Multi-line grep

Usage: mlgrep -h
       -h:  help

       mlgrep [-{igjcsewoaN}] [-n|-l] [-q[<len>]] [-{CHJLMPRS}|-r <pattern> ...] \
              [-X] [-x <regexp> ...] <regexp> [<files ...>]

Command flag	Description
-a	use absolute pathnames in printouts of matches
-c	no comments (don't search within C/C++/Java/Ruby comments)
-e	exclude self, i.e. don't search in files whose name matches the regexp we're searching for
-g	ignore errors (file not found, encoding, etc)
-i	ignore case
-j	invoke IntelliJ IDEA and wait for user input between each match
-k	print statistics about occurrences of the regexp
-l	list (just print names of files where a match was found)
-n	line mode (a dot (.) or \s in regexp doesn't match newline and the whole line where a match was found is printed)
-o	only match. If the regexp contains a group (a parenthesized expression), only text matching that group, rather than the entire regexp match, will be printed.
-r <pattern>	search in files matching the pattern (e.g. -r "*.skel") The directory tree starting at current directory is searched.
-s	no strings (don't search within double quoted strings)
-w	match whole words only
-x <regexp>	exclude files whose names match the regexp
-X	exclude files according to 'exclude' property in ~/.mlgrep.yml
-C	equivalent to -r '*.{cc,c}' (C, C++)
-H	equivalent to -r '*.{hh,h}' (C/C++ headers)
-J	equivalent to -r '*.java' (Java)
-L	equivalent to -r '*.{pl,PL,pm,pod,t}' (Perl)
-M	equivalent to -r '{*.cmake,CMakeLists.txt}' (CMake)
-P	equivalent to -r '*.py' (Python)
-R	equivalent to -r '*.rb' (Ruby)
-S	equivalent to -r <'source' pattern defined in ~/.mlgreprc>
-T	equivalent to -r '*.{html,htm,xhtml,xml,css}'
-q[<len>]	quiet - output first and last characters, default is 20
-N	don't print file names and line numbers

While the unix grep is line-based, this command can search over line breaks. Rather than printing lines where a match was found, mlgrep prints the matching part of the file (replacing each sequence of whitespace with a single space). Spaces in the regexp can match any combination of whitespace in the file, and . (dot) matches any character including newline. To match exactly one space, escape it with a backslash or put it inside square brackets. To match any character except newline, use [^\n].

Option flags can be compounded. I.e., -ics means -i -c -s.

Special regexp functionality

\uX (where X is an optionally backslash-escaped single character or a class of characters within square brackets) matches everything up to the first occurrence of X (\u is for until). For example, /\u;/ is replaced by /[^;]*;/ and /\u[123]/ is replaced by /[^123]*[123]/.

Pattern options (-r, -C, -H, -J, -L, -M, -P, -S) may be combined with each other and with files....

Examples

Search all source code for single statements inside braces:

$ mlgrep -Ssc '[^(]\) \{[^;{}]*; \}'

Search in C++ files - excluding generated code - for null pointer check after allocation (which is pointless in Dicos):

$ mlgrep -x '/(generated|user|provider)' -CHc '(\w+) = (new|\w+ :: construct \()\u; [^}]* if \( \1 == (0|NULL)'

Strip comments from cc files and write them to temporary files:

$ for i in *.cc; do mlgrep -Nnc . $i > /tmp/$i; done

Example of ~/.mlgrep.yml:

exclude: '/(test|user|generated|provider|included|.deps|delosInfo|obj\.)'