sed find an replace.txt

sed -i 's/original/new/g' file.txt

SED WILL NOT WORK WITHOUT THE SINGLE QUOTES
JUST LIKE GREP USE SINGLE QUOTE AND NEVER USE DOUBLE QUOTES


There is only one way to replace multiple strings in a file using sed. Or a multiple file s if you use the find command. 

sed -i -e 's/abc/new/g' -e 's/nothing/new/g' theisatest.txt

You can use the following commands when piping from less or cat, but you cannot overwrite the file, you can just write the output to another file

1- Pipe sed into another sed 

sed s/1/0/ | sed s/true/false/ (just piping)
sed 's/abc/new/g' theisatest.txt | sed 's/nothing/new/g'  (piping file no need for cat or less. The name of the file must only be added at the beginning of the pipe)
sed 's/abc/new/g' theisatest.txt | sed 's/nothing/new/g' > theisatest1.txt
sed 's/abc/new/g' theisatest.txt | sed 's/nothing/new/g' | tee th444eisatest1.txt

>> append at the end of file 
> overwrite the file


2-separate the patterns using -e

sed -e 's/1/0/g' -e 's/true/false/g'

-i argument makes sed Overwrite the file. But when piping it is useless, so do not use it when piping



Everything between the start of range, and the end of range can be changed

sed is greedy, it does not support the non-greedy [?] as a workaround you need to be very specific about the end of range

if the start of range and end of range are not at the beginning of the line [.*] will help you a lot



sed '/start_of_range/,/end_of_range/ s/find/replace/' file
sed '/start_of_range/{:a;n;/^end_of_range/!ba;s#"find"#"replace"#}' file


use the -E flag when dealing with complex regex. And a fun fact, grep also has -E but it is still weaker than the -P
-e is only used when specifying two regexes.
-e only expect another -e after it, so you can not use the -E with -e
and if you are replacing in a file, the first flag must be -i
sed -e 's/string1/string2/g' -e 's/string4/string5/g'

sed does not support all the “character classes”, or also called “characters sets” 
you will need to use the alternative form
here is the table 
note that dot [.] Wildcard is supported. i will not write the alternative form to \s or \S because it is so long and \s or \s are kind of niche
\d	Digit	    [0-9] - matches a digit
\D	Non-digit	[^0-9] - matches all but a digit
\w	Word	    [a-zA-Z0-9_] - matches a word character
\W	Non-word	[^a-zA-Z0-9_] - matches all but a word character

If you want to use back-references in sed the flag -E is a must
in other regex engines $1 means return the first group
in sed it is \1 like this

echo "one two three four" | sed -E 's/(\w+) (\w+)/\2 \1/g'