Upgrade to latest gitweb
authorBernie Innocenti <bernie@codewiz.org>
Tue, 17 Feb 2009 16:59:22 +0000 (17:59 +0100)
committerBernie Innocenti <bernie@codewiz.org>
Tue, 17 Feb 2009 16:59:22 +0000 (17:59 +0100)
gitweb.cgi
gitweb.css
gitweb_config.perl

index 2b78db3844b31957a2b2cec485342466b15ab647..643d43a5777dee77e6e865beb7c43ae0dff0ff31 100755 (executable)
@@ -23,10 +23,35 @@ BEGIN {
 }
 
 our $cgi = new CGI;
-our $version = "1.5.6.5";
+our $version = "1.6.2.rc1.3.g81d3f";
 our $my_url = $cgi->url();
 our $my_uri = $cgi->url(-absolute => 1);
 
+# 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) {
+       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
 # this can just be "git" if your webserver has a sensible PATH
 our $GIT = "/usr/bin/git";
@@ -43,7 +68,7 @@ our $project_maxdepth = 2007;
 our $home_link = $my_uri || "/";
 
 # string of the home link on top of all pages
-our $home_link_str = $ENV{'SERVER_NAME'} ? "git://" . $ENV{'SERVER_NAME'} : "projects";
+our $home_link_str = "projects";
 
 # name of your site or organization to appear in page titles
 # replace this with something more descriptive for clearer bookmarks
@@ -86,6 +111,11 @@ our $default_projects_order = "project";
 # (only effective if this variable evaluates to true)
 our $export_ok = "";
 
+# show repository only if this subroutine returns true
+# when given the path to the project, for example:
+#    sub { return -e "$_[0]/git-daemon-export-ok"; }
+our $export_auth_hook = undef;
+
 # only allow viewing of repositories also shown on the overview page
 our $strict_export = "";
 
@@ -118,6 +148,10 @@ our $fallback_encoding = 'latin1';
 # - 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 => {
@@ -176,7 +210,9 @@ our %feature = (
        # if there is no 'sub' key (no feature-sub), then feature cannot be
        # overriden
        #
-       # use gitweb_check_feature(<feature>) to check if <feature> is enabled
+       # use gitweb_get_feature(<feature>) to retrieve the <feature> value
+       # (an array) or gitweb_check_feature(<feature>) to check if <feature>
+       # is enabled
 
        # Enable the 'blame' blob view, showing the last commit that modified
        # each line in the file. This can be very CPU-intensive.
@@ -187,7 +223,7 @@ our %feature = (
        # $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]},
 
@@ -225,6 +261,7 @@ our %feature = (
        # $feature{'grep'}{'override'} = 1;
        # and in project config gitweb.grep = 0|1;
        'grep' => {
+               'sub' => sub { feature_bool('grep', @_) },
                'override' => 0,
                'default' => [1]},
 
@@ -238,7 +275,7 @@ our %feature = (
        # $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]},
 
@@ -275,9 +312,62 @@ our %feature = (
        'forks' => {
                'override' => 0,
                'default' => [0]},
+
+       # Insert custom links to the action bar of all project pages.
+       # This enables you mainly to link to third-party scripts integrating
+       # into gitweb; e.g. git-browser for graphical history representation
+       # or custom web-based repository administration interface.
+
+       # The 'default' value consists of a list of triplets in the form
+       # (label, link, position) where position is the label after which
+       # to insert the link and link is a format string where %n expands
+       # to the project name, %f to the project path within the filesystem,
+       # %h to the current hash (h gitweb parameter) and %b to the current
+       # hash base (hb gitweb parameter); %% expands to %.
+
+       # To enable system wide have in $GITWEB_CONFIG e.g.
+       # $feature{'actions'}{'default'} = [('graphiclog',
+       #       '/git-browser/by-commit.html?r=%n', 'summary')];
+       # Project specific override is not supported.
+       'actions' => {
+               'override' => 0,
+               'default' => []},
+
+       # Allow gitweb scan project content tags described in ctags/
+       # of project repository, and display the popular Web 2.0-ish
+       # "tag cloud" near the project list. Note that this is something
+       # COMPLETELY different from the normal Git tags.
+
+       # gitweb by itself can show existing tags, but it does not handle
+       # tagging itself; you need an external application for that.
+       # For an example script, check Girocco's cgi/tagproj.cgi.
+       # You may want to install the HTML::TagCloud Perl module to get
+       # a pretty tag cloud instead of just a list of tags.
+
+       # To enable system wide have in $GITWEB_CONFIG
+       # $feature{'ctags'}{'default'} = ['path_to_tag_script'];
+       # Project specific override is not supported.
+       '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_check_feature {
+sub gitweb_get_feature {
        my ($name) = @_;
        return unless exists $feature{$name};
        my ($sub, $override, @defaults) = (
@@ -292,16 +382,33 @@ sub gitweb_check_feature {
        return $sub->(@defaults);
 }
 
-sub feature_blame {
-       my ($val) = git_get_project_config('blame', '--bool');
+# A wrapper to check if a given feature is enabled.
+# With this, you can say
+#
+#   my $bool_feat = gitweb_check_feature('bool_feat');
+#   gitweb_check_feature('bool_feat') or somecode;
+#
+# instead of
+#
+#   my ($bool_feat) = gitweb_get_feature('bool_feat');
+#   (gitweb_get_feature('bool_feat'))[0] or somecode;
+#
+sub gitweb_check_feature {
+       return (gitweb_get_feature(@_))[0];
+}
+
+
+sub feature_bool {
+       my $key = shift;
+       my ($val) = git_get_project_config($key, '--bool');
 
        if ($val eq 'true') {
-               return 1;
+               return (1);
        } elsif ($val eq 'false') {
-               return 0;
+               return (0);
        }
 
-       return $_[0];
+       return ($_[0]);
 }
 
 sub feature_snapshot {
@@ -316,25 +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 ($_[0]);
@@ -353,7 +446,8 @@ sub check_head_link {
 sub check_export_ok {
        my ($dir) = @_;
        return (check_head_link($dir) &&
-               (!$export_ok || -e "$dir/$export_ok"));
+               (!$export_ok || -e "$dir/$export_ok") &&
+               (!$export_auth_hook || $export_auth_hook->($dir)));
 }
 
