Merge branch 'en/maint-hash-object' into maint
[gitweb.git] / gitweb / gitweb.perl
index 99f71b47c2a6b53bb52ce29e96361e7c2acbe19d..33ef190ceb99565f804783ba9fe76987ad6bea89 100755 (executable)
@@ -27,13 +27,29 @@ BEGIN
 our $my_url = $cgi->url();
 our $my_uri = $cgi->url(-absolute => 1);
 
-# if we're called with PATH_INFO, we have to strip that
-# from the URL to find our real URL
-# we make $path_info global because it's also used later on
+# Base URL for relative URLs in gitweb ($logo, $favicon, ...),
+# needed and used only for URLs with nonempty PATH_INFO
+our $base_url = $my_url;
+
+# When the script is used as DirectoryIndex, the URL does not contain the name
+# of the script file itself, and $cgi->url() fails to strip PATH_INFO, so we
+# have to do it ourselves. We make $path_info global because it's also used
+# later on.
+#
+# Another issue with the script being the DirectoryIndex is that the resulting
+# $my_url data is not the full script URL: this is good, because we want
+# generated links to keep implying the script name if it wasn't explicitly
+# indicated in the URL we're handling, but it means that $my_url cannot be used
+# as base URL.
+# Therefore, if we needed to strip PATH_INFO, then we know that we have
+# to build the base URL ourselves:
 our $path_info = $ENV{"PATH_INFO"};
 if ($path_info) {
-       $my_url =~ s,\Q$path_info\E$,,;
-       $my_uri =~ s,\Q$path_info\E$,,;
+       if ($my_url =~ s,\Q$path_info\E$,, &&
+           $my_uri =~ s,\Q$path_info\E$,, &&
+           defined $ENV{'SCRIPT_NAME'}) {
+               $base_url = $cgi->url(-base => 1) . $ENV{'SCRIPT_NAME'};
+       }
 }
 
 # core git executable to use
@@ -132,6 +148,10 @@ BEGIN
 # - one might want to include '-B' option, e.g. '-B', '-M'
 our @diff_opts = ('-M'); # taken from git_commit
 
+# Disables features that would allow repository owners to inject script into
+# the gitweb domain.
+our $prevent_xss = 0;
+
 # information about snapshot formats that gitweb is capable of serving
 our %known_snapshot_formats = (
        # name => {
@@ -203,7 +223,7 @@ BEGIN
        # $feature{'blame'}{'override'} = 1;
        # and in project config gitweb.blame = 0|1;
        'blame' => {
-               'sub' => \&feature_blame,
+               'sub' => sub { feature_bool('blame', @_) },
                'override' => 0,
                'default' => [0]},
 
@@ -241,7 +261,7 @@ BEGIN
        # $feature{'grep'}{'override'} = 1;
        # and in project config gitweb.grep = 0|1;
        'grep' => {
-               'sub' => \&feature_grep,
+               'sub' => sub { feature_bool('grep', @_) },
                'override' => 0,
                'default' => [1]},
 
@@ -255,7 +275,7 @@ BEGIN
        # $feature{'pickaxe'}{'override'} = 1;
        # and in project config gitweb.pickaxe = 0|1;
        'pickaxe' => {
-               'sub' => \&feature_pickaxe,
+               'sub' => sub { feature_bool('pickaxe', @_) },
                'override' => 0,
                'default' => [1]},
 
@@ -330,6 +350,21 @@ BEGIN
        'ctags' => {
                'override' => 0,
                'default' => [0]},
+
+       # The maximum number of patches in a patchset generated in patch
+       # view. Set this to 0 or undef to disable patch view, or to a
+       # negative number to remove any limit.
+
+       # To disable system wide have in $GITWEB_CONFIG
+       # $feature{'patches'}{'default'} = [0];
+       # To have project specific config enable override in $GITWEB_CONFIG
+       # $feature{'patches'}{'override'} = 1;
+       # and in project config gitweb.patches = 0|n;
+       # where n is the maximum number of patches allowed in a patchset.
+       'patches' => {
+               'sub' => \&feature_patches,
+               'override' => 0,
+               'default' => [16]},
 );
 
 sub gitweb_get_feature {
@@ -363,16 +398,17 @@ sub gitweb_check_feature {
 }
 
 
-sub feature_blame {
-       my ($val) = git_get_project_config('blame', '--bool');
+sub feature_bool {
+       my $key = shift;
+       my ($val) = git_get_project_config($key, '--bool');
 
