lib/Frey/SVN.pm

package Frey::SVN;
use Moose;

# Convert output from svn log to html page (with some formatting of
# commit messages)
#
# 2004-04-28 Dobrica Pavlinusic <dpavlin@rot13.org>

extends 'Frey';
with 'Frey::Web';
with 'Frey::Storage';
with 'Frey::HTML::Diff';

use XML::Simple;
use DateTimeX::Easy;
use Text::Diff::Parser;
use File::Path qw/mkpath/;

has repository => (
        is => 'rw',
        isa => 'Str',
        required => 1,
        default => 'file:///home/dpavlin/private/svn/Frey',
);

has path => (
        is => 'rw',
        isa => 'Str'
);

has limit => (
        is => 'rw',
        isa => 'Int',
        default => 50,
);

has include_diff => (
        is => 'ro',
        isa => 'Bool',
        default => 0,
);

has file_stats => (
        is => 'ro',
        isa => 'Bool',
        default => 1,
);

sub iterator {
        my ($self,$coderef) = @_;
        
        sub sh_regex($$) {
                my ($cmd,$regex) = @_;
                open(my $sh, $cmd . ' |') || die "sh_regex failed on $cmd: $!";
                while(my $l = <$sh>) {
                        chomp($l);
                        if ($l =~ $regex) {
                                if ($1 && $2) {
                                        return ($1,$2);
                                } elsif ($1) {
                                        return $1;
                                } else {
                                        return $l;
                                }
                        }
                }
                #warn "can't find $regex in output of $cmd\n";
                return;
        }

        my $path = $self->repository . $self->path;
        warn "# path $path\n";

        my $cmd;
        my $svn_path = $path;

        if ($path =~ m#file://# || -e "$path/.svn") {
                $cmd = "svn log -v --xml $path";
        } else {
                $svn_path = sh_regex('svk info', qr#Mirrored From:\s+([^,]+)#i);

                if (! $svn_path) {

                        my $svk_depot = sh_regex('svk info', qr#Depot Path: (/.+)#i);

                        my $depot = $svk_depot;
                        my $rel_path;

                        my $path = sh_regex('svk depot --list', qr/^$depot\s+(\S+)/i);

                        while (! $path && $depot =~ s{^(/.*/)([^/]+)/?$}{$1} ) {
                                $rel_path = "$2/$rel_path";
                                $path = sh_regex('svk depot --list', qr/^$depot\s+(\S+)/i);
                        }

                        die "can't find depot path '$svk_depot' in svk depot --list\n" unless ($path);
                        $svn_path = "file:///$path/$rel_path";
                }

                $cmd = "svn log -v --xml $svn_path";
        }

        $cmd .= " --limit " . $self->limit if $self->limit;

        warn "# $cmd\n";
        open(my $fh, $cmd .' |') || die "failed $cmd: $!";
        my $log;
        while(<$fh>) {
                $log .= $_;
        }
        close($fh);

        my $xml = XMLin($log, ForceArray => [ 'logentry', 'path' ]);

        foreach my $e (@{$xml->{'logentry'}}) {
                warn "# e = ",$self->dump( $e ) if $self->debug;

                if ( $self->include_diff || $self->file_stats ) {
                        my $rev = $e->{'revision'};

                        $e->{diff_paths}->{rev} = $rev; # XXX debug

                        my $diff_file = $svn_path;
                        $diff_file =~ s{^\w+:/+}{};
                        mkpath "var/svn/$diff_file" unless -e "var/svn/$diff_file";
                        $diff_file = "var/svn/$diff_file/$rev.diff";
                        my $diff_fh;
                        my $diff_out;
                        if ( ! -e $diff_file || ! -s $diff_file ) {
                                my $cmd = "svn diff -c $rev $svn_path";
                                open($diff_fh, '-|', $cmd )  || die "can't open pipe from $cmd: $!";
                                open($diff_out,'>' , $diff_file ) || die "can't write $diff_file: $!";
                                warn "# creating $diff_file from $cmd\n";
                        } else {
                                open($diff_fh, '<' , $diff_file ) || die "can't read $diff_file: $!";
                        }

                        my $diff_path;

                        while( <$diff_fh> ) {
                                $e->{diff} .= $_ if $self->include_diff;
                                print $diff_out $_ if defined $diff_out;

