Code

Limit file descriptors used by packs
[git.git] / gitweb / gitweb.perl
index 62b2d21b846d8aaa850a451c0248cb71bb6b0741..1b9369d1a72f625bcb66e021bdab8559930cb0eb 100755 (executable)
@@ -250,13 +250,14 @@ our %highlight_ext = (
        # main extensions, defining name of syntax;
        # see files in /usr/share/highlight/langDefs/ directory
        map { $_ => $_ }
-               qw(py c cpp rb java css php sh pl js tex bib xml awk bat ini spec tcl),
+               qw(py c cpp rb java css php sh pl js tex bib xml awk bat ini spec tcl sql make),
        # alternate extensions, see /etc/highlight/filetypes.conf
        'h' => 'c',
+       map { $_ => 'sh'  } qw(bash zsh ksh),
        map { $_ => 'cpp' } qw(cxx c++ cc),
-       map { $_ => 'php' } qw(php3 php4),
+       map { $_ => 'php' } qw(php3 php4 php5 phps),
        map { $_ => 'pl'  } qw(perl pm), # perhaps also 'cgi'
-       'mak' => 'make',
+       map { $_ => 'make'} qw(mak mk),
        map { $_ => 'xml' } qw(xhtml html htm),
 );
 
@@ -1230,7 +1231,7 @@ sub href {
                $href =~ s,/$,,;
 
                # Then add the project name, if present
-               $href .= "/".esc_url($params{'project'});
+               $href .= "/".esc_path_info($params{'project'});
                delete $params{'project'};
 
                # since we destructively absorb parameters, we keep this
@@ -1240,7 +1241,8 @@ sub href {
                # Summary just uses the project path URL, any other action is
                # added to the URL
                if (defined $params{'action'}) {
-                       $href .= "/".esc_url($params{'action'}) unless $params{'action'} eq 'summary';
+                       $href .= "/".esc_path_info($params{'action'})
+                               unless $params{'action'} eq 'summary';
                        delete $params{'action'};
                }
 
@@ -1250,13 +1252,13 @@ sub href {
                        || $params{'hash_parent'} || $params{'hash'});
                if (defined $params{'hash_base'}) {
                        if (defined $params{'hash_parent_base'}) {
-                               $href .= esc_url($params{'hash_parent_base'});
+                               $href .= esc_path_info($params{'hash_parent_base'});
                                # skip the file_parent if it's the same as the file_name
                                if (defined $params{'file_parent'}) {
                                        if (defined $params{'file_name'} && $params{'file_parent'} eq $params{'file_name'}) {
                                                delete $params{'file_parent'};
                                        } elsif ($params{'file_parent'} !~ /\.\./) {
-                                               $href .= ":/".esc_url($params{'file_parent'});
+                                               $href .= ":/".esc_path_info($params{'file_parent'});
                                                delete $params{'file_parent'};
                                        }
                                }
@@ -1264,19 +1266,19 @@ sub href {
                                delete $params{'hash_parent'};
                                delete $params{'hash_parent_base'};
                        } elsif (defined $params{'hash_parent'}) {
-                               $href .= esc_url($params{'hash_parent'}). "..";
+                               $href .= esc_path_info($params{'hash_parent'}). "..";
                                delete $params{'hash_parent'};
                        }
 
-                       $href .= esc_url($params{'hash_base'});
+                       $href .= esc_path_info($params{'hash_base'});
                        if (defined $params{'file_name'} && $params{'file_name'} !~ /\.\./) {
-                               $href .= ":/".esc_url($params{'file_name'});
+                               $href .= ":/".esc_path_info($params{'file_name'});
                                delete $params{'file_name'};
                        }
                        delete $params{'hash'};
                        delete $params{'hash_base'};
                } elsif (defined $params{'hash'}) {
-                       $href .= esc_url($params{'hash'});
+                       $href .= esc_path_info($params{'hash'});
                        delete $params{'hash'};
                }
 
@@ -1309,6 +1311,9 @@ sub href {
        }
        $href .= "?" . join(';', @result) if scalar @result;
 
+       # final transformation: trailing spaces must be escaped (URI-encoded)
+       $href =~ s/(\s+)$/CGI::escape($1)/e;
+
        return $href;
 }
 