 # process alternate names for backward compatibility
@@ -383,191 +477,343 @@ $projects_list ||= $projectroot;
 
 # ======================================================================
 # input validation and dispatch
-our $action = $cgi->param('a');
+
+# input parameters can be collected from a variety of sources (presently, CGI
+# and PATH_INFO), so we define an %input_params hash that collects them all
+# together during validation: this allows subsequent uses (e.g. href()) to be
+# agnostic of the parameter origin
+
+our %input_params = ();
+
+# input parameters are stored with the long parameter name as key. This will
+# also be used in the href subroutine to convert parameters to their CGI
+# equivalent, and since the href() usage is the most frequent one, we store
+# the name -> CGI key mapping here, instead of the reverse.
+#
+# XXX: Warning: If you touch this, check the search form for updating,
+# too.
+
+our @cgi_param_mapping = (
+       project => "p",
+       action => "a",
+       file_name => "f",
+       file_parent => "fp",
+       hash => "h",
+       hash_parent => "hp",
+       hash_base => "hb",
+       hash_parent_base => "hpb",
+       page => "pg",
+       order => "o",
+       searchtext => "s",
+       searchtype => "st",
+       snapshot_format => "sf",
+       extra_options => "opt",
+       search_use_regexp => "sr",
+);
+our %cgi_param_mapping = @cgi_param_mapping;
+
+# we will also need to know the possible actions, for validation
+our %actions = (
+       "blame" => \&git_blame,
+       "blobdiff" => \&git_blobdiff,
+       "blobdiff_plain" => \&git_blobdiff_plain,
+       "blob" => \&git_blob,
+       "blob_plain" => \&git_blob_plain,
+       "commitdiff" => \&git_commitdiff,
+       "commitdiff_plain" => \&git_commitdiff_plain,
+       "commit" => \&git_commit,
+       "forks" => \&git_forks,
+       "heads" => \&git_heads,
+       "history" => \&git_history,
+       "log" => \&git_log,
+       "patch" => \&git_patch,
+       "patches" => \&git_patches,
+       "rss" => \&git_rss,
+       "atom" => \&git_atom,
+       "search" => \&git_search,
+       "search_help" => \&git_search_help,
+       "shortlog" => \&git_shortlog,
+       "summary" => \&git_summary,
+       "tag" => \&git_tag,
+       "tags" => \&git_tags,
+       "tree" => \&git_tree,
+       "snapshot" => \&git_snapshot,
+       "object" => \&git_object,
+       # those below don't need $project
+       "opml" => \&git_opml,
+       "project_list" => \&git_project_list,
+       "project_index" => \&git_project_index,
+);
+
+# finally, we have the hash of allowed extra_options for the commands that
+# allow them
+our %allowed_options = (
+       "--no-merges" => [ qw(rss atom log shortlog history) ],
+);
+
+# fill %input_params with the CGI parameters. All values except for 'opt'
+# should be single values, but opt can be an array. We should probably
+# build an array of parameters that can be multi-valued, but since for the time
+# being it's only this one, we just single it out
+while (my ($name, $symbol) = each %cgi_param_mapping) {
+       if ($symbol eq 'opt') {
+               $input_params{$name} = [ $cgi->param($symbol) ];
+       } else {
+               $input_params{$name} = $cgi->param($symbol);
+       }
+}
+
+# now read PATH_INFO and update the parameter list for missing parameters
+sub evaluate_path_info {
+       return if defined $input_params{'project'};
+       return if !$path_info;
+       $path_info =~ s,^/+,,;
+       return if !$path_info;
+
+       # find which part of PATH_INFO is project
+       my $project = $path_info;
+       $project =~ s,/+$,,;
+       while ($project && !check_head_link("$projectroot/$project")) {
+               $project =~ s,/*[^/]*$,,;
+       }
+       return unless $project;
+       $input_params{'project'} = $project;
+
+       # do not change any parameters if an action is given using the query string
+       return if $input_params{'action'};
+       $path_info =~ s,^\Q$project\E/*,,;
+
+       # next, check if we have an action
+       my $action = $path_info;
+       $action =~ s,/.*$,,;
+       if (exists $actions{$action}) {
+               $path_info =~ s,^$action/*,,;
+               $input_params{'action'} = $action;
+       }
+
+       # list of actions that want hash_base instead of hash, but can have no
+       # pathname (f) parameter
+       my @wants_base = (
+               'tree',
+               'history',
+       );
+
+       # we want to catch
+       # [$hash_parent_base[:$file_parent]..]$hash_parent[:$file_name]
+       my ($parentrefname, $parentpathname, $refname, $pathname) =
+               ($path_info =~ /^(?:(.+?)(?::(.+))?\.\.)?(.+?)(?::(.+))?$/);
+
+       # first, analyze the 'current' part
+       if (defined $pathname) {
+               # we got "branch:filename" or "branch:dir/"
+               # we could use git_get_type(branch:pathname), but:
+               # - it needs $git_dir
+               # - it does a git() call
+               # - the convention of terminating directories with a slash
+               #   makes it superfluous
+               # - embedding the action in the PATH_INFO would make it even
+               #   more superfluous
+               $pathname =~ s,^/+,,;
+               if (!$pathname || substr($pathname, -1) eq "/") {
+                       $input_params{'action'} ||= "tree";
+                       $pathname =~ s,/$,,;
+               } else {
+                       # the default action depends on whether we had parent info
+                       # or not
+                       if ($parentrefname) {
+                               $input_params{'action'} ||= "blobdiff_plain";
+                       } else {
+                               $input_params{'action'} ||= "blob_plain";
+                       }
+               }
+               $input_params{'hash_base'} ||= $refname;
+               $input_params{'file_name'} ||= $pathname;
+       } elsif (defined $refname) {
+               # we got "branch". In this case we have to choose if we have to
+               # set hash or hash_base.
+               #
+               # Most of the actions without a pathname only want hash to be
+               # set, except for the ones specified in @wants_base that want
+               # hash_base instead. It should also be noted that hand-crafted
+               # links having 'history' as an action and no pathname or hash
+               # set will fail, but that happens regardless of PATH_INFO.
+               $input_params{'action'} ||= "shortlog";
+               if (grep { $_ eq $input_params{'action'} } @wants_base) {
+                       $input_params{'hash_base'} ||= $refname;
+               } else {
+                       $input_params{'hash'} ||= $refname;
+               }
+       }
+
+       # next, handle the 'parent' part, if present
+       if (defined $parentrefname) {
+               # a missing pathspec defaults to the 'current' filename, allowing e.g.
+               # someproject/blobdiff/oldrev..newrev:/filename
+               if ($parentpathname) {
+                       $parentpathname =~ s,^/+,,;
+                       $parentpathname =~ s,/$,,;
+                       $input_params{'file_parent'} ||= $parentpathname;
+               } else {
+                       $input_params{'file_parent'} ||= $input_params{'file_name'};
+               }
+               # we assume that hash_parent_base is wanted if a path was specified,
+               # or if the action wants hash_base instead of hash
+               if (defined $input_params{'file_parent'} ||
+                       grep { $_ eq $input_params{'action'} } @wants_base) {
+                       $input_params{'hash_parent_base'} ||= $parentrefname;
+               } else {
+                       $input_params{'hash_parent'} ||= $parentrefname;
+               }
+       }
+
+       # for the snapshot action, we allow URLs in the form
+       # $project/snapshot/$hash.ext
+       # where .ext determines the snapshot and gets removed from the
+       # passed $refname to provide the $hash.
+       #
+       # To be able to tell that $refname includes the format extension, we
+       # require the following two conditions to be satisfied:
+       # - the hash input parameter MUST have been set from the $refname part
+       #   of the URL (i.e. they must be equal)
+       # - the snapshot format MUST NOT have been defined already (e.g. from
+       #   CGI parameter sf)
+       # It's also useless to try any matching unless $refname has a dot,
+       # so we check for that too
+       if (defined $input_params{'action'} &&
+               $input_params{'action'} eq 'snapshot' &&
+               defined $refname && index($refname, '.') != -1 &&
+               $refname eq $input_params{'hash'} &&
+               !defined $input_params{'snapshot_format'}) {
+               # We loop over the known snapshot formats, checking for
+               # extensions. Allowed extensions are both the defined suffix
+               # (which includes the initial dot already) and the snapshot
+               # format key itself, with a prepended dot
+               while (my ($fmt, %opt) = each %known_snapshot_formats) {
+                       my $hash = $refname;
+                       my $sfx;
+                       $hash =~ s/(\Q$opt{'suffix'}\E|\Q.$fmt\E)$//;
+                       next unless $sfx = $1;
+                       # a valid suffix was found, so set the snapshot format
+                       # and reset the hash parameter
+                       $input_params{'snapshot_format'} = $fmt;
+                       $input_params{'hash'} = $hash;
+                       # we also set the format suffix to the one requested
+                       # in the URL: this way a request for e.g. .tgz returns
+                       # a .tgz instead of a .tar.gz
+                       $known_snapshot_formats{$fmt}{'suffix'} = $sfx;
+                       last;
+               }
+       }
+}
+evaluate_path_info();
+
+our $action = $input_params{'action'};
 if (defined $action) {
-       if ($action =~ m/[^0-9a-zA-Z\.\-_]/) {
-               die_error(undef, "Invalid action parameter");
+       if (!validate_action($action)) {
+               die_error(400, "Invalid action parameter");
        }
 }
 
