed s/\ .e/\&/g simple Here is a file containing some words. A total of 3 lines have been used. This line makes the last one right.	Whenever e occurs as the second character within a word (something preceded by a blank), append an asterisk to it. [Note: '\ ' matches the space character, while \& refers to whatever pattern was found on the left.]
sed s/\ .e//g simple Here is a file containing some words. A total of 3 lines haveen used. This line makes the last one right.	The same as above, but without the recall (through '&') of the matched string.
$ sed s/.t/\&\&/g simple Here is a file contntaining some words. A t tototal of 3 lines have been used. This line makes t the lastst one rightht.	Whenever t occurs, rewrite it and the character which precedes it twice.
$ cat f4 audiogalaxy.jpg::pic1.jpg weather.jpg::pic2.gif dialpad.jpg::wp156.htm scour.jpg::wp156.htm dailey.jpg::Royas11cpsc130sect2.html mellon.jpg::pic1.jpg hotmail.jpg::130.2assignment11.html napster.jpg::j1.gif	Here's a file in which each line has material to the left and right of two colons.
$ sed 's/$.$::$.$/\2\1/' f4 pic1.jpgaudiogalaxy.jpg pic2.gifweather.jpg wp156.htmdialpad.jpg wp156.htmscour.jpg Royas11cpsc130sect2.htmldailey.jpg pic1.jpgmellon.jpg 130.2assignment11.htmlhotmail.jpg j1.gif**napster.jpg	We switch the order of the material, by remembering strings. (string) remembers a string in the pattern space. The first of these can later be recalled as \1 (or \n for the nth remembered string). Since '(' is a metacharacter, we have to use '\(' to refer to it. For fun we replace '::' by '**'.
$head -5 common5 abide abode about above abuse $sed '4s/.*/ha!/g' common5\|head -5 abide abode about ha! abuse	Replacing the characters in the 4th line of a file by a particular string.
$sed -n 1~2p common5\|head -5 abide about abuse acrid acute	Printing every other line of a file.(m~n means print every nth line starting at line m).

Multi-line replacements

Command	Description
$cat simple Here is a file containing some words. A total of 3 lines have been used. This line makes the last one right. $sed 's/\ /\ > /g' simple\|head -7 Here is a file containing some words.	Putting each word of a file on a separate line: We substitute the literal character 'space' by the literal character 'newline' (signalled by actually typing a carriage return). By putting the substitute part of the sed command inside single quotes, allows the command to straddle more than a single line and hence for the carriage return to be seen as a literal character, rather than an end-of-command mark.
$cat simple Here is a file containing some words. A total of 3 lines have been used. This line makes the last one right. $sed 'N;s/\n/ /' simple Here is a file containing some words. A total of 3 lines have been used. This line makes the last one right.	In order to insert the newline character (matched in the pattern space as \n), one has to look across multiple lines, by using the command N -- meaning load next line, as well as the current, into the pattern space. sed 's/\n/ /' simple won't do it.
cat alph10 a b c d e f g h i j $ sed 'N;N;N;N;s/\n//g' alph10 abcde fghij	reapply the above principle several times, placing five lines into a single buffer, prior to the replacement of the newline character.
Sample