@@ -1391,6 +1396,17 @@ sub esc_param {
        return $str;
 }
 
+# the quoting rules for path_info fragment are slightly different
+sub esc_path_info {
+       my $str = shift;
+       return undef unless defined $str;
+
+       # path_info doesn't treat '+' as space (specially), but '?' must be escaped
+       $str =~ s/([^A-Za-z0-9\-_.~();\/;:@&= +]+)/CGI::escape($1)/eg;
+
+       return $str;
+}
+
 # quote unsafe chars in whole URL, so some characters cannot be quoted
 sub esc_url {
        my $str = shift;
@@ -1400,6 +1416,13 @@ sub esc_url {
        return $str;
 }
 
+# quote unsafe characters in HTML attributes
+sub esc_attr {
+
+       # for XHTML conformance escaping '"' to '"' is not enough
+       return esc_html(@_);
+}
+
 # replace invalid utf8 character with SUBSTITUTION sequence
 sub esc_html {
        my $str = shift;
@@ -1805,7 +1828,7 @@ sub format_ref_marker {
                                        hash=>$dest
                                )}, $name);
 
-                       $markers .= " <span class=\"$class\" title=\"$ref\">" .
+                       $markers .= " <span class=\"".esc_attr($class)."\" title=\"".esc_attr($ref)."\">" .
                                $link . "</span>";
                }
        }
@@ -1889,7 +1912,7 @@ sub git_get_avatar {
                return $pre_white .
                       "<img width=\"$size\" " .
                            "class=\"avatar\" " .
-                           "src=\"$url\" " .
+                           "src=\"".esc_url($url)."\" " .
                            "alt=\"\" " .
                       "/>" . $post_white;
        } else {
@@ -2600,7 +2623,7 @@ sub git_show_project_tagcloud {
        } else {
                my @tags = sort { $cloud->{$a}->{count} <=> $cloud->{$b}->{count} } keys %$cloud;
                return '<p align="center">' . join (', ', map {
-                       "<a href=\"$home_link?by_tag=$_\">$cloud->{$_}->{topname}</a>"
+                       $cgi->a({-href=>"$home_link?by_tag=$_"}, $cloud->{$_}->{topname})
                } splice(@tags, 0, $count)) . '</p>';
        }
 }
@@ -3442,11 +3465,10 @@ sub run_highlighter {
        my ($fd, $highlight, $syntax) = @_;
        return $fd unless ($highlight && defined $syntax);
 
-       close $fd
-               or die_error(404, "Reading blob failed");
+       close $fd;
        open $fd, quote_command(git_cmd(), "cat-file", "blob", $hash)." | ".
                  quote_command($highlight_bin).
-                 " --xhtml --fragment --syntax $syntax |"
+                 " --fragment --syntax $syntax |"
                or die_error(500, "Couldn't open file or run syntax highlighter");
        return $fd;
 }
@@ -3472,6 +3494,51 @@ sub get_page_title {
        return $title;
 }
 