 # parameters which are pathnames
-our $project = $cgi->param('p');
+our $project = $input_params{'project'};
 if (defined $project) {
-       if (!validate_pathname($project) ||
-           !(-d "$projectroot/$project") ||
-           !check_head_link("$projectroot/$project") ||
-           ($export_ok && !(-e "$projectroot/$project/$export_ok")) ||
-           ($strict_export && !project_in_list($project))) {
+       if (!validate_project($project)) {
                undef $project;
-               die_error(undef, "No such project");
+               die_error(404, "No such project");
        }
 }
 
-our $file_name = $cgi->param('f');
+our $file_name = $input_params{'file_name'};
 if (defined $file_name) {
        if (!validate_pathname($file_name)) {
-               die_error(undef, "Invalid file parameter");
+               die_error(400, "Invalid file parameter");
        }
 }
 
-our $file_parent = $cgi->param('fp');
+our $file_parent = $input_params{'file_parent'};
 if (defined $file_parent) {
        if (!validate_pathname($file_parent)) {
-               die_error(undef, "Invalid file parent parameter");
+               die_error(400, "Invalid file parent parameter");
        }
 }
 
 # parameters which are refnames
-our $hash = $cgi->param('h');
+our $hash = $input_params{'hash'};
 if (defined $hash) {
        if (!validate_refname($hash)) {
-               die_error(undef, "Invalid hash parameter");
+               die_error(400, "Invalid hash parameter");
        }
 }
 
-our $hash_parent = $cgi->param('hp');
+our $hash_parent = $input_params{'hash_parent'};
 if (defined $hash_parent) {
        if (!validate_refname($hash_parent)) {
-               die_error(undef, "Invalid hash parent parameter");
+               die_error(400, "Invalid hash parent parameter");
        }
 }
 
-our $hash_base = $cgi->param('hb');
+our $hash_base = $input_params{'hash_base'};
 if (defined $hash_base) {
        if (!validate_refname($hash_base)) {
-               die_error(undef, "Invalid hash base parameter");
+               die_error(400, "Invalid hash base parameter");
        }
 }
 
-my %allowed_options = (
-       "--no-merges" => [ qw(rss atom log shortlog history) ],
-);
-
-our @extra_options = $cgi->param('opt');
-if (defined @extra_options) {
-       foreach my $opt (@extra_options) {
-               if (not exists $allowed_options{$opt}) {
-                       die_error(undef, "Invalid option parameter");
-               }
-               if (not grep(/^$action$/, @{$allowed_options{$opt}})) {
-                       die_error(undef, "Invalid option parameter for this action");
-               }
+our @extra_options = @{$input_params{'extra_options'}};
+# @extra_options is always defined, since it can only be (currently) set from
+# CGI, and $cgi->param() returns the empty array in array context if the param
+# is not set
+foreach my $opt (@extra_options) {
+       if (not exists $allowed_options{$opt}) {
+               die_error(400, "Invalid option parameter");
+       }
+       if (not grep(/^$action$/, @{$allowed_options{$opt}})) {
+               die_error(400, "Invalid option parameter for this action");
        }
 }
 
-our $hash_parent_base = $cgi->param('hpb');
+our $hash_parent_base = $input_params{'hash_parent_base'};
 if (defined $hash_parent_base) {
        if (!validate_refname($hash_parent_base)) {
-               die_error(undef, "Invalid hash parent base parameter");
+               die_error(400, "Invalid hash parent base parameter");
        }
 }
 
