Sed out word after specific word and /

问题:

I am using grep to search out /Users/ in a text file. This is the output:

'dir': u'/Users/dlee/Desktop',

Is there a way to sed out the word after Users/? That is, to get dlee as the output?

回答1:

This should do the job:

sed -n '\@.*/Users/\([^/]*\)/.*@ s//\1/p' text.file

The -n suppresses output unless data is printed explicitly. The \@ notation replaces the search character (default /) with @. Then the search pattern looks for /Users/ followed by a sequence of non-slashes and a slash, capturing the non-slashes (user name); it matches everything else on the line too (the .* at beginning and end). The s//\1/p` command replaces what was matched (the whole line) with what was captured (the user name), and prints the information.

Alternatively, you can use a backslash-slash sequence to match the slashes in the data:

sed -n '/.*\/Users\/\([^/]*\)\/.*/ s//\1/p' text.file

You do not need a backslash before the slash in the character class, but the other backslashes are needed. If you were unwise enough to have a user name d\lee and you included a backslash in the character class, you would not see the name d\lee in the output.

回答2:

Potentially, you could also use awk to do achieve the same.

awk -F"/" '/Users/{print $2}' file

The delimiter has been set to / which will print the second column on rows containing the string "Users".

回答3:

With grep and Perl regex:

$ grep -Po '(?<=/Users/)[^/]*' <<< "'dir': u'/Users/dlee/Desktop',"
dlee

• -o returns only the match
• (?<=/Users/) is a "positive look-behind", i.e., the match has to be preceded by it but it is not included
• [^/]* is any number of characters except forward slashes