X-Git-Url: http://git.shiar.nl/git-grep-footer.git/blobdiff_plain/784eaf272112bcfcd72257efb7bb796d34b4f922..b1c0ccc9afe38e48b1d2b704d1cacc9db4f90132:/git-grep-footer

diff --git a/git-grep-footer b/git-grep-footer
index 1e60d4e..ff279a6 100755
--- a/git-grep-footer
+++ b/git-grep-footer
@@ -1,12 +1,44 @@
-#!/usr/bin/perl -0 -CS
+#!/usr/bin/perl
 use 5.010;
 use strict;
 use warnings;
+use open ':std', OUT => ':utf8';
+use Encode 'decode';
 use Data::Dump 'pp';
+use Getopt::Long qw(:config bundling);
+
+GetOptions(\my %opt,
+	'debug!',
+	'simplify|s:s',
+	'ignore-case|i!',
+	'min|min-count|unique|u:i',
+	'max|max-count|show|n:i',
+	'version|V'  => sub { Getopt::Long::VersionMessage() },
+	'usage|h'    => sub { Getopt::Long::HelpMessage() },
+	'help|man|?' => sub { Getopt::Long::HelpMessage(-verbose => 2) },
+) or exit 129;
+
+local $| = 1;
+local $/ = "\0";
 
 my $HEADERMATCH = qr/ [a-z]+ (?: (?:-\w+)+ | \ by ) /ix;
 
 while (readline) {
+	s/(.+)\n//m;
+	my $hash = $1;
+
+	# strip commit seperator
+	chomp;
+	# skip expensive checks without potential identifier
+	m/:/ or next;
+	# try to parse as UTF-8
+	eval { $_ = decode(utf8   => $_, Encode::FB_CROAK()) };
+	# if invalid, assume it's latin1
+	       $_ = decode(cp1252 => $_) if $@;
+
+	my $prefix = 0;
+	my %attr;
+
 	BLOCK:
 	for (reverse split /\n\n/) {
 		my @headers;
@@ -20,17 +52,143 @@ while (readline) {
 				: \s*
 				(?<val> \S .+)
 				$
-			}imx or next LINE;
+			}imx or do {
+				$prefix++;
+				next LINE;
+			};
+
+			push @header, $_ if defined $opt{max};
+
+			given ($opt{simplify} // 'none') {
+				when (['email', 'authors']) {
+					$header[1] =~ s{
+						\A
+						(?: [^:;]+ )?
+						< [^@>]+ (?: @ | \h?\W? at \W?\h? ) [a-z0-9.-]+ >
+						\Z
+					}{<...>}imsx;
+				}
+				when (['var', 'vars', '']) {
+					when ($header[0] =~ /[ _-] (?: by | to ) $/imsx) {
+						$header[1] = undef;
+					}
+					for ($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', 'contents']) {
+					$header[1] = undef;
+				}
+				when (['none', 'no', '0']) {
+				}
+				default {
+					die "Unknown simplify option: '$_'\n";
+				}
+			}
+
+			if ($opt{'ignore-case'}) {
+				$_ = lc for $header[0], $header[1] // ();
+			}
+
+			pop @header if not defined $header[-1];
 
 			push @headers, \@header;
 		}
 
 		next BLOCK if not @headers;
 
+		if ($opt{debug} and $prefix) {
+			say "infix junk in commit $hash";
+		}
+
 		for (@headers) {
-			say join ': ', @$_;
+			if (defined $opt{min} or $opt{max}) {
+				state $seen;
+				my $count = $seen->{ $_->[0] }->{ $_->[1] // '' }++;
+				next if $count >= ($opt{min} // 0) + ($opt{max} || 1);
+				next if $count < ($opt{min} // 0);
+			}
+			say $_->[2] // join(': ', @$_);
 		}
 
 		last BLOCK;
 	}
 }
+
+__END__
+
+=head1 NAME
+
+git-grep-footer - Find custom header lines in commit messages
+
+=head1 SYNOPSIS
+
+F<git> log --pretty=%b%x00 | F<git-grep-footer> [OPTIONS]
+
+=head1 DESCRIPTION
+
+Filters out header sections near the end of a commit body,
+a common convention to list custom metadata such as
+C<Signed-off-by> and C<Acked-by>.
+
+Sections are identified by at least one leading keyword containing a dash
+followed by a colon.
+
+=head1 OPTIONS
+
+=over
+
+=item -i, --ignore-case
+
+Lowercases everything.
+
+=item -s, --simplify[=<rule>]
+
+Modifies values to hide specific details.
+Several different rules are supported:
+
+=over
+
+=item I<var> (default)
+
+Replaces highly variable contents such as numbers, hashes, and addresses,
+leaving only exceptional annotations as distinct text.
+Attributes ending in I<-to> or I<-by> are assumed variable author names
+and omitted entirely,
+unless they contain a colon indicating possible attribute exceptions.
+
+=item I<email>
+
+Filters out author lines following the git signoff convention,
+i.e. an <email address> optionally preceded by a name.
+
+=item I<all>
+
+Values will be hidden entirely, so only attribute names remain.
+
+=back
+
+=item -u, --unique[=<threshold>]
+
+Each match is only shown once,
+optionally after it has already occurred a given amount of times.
+
+=item -n, --show[=<limit>]
+
+The original line is given for each match,
+but simplifications still apply for duplicate determination.
+Additional samples are optionally given upto the given maximum.
+
+=back
+
+=head1 AUTHOR
+
+Mischa POSLAWSKY <perl@shiar.org>
+
+=head1 LICENSE
+
+Copyright. All rights reserved.
+