 # other parameters
-our $page = $cgi->param('pg');
+our $page = $input_params{'page'};
 if (defined $page) {
        if ($page =~ m/[^0-9]/) {
-               die_error(undef, "Invalid page parameter");
+               die_error(400, "Invalid page parameter");
        }
 }
 
-our $searchtype = $cgi->param('st');
+our $searchtype = $input_params{'searchtype'};
 if (defined $searchtype) {
        if ($searchtype =~ m/[^a-z]/) {
-               die_error(undef, "Invalid searchtype parameter");
+               die_error(400, "Invalid searchtype parameter");
        }
 }
 
-our $search_use_regexp = $cgi->param('sr');
+our $search_use_regexp = $input_params{'search_use_regexp'};
 
-our $searchtext = $cgi->param('s');
+our $searchtext = $input_params{'searchtext'};
 our $search_regexp;
 if (defined $searchtext) {
        if (length($searchtext) < 2) {
-               die_error(undef, "At least two characters are required for search parameter");
+               die_error(403, "At least two characters are required for search parameter");
        }
        $search_regexp = $search_use_regexp ? $searchtext : quotemeta $searchtext;
 }
 
-# now read PATH_INFO and use it as alternative to parameters
-sub evaluate_path_info {
-       return if defined $project;
-       my $path_info = $ENV{"PATH_INFO"};
-       return if !$path_info;
-       $path_info =~ s,^/+,,;
-       return if !$path_info;
-       # find which part of PATH_INFO is project
-       $project = $path_info;
-       $project =~ s,/+$,,;
-       while ($project && !check_head_link("$projectroot/$project")) {
-               $project =~ s,/*[^/]*$,,;
-       }
-       # validate project
-       $project = validate_pathname($project);
-       if (!$project ||
-           ($export_ok && !-e "$projectroot/$project/$export_ok") ||
-           ($strict_export && !project_in_list($project))) {
-               undef $project;
-               return;
-       }
-       # do not change any parameters if an action is given using the query string
-       return if $action;
-       $path_info =~ s,^\Q$project\E/*,,;
-       my ($refname, $pathname) = split(/:/, $path_info, 2);
-       if (defined $pathname) {
-               # we got "project.git/branch:filename" or "project.git/branch:dir/"
-               # we could use git_get_type(branch:pathname), but it needs $git_dir
-               $pathname =~ s,^/+,,;
-               if (!$pathname || substr($pathname, -1) eq "/") {
-                       $action  ||= "tree";
-                       $pathname =~ s,/$,,;
-               } else {
-                       $action  ||= "blob_plain";
-               }
-               $hash_base ||= validate_refname($refname);
-               $file_name ||= validate_pathname($pathname);
-       } elsif (defined $refname) {
-               # we got "project.git/branch"
-               $action ||= "shortlog";
-               $hash   ||= validate_refname($refname);
-       }
-}
-evaluate_path_info();
-
 # path to the current git repository
 our $git_dir;
 $git_dir = "$projectroot/$project" if $project;
 
-# dispatch
-my %actions = (
-       "blame" => \&git_blame2,
-       "blobdiff" => \&git_blobdiff,
-       "blobdiff_plain" => \&git_blobdiff_plain,
-       "blob" => \&git_blob,
-       "blob_plain" => \&git_blob_plain,
-       "commitdiff" => \&git_commitdiff,
-       "commitdiff_plain" => \&git_commitdiff_plain,
-       "commit" => \&git_commit,
-       "forks" => \&git_forks,
-       "heads" => \&git_heads,
-       "history" => \&git_history,
-       "log" => \&git_log,
-       "rss" => \&git_rss,
-       "atom" => \&git_atom,
-       "search" => \&git_search,
-       "search_help" => \&git_search_help,
-       "shortlog" => \&git_shortlog,
-       "summary" => \&git_summary,
-       "tag" => \&git_tag,
-       "tags" => \&git_tags,
-       "tree" => \&git_tree,
-       "snapshot" => \&git_snapshot,
-       "object" => \&git_object,
-       # those below don't need $project
-       "opml" => \&git_opml,
-       "project_list" => \&git_project_list,
-       "project_index" => \&git_project_index,
-);
+# list of supported snapshot formats
+our @snapshot_fmts = gitweb_get_feature('snapshot');
+@snapshot_fmts = filter_snapshot_fmts(@snapshot_fmts);
 
+# dispatch
 if (!defined $action) {
        if (defined $hash) {
                $action = git_get_type($hash);
@@ -580,11 +826,11 @@ if (!defined $action) {
        }
 }
 if (!defined($actions{$action})) {
-       die_error(undef, "Unknown action");
+       die_error(400, "Unknown action");
 }
 if ($action !~ m/^(opml|project_list|project_index)$/ &&
     !$project) {
-       die_error(undef, "Project needed");
+       die_error(400, "Project needed");
 }
 $actions{$action}->();
 exit;
@@ -597,55 +843,96 @@ sub href (%) {
        # default is to use -absolute url() i.e. $my_uri
        my $href = $params{-full} ? $my_url : $my_uri;
 
-       # XXX: Warning: If you touch this, check the search form for updating,
-       # too.
-
-       my @mapping = (
-               project => "p",
-               action => "a",
-               file_name => "f",
-               file_parent => "fp",
-               hash => "h",
-               hash_parent => "hp",
-               hash_base => "hb",
-               hash_parent_base => "hpb",
-               page => "pg",
-               order => "o",
-               searchtext => "s",
-               searchtype => "st",
-               snapshot_format => "sf",
-               extra_options => "opt",
-               search_use_regexp => "sr",
-       );
-       my %mapping = @mapping;
-
        $params{'project'} = $project unless exists $params{'project'};
 
        if ($params{-replay}) {
-               while (my ($name, $symbol) = each %mapping) {
+               while (my ($name, $symbol) = each %cgi_param_mapping) {
                        if (!exists $params{$name}) {
-                               # to allow for multivalued params we use arrayref form
-                               $params{$name} = [ $cgi->param($symbol) ];
+                               $params{$name} = $input_params{$name};
                        }
                }
        }
 