-       if ($val eq 'true') {
-               return 1;
+       if (!defined $val) {
+               return ($_[0]);
+       } elsif ($val eq 'true') {
+               return (1);
        } elsif ($val eq 'false') {
-               return 0;
+               return (0);
        }
-
-       return $_[0];
 }
 
 sub feature_snapshot {
@@ -387,25 +423,11 @@ sub feature_snapshot {
        return @fmts;
 }
 
-sub feature_grep {
-       my ($val) = git_get_project_config('grep', '--bool');
+sub feature_patches {
+       my @val = (git_get_project_config('patches', '--int'));
 
-       if ($val eq 'true') {
-               return (1);
-       } elsif ($val eq 'false') {
-               return (0);
-       }
-
-       return ($_[0]);
-}
-
-sub feature_pickaxe {
-       my ($val) = git_get_project_config('pickaxe', '--bool');
-
-       if ($val eq 'true') {
-               return (1);
-       } elsif ($val eq 'false') {
-               return (0);
+       if (@val) {
+               return @val;
        }
 
        return ($_[0]);
@@ -504,6 +526,8 @@ sub filter_snapshot_fmts {
        "heads" => \&git_heads,
        "history" => \&git_history,
        "log" => \&git_log,
+       "patch" => \&git_patch,
+       "patches" => \&git_patches,
        "rss" => \&git_rss,
        "atom" => \&git_atom,
        "search" => \&git_search,
@@ -830,7 +854,7 @@ (%)
        }
 
        my $use_pathinfo = gitweb_check_feature('pathinfo');
-       if ($use_pathinfo) {
+       if ($use_pathinfo and defined $params{'project'}) {
                # try to put as many parameters as possible in PATH_INFO:
                #   - project name
                #   - action
@@ -845,7 +869,7 @@ (%)
                $href =~ s,/$,,;
 
                # Then add the project name, if present
-               $href .= "/".esc_url($params{'project'}) if defined $params{'project'};
+               $href .= "/".esc_url($params{'project'});
                delete $params{'project'};
 
                # since we destructively absorb parameters, we keep this
@@ -1360,13 +1384,11 @@ sub format_log_line_html {
        my $line = shift;
 
        $line = esc_html($line, -nbsp=>1);
-       if ($line =~ m/([0-9a-fA-F]{8,40})/) {
-               my $hash_text = $1;
-               my $link =
-                       $cgi->a({-href => href(action=>"object", hash=>$hash_text),
-                               -class => "text"}, $hash_text);
-               $line =~ s/$hash_text/$link/;
-       }
+       $line =~ s{\b([0-9a-fA-F]{8,40})\b}{
+               $cgi->a({-href => href(action=>"object", hash=>$1),
+                                       -class => "text"}, $1);
+       }eg;
+
        return $line;
 }
 
@@ -1890,18 +1912,19 @@ sub git_parse_project_config {
        return %config;
 }
 
-# convert config value to boolean, 'true' or 'false'
+# convert config value to boolean: 'true' or 'false'
 # no value, number > 0, 'true' and 'yes' values are true
 # rest of values are treated as false (never as error)
 sub config_to_bool {
        my $val = shift;
 
+       return 1 if !defined $val;             # section.key
+
        # strip leading and trailing whitespace
        $val =~ s/^\s+//;
        $val =~ s/\s+$//;
 
-       return (!defined $val ||               # section.key
-               ($val =~ /^\d+$/ && $val) ||   # section.key = 1
+       return (($val =~ /^\d+$/ && $val) ||   # section.key = 1
                ($val =~ /^(?:true|yes)$/i));  # section.key = true
 }
 
@@ -1954,6 +1977,9 @@ sub git_get_project_config {
                $config_file = "$git_dir/config";
        }
 
