sed replace two double quotes and keep text

问题:

Text to be processed :

""text""
"text"
""

output desired :

"text"
"text"
""

Tried with :

 echo -e '""text""\n"text"\n""' | sed -e 's/"".*""/".*"/g'

But obviously no luck.

Cheers,

回答1:

You want to use a backreference (something that refers to a previously matched part) in your sed command :

echo -e '""text""\n"text"\n""' | sed -E -e 's/""(.*)""/"\1"/g'

Here are the modifications I did :

• I grouped what was inside the double-quote in the matching pattern with (...)
• I referenced that matching group in the replacement pattern with \1
• I told sed to use -Extended regex so I wouldn't have to escape the grouping parenthesis

回答2:

Based on your sample input, you could just use this:

sed '/^""$/! s/""/"/g' file

On lines which don't only contain two double quotes, globally replace all pairs of double quotes with one double quote.

回答3:

$ sed 's/"\("[^"]*"\)"/\1/' file
"text"
"text"
""

回答4:

Could you please try following awk command and let me know if this helps you.

awk -v s1="\"" '!/^\"\"$/{gsub(/\"\"/,s1)} 1'  Input_file

回答5:

Another approach

cat file | sed 's/"//g' | sed 's/^/"/g' | sed 's/$/"/g'

Details

sed 's/"//g' remove all double quote

sed 's/^/"/g' put a double quote at the beginning of each line

sed 's/$/"/g' put a double quote at the end of each line

Note

This approach will work only if there is one word per line, like in your example.

回答6:

Answer by Aaron is also fine but back-referencing is costlier in general. Why not use a simpler command:

echo -e '""text""\n"text"\n""' | sed 's/""*/"/g'

Regex to replace more than one double quote to single one.

回答7:

echo -e '""text""\n"text"\n""' |awk '/""text""/{gsub(/""/,"\42")}1'

"text"
"text"
""