                                if ( m{^\+\+\+ (\S+)} ) {
                                        $diff_path = "/$1"; # subversion paths start with /
                                } elsif ( m{^\+} && $diff_path ) {
                                        $e->{diff_paths}->{$diff_path}->{added}++;
                                } elsif ( m{^-} && $diff_path ) {
                                        $e->{diff_paths}->{$diff_path}->{removed}++;
                                }
                        }
                }

                $coderef->($e);
        }
}

sub as_markup {
        my ($self) = @_;

        # extract svk revision: r113@athlon (orig r999): dpavlin | 2005-09-01 20:38:07 +0200
        our $svk_rev_re = '\s+(r\d+@\w+(\s+\(orig\s+r\d+\))*:\s+\w+\s+\|\s+\d{4}-\d{2}-\d{2}\s+\d{2}:\d{2}:\d{2}\s+\+\d+)\s*';

        sub encode {
                my $foo = shift;
                $foo =~ s/$svk_rev_re//gsm;
                $foo =~ s/</&lt;/g;
                $foo =~ s/>/&gt;/g;
                $foo =~ s/"/&quot;/g;
#               $foo =~ s/([\n\r][\n\r]+)/$1<br\/>/gis;
                $foo =~ s/([\n\r]+)([\-\*]\s+)/$1<br\/>$2/gis;
                $foo =~ s/([\n\r]+)(r\d+:\s+)/$1<br\/>$2/gis;
                $foo =~ s/([\n\r]+)(\s+r\d+@)/$1<br\/>$2/gis;           # svk
                return $foo;
        }

        my $repository = $self->repository;
        my $path = $self->path;

        our $html = qq|
                <h1><a href="?repository=$repository">$repository</a></h1>
                <h2>$path</h2>
        |;

        $self->add_css(qq|
                .commit {
                        clear: both;
                        padding-top: 1em;
                        padding-bottom: 1em;
                        border-top: 1px dashed #ccc;
                }
                .files {
                        color: #888;
                        font-family: monospace;
                        font-size: 80%;
                        float: right;
                        padding-bottom: 1.2em; /* fix 80% back to original 1em */
                }
                .files a {
                        text-decoration: none;
                        color: #888;
                }
                .date, .revision { color: #666; }
                .message {
                        padding-top: 0.5em;
                        padding-left: 2em; /* like blockquote */
                        white-space: pre-wrap;
                }

                ins { color: #8c8 }
                del { color: #c88 }
        |);

        my $max_path_len = 0;
        my $path_count;

        $self->iterator( sub {
                my $e = shift;

                my $rev = $e->{'revision'};
                my $date = $e->{'date'};

                $date =~ s/T/ /;
                $date =~ s/\.\d+Z$//;

                my $msg = $e->{'msg'};
                $msg = '' if ref($msg); # FIXME why do I need this, dammit?
                if ( $msg ) {
                        $msg = encode( $msg );
                        $msg = qq|<div class="message">$msg</div>|;
                }

                my @files;

                foreach my $p (@{$e->{'paths'}->{'path'}}) {
                        my ($action,$path) = ($p->{'action'},$p->{'content'});

                        if ($action eq "A") {
                                push @files, "<ins>$path</ins>";
                        } elsif ($action eq "D") {
                                push @files, "<del>$path</del>";
                        } else {
                                push @files, $path;
                        }

                        $max_path_len = length $path if length $path > $max_path_len;
                        $path_count->{$path}++;
                }

                my $diff = $self->html_diff( $e->{diff} ) if $e->{diff};

                $html .= $self->dump( $e->{diff_paths} );

                $html .= qq|
                        <div class="commit">
                                <span class="date">$date</span>
                                <em>$e->{author}</em>
                                <span class="revision">$e->{revision}</span>
                                <div class="files">\n
                                |
                                . join("<br>\n",
                                        map {
                                                my $path = $_;
                                                $path =~ s{<[^>]+>}{}g;
                                                qq|<a href="?repository=$repository;path=$path" title="$path ##">$_</a>|
                                        } @files
                                )
                                . qq|
                                </div>
                                $msg
                                $diff
                        </div>
                |;

        });

        $self->add_css(qq|
                .files {
                        width: ${max_path_len}ex;
                }
        |);

        $html =~ s[title="(\S+) ##"]['title="' . $path_count->{$1} . '"']gse;

        return $html;
}

sub codeswarm_as_markup {
        my ($self) = @_;

        $self->content_type('text/xml');

        my $file_events = '';

