Skip to content

Latest commit

 

History

History
203 lines (176 loc) · 8.07 KB

'sed'.md

File metadata and controls

203 lines (176 loc) · 8.07 KB
Raw

sed

Default:

  • Search line: Process all lines - limit by range /a.../,/b.../ <x>
  • Search string: Process all lines but only 1st matched of each line - global match by s/a.../b.../g
  • Search 1st matched string only: 0,/.../ s/a.../b.../

escaped characters

  • 's/ \| \[ \] \/ \\ \$ \* \. \^ /abc/g' - escape literal in pattern: | [ ] / \ $ * . ^
  • 's/abc/ \\ \/ \& /g' - escape in replace: \ / &
  • 's/abc/ a \' \" ' - escape in insert/append: ' "

variables / quotes

  • $'/...\'...\'.../' : $ + escaped ' inside single quote (or use . wildcard if posible)
  • '/...\x27...\x27.../ : ' hex character
  • "/...\"...\".../" : escaped " inside double quote
  • '/...\x22...\x22.../ : " hex character
  • "/...$var.../" : variable inside double quotes
  • '/...'"$var"'.../' : variable outside quotes (double quote variable with spaces)

search address

  • string / line number / regex

flags

  • -i : in-place edit file
  • -i '.bak' : in-place edit file with optional backup file.bak
  • -e : expressions in sequence
  • -n : line indication

commands

  • '/.../ i\ ...' : i insert before line matched (\ needed for escaped every new line)
  • '/.../ a\ ...' : a append after line matched (\ needed for escaped every new line)
  • '/.../ c\ ...' : c change line matched (\ needed for escaped every new line)
  • '/.../ r <file> : r read file to append after line matched
  • '/.../ d : d delete line matched
  • -n '/.../ p : p print line matched
  • 's/.../.../' : substitute 1st matched (delimiter can be any symbol or character)

comments

  • between each command, the end of lines or between lines
  • not inside new insert/replace string

multiple commands

  • ; : new line
  • \ : escaped new line (includes blank lines, optional last line)
sed 's/.../.../; s/.../.../' file         # optional {...}
sed '/.../ s/.../.../; s/.../.../' file   # limit to matched lines - replace before prepend/append
sed '/.../ { # comment                    # required {...}
# comment
s/.../.../   # comment
a\
...\
\
...
}
' <file>

swap lines - use ex ( line-editor mode of vi )

  • printf '%s\n' 'NmN' 'wq' | ex -s <file> : NmN from line #N to after line #N (N start at 1); - -s silent
  • printf '%s\n' '1m3' 'wq' | ex <file> : move line #1 to after line #3
  • printf '%s\n' '5m0' 'wq' | ex <file> : move line #5 to after line #0 / top

line operations (all matched)
(first matched line only)
* <x> : i, a, d, p, r, c
* <n> : number

  • /.../ : default delimiter

  • \|...|, \#...# : delimiters can be any (but in line search, none default / must be escaped)

  • -n '/.../p' : print line string by matched

  • -n '/.../p; q' : print only 1st line string by matched

  • '<n>q;d' : get line string by line number

  • -n '/.../=' : get line number

  • /.../ <x> : line matched (left escaped delimiter \| for \|...|)

  • '/^.../' : ^ line start with

  • /...$/ : $ line end with

  • /^$/ : blank line

  • '<n> <x>' : <n>th line

  • '1 <x>' : 1st line

  • '$ <x>' : last line

  • '/^a.*b$/ <x>' : start with a and end with b

  • '/^\s*$/ <x>' : blank lines

  • /<1st...>/, /<last...>/ <x> : matched line range by matched (non-greedy)

  • N, N <x> : matched line range by line number

  • 'N, /.../ <x>' : N to 1st matched line range (also for limit only 1st occurrence)

  • /.../, +<n> <x> : line range from matched to next +<n> line

  • /a...\|b...\|c.../ <x> : line matched or \| (between / only)

  • q : print 1st line (then quit)

  • <n>q : print 1st to <n>th line

'/.../{N; <x>
}'
  • : line matched includes Next line (for i, a)
'/.../{n; <x>
}'

