带有捕获功能的sed / awk正则表达式-如何使一个简单的正则表达式起作用？

Question

Sample contents of FILE.txt are shown below. How would I modify the regex used by SED to do a capture which results in the desired output section? Prefer to do this with POSIX awk or sed functions if possible. I've looked into doing this solely with AWK, but am not following how you create the same behavior as a capture with the options it provides.

One of the problems I've run into trying various solutions is how to make the double-quotes optional.

sed -e 's/.Include .*"*\(.*\)"*/\1/g' FILE.txt


FILE.txt
##########################################################################
# Indexes Includes FollowSymLinks SymLinksifOwnerMatch ExecCGI Multiviews
# Options MultiViews Indexes SymLinksIfOwnerMatch IncludesNoExec
# Possible values include: debug, info, notice, warn, error, crit,
# does not include the trailing slash. 
AddOutputFilter INCLUDES .shtml .html
    Options -Indexes FollowSymLinks Includes
LoadModule include_module modules/mod_include.so
Include /opt/file.conf
Include "/opt/file.conf"
Include /usr/bin/abcOutput.conf
Include /usr/bin/ed_Output.conf
###########################################################################


**Desired Output:**
/opt/file.conf
/opt/file.conf
/usr/bin/abcOutput.conf
/usr/bin/ed_Output.conf

Answer 1

Use extended regular expressions to avoid the unnecessary \\( and \\) and use [[:blank:]] as a character class that includes space and tab. The ? indicates 0 or 1 matches. The + indicates 1 or more.

sed -rn 's/^Include[[:blank:]]+"?([^"]+)"?/\\1/p'

Note: The -E option for extended regular expression (instead of -r ) makes it compatible with older versions of sed

Answer 2

NOTE: Assuming you don't want the '#############' strings in the output, and based solely on the example you've provided ...

How about a awk/sed combo:

$ awk '/^Include/ { print $2 } ' FILE.txt | sed 's/\"//g'
/opt/file.conf
/opt/file.conf
/usr/bin/abcOutput.conf
/usr/bin/ed_Output.conf

Perhaps not as efficient as a single sed command, but easier to understand/maintain (KISS) and unless you're calling this kind of construct a LOT then the performance difference will be negligible.

Or a purely awk example (and a bit more efficient than the awk/sed idea):

$ awk '/^Include/ { gsub("\"","") ; print $2 } ' FILE.txt
/opt/file.conf
/opt/file.conf
/usr/bin/abcOutput.conf
/usr/bin/ed_Output.conf