        $self->iterator( sub {
                my $e = shift;
                
                my $rev = $e->{'revision'};
                my $date = DateTimeX::Easy->new( $e->{'date'} )->epoch . '000'; # ms
                my $author = $e->{'author'};

                foreach my $p (@{$e->{'paths'}->{'path'}}) {
                        my ($action,$path) = ($p->{'action'},$p->{'content'});
                        my $weight = '';
                        if ( my $s = $e->{diff_paths}->{$path} ) {
                                $weight  = $s->{removed}   if $s->{removed};
                                $weight += $s->{added} * 2 if $s->{added};
                                $weight  = qq| weight="$weight" |;
                        }
                        $file_events .= qq|\t<event filename="$path" date="$date" author="$author"$weight/>\n|;
                }

        });

        return qq|<?xml version="1.0"?>
        <!-- One commit per day for one month by a documenter and programmer. -->
        <file_events>
        $file_events
        </file_events>
        |;

}

1;
1	dpavlin	769	package Frey::SVN;
2			use Moose;
3
4			# Convert output from svn log to html page (with some formatting of
5			# commit messages)
6			#
7			# 2004-04-28 Dobrica Pavlinusic <dpavlin@rot13.org>
8
9			extends 'Frey';
10			with 'Frey::Web';
11	dpavlin	1066	with 'Frey::Storage';
12			with 'Frey::HTML::Diff';
13	dpavlin	769
14			use XML::Simple;
15	dpavlin	807	use DateTimeX::Easy;
16	dpavlin	1066	use Text::Diff::Parser;
17	dpavlin	1068	use File::Path qw/mkpath/;
18	dpavlin	769
19			has repository => (
20			is => 'rw',
21			isa => 'Str',
22			required => 1,
23			default => 'file:///home/dpavlin/private/svn/Frey',
24			);
25
26	dpavlin	987	has path => (
27			is => 'rw',
28			isa => 'Str'
29			);
30
31	dpavlin	770	has limit => (
32			is => 'rw',
33			isa => 'Int',
34			default => 50,
35			);
36
37	dpavlin	1066	has include_diff => (
38			is => 'ro',
39			isa => 'Bool',
40	dpavlin	1068	default => 0,
41	dpavlin	1066	);
42
43	dpavlin	1067	has file_stats => (
44			is => 'ro',
45			isa => 'Bool',
46			default => 1,
47			);
48
49	dpavlin	804	sub iterator {
50			my ($self,$coderef) = @_;
51
52	dpavlin	769	sub sh_regex($$) {
53			my ($cmd,$regex) = @_;
54			open(my $sh, $cmd . ' \|') \|\| die "sh_regex failed on $cmd: $!";
55			while(my $l = <$sh>) {
56			chomp($l);
57			if ($l =~ $regex) {
58			if ($1 && $2) {
59			return ($1,$2);
60			} elsif ($1) {
61			return $1;
62			} else {
63			return $l;
64			}
65			}
66			}
67			#warn "can't find $regex in output of $cmd\n";
68			return;
69			}
70
71	dpavlin	987	my $path = $self->repository . $self->path;
72	dpavlin	769	warn "# path $path\n";
73
74			my $cmd;
75	dpavlin	1066	my $svn_path = $path;
76
77	dpavlin	769	if ($path =~ m#file://# \|\| -e "$path/.svn") {
78			$cmd = "svn log -v --xml $path";
79			} else {
80	dpavlin	1066	$svn_path = sh_regex('svk info', qr#Mirrored From:\s+([^,]+)#i);
81	dpavlin	769
82			if (! $svn_path) {
83
84			my $svk_depot = sh_regex('svk info', qr#Depot Path: (/.+)#i);
85
86			my $depot = $svk_depot;
87			my $rel_path;
88
89			my $path = sh_regex('svk depot --list', qr/^$depot\s+(\S+)/i);
90
91			while (! $path && $depot =~ s{^(/.*/)([^/]+)/?$}{$1} ) {
92			$rel_path = "$2/$rel_path";
93			$path = sh_regex('svk depot --list', qr/^$depot\s+(\S+)/i);
94			}
95