+       # check if config variable (key) exists
+       return unless exists $config{"gitweb.$key"};
+
        # ensure given type
        if (!defined $type) {
                return $config{"gitweb.$key"};
@@ -2897,9 +2923,14 @@ sub git_header_html {
 <meta name="robots" content="index, nofollow"/>
 <title>$title</title>
 EOF
-# print out each stylesheet that exist
+       # the stylesheet, favicon etc urls won't work correctly with path_info
+       # unless we set the appropriate base URL
+       if ($ENV{'PATH_INFO'}) {
+               print "<base href=\"".esc_url($base_url)."\" />\n";
+       }
+       # print out each stylesheet that exist, providing backwards capability
+       # for those people who defined $stylesheet in a config file
        if (defined $stylesheet) {
-#provides backwards capability for those people who define style sheet in a config file
                print '<link rel="stylesheet" type="text/css" href="'.$stylesheet.'"/>'."\n";
        } else {
                foreach my $stylesheet (@stylesheets) {
@@ -4494,7 +4525,9 @@ sub git_summary {
 
        print "</table>\n";
 
-       if (-s "$projectroot/$project/README.html") {
+       # If XSS prevention is on, we don't include README.html.
+       # TODO: Allow a readme in some safe format.
+       if (!$prevent_xss && -s "$projectroot/$project/README.html") {
                print "<div class=\"title\">readme</div>\n" .
                      "<div class=\"readme\">\n";
                insert_file("$projectroot/$project/README.html");
@@ -4576,28 +4609,33 @@ sub git_tag {
 }
 
 sub git_blame {
-       my $fd;
-       my $ftype;
-
+       # permissions
        gitweb_check_feature('blame')
-           or die_error(403, "Blame view not allowed");
+               or die_error(403, "Blame view not allowed");
 
+       # error checking
        die_error(400, "No file name given") unless $file_name;
        $hash_base ||= git_get_head_hash($project);
-       die_error(404, "Couldn't find base commit") unless ($hash_base);
+       die_error(404, "Couldn't find base commit") unless $hash_base;
        my %co = parse_commit($hash_base)
                or die_error(404, "Commit not found");
+       my $ftype = "blob";
        if (!defined $hash) {
                $hash = git_get_hash_by_path($hash_base, $file_name, "blob")
                        or die_error(404, "Error looking up file");
+       } else {
+               $ftype = git_get_type($hash);
+               if ($ftype !~ "blob") {
+                       die_error(400, "Object is not a blob");
+               }
        }
-       $ftype = git_get_type($hash);
-       if ($ftype !~ "blob") {
-               die_error(400, "Object is not a blob");
-       }
-       open ($fd, "-|", git_cmd(), "blame", '-p', '--',
-             $file_name, $hash_base)
+
+       # run git-blame --porcelain
+       open my $fd, "-|", git_cmd(), "blame", '-p',
+               $hash_base, '--', $file_name
                or die_error(500, "Open git-blame failed");
+
+       # page header
        git_header_html();
        my $formats_nav =
                $cgi->a({-href => href(action=>"blob", -replay=>1)},
@@ -4611,42 +4649,46 @@ sub git_blame {
        git_print_page_nav('','', $hash_base,$co{'tree'},$hash_base, $formats_nav);
        git_print_header_div('commit', esc_html($co{'title'}), $hash_base);
        git_print_page_path($file_name, $ftype, $hash_base);
-       my @rev_color = (qw(light2 dark2));
+
+       # page body
+       my @rev_color = qw(light2 dark2);
        my $num_colors = scalar(@rev_color);
        my $current_color = 0;
-       my $last_rev;
+       my %metainfo = ();
+
        print <<HTML;
 <div class="page_body">
 <table class="blame">
 <tr><th>Commit</th><th>Line</th><th>Data</th></tr>
 HTML
-       my %metainfo = ();
-       while (1) {
-               $_ = <$fd>;
-               last unless defined $_;
+ LINE:
+       while (my $line = <$fd>) {
+               chomp $line;
+               # the header: <SHA-1> <src lineno> <dst lineno> [<lines in group>]
+               # no <lines in group> for subsequent lines in group of lines
                my ($full_rev, $orig_lineno, $lineno, $group_size) =
-                   /^([0-9a-f]{40}) (\d+) (\d+)(?: (\d+))?$/;
+                  ($line =~ /^([0-9a-f]{40}) (\d+) (\d+)(?: (\d+))?$/);
                if (!exists $metainfo{$full_rev}) {
                        $metainfo{$full_rev} = {};
                }
                my $meta = $metainfo{$full_rev};
-               while (<$fd>) {
-                       last if (s/^\t//);
-                       if (/^(\S+) (.*)$/) {
+               my $data;
+               while ($data = <$fd>) {
+                       chomp $data;
+                       last if ($data =~ s/^\t//); # contents of line
+                       if ($data =~ /^(\S+) (.*)$/) {
                                $meta->{$1} = $2;
                        }
                }
-               my $data = $_;
-               chomp $data;
-               my $rev = substr($full_rev, 0, 8);
+               my $short_rev = substr($full_rev, 0, 8);
                my $author = $meta->{'author'};
-               my %date = parse_date($meta->{'author-time'},
-                                     $meta->{'author-tz'});
+               my %date =
+                       parse_date($meta->{'author-time'}, $meta->{'author-tz'});
                my $date = $date{'iso-tz'};
                if ($group_size) {
-                       $current_color = ++$current_color % $num_colors;
+                       $current_color = ($current_color + 1) % $num_colors;
                }
-               print "<tr class=\"$rev_color[$current_color]\">\n";
+               print "<tr id=\"l$lineno\" class=\"$rev_color[$current_color]\">\n";
                if ($group_size) {
                        print "<td class=\"sha1\"";
                        print " title=\"". esc_html($author) . ", $date\"";
@@ -4655,20 +4697,25 @@ sub git_blame {
                        print $cgi->a({-href => href(action=>"commit",
                                                     hash=>$full_rev,
                                                     file_name=>$file_name)},
-                                     esc_html($rev));
+                                     esc_html($short_rev));
                        print "</td>\n";
                }
-               open (my $dd, "-|", git_cmd(), "rev-parse", "$full_rev^")
-                       or die_error(500, "Open git-rev-parse failed");
-               my $parent_commit = <$dd>;
-               close $dd;
-               chomp($parent_commit);
+               my $parent_commit;
+               if (!exists $meta->{'parent'}) {
+                       open (my $dd, "-|", git_cmd(), "rev-parse", "$full_rev^")
+                               or die_error(500, "Open git-rev-parse failed");
+                       $parent_commit = <$dd>;
+                       close $dd;
+                       chomp($parent_commit);
+                       $meta->{'parent'} = $parent_commit;
+               } else {
+                       $parent_commit = $meta->{'parent'};
+               }
                my $blamed = href(action => 'blame',
                                  file_name => $meta->{'filename'},
                                  hash_base => $parent_commit);
                print "<td class=\"linenr\">";
                print $cgi->a({ -href => "$blamed#l$orig_lineno",
-                               -id => "l$lineno",
                                -class => "linenr" },
                              esc_html($lineno));
                print "</td>";
@@ -4679,6 +4726,8 @@ sub git_blame {
        print "</div>";
        close $fd
                or print "Reading blob failed\n";
+
+       # page footer
        git_footer_html();
 }
 