-       my ($use_pathinfo) = gitweb_check_feature('pathinfo');
-       if ($use_pathinfo) {
-               # use PATH_INFO for project name
-               $href .= "/".esc_url($params{'project'}) if defined $params{'project'};
+       my $use_pathinfo = gitweb_check_feature('pathinfo');
+       if ($use_pathinfo and defined $params{'project'}) {
+               # try to put as many parameters as possible in PATH_INFO:
+               #   - project name
+               #   - action
+               #   - hash_parent or hash_parent_base:/file_parent
+               #   - hash or hash_base:/filename
+               #   - the snapshot_format as an appropriate suffix
+
+               # When the script is the root DirectoryIndex for the domain,
+               # $href here would be something like http://gitweb.example.com/
+               # Thus, we strip any trailing / from $href, to spare us double
+               # slashes in the final URL
+               $href =~ s,/$,,;
+
+               # Then add the project name, if present
+               $href .= "/".esc_url($params{'project'});
                delete $params{'project'};
 
-               # Summary just uses the project path URL
-               if (defined $params{'action'} && $params{'action'} eq 'summary') {
+               # since we destructively absorb parameters, we keep this
+               # boolean that remembers if we're handling a snapshot
+               my $is_snapshot = $params{'action'} eq 'snapshot';
+
+               # 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';
                        delete $params{'action'};
                }
+
+               # Next, we put hash_parent_base:/file_parent..hash_base:/file_name,
+               # stripping nonexistent or useless pieces
+               $href .= "/" if ($params{'hash_base'} || $params{'hash_parent_base'}
+                       || $params{'hash_parent'} || $params{'hash'});
+               if (defined $params{'hash_base'}) {
+                       if (defined $params{'hash_parent_base'}) {
+                               $href .= esc_url($params{'hash_parent_base'});
+                               # skip the file_parent if it's the same as the file_name
+                               delete $params{'file_parent'} if $params{'file_parent'} eq $params{'file_name'};
+                               if (defined $params{'file_parent'} && $params{'file_parent'} !~ /\.\./) {
+                                       $href .= ":/".esc_url($params{'file_parent'});
+                                       delete $params{'file_parent'};
+                               }
+                               $href .= "..";
+                               delete $params{'hash_parent'};
+                               delete $params{'hash_parent_base'};
+                       } elsif (defined $params{'hash_parent'}) {
+                               $href .= esc_url($params{'hash_parent'}). "..";
+                               delete $params{'hash_parent'};
+                       }
+
+                       $href .= esc_url($params{'hash_base'});
+                       if (defined $params{'file_name'} && $params{'file_name'} !~ /\.\./) {
+                               $href .= ":/".esc_url($params{'file_name'});
+                               delete $params{'file_name'};
+                       }
+                       delete $params{'hash'};
+                       delete $params{'hash_base'};
+               } elsif (defined $params{'hash'}) {
+                       $href .= esc_url($params{'hash'});
+                       delete $params{'hash'};
+               }
+
+               # If the action was a snapshot, we can absorb the
+               # snapshot_format parameter too
+               if ($is_snapshot) {
+                       my $fmt = $params{'snapshot_format'};
+                       # snapshot_format should always be defined when href()
+                       # is called, but just in case some code forgets, we
+                       # fall back to the default
+                       $fmt ||= $snapshot_fmts[0];
+                       $href .= $known_snapshot_formats{$fmt}{'suffix'};
+                       delete $params{'snapshot_format'};
+               }
        }
 
        # now encode the parameters explicitly
        my @result = ();
-       for (my $i = 0; $i < @mapping; $i += 2) {
-               my ($name, $symbol) = ($mapping[$i], $mapping[$i+1]);
+       for (my $i = 0; $i < @cgi_param_mapping; $i += 2) {
+               my ($name, $symbol) = ($cgi_param_mapping[$i], $cgi_param_mapping[$i+1]);
                if (defined $params{$name}) {
                        if (ref($params{$name}) eq "ARRAY") {
                                foreach my $par (@{$params{$name}}) {
@@ -665,6 +952,24 @@ sub href (%) {
 ## ======================================================================
 ## validation, quoting/unquoting and escaping
 
+sub validate_action {
+       my $input = shift || return undef;
+       return undef unless exists $actions{$input};
+       return $input;
+}
+
+sub validate_project {
+       my $input = shift || return undef;
+       if (!validate_pathname($input) ||
+               !(-d "$projectroot/$input") ||
+               !check_export_ok("$projectroot/$input") ||
+               ($strict_export && !project_in_list($input))) {
+               return undef;
+       } else {
+               return $input;
+       }
+}
+
 sub validate_pathname {
        my $input = shift || return undef;
 
@@ -775,7 +1080,7 @@ sub quot_cec {
        );
        my $chr = ( (exists $es{$cntrl})
                    ? $es{$cntrl}
-                   : sprintf('\%03o', ord($cntrl)) );
+                   : sprintf('\%2x', ord($cntrl)) );
        if ($opts{-nohtml}) {
                return $chr;
        } else {
@@ -1079,7 +1384,7 @@ sub format_log_line_html {
        my $line = shift;
 
        $line = esc_html($line, -nbsp=>1);
-       if ($line =~ m/([0-9a-fA-F]{8,40})/) {
+       if ($line =~ m/\b([0-9a-fA-F]{8,40})\b/) {
                my $hash_text = $1;
                my $link =
                        $cgi->a({-href => href(action=>"object", hash=>$hash_text),
@@ -1090,13 +1395,23 @@ sub format_log_line_html {
 }
 
 # format marker of refs pointing to given object
+
+# the destination action is chosen based on object type and current context:
+# - for annotated tags, we choose the tag view unless it's the current view
+#   already, in which case we go to shortlog view
+# - for other refs, we keep the current view if we're in history, shortlog or
+#   log view, and select shortlog otherwise
 sub format_ref_marker {
        my ($refs, $id) = @_;
        my $markers = '';
 
        if (defined $refs->{$id}) {
                foreach my $ref (@{$refs->{$id}}) {
+                       # this code exploits the fact that non-lightweight tags are the
+                       # only indirect objects, and that they are the only objects for which
+                       # we want to use tag instead of shortlog as action
                        my ($type, $name) = qw();
+                       my $indirect = ($ref =~ s/\^\{\}$//);
                        # e.g. tags/v2.6.11 or heads/next
                        if ($ref =~ m!^(.*?)s?/(.*)$!) {
                                $type = $1;
@@ -1106,8 +1421,29 @@ sub format_ref_marker {
                                $name = $ref;
                        }
 