+sub print_feed_meta {
+       if (defined $project) {
+               my %href_params = get_feed_info();
+               if (!exists $href_params{'-title'}) {
+                       $href_params{'-title'} = 'log';
+               }
+
+               foreach my $format (qw(RSS Atom)) {
+                       my $type = lc($format);
+                       my %link_attr = (
+                               '-rel' => 'alternate',
+                               '-title' => esc_attr("$project - $href_params{'-title'} - $format feed"),
+                               '-type' => "application/$type+xml"
+                       );
+
+                       $href_params{'action'} = $type;
+                       $link_attr{'-href'} = href(%href_params);
+                       print "<link ".
+                             "rel=\"$link_attr{'-rel'}\" ".
+                             "title=\"$link_attr{'-title'}\" ".
+                             "href=\"$link_attr{'-href'}\" ".
+                             "type=\"$link_attr{'-type'}\" ".
+                             "/>\n";
+
+                       $href_params{'extra_options'} = '--no-merges';
+                       $link_attr{'-href'} = href(%href_params);
+                       $link_attr{'-title'} .= ' (no merges)';
+                       print "<link ".
+                             "rel=\"$link_attr{'-rel'}\" ".
+                             "title=\"$link_attr{'-title'}\" ".
+                             "href=\"$link_attr{'-href'}\" ".
+                             "type=\"$link_attr{'-type'}\" ".
+                             "/>\n";
+               }
+
+       } else {
+               printf('<link rel="alternate" title="%s projects list" '.
+                      'href="%s" type="text/plain; charset=utf-8" />'."\n",
+                      esc_attr($site_name), href(project=>undef, action=>"project_index"));
+               printf('<link rel="alternate" title="%s projects feeds" '.
+                      'href="%s" type="text/x-opml" />'."\n",
+                      esc_attr($site_name), href(project=>undef, action=>"opml"));
+       }
+}
+
 sub git_header_html {
        my $status = shift || "200 OK";
        my $expires = shift;
@@ -3514,57 +3581,17 @@ EOF
        # print out each stylesheet that exist, providing backwards capability
        # for those people who defined $stylesheet in a config file
        if (defined $stylesheet) {
-               print '<link rel="stylesheet" type="text/css" href="'.$stylesheet.'"/>'."\n";
+               print '<link rel="stylesheet" type="text/css" href="'.esc_url($stylesheet).'"/>'."\n";
        } else {
                foreach my $stylesheet (@stylesheets) {
                        next unless $stylesheet;
-                       print '<link rel="stylesheet" type="text/css" href="'.$stylesheet.'"/>'."\n";
-               }
-       }
-       if (defined $project) {
-               my %href_params = get_feed_info();
-               if (!exists $href_params{'-title'}) {
-                       $href_params{'-title'} = 'log';
-               }
-
-               foreach my $format qw(RSS Atom) {
-                       my $type = lc($format);
-                       my %link_attr = (
-                               '-rel' => 'alternate',
-                               '-title' => "$project - $href_params{'-title'} - $format feed",
-                               '-type' => "application/$type+xml"
-                       );
-
-                       $href_params{'action'} = $type;
-                       $link_attr{'-href'} = href(%href_params);
-                       print "<link ".
-                             "rel=\"$link_attr{'-rel'}\" ".
-                             "title=\"$link_attr{'-title'}\" ".
-                             "href=\"$link_attr{'-href'}\" ".
-                             "type=\"$link_attr{'-type'}\" ".
-                             "/>\n";
-
-                       $href_params{'extra_options'} = '--no-merges';
-                       $link_attr{'-href'} = href(%href_params);
-                       $link_attr{'-title'} .= ' (no merges)';
-                       print "<link ".
-                             "rel=\"$link_attr{'-rel'}\" ".
-                             "title=\"$link_attr{'-title'}\" ".
-                             "href=\"$link_attr{'-href'}\" ".
-                             "type=\"$link_attr{'-type'}\" ".
-                             "/>\n";
+                       print '<link rel="stylesheet" type="text/css" href="'.esc_url($stylesheet).'"/>'."\n";
                }
-
-       } else {
-               printf('<link rel="alternate" title="%s projects list" '.
-                      'href="%s" type="text/plain; charset=utf-8" />'."\n",
-                      $site_name, href(project=>undef, action=>"project_index"));
-               printf('<link rel="alternate" title="%s projects feeds" '.
-                      'href="%s" type="text/x-opml" />'."\n",
-                      $site_name, href(project=>undef, action=>"opml"));
        }
+       print_feed_meta()
+               if ($status eq '200 OK');
        if (defined $favicon) {
-               print qq(<link rel="shortcut icon" href="$favicon" type="image/png" />\n);
+               print qq(<link rel="shortcut icon" href=").esc_url($favicon).qq(" type="image/png" />\n);
        }
 
        print "</head>\n" .
@@ -3574,10 +3601,15 @@ EOF
                insert_file($site_header);
        }
 
-       print "<div class=\"page_header\">\n" .
-             $cgi->a({-href => esc_url($logo_url),
-                      -title => $logo_label},
-                     qq(<img src="$logo" width="72" height="27" alt="git" class="logo"/>));
+       print "<div class=\"page_header\">\n";
+       if (defined $logo) {
+               print $cgi->a({-href => esc_url($logo_url),
+                              -title => $logo_label},
+                             $cgi->img({-src => esc_url($logo),
+                                        -width => 72, -height => 27,
+                                        -alt => "git",
+                                        -class => "logo"}));
+       }
        print $cgi->a({-href => esc_url($home_link)}, $home_link_str) . " / ";
        if (defined $project) {
                print $cgi->a({-href => href(action=>"summary")}, esc_html($project));
@@ -3650,7 +3682,7 @@ sub git_footer_html {
                }
                $href_params{'-title'} ||= 'log';
 