@@ -4739,10 +4788,21 @@ sub git_blob_plain {
                $save_as .= '.txt';
        }
 
+       # With XSS prevention on, blobs of all types except a few known safe
+       # ones are served with "Content-Disposition: attachment" to make sure
+       # they don't run in our security domain.  For certain image types,
+       # blob view writes an <img> tag referring to blob_plain view, and we
+       # want to be sure not to break that by serving the image as an
+       # attachment (though Firefox 3 doesn't seem to care).
+       my $sandbox = $prevent_xss &&
+               $type !~ m!^(?:text/plain|image/(?:gif|png|jpeg))$!;
+
        print $cgi->header(
                -type => $type,
                -expires => $expires,
-               -content_disposition => 'inline; filename="' . $save_as . '"');
+               -content_disposition =>
+                       ($sandbox ? 'attachment' : 'inline')
+                       . '; filename="' . $save_as . '"');
        undef $/;
        binmode STDOUT, ':raw';
        print <$fd>;
@@ -4998,6 +5058,15 @@ sub git_log {
 
        my $paging_nav = format_paging_nav('log', $hash, $head, $page, $#commitlist >= 100);
 
+       my ($patch_max) = gitweb_get_feature('patches');
+       if ($patch_max) {
+               if ($patch_max < 0 || @commitlist <= $patch_max) {
+                       $paging_nav .= " &sdot; " .
+                               $cgi->a({-href => href(action=>"patches", -replay=>1)},
+                                       "patches");
+               }
+       }
+
        git_header_html();
        git_print_page_nav('log','', $hash,undef,undef, $paging_nav);
 