96			die "can't find depot path '$svk_depot' in svk depot --list\n" unless ($path);
97			$svn_path = "file:///$path/$rel_path";
98			}
99
100			$cmd = "svn log -v --xml $svn_path";
101			}
102
103	dpavlin	770	$cmd .= " --limit " . $self->limit if $self->limit;
104
105	dpavlin	769	warn "# $cmd\n";
106			open(my $fh, $cmd .' \|') \|\| die "failed $cmd: $!";
107			my $log;
108			while(<$fh>) {
109			$log .= $_;
110			}
111			close($fh);
112
113			my $xml = XMLin($log, ForceArray => [ 'logentry', 'path' ]);
114
115	dpavlin	804	foreach my $e (@{$xml->{'logentry'}}) {
116	dpavlin	875	warn "# e = ",$self->dump( $e ) if $self->debug;
117	dpavlin	1066
118	dpavlin	1067	if ( $self->include_diff \|\| $self->file_stats ) {
119	dpavlin	1066	my $rev = $e->{'revision'};
120	dpavlin	1068
121			$e->{diff_paths}->{rev} = $rev; # XXX debug
122
123			my $diff_file = $svn_path;
124			$diff_file =~ s{^\w+:/+}{};
125			mkpath "var/svn/$diff_file" unless -e "var/svn/$diff_file";
126			$diff_file = "var/svn/$diff_file/$rev.diff";
127			my $diff_fh;
128			my $diff_out;
129			if ( ! -e $diff_file \|\| ! -s $diff_file ) {
130			my $cmd = "svn diff -c $rev $svn_path";
131			open($diff_fh, '-\|', $cmd ) \|\| die "can't open pipe from $cmd: $!";
132			open($diff_out,'>' , $diff_file ) \|\| die "can't write $diff_file: $!";
133			warn "# creating $diff_file from $cmd\n";
134			} else {
135			open($diff_fh, '<' , $diff_file ) \|\| die "can't read $diff_file: $!";
136	dpavlin	1066	}
137	dpavlin	1067
138	dpavlin	1068	my $diff_path;
139	dpavlin	1067
140	dpavlin	1068	while( <$diff_fh> ) {
141			$e->{diff} .= $_ if $self->include_diff;
142			print $diff_out $_ if defined $diff_out;
143	dpavlin	1067
144	dpavlin	1068	if ( m{^\+\+\+ (\S+)} ) {
145	dpavlin	1067	$diff_path = "/$1"; # subversion paths start with /
146	dpavlin	1068	} elsif ( m{^\+} && $diff_path ) {
147	dpavlin	1067	$e->{diff_paths}->{$diff_path}->{added}++;
148	dpavlin	1068	} elsif ( m{^-} && $diff_path ) {
149	dpavlin	1067	$e->{diff_paths}->{$diff_path}->{removed}++;
150			}
151			}
152	dpavlin	1066	}
153
154	dpavlin	804	$coderef->($e);
155			}
156			}
157
158			sub as_markup {
159			my ($self) = @_;
160
161			# extract svk revision: r113@athlon (orig r999): dpavlin \| 2005-09-01 20:38:07 +0200
162			our $svk_rev_re = '\s+(r\d+@\w+(\s+\(orig\s+r\d+\)):\s+\w+\s+\\|\s+\d{4}-\d{2}-\d{2}\s+\d{2}:\d{2}:\d{2}\s+\+\d+)\s';
163
164			sub encode {
165			my $foo = shift;
166			$foo =~ s/$svk_rev_re//gsm;
167			$foo =~ s/</</g;
168			$foo =~ s/>/>/g;
169			$foo =~ s/"/"/g;
170	dpavlin	909	# $foo =~ s/([\n\r][\n\r]+)/$1<br\/>/gis;
171	dpavlin	804	$foo =~ s/([\n\r]+)([\-\*]\s+)/$1<br\/>$2/gis;
172			$foo =~ s/([\n\r]+)(r\d+:\s+)/$1<br\/>$2/gis;
173			$foo =~ s/([\n\r]+)(\s+r\d+@)/$1<br\/>$2/gis; # svk
174			return $foo;
175			}
176
177	dpavlin	987	my $repository = $self->repository;
178			my $path = $self->path;
179	dpavlin	769
180	dpavlin	987	our $html = qq\|
181			<h1><a href="?repository=$repository">$repository</a></h1>
182			<h2>$path</h2>
183			\|;
184
185	dpavlin	852	$self->add_css(qq\|
186	dpavlin	909	.commit {
187			clear: both;
188			padding-top: 1em;
189			padding-bottom: 1em;
190			border-top: 1px dashed #ccc;
191			}
192			.files {
193			color: #888;
194			font-family: monospace;
195			font-size: 80%;
196			float: right;
197			padding-bottom: 1.2em; /* fix 80% back to original 1em */
