sed

source (C)	https://github.com/mirror/sed
home	https://www.gnu.org/software/sed/
manual	https://www.gnu.org/software/sed/manual/sed.html
wiki	https://en.wikibooks.org/wiki/Sed
wiki	https://en.wikipedia.org/wiki/Sed

language

#!/bin/sed -f
s/Hello/Hell/g

"Stream EDitor"
Author: Lee E. McMahon
1974-
line-oriented text processing utility
reads text, line by line, into a temporary buffer called pattern space
each line read starts a cycle
there is a special buffer called hold space used by some commands to hold and accumulate text between cycles
https://en.wikipedia.org/wiki/Leaning_toothpick_syndrome aka when a quoted expresion becomes unreadable because it contains a large number of escape characters
can separate commands with either
- cmd1;cmd2
- -e cmd2 -e cmd2
[address]command address { cmd1; cmd2; } # cannot have a space before ";"

If
- NO address, command is applied to each line
- has address, applies to the line matching
  - address is followed by a (!), inverse the address match
Can be
- 1,2 where 1=startline and 2=endline a comma separated tuple, command is applied from and upto including the matched line "You can think of the 1st address as enabling the action and the second address as disabling it."
- 1~2 where 1=startline and 2=step
- 1,+2 where 1=startline and +2=distance/length
- aregex
  - also supports arbitrary characters as delimitators, with a backslash for the left one
  - \:aregex:
- NUMBER of line (not reset by different files)
- $ represents the last line
- {} to group different commands to the same OR different address
- line addressing symbol
- can mix up line number and regex

<r>	name	description
q	quit
d	delete	the pattern space
D	delete	only the first line from the pattern space, and run again
n	next	reads the next line into the pattern space, and prints previous value there
N	next	appends the next line into the pattern space, in a new line
p	print	the pattern space
P	print	the first line on the pattern space
=		prints the current line number plus a new line
y		srcstr/dststr (tr)ansliterates src by dst
l		list the pattern space unambigously, nice for debugging
r	reads	from provided filename
w	writes	to provided filename
t		jump to given tag if previous substitution succeded
T		jump to given tag if previous substitution failed
[line]a		appends, after current pattern space
[line]i		inserts, before current pattern space
[addr]c		change, replaces current pattern space with given argument
s		/pattern/replacement/flags
		uses address for pattern if missing
		replacement can be ("&" pattern matched, or \N where N is the N-match group)
		[w FILE,g,p,I,NUMBEROFMATCH]
		hk
h	hold	copy pattern -> hold
H	hold	append \npattern -> hold
g	get	copy hold -> pattern
G	get	append \nhold -> pattern
x	exchange	aka swap content between hold space and pattern space

-e script	inline script
-f script-file	path to script-file
-n	only output via the p command
-i	inplace edit
-i.bkp	inplace edit, saves old version with the .bkp extension

	flag	description
p		prints every line twice
p	-n	prints every line
1p	-n	prints first line
$p	-n	prints last line
1,3p	-n	prints lines 1 through 3
20,$p	-n	prints from line 20 to end
#,$p	-n	remove before comment (#)
=;n		interleaves printing line number, every 2 numbers (1,3..) and each line
$=	-n	prints the number of lines (slow?)
$a 8.8.8.8 google.com	-i /etc/hosts	appends ip/hostname at the end
1i #Managed by sed	-i /etc/hosts	inserts before line 1, a comment
$d	-i /etc/hosts	deletes last line
2d		delete line 2
1,10d		remove the first 10 lines
/^ /d		filters out lines starting with space
/^ *$/d		filters out lines containing only spaces
50,$d		deletes from line 50 to the end
/needle/d		deletes lines containing "needle"
1,/^$/d		deletes from 1st line to the first blank line
/^(#¦$)/d	-E	remove comments and empty lines
/^#/d;/^$/d		remove comments and empty lines
/^\s*(#¦$)/d	-E	remove comments, indentend comments, and empty lines
---!s/–/\\(em/g		on all lines that do not have 3(-), replace 2(-)
s/.*/Hello/;q		reads 1st line of input and prints "Hello"
s/needle//g		deletes "needle" from lines
s/.$//		dos2unix, aka CRLF to LF
/ant/s/needle//g		delete needle on lines containing "ant"