@@ -5077,6 +5146,11 @@ sub git_commit {
                        } @$parents ) .
                        ')';
        }
+       if (gitweb_check_feature('patches')) {
+               $formats_nav .= " | " .
+                       $cgi->a({-href => href(action=>"patch", -replay=>1)},
+                               "patch");
+       }
 
        if (!defined $parent) {
                $parent = "--root";
@@ -5353,7 +5427,14 @@ sub git_blobdiff_plain {
 }
 
 sub git_commitdiff {
-       my $format = shift || 'html';
+       my %params = @_;
+       my $format = $params{-format} || 'html';
+
+       my ($patch_max) = gitweb_get_feature('patches');
+       if ($format eq 'patch') {
+               die_error(403, "Patch view not allowed") unless $patch_max;
+       }
+
        $hash ||= $hash_base || "HEAD";
        my %co = parse_commit($hash)
            or die_error(404, "Unknown commit object");
@@ -5368,6 +5449,11 @@ sub git_commitdiff {
                $formats_nav =
                        $cgi->a({-href => href(action=>"commitdiff_plain", -replay=>1)},
                                "raw");
+               if ($patch_max) {
+                       $formats_nav .= " | " .
+                               $cgi->a({-href => href(action=>"patch", -replay=>1)},
+                                       "patch");
+               }
 
                if (defined $hash_parent &&
                    $hash_parent ne '-c' && $hash_parent ne '--cc') {
@@ -5451,7 +5537,31 @@ sub git_commitdiff {
                open $fd, "-|", git_cmd(), "diff-tree", '-r', @diff_opts,
                        '-p', $hash_parent_param, $hash, "--"
                        or die_error(500, "Open git-diff-tree failed");
-
+       } elsif ($format eq 'patch') {
+               # For commit ranges, we limit the output to the number of
+               # patches specified in the 'patches' feature.
+               # For single commits, we limit the output to a single patch,
+               # diverging from the git-format-patch default.
+               my @commit_spec = ();
+               if ($hash_parent) {
+                       if ($patch_max > 0) {
+                               push @commit_spec, "-$patch_max";
+                       }
+                       push @commit_spec, '-n', "$hash_parent..$hash";
+               } else {
+                       if ($params{-single}) {
+                               push @commit_spec, '-1';
+                       } else {
+                               if ($patch_max > 0) {
+                                       push @commit_spec, "-$patch_max";
+                               }
+                               push @commit_spec, "-n";
+                       }
+                       push @commit_spec, '--root', $hash;
+               }
+               open $fd, "-|", git_cmd(), "format-patch", '--encoding=utf8',
+                       '--stdout', @commit_spec
+                       or die_error(500, "Open git-format-patch failed");
        } else {
                die_error(400, "Unknown commitdiff format");
        }
@@ -5500,6 +5610,14 @@ sub git_commitdiff {
                        print to_utf8($line) . "\n";
                }
                print "---\n\n";
+       } elsif ($format eq 'patch') {
+               my $filename = basename($project) . "-$hash.patch";
+
+               print $cgi->header(
+                       -type => 'text/plain',
+                       -charset => 'utf-8',
+                       -expires => $expires,
+                       -content_disposition => 'inline; filename="' . "$filename" . '"');
        }
 
        # write patch
@@ -5521,11 +5639,25 @@ sub git_commitdiff {
                print <$fd>;
                close $fd
                        or print "Reading git-diff-tree failed\n";
+       } elsif ($format eq 'patch') {
+               local $/ = undef;
+               print <$fd>;
+               close $fd
+                       or print "Reading git-format-patch failed\n";
        }
 }
 
 sub git_commitdiff_plain {
-       git_commitdiff('plain');
+       git_commitdiff(-format => 'plain');
+}
+
+# format-patch-style patches
+sub git_patch {
+       git_commitdiff(-format => 'patch', -single=> 1);
+}
+
+sub git_patches {
+       git_commitdiff(-format => 'patch');
 }
 