198			}
199	dpavlin	987	.files a {
200			text-decoration: none;
201			color: #888;
202			}
203	dpavlin	852	.date, .revision { color: #666; }
204	dpavlin	909	.message {
205			padding-top: 0.5em;
206			padding-left: 2em; /* like blockquote */
207			white-space: pre-wrap;
208			}
209	dpavlin	852
210			ins { color: #8c8 }
211			del { color: #c88 }
212			\|);
213
214	dpavlin	1013	my $max_path_len = 0;
215			my $path_count;
216
217	dpavlin	804	$self->iterator( sub {
218			my $e = shift;
219
220	dpavlin	769	my $rev = $e->{'revision'};
221			my $date = $e->{'date'};
222
223			$date =~ s/T/ /;
224			$date =~ s/\.\d+Z$//;
225
226	dpavlin	852	my $msg = $e->{'msg'};
227			$msg = '' if ref($msg); # FIXME why do I need this, dammit?
228			if ( $msg ) {
229			$msg = encode( $msg );
230			$msg = qq\|<div class="message">$msg</div>\|;
231			}
232
233	dpavlin	769	my @files;
234
235			foreach my $p (@{$e->{'paths'}->{'path'}}) {
236			my ($action,$path) = ($p->{'action'},$p->{'content'});
237
238			if ($action eq "A") {
239			push @files, "<ins>$path</ins>";
240			} elsif ($action eq "D") {
241			push @files, "<del>$path</del>";
242	dpavlin	851	} else {
243	dpavlin	769	push @files, $path;
244			}
245	dpavlin	1013
246			$max_path_len = length $path if length $path > $max_path_len;
247			$path_count->{$path}++;
248	dpavlin	769	}
249
250	dpavlin	1066	my $diff = $self->html_diff( $e->{diff} ) if $e->{diff};
251
252	dpavlin	1067	$html .= $self->dump( $e->{diff_paths} );
253
254	dpavlin	909	$html .= qq\|
255			<div class="commit">
256	dpavlin	987	<span class="date">$date</span>
257			<em>$e->{author}</em>
258			<span class="revision">$e->{revision}</span>
259			<div class="files">\n
260			\|
261			. join("<br>\n",
262			map {
263	dpavlin	1013	my $path = $_;
264			$path =~ s{<[^>]+>}{}g;
265			qq\|<a href="?repository=$repository;path=$path" title="$path ##">$_</a>\|
266	dpavlin	987	} @files
267			)
268			. qq\|
269			</div>
270			$msg
271	dpavlin	1066	$diff
272	dpavlin	909	</div>
273			\|;
274	dpavlin	769
275	dpavlin	804	});
276	dpavlin	769
277	dpavlin	1013	$self->add_css(qq\|
278			.files {
279			width: ${max_path_len}ex;
280			}
281			\|);
282
283			$html =~ s[title="(\S+) ##"]['title="' . $path_count->{$1} . '"']gse;
284
285	dpavlin	769	return $html;
286			}
287
288	dpavlin	805	sub codeswarm_as_markup {
289			my ($self) = @_;
290
291			$self->content_type('text/xml');
292
293			my $file_events = '';
294
295			$self->iterator( sub {
296			my $e = shift;
297
298			my $rev = $e->{'revision'};
299	dpavlin	807	my $date = DateTimeX::Easy->new( $e->{'date'} )->epoch . '000'; # ms
300	dpavlin	805	my $author = $e->{'author'};
301
302			foreach my $p (@{$e->{'paths'}->{'path'}}) {
303			my ($action,$path) = ($p->{'action'},$p->{'content'});
304	dpavlin	1067	my $weight = '';
305	dpavlin	1068	if ( my $s = $e->{diff_paths}->{$path} ) {
306			$weight = $s->{removed} if $s->{removed};
307			$weight += $s->{added} * 2 if $s->{added};
308	dpavlin	1067	$weight = qq\| weight="$weight" \|;
309			}
310			$file_events .= qq\|\t<event filename="$path" date="$date" author="$author"$weight/>\n\|;
311	dpavlin	805	}
312
313			});
314
315			return qq\|<?xml version="1.0"?>
316			<!-- One commit per day for one month by a documenter and programmer. -->
317			<file_events>
318			$file_events
319			</file_events>
320			\|;
321
322			}
323
324	dpavlin	769	1;