convert non-utf8; avoid buffering; debug unmatched values

diff --git a/git-grep-footer b/git-grep-footer
index 1e60d4e840c2dc508b55265485c4cdb9db2843bf..6952508cdb94a4ca84c97857fe60fbbc8c481b5c 100755 (executable)
--- 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 $hash";
+               }
+
                for (@headers) {
                        say join ': ', @$_;
                }