From c0314d02f10951833a0879a0755010a998c4b76c Mon Sep 17 00:00:00 2001
From: Mischa POSLAWSKY <perl@shiar.org>
Date: Thu, 3 Feb 2011 19:56:28 +0100
Subject: [PATCH] convert non-utf8; avoid buffering; debug unmatched values

---
 git-grep-footer | 34 ++++++++++++++++++++++++++++++++--
 1 file changed, 32 insertions(+), 2 deletions(-)

diff --git a/git-grep-footer b/git-grep-footer
index 1e60d4e..6952508 100755
--- a/git-grep-footer
+++ b/git-grep-footer
@@ -1,12 +1,35 @@
-#!/usr/bin/perl -0 -CS
+#!/usr/bin/perl -0 -CO
 use 5.010;
 use strict;
 use warnings;
+use Encode 'decode';
 use Data::Dump 'pp';
+use Getopt::Long;
+
+GetOptions(\my %opt,
+	'debug!',
+) or die;
+
+local $| = 1;
 
 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;
@@ -20,13 +43,20 @@ while (readline) {
 				: \s*
 				(?<val> \S .+)
 				$
-			}imx or next LINE;
+			}imx or do {
+				$prefix++;
+				next LINE;
+			};
 
 			push @headers, \@header;
 		}
 
 		next BLOCK if not @headers;
 
+		if ($opt{debug} and $prefix) {
+			say "infix junk in commit $hash";
+		}
+
 		for (@headers) {
 			say join ': ', @$_;
 		}
-- 
2.30.0