: line next to matched

'/.../{n;n; <x>
}'
  • : 2nd line next to matched
-n '/.../{x;p;d
};x

  • : print 1 line above matched

  • $(( $( sed -n '/.../=' $file ) - <n> ))' <x>' : <n> line prior to matched

  • sed -i -e :a -e '/^\n*$/{$d;N;};/\n$/ba' : remove all trailing new lines
    or printf '%s\n' "$( cat file )"

  • sed -i -e '$a\' : add a trailing new line if not exist

string operations
alternative: simple brace expansion

  • /.../ : word

  • /[...]/ : character

  • s/.../.../, s|...|...|,s#...#...# : delimiter (no escape needed)

  • -E 's/.{n}/.../, s/(...){n}/.../' : match repeating characters or word(...)
    all lines - first matched string of each line unless with g

  • non-greedy

    • echo 192.168.1.123 | sed 's/^\(.*\.\)[^.]*$/\1/' : 192.168.1.
    • echo var=abc=de | sed 's/[^=]*=//' : abc=de
    • echo fa fa-list 2x red hide | sed 's/.* \(fa-[^ ]*\).*/\1/' : fa-list
    • cat /path/file | grep 'fa-' | sed 's/.* \(fa-[a-z][^ '"'"'"]*\).*/\1/' | sort -u : list of fa-xxx sorted and -u no duplicates
    • echo 123456789012345 | perl -pe 's/2(.*?)5/a\1b/' : perl - 1a34b6789012345 - ?=shortest

  • 's/a.../b.../' : s substitute 1st matched a... with b...

  • 's/a.../b.../<n>' : substitute <n>th matched

  • 's/a.../b.../g' : g global - substitute all in that line

  • '0,/a.../ s/a.../b.../' : substitute only first matched (range line 1 to matched)

  • 's/a.../b.../i' : i case insensitive

  • '/^a.../ s/^/b.../' : prepend b... to the ^ begining of line start with a...

  • '/^a.../ s/$/b.../' : append b... to the $ end of line start with a...

  • 's/a.../&b.../' : reuse input as & = a...b...

  • 's/a.../b...&/' : b...a...

  • 's|--\(aaa\)--\(bbb)--|==\1\2==|' : reuse input parts \(...\) as \n, --aaa--bbb-- to ==aaabbb==

  • 's/a.../\L&/' : change to lowercase

  • 's/a.../\U&/' : change to uppercase

  • replace each character respectively

sed -i 'y/āáǎàēéěèīíǐìōóǒòūúǔùǖǘǚǜĀÁǍÀĒÉĚÈĪÍǏÌŌÓǑÒŪÚǓÙǕǗǙǛ/aaaaeeeeiiiioooouuuuüüüüAAAAEEEEIIIIOOOOUUUUÜÜÜÜ/' $file
  • 's|/[^/]*$||' : get directory path from from file path
  • sed 's|.*/||' : get filename from file path

wildcards

  • . : single character or space wildcard
  • .* : multiple character wildcard
  • * : 0 or more character
  • \? : 0 or 1 character
  • \+ : 1 or more character

regex list

  • '/[...]/' : any characters in [...]
  • '/[^...]/' : [^ not any characters in ...]
  • '/[^ ]*/' : * zero or more characters that ^ are not space
  • '/^[...]/' : start with any character in [ ]
  • '/[...]$/' : end with any character in [ ]
  • '/a...\|b.../ <x>' : string a... or b... (\ needed for escaped | and cannot use | as delimiter)

mixed regex

  • '/[...]\|.../' : any characters in [...] or pattern ... (\ needed for escaped |)
  • '[^...\{...\}]$' : [^ not $ end with any characters in [... or pattern in \{...\}

escaped new lines

  • ...\ : escaped new lines within single quote
  • ...\\ : escaped new lines within double quote (escaped backslash \)

escaped characters

  • \s : space or tab
  • \t : tab (GNU sed only)
  • \n : new line
  • \r : return
  • $'...\t...\n\r' : $ escapes \t \n \r to t n r

multiline variable

var=$( echo "$var" | sed '$!s/$/\\/' )
sed "/.../ a$var" /path/to/file

multiline insert after matched

# heredoc - `'` or `"` quote leading 'EOF' = no $ expansion
cat << 'EOF' | sed '/.../ r /dev/stdin' /path/to/file
use any characters or symbols without escape
@#&*()'"%-+=/;:!?_^[]{}\|~\<>
except ^ as 1st character of each line
closing heredoc line must contains NO other characters
EOF