-               foreach my $format qw(RSS Atom) {
+               foreach my $format (qw(RSS Atom)) {
                        $href_params{'action'} = lc($format);
                        print $cgi->a({-href => href(%href_params),
                                      -title => "$href_params{'-title'} $format feed",
@@ -3683,7 +3715,7 @@ sub git_footer_html {
                insert_file($site_footer);
        }
 
-       print qq!<script type="text/javascript" src="$javascript"></script>\n!;
+       print qq!<script type="text/javascript" src="!.esc_url($javascript).qq!"></script>\n!;
        if (defined $action &&
            $action eq 'blame_incremental') {
                print qq!<script type="text/javascript">\n!.
@@ -4380,7 +4412,7 @@ sub git_difftree_body {
                }
                if ($diff->{'from_mode'} ne ('0' x 6)) {
                        $from_mode_oct = oct $diff->{'from_mode'};
-                       if (S_ISREG($to_mode_oct)) { # only for regular file
+                       if (S_ISREG($from_mode_oct)) { # only for regular file
                                $from_mode_str = sprintf("%04o", $from_mode_oct & 0777); # permission bits
                        }
                        $from_file_type = file_type($diff->{'from_mode'});
@@ -5894,14 +5926,14 @@ sub git_blob {
        } else {
                print "<div class=\"page_nav\">\n" .
                      "<br/><br/></div>\n" .
-                     "<div class=\"title\">$hash</div>\n";
+                     "<div class=\"title\">".esc_html($hash)."</div>\n";
        }
        git_print_page_path($file_name, "blob", $hash_base);
        print "<div class=\"page_body\">\n";
        if ($mimetype =~ m!^image/!) {
-               print qq!<img type="$mimetype"!;
+               print qq!<img type="!.esc_attr($mimetype).qq!"!;
                if ($file_name) {
-                       print qq! alt="$file_name" title="$file_name"!;
+                       print qq! alt="!.esc_attr($file_name).qq!" title="!.esc_attr($file_name).qq!"!;
                }
                print qq! src="! .
                      href(action=>"blob_plain", hash=>$hash,
@@ -5914,7 +5946,7 @@ sub git_blob {
                        $nr++;
                        $line = untabify($line);
                        printf qq!<div class="pre"><a id="l%i" href="%s#l%i" class="linenr">%4i</a> %s</div>\n!,
-                              $nr, href(-replay => 1), $nr, $nr, $syntax ? $line : esc_html($line, -nbsp=>1);
+                              $nr, esc_attr(href(-replay => 1)), $nr, $nr, $syntax ? $line : esc_html($line, -nbsp=>1);
                }
        }
        close $fd
@@ -5976,7 +6008,7 @@ sub git_tree {
                undef $hash_base;
                print "<div class=\"page_nav\">\n";
                print "<br/><br/></div>\n";
-               print "<div class=\"title\">$hash</div>\n";
+               print "<div class=\"title\">".esc_html($hash)."</div>\n";
        }
        if (defined $file_name) {
                $basedir = $file_name;
@@ -6444,7 +6476,7 @@ sub git_blobdiff {
                        git_print_header_div('commit', esc_html($co{'title'}), $hash_base);
                } else {
                        print "<div class=\"page_nav\"><br/>$formats_nav<br/></div>\n";
-                       print "<div class=\"title\">$hash vs $hash_parent</div>\n";
+                       print "<div class=\"title\">".esc_html("$hash vs $hash_parent")."</div>\n";
                }
                if (defined $file_name) {
                        git_print_page_path($file_name, "blob", $hash_base);
@@ -7142,7 +7174,7 @@ XML
                if (defined $favicon) {
                        print "<icon>" . esc_url($favicon) . "</icon>\n";
                }
-               if (defined $logo_url) {
+               if (defined $logo) {
                        # not twice as wide as tall: 72 x 27 pixels
                        print "<logo>" . esc_url($logo) . "</logo>\n";
                }