 sub git_history {
@@ -5878,6 +6010,14 @@ sub git_shortlog {
                        $cgi->a({-href => href(-replay=>1, page=>$page+1),
                                 -accesskey => "n", -title => "Alt-n"}, "next");
        }
+       my $patch_max = gitweb_check_feature('patches');
+       if ($patch_max) {
+               if ($patch_max < 0 || @commitlist <= $patch_max) {
+                       $paging_nav .= " &sdot; " .
+                               $cgi->a({-href => href(action=>"patches", -replay=>1)},
+                                       "patches");
+               }
+       }
 
        git_header_html();
        git_print_page_nav('shortlog','', $hash,$hash,$hash, $paging_nav);
@@ -5915,7 +6055,25 @@ sub git_feed {
        }
        if (defined($commitlist[0])) {
                %latest_commit = %{$commitlist[0]};
-               %latest_date   = parse_date($latest_commit{'author_epoch'});
+               my $latest_epoch = $latest_commit{'committer_epoch'};
+               %latest_date   = parse_date($latest_epoch);
+               my $if_modified = $cgi->http('IF_MODIFIED_SINCE');
+               if (defined $if_modified) {
+                       my $since;
+                       if (eval { require HTTP::Date; 1; }) {
+                               $since = HTTP::Date::str2time($if_modified);
+                       } elsif (eval { require Time::ParseDate; 1; }) {
+                               $since = Time::ParseDate::parsedate($if_modified, GMT => 1);
+                       }
+                       if (defined $since && $latest_epoch <= $since) {
+                               print $cgi->header(
+                                       -type => $content_type,
+                                       -charset => 'utf-8',
+                                       -last_modified => $latest_date{'rfc2822'},
+                                       -status => '304 Not Modified');
+                               return;
+                       }
+               }
                print $cgi->header(
                        -type => $content_type,
                        -charset => 'utf-8',
@@ -5974,7 +6132,24 @@ sub git_feed {
                print "<title>$title</title>\n" .
                      "<link>$alt_url</link>\n" .
                      "<description>$descr</description>\n" .
-                     "<language>en</language>\n";
+                     "<language>en</language>\n" .
+                     # project owner is responsible for 'editorial' content
+                     "<managingEditor>$owner</managingEditor>\n";
+               if (defined $logo || defined $favicon) {
+                       # prefer the logo to the favicon, since RSS
+                       # doesn't allow both
+                       my $img = esc_url($logo || $favicon);
+                       print "<image>\n" .
+                             "<url>$img</url>\n" .
+                             "<title>$title</title>\n" .
+                             "<link>$alt_url</link>\n" .
+                             "</image>\n";
+               }
+               if (%latest_date) {
+                       print "<pubDate>$latest_date{'rfc2822'}</pubDate>\n";
+                       print "<lastBuildDate>$latest_date{'rfc2822'}</lastBuildDate>\n";
+               }
+               print "<generator>gitweb v.$version/$git_version</generator>\n";
        } elsif ($format eq 'atom') {
                print <<XML;
 <feed xmlns="http://www.w3.org/2005/Atom">
@@ -6001,6 +6176,7 @@ sub git_feed {
                } else {
                        print "<updated>$latest_date{'iso-8601'}</updated>\n";
                }
+               print "<generator version='$version/$git_version'>gitweb</generator>\n";
        }
 
        # contents
@@ -6122,7 +6298,11 @@ sub git_atom {
 sub git_opml {
        my @list = git_get_projects_list();
 
-       print $cgi->header(-type => 'text/xml', -charset => 'utf-8');
+       print $cgi->header(
+               -type => 'text/xml',
+               -charset => 'utf-8',
+               -content_disposition => 'inline; filename="opml.xml"');
+
        print <<XML;
 <?xml version="1.0" encoding="utf-8"?>
 <opml version="1.0">
@@ -6146,8 +6326,8 @@ sub git_opml {
                }
 
                my $path = esc_html(chop_str($proj{'path'}, 25, 5));
-               my $rss  = "$my_url?p=$proj{'path'};a=rss";
-               my $html = "$my_url?p=$proj{'path'};a=summary";
+               my $rss  = href('project' => $proj{'path'}, 'action' => 'rss', -full => 1);
+               my $html = href('project' => $proj{'path'}, 'action' => 'summary', -full => 1);
                print "<outline type=\"rss\" text=\"$path\" title=\"$path\" xmlUrl=\"$rss\" htmlUrl=\"$html\"/>\n";
        }
        print <<XML;