options to hide duplicates (simplify, ignore-case, unique)
authorMischa POSLAWSKY <perl@shiar.org>
Thu, 3 Feb 2011 21:40:27 +0000 (22:40 +0100)
committerMischa POSLAWSKY <perl@shiar.org>
Thu, 3 Feb 2011 21:40:27 +0000 (22:40 +0100)
git-grep-footer

index 6952508cdb94a4ca84c97857fe60fbbc8c481b5c..dcb24d4e35e36d8f812c856951d2dc4788f09e60 100755 (executable)
@@ -8,6 +8,9 @@ use Getopt::Long;
 
 GetOptions(\my %opt,
        'debug!',
+       'simplify|s:s',
+       'unique|u!',
+       'ignore-case|i!',
 ) or die;
 
 local $| = 1;
@@ -48,6 +51,40 @@ while (readline) {
                                next LINE;
                        };
 
+                       given ($opt{simplify} // 'no') {
+                               when ('strict') {
+                                       $header[1] =~ s{
+                                               \A
+                                               (?: [^:]+ )?
+                                               < [^@>]+ (?: @ | \h?\W? at \W?\h? ) [a-z0-9.-]+ >
+                                               \Z
+                                       }{<...>}imsx;
+                               }
+                               when (['text', '']) {
+                                       when ($header[0] =~ /[ _-] (?: by | to ) $/imsx) {
+                                               pop @header;
+                                       }
+                                       for ($header[1]) {
+                                               s{\b (https?)://\S+ }{[$1]}gmsx;  # url
+                                               s{(?: < | \A ) [^@>\s]+ @ [^>]+ (?: > | \Z )}{<...>}igmsx;  # address
+                                               s{\b [0-9]+ \b}{[num]}gmsx;  # number
+                                               s{\b I? [0-9a-f]{40} \b}{[sha1]}gmsx;  # hash
+                                       }
+                               }
+                               when (['all', 'any']) {
+                                       pop @header;
+                               }
+                               when ('no') {
+                               }
+                               default {
+                                       die "Unknown simplify option: '$_'\n";
+                               }
+                       }
+
+                       if ($opt{'ignore-case'}) {
+                               $_ = lc for @header;
+                       }
+
                        push @headers, \@header;
                }
 
@@ -58,6 +95,10 @@ while (readline) {
                }
 
                for (@headers) {
+                       if ($opt{unique}) {
+                               state $seen;
+                               next if $seen->{ $_->[0] }->{ $_->[1] // '' }++;
+                       }
                        say join ': ', @$_;
                }