-                       $markers .= " <span class=\"$type\" title=\"$ref\">" .
-                                   esc_html($name) . "</span>";
+                       my $class = $type;
+                       $class .= " indirect" if $indirect;
+
+                       my $dest_action = "shortlog";
+
+                       if ($indirect) {
+                               $dest_action = "tag" unless $action eq "tag";
+                       } elsif ($action =~ /^(history|(short)?log)$/) {
+                               $dest_action = $action;
+                       }
+
+                       my $dest = "";
+                       $dest .= "refs/" unless $ref =~ m!^refs/!;
+                       $dest .= $ref;
+
+                       my $link = $cgi->a({
+                               -href => href(
+                                       action=>$dest_action,
+                                       hash=>$dest
+                               )}, $name);
+
+                       $markers .= " <span class=\"$class\" title=\"$ref\">" .
+                               $link . "</span>";
                }
        }
 
@@ -1419,8 +1755,6 @@ sub format_diff_line {
 # linked.  Pass the hash of the tree/commit to snapshot.
 sub format_snapshot_links {
        my ($hash) = @_;
-       my @snapshot_fmts = gitweb_check_feature('snapshot');
-       @snapshot_fmts = filter_snapshot_fmts(@snapshot_fmts);
        my $num_fmts = @snapshot_fmts;
        if ($num_fmts > 1) {
                # A parenthesized list of links bearing format names.
@@ -1665,7 +1999,7 @@ sub git_get_hash_by_path {
        $path =~ s,/+$,,;
 
        open my $fd, "-|", git_cmd(), "ls-tree", $base, "--", $path
-               or die_error(undef, "Open git-ls-tree failed");
+               or die_error(500, "Open git-ls-tree failed");
        my $line = <$fd>;
        close $fd or return undef;
 
@@ -1724,6 +2058,71 @@ sub git_get_project_description {
        return $descr;
 }
 
+sub git_get_project_ctags {
+       my $path = shift;
+       my $ctags = {};
+
+       $git_dir = "$projectroot/$path";
+       unless (opendir D, "$git_dir/ctags") {
+               return $ctags;
+       }
+       foreach (grep { -f $_ } map { "$git_dir/ctags/$_" } readdir(D)) {
+               open CT, $_ or next;
+               my $val = <CT>;
+               chomp $val;
+               close CT;
+               my $ctag = $_; $ctag =~ s#.*/##;
+               $ctags->{$ctag} = $val;
+       }
+       closedir D;
+       $ctags;
+}
+
+sub git_populate_project_tagcloud {
+       my $ctags = shift;
+
+       # First, merge different-cased tags; tags vote on casing
+       my %ctags_lc;
+       foreach (keys %$ctags) {
+               $ctags_lc{lc $_}->{count} += $ctags->{$_};
+               if (not $ctags_lc{lc $_}->{topcount}
+                   or $ctags_lc{lc $_}->{topcount} < $ctags->{$_}) {
+                       $ctags_lc{lc $_}->{topcount} = $ctags->{$_};
+                       $ctags_lc{lc $_}->{topname} = $_;
+               }
+       }
+
+       my $cloud;
+       if (eval { require HTML::TagCloud; 1; }) {
+               $cloud = HTML::TagCloud->new;
+               foreach (sort keys %ctags_lc) {
+                       # Pad the title with spaces so that the cloud looks
+                       # less crammed.
+                       my $title = $ctags_lc{$_}->{topname};
+                       $title =~ s/ /&nbsp;/g;
+                       $title =~ s/^/&nbsp;/g;
+                       $title =~ s/$/&nbsp;/g;
+                       $cloud->add($title, $home_link."?by_tag=".$_, $ctags_lc{$_}->{count});
+               }
+       } else {
+               $cloud = \%ctags_lc;
+       }
+       $cloud;
+}
+
+sub git_show_project_tagcloud {
+       my ($cloud, $count) = @_;
+       print STDERR ref($cloud)."..\n";
+       if (ref $cloud eq 'HTML::TagCloud') {
+               return $cloud->html_and_css($count);
+       } 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>"
+               } splice(@tags, 0, $count)) . '</p>';
+       }
+}
+
 sub git_get_project_url_list {
        my $path = shift;
 
@@ -1745,7 +2144,7 @@ sub git_get_projects_list {
        $filter ||= '';
        $filter =~ s/\.git$//;
 
-       my ($check_forks) = gitweb_check_feature('forks');
+       my $check_forks = gitweb_check_feature('forks');
 
        if (-d $projects_list) {
                # search in directory
@@ -1772,10 +2171,9 @@ sub git_get_projects_list {
 
                                my $subdir = substr($File::Find::name, $pfxlen + 1);
                                # we check related file in $projectroot
-                               if ($check_forks and $subdir =~ m#/.#) {
-                                       $File::Find::prune = 1;
-                               } elsif (check_export_ok("$projectroot/$filter/$subdir")) {
-                                       push @list, { path => ($filter ? "$filter/" : '') . $subdir };
+                               my $path = ($filter ? "$filter/" : '') . $subdir;
+                               if (check_export_ok("$projectroot/$path")) {
+                                       push @list, { path => $path };
                                        $File::Find::prune = 1;
                                }
                        },
@@ -1918,7 +2316,7 @@ sub git_get_references {
 
        while (my $line = <$fd>) {
                chomp $line;
-               if ($line =~ m!^([0-9a-fA-F]{40})\srefs/($type/?[^^]+)!) {
+               if ($line =~ m!^([0-9a-fA-F]{40})\srefs/($type.*)$!) {
                        if (defined $refs{$1}) {
                                push @{$refs{$1}}, $2;
                        } else {
@@ -2092,7 +2490,7 @@ sub parse_commit_text {
                        last;
                }
        }
-       if ($co{'title'} eq "") {
+       if (! defined $co{'title'} || $co{'title'} eq "") {
                $co{'title'} = $co{'title_short'} = '(no commit message)';
        }
        # remove added spaces
@@ -2127,7 +2525,7 @@ sub parse_commit {
                "--max-count=1",
                $commit_id,
                "--",
-               or die_error(undef, "Open git-rev-list failed");
+               or die_error(500, "Open git-rev-list failed");
        %co = parse_commit_text(<$fd>, 1);
        close $fd;
 
@@ -2152,7 +2550,7 @@ sub parse_commits {
                $commit_id,
                "--",
                ($filename ? ($filename) : ())
-               or die_error(undef, "Open git-rev-list failed");
+               or die_error(500, "Open git-rev-list failed");
        while (my $line = <$fd>) {
                my %co = parse_commit_text($line);
                push @cos, \%co;
@@ -2386,6 +2784,15 @@ sub get_file_owner {
        return to_utf8($owner);
 }
 
