Sed_Cheatsheet.txt

Sed cheat sheet

sed 's/str/repstr/g' file.txt
cat file | sed s 's///g'
sed 's/t/T/g' file > new.txt  -> Replace all lowercase t with Uppercase T and
				put it in a new file.

substitute / this / for that/
substitute / this / for that/ globally

Delimiters do not need to be / they can be any char, as long as it's
consistent. 
	sed 's#t#T#g' file 
works just like 
	sed 's/t/T/g/ file 
Handy if you dealing with directory paths.

^ -> begins with
$ -> ends with

& is the matched string, so:
	sed 's/is/*&*/g' file
		is the same as
	sed 's/is/*is*/g' file
Saves you having to type a long string twice.

sed -i -> Edit inline, overwrite the file
		Without that, the file is untouched and it's just what's being
		echoed out to the console that is being effected.

sed 's/[0-9]/*/g' file -> Replace all numbers
sed 's/[^0-9]/*/g' file -> Replace everything but numbers
    's/[^a-z]/*/g'	Everything but lowercase letters
    's/[^A-Z]/*/g'      Everything but Uppercase letters.
    's/[^A-z]/*/g'	Everything but Alphabetic chars (Note: also excludes some punctuation \[])
    's/[^a-zA-Z ]/*/g'	Everything but Alphabetic letters and spaces.
    's/[^a-zA-Z\]/*/g'	  	"	"	"	and \
    's/[0-z/ ]/*/g'	This should grab EVERYTHING

^ means everything but. If you want to make it just those chars, remove the carat.
If you want to use a regex to include [] do it like [][a-z ] 
	(ie, open regex, closing square bracket, opening square bracket, close regex.)


Making more than one edit across one file/stream
	sed 's/Twice/None/g;s/Three/Two/g' file.txt
	sed 's/Twice/None/g;s/Three/Two/g;s/something/else/g' file.txt

The above saves you from having to pipe sed into sed. As long as it's between
the two ' and separated by a ; it will be treated fine without spinning up a
separate process.

sed 's/\w*.//' text -> Replace the first word in a line with nothing.
	\w* Find the first word
	. include the trailing space aswell

sed 's/\w*$//' text -> Replace the last word in a line with nothing.
	This will struggle with special characters and numbers.

sed -n -> Silent mode. 
sed -n 's/T/t/p' text (Run in silent mode, but the p is showing the matching
			lines. Can be combined with a g for global.)
sed 's/that/*/Ig' file -> I ignores case.

Get SED to read in commands from a file.
	In file mysedscript
		s/T/t/g
		s/e/E/g
		s/\w*.//
	
	sed -f mysedscript text

Replace a specific word:
	sed 's/\bnew\b/old/g'
	\b (bounds) defines the limits of a word. Prevents you from having to define
each specific instance of a word. (ie: s/old / and s/old$/)

Remove a line based on a string:
	sed '/new/d' text2 (if you want case insensitive I has to go before d.)

sed '4,$ d' text2 -> display the first 3 lines of a file (like head -n 3 text2)
		go to line 4, find the end of line character, and delete all
		the subsequent lines.
sed '4 q' text2 -> do whatever, until the 4th line and then quit.

sed '!1{/^#/d}' Skip the first line, remove all lines begining with #
sed '!2{//}' skip the second line. (not two lines, line 2)

sed -n 'p;n' text2 -> prints every other line, starting from line 1.
sed -n '1!n;d' -> prints every even line (but deletes the other ones when
		displaying