improve regexp

[ikiwiki.git] / IkiWiki / Rcs / svn.pm

diff --git a/IkiWiki/Rcs/svn.pm b/IkiWiki/Rcs/svn.pm
index be4106a02d143704f1e9bf2496e1e0b56dcf43b2..71189c82eb1766998bb90f1f3f4cdc0e0d22f2de 100644 (file)
--- a/IkiWiki/Rcs/svn.pm
+++ b/IkiWiki/Rcs/svn.pm
@@ -1,13 +1,32 @@
 #!/usr/bin/perl
-# For subversion support.
 
 use warnings;
 use strict;
 use IkiWiki;
+use POSIX qw(setlocale LC_CTYPE);
 
 package IkiWiki;
                
-my $svn_webcommit=qr/^web commit by (\w+):?(.*)/;
+# svn needs LC_CTYPE set to a UTF-8 locale, so try to find one. Any will do.
+sub find_lc_ctype() {
+       my $current = setlocale(LC_CTYPE());
+       return $current if $current =~ m/UTF-?8$/i;
+
+       # Make some obvious attempts to avoid calling `locale -a`
+       foreach my $locale ("$current.UTF-8", "en_US.UTF-8", "en_GB.UTF-8") {
+               return $locale if setlocale(LC_CTYPE(), $locale);
+       }
+
+       # Try to get all available locales and pick the first UTF-8 one found.
+       if (my @locale = grep(/UTF-?8$/i, `locale -a`)) {
+               chomp @locale;
+               return $locale[0] if setlocale(LC_CTYPE(), $locale[0]);
+       }
+
+       # fallback to the current locale
+       return $current;
+} # }}}
+$ENV{LC_CTYPE} = $ENV{LC_CTYPE} || find_lc_ctype();
 
 sub svn_info ($$) { #{{{
        my $field=shift;
@@ -101,13 +120,28 @@ sub rcs_recentchanges ($) { #{{{
        
        return unless -d "$config{srcdir}/.svn";
 
-       eval q{use CGI 'escapeHTML'};
-       eval q{use Date::Parse};
-       eval q{use Time::Duration};
-       eval q{use XML::Simple};
-       
+       eval q{
+               use Date::Parse;
+               use Time::Duration;
+               use XML::SAX;
+               use XML::Simple;
+       };
+       error($@) if $@;
+
+       # avoid using XML::SAX::PurePerl, it's buggy with UTF-8 data
+       my @parsers = map { ${$_}{Name} } @{XML::SAX->parsers()};
+       do {
+               $XML::Simple::PREFERRED_PARSER = pop @parsers;
+       } until $XML::Simple::PREFERRED_PARSER ne 'XML::SAX::PurePerl';
+
+       # --limit is only supported on Subversion 1.2.0+
+       my $svn_version=`svn --version -q`;
+       my $svn_limit='';
+       $svn_limit="--limit $num"
+               if $svn_version =~ /\d\.(\d)\.\d/ && $1 >= 2;
+
        my $svn_url=svn_info("URL", $config{srcdir});
-       my $xml = XMLin(scalar `svn --xml -v log '$svn_url'`,
+       my $xml = XMLin(scalar `svn $svn_limit --xml -v log '$svn_url'`,
                ForceArray => [ 'logentry', 'path' ],
                GroupTags => { paths => 'path' },
                KeyAttr => { path => 'content' },
@@ -118,20 +152,17 @@ sub rcs_recentchanges ($) { #{{{
                my $rev = $logentry->{revision};
                my $user = $logentry->{author};
 
-               my $date = $logentry->{date};
-               $date =~ s/T/ /;
-               $date =~ s/\.\d+Z$//;
-               my $when=concise(ago(time - str2time($date, 'UTC')));
+               my $when=time - str2time($logentry->{date}, 'UTC');
 
                foreach my $msgline (split(/\n/, $logentry->{msg})) {
-                       push @message, { line => escapeHTML($msgline) };
+                       push @message, { line => $msgline };
                }
 
                my $committype="web";
                if (defined $message[0] &&
-                   $message[0]->{line}=~/$svn_webcommit/) {
-                       $user="$1";
-                       $message[0]->{line}=$2;
+                   $message[0]->{line}=~/$config{web_commit_regexp}/) {
+                       $user=defined $2 ? "$2" : "$3";
+                       $message[0]->{line}=$4;
                }
                else {
                        $committype="svn";
@@ -145,12 +176,12 @@ sub rcs_recentchanges ($) { #{{{
                        $diffurl=~s/\[\[r1\]\]/$rev - 1/eg;
                        $diffurl=~s/\[\[r2\]\]/$rev/g;
                        push @pages, {
-                               link => htmllink("", "", pagename($file), 1),
+                               page => pagename($file),
                                diffurl => $diffurl,
                        } if length $file;
                }
                push @ret, { rev => $rev,
-                       user => htmllink("", "", $user, 1),
+                       user => $user,
                        committype => $committype,
                        when => $when,
                        message => [@message],
@@ -171,9 +202,9 @@ sub rcs_notify () { #{{{
        my $user=`svnlook author $config{svnrepo} -r $rev`;
        chomp $user;
        my $message=`svnlook log $config{svnrepo} -r $rev`;
-       if ($message=~/$svn_webcommit/) {
-               $user="$1";
-               $message=$2;
+       if ($message=~/$config{web_commit_regexp}/) {
+               $user=defined $2 ? "$2" : "$3";
+               $message=$4;
        }
 
        my @changed_pages;
@@ -211,6 +242,7 @@ sub rcs_notify () { #{{{
                );
                
                eval q{use Mail::Sendmail};
+               error($@) if $@;
                foreach my $email (@email_recipients) {
                        sendmail(
                                To => $email,
@@ -224,7 +256,6 @@ sub rcs_notify () { #{{{
 
 sub rcs_getctime ($) { #{{{
        my $file=shift;
-       eval q{use Date::Parse};
 
        my $svn_log_infoline=qr/^r\d+\s+\|\s+[^\s]+\s+\|\s+(\d+-\d+-\d+\s+\d+:\d+:\d+\s+[-+]?\d+).*/;
                
@@ -246,6 +277,8 @@ sub rcs_getctime ($) { #{{{
                return 0;
        }
                
+       eval q{use Date::Parse};
+       error($@) if $@;
        $date=str2time($date);
        debug("found ctime ".localtime($date)." for $file");
        return $date;