-#!/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;
: \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 $hash";
+ }
+
for (@headers) {
say join ': ', @$_;
}