+# assume that file exists
+sub insert_file {
+       my $filename = shift;
+
+       open my $fd, '<', $filename;
+       print map { to_utf8($_) } <$fd>;
+       close $fd;
+}
+
 ## ......................................................................
 ## mimetype related functions
 
@@ -2514,9 +2921,14 @@ 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) {
@@ -2574,9 +2986,7 @@ EOF
              "<body>\n";
 
        if (-f $site_header) {
-               open (my $fd, $site_header);
-               print <$fd>;
-               close $fd;
+               insert_file($site_header);
        }
 
        print "<div class=\"page_header\">\n" .
@@ -2593,7 +3003,7 @@ EOF
        }
        print "</div>\n";
 
-       my ($have_search) = gitweb_check_feature('search');
+       my $have_search = gitweb_check_feature('search');
        if (defined $project && $have_search) {
                if (!defined $searchtext) {
                        $searchtext = "";
@@ -2607,7 +3017,7 @@ EOF
                        $search_hash = "HEAD";
                }
                my $action = $my_uri;
-               my ($use_pathinfo) = gitweb_check_feature('pathinfo');
+               my $use_pathinfo = gitweb_check_feature('pathinfo');
                if ($use_pathinfo) {
                        $action .= "/".esc_url($project);
                }
@@ -2663,20 +3073,33 @@ sub git_footer_html {
        print "</div>\n"; # class="page_footer"
 
        if (-f $site_footer) {
-               open (my $fd, $site_footer);
-               print <$fd>;
-               close $fd;
+               insert_file($site_footer);
        }
 
        print "</body>\n" .
              "</html>";
 }
 
+# die_error(<http_status_code>, <error_message>)
+# Example: die_error(404, 'Hash not found')
+# By convention, use the following status codes (as defined in RFC 2616):
+# 400: Invalid or missing CGI parameters, or
+#      requested object exists but has wrong type.
+# 403: Requested feature (like "pickaxe" or "snapshot") not enabled on
+#      this server or project.
+# 404: Requested object/revision/project doesn't exist.
+# 500: The server isn't configured properly, or
+#      an internal error occurred (e.g. failed assertions caused by bugs), or
+#      an unknown error occurred (e.g. the git binary died unexpectedly).
 sub die_error {
-       my $status = shift || "403 Forbidden";
-       my $error = shift || "Malformed query, file missing or permission denied";
-
-       git_header_html($status);
+       my $status = shift || 500;
+       my $error = shift || "Internal server error";
+
+       my %http_responses = (400 => '400 Bad Request',
+                             403 => '403 Forbidden',
+                             404 => '404 Not Found',
+                             500 => '500 Internal Server Error');
+       git_header_html($http_responses{$status});
        print <<EOF;
 <div class="page_body">
 <br /><br />
@@ -2711,13 +3134,31 @@ sub git_print_page_nav {
                        }
                }
        }
+
        $arg{'tree'}{'hash'} = $treehead if defined $treehead;
        $arg{'tree'}{'hash_base'} = $treebase if defined $treebase;
 
+       my @actions = gitweb_get_feature('actions');
+       my %repl = (
+               '%' => '%',
+               'n' => $project,         # project name
+               'f' => $git_dir,         # project path within filesystem
+               'h' => $treehead || '',  # current hash ('h' parameter)
+               'b' => $treebase || '',  # hash base ('hb' parameter)
+       );
+       while (@actions) {
+               my ($label, $link, $pos) = splice(@actions,0,3);
+               # insert
+               @navs = map { $_ eq $pos ? ($_, $label) : $_ } @navs;
+               # munch munch
+               $link =~ s/%([%nfhb])/$repl{$1}/g;
+               $arg{$label}{'_href'} = $link;
+       }
+
        print "<div class=\"page_nav\">\n" .
                (join " | ",
                 map { $_ eq $current ?
-                      $_ : $cgi->a({-href => href(%{$arg{$_}})}, "$_")
+                      $_ : $cgi->a({-href => ($arg{$_}{_href} ? $arg{$_}{_href} : href(%{$arg{$_}}))}, "$_")
                 } @navs);
        print "<br/>\n$extra<br/>\n" .
              "</div>\n";
@@ -3067,7 +3508,7 @@ sub is_patch_split {
 sub git_difftree_body {
        my ($difftree, $hash, @parents) = @_;
        my ($parent) = $parents[0];
-       my ($have_blame) = gitweb_check_feature('blame');
+       my $have_blame = gitweb_check_feature('blame');
        print "<div class=\"list_head\">\n";
        if ($#{$difftree} > 10) {
                print(($#{$difftree} + 1) . " files changed:\n");
@@ -3520,21 +3961,25 @@ sub git_patchset_body {
 
 # . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
 
-sub git_project_list_body {
-       my ($projlist, $order, $from, $to, $extra, $no_header) = @_;
-
-       my ($check_forks) = gitweb_check_feature('forks');
-
+# fills project list info (age, description, owner, forks) for each
+# project in the list, removing invalid projects from returned list
+# NOTE: modifies $projlist, but does not remove entries from it
+sub fill_project_list_info {
+       my ($projlist, $check_forks) = @_;
        my @projects;
+
+       my $show_ctags = gitweb_check_feature('ctags');
+ PROJECT:
        foreach my $pr (@$projlist) {
-               my (@aa) = git_get_last_activity($pr->{'path'});
-               unless (@aa) {
-                       next;
+               my (@activity) = git_get_last_activity($pr->{'path'});
+               unless (@activity) {
+                       next PROJECT;
                }
-               ($pr->{'age'}, $pr->{'age_string'}) = @aa;
+               ($pr->{'age'}, $pr->{'age_string'}) = @activity;
                if (!defined $pr->{'descr'}) {
                        my $descr = git_get_project_description($pr->{'path'}) || "";
-                       $pr->{'descr_long'} = to_utf8($descr);
+                       $descr = to_utf8($descr);
+                       $pr->{'descr_long'} = $descr;
                        $pr->{'descr'} = chop_str($descr, $projects_list_description_width, 5);
                }
                if (!defined $pr->{'owner'}) {
@@ -3546,66 +3991,98 @@ sub git_project_list_body {
                            ($pname !~ /\/$/) &&
                            (-d "$projectroot/$pname")) {
                                $pr->{'forks'} = "-d $projectroot/$pname";
-                       }
-                       else {
+                       }       else {
                                $pr->{'forks'} = 0;
                        }
                }
+               $show_ctags and $pr->{'ctags'} = git_get_project_ctags($pr->{'path'});
                push @projects, $pr;
        }
 
+       return @projects;
+}
+
+# print 'sort by' <th> element, generating 'sort by $name' replay link
+# if that order is not selected
+sub print_sort_th {
+       my ($name, $order, $header) = @_;
+       $header ||= ucfirst($name);
+
+       if ($order eq $name) {
+               print "<th>$header</th>\n";
+       } else {
+               print "<th>" .
+                     $cgi->a({-href => href(-replay=>1, order=>$name),
+                              -class => "header"}, $header) .
+                     "</th>\n";
+       }
+}
+
+sub git_project_list_body {
+       # actually uses global variable $project
+       my ($projlist, $order, $from, $to, $extra, $no_header) = @_;
+
+       my $check_forks = gitweb_check_feature('forks');
+       my @projects = fill_project_list_info($projlist, $check_forks);
+
        $order ||= $default_projects_order;
        $from = 0 unless defined $from;
        $to = $#projects if (!defined $to || $#projects < $to);
 
+       my %order_info = (
+               project => { key => 'path', type => 'str' },
+               descr => { key => 'descr_long', type => 'str' },
+               owner => { key => 'owner', type => 'str' },
+               age => { key => 'age', type => 'num' }
+       );
+       my $oi = $order_info{$order};
+       if ($oi->{'type'} eq 'str') {
+               @projects = sort {$a->{$oi->{'key'}} cmp $b->{$oi->{'key'}}} @projects;
+       } else {
+               @projects = sort {$a->{$oi->{'key'}} <=> $b->{$oi->{'key'}}} @projects;
+       }
+
+       my $show_ctags = gitweb_check_feature('ctags');
+       if ($show_ctags) {
+               my %ctags;
+               foreach my $p (@projects) {
+                       foreach my $ct (keys %{$p->{'ctags'}}) {
+                               $ctags{$ct} += $p->{'ctags'}->{$ct};
+                       }
+               }
+               my $cloud = git_populate_project_tagcloud(\%ctags);
+               print git_show_project_tagcloud($cloud, 64);
+       }
+
        print "<table class=\"project_list\">\n";
        unless ($no_header) {
                print "<tr>\n";
                if ($check_forks) {
                        print "<th></th>\n";
                }
-               if ($order eq "project") {
-                       @projects = sort {$a->{'path'} cmp $b->{'path'}} @projects;
-                       print "<th>Project</th>\n";
-               } else {
-                       print "<th>" .
-                             $cgi->a({-href => href(project=>undef, order=>'project'),
-                                      -class => "header"}, "Project") .
-                             "</th>\n";
-               }
-               if ($order eq "descr") {
-                       @projects = sort {$a->{'descr'} cmp $b->{'descr'}} @projects;
-                       print "<th>Description</th>\n";
-               } else {
-                       print "<th>" .
-                             $cgi->a({-href => href(project=>undef, order=>'descr'),
-                                      -class => "header"}, "Description") .
-                             "</th>\n";
-               }
-               if ($order eq "owner") {
-                       @projects = sort {$a->{'owner'} cmp $b->{'owner'}} @projects;
-                       print "<th>Owner</th>\n";
-               } else {
-                       print "<th>" .
-                             $cgi->a({-href => href(project=>undef, order=>'owner'),
-                                      -class => "header"}, "Owner") .
-                             "</th>\n";
-               }
-               if ($order eq "age") {
-                       @projects = sort {$a->{'age'} <=> $b->{'age'}} @projects;
-                       print "<th>Last Change</th>\n";
-               } else {
-                       print "<th>" .
-                             $cgi->a({-href => href(project=>undef, order=>'age'),
-                                      -class => "header"}, "Last Change") .
-                             "</th>\n";
-               }
-               print "<th></th>\n" .
+               print_sort_th('project', $order, 'Project');
+               print_sort_th('descr', $order, 'Description');
+               print_sort_th('owner', $order, 'Owner');
+               print_sort_th('age', $order, 'Last Change');
+               print "<th></th>\n" . # for links
                      "</tr>\n";
        }
        my $alternate = 1;
+       my $tagfilter = $cgi->param('by_tag');
        for (my $i = $from; $i <= $to; $i++) {
                my $pr = $projects[$i];
+
+               next if $tagfilter and $show_ctags and not grep { lc $_ eq lc $tagfilter } keys %{$pr->{'ctags'}};
+               next if $searchtext and not $pr->{'path'} =~ /$searchtext/
+                       and not $pr->{'descr_long'} =~ /$searchtext/;
+               # Weed out forks or non-matching entries of search
+               if ($check_forks) {
+                       my $forkbase = $project; $forkbase ||= ''; $forkbase =~ s#\.git$#/#;
+                       $forkbase="^$forkbase" if $forkbase;
+                       next if not $searchtext and not $tagfilter and $show_ctags
+                               and $pr->{'path'} =~ m#$forkbase.*/.*#; # regexp-safe
+               }
+
                if ($alternate) {
                        print "<tr class=\"dark\">\n";
                } else {
@@ -3922,37 +4399,40 @@ sub git_search_grep_body {
 ## actions
 
 sub git_project_list {
-       my $order = $cgi->param('o');
+       my $order = $input_params{'order'};
        if (defined $order && $order !~ m/none|project|descr|owner|age/) {
-               die_error(undef, "Unknown order parameter");
+               die_error(400, "Unknown order parameter");
        }
 
        my @list = git_get_projects_list();
        if (!@list) {
-               die_error(undef, "No projects found");
+               die_error(404, "No projects found");
        }
 
        git_header_html();
        if (-f $home_text) {
                print "<div class=\"index_include\">\n";
-               open (my $fd, $home_text);
-               print <$fd>;
-               close $fd;
+               insert_file($home_text);
                print "</div>\n";
        }
+       print $cgi->startform(-method => "get") .
+             "<p class=\"projsearch\">Search:\n" .
+             $cgi->textfield(-name => "s", -value => $searchtext) . "\n" .
+             "</p>" .
+             $cgi->end_form() . "\n";
        git_project_list_body(\@list, $order);
        git_footer_html();
 }
 
 sub git_forks {
-       my $order = $cgi->param('o');
+       my $order = $input_params{'order'};
        if (defined $order && $order !~ m/none|project|descr|owner|age/) {
-               die_error(undef, "Unknown order parameter");
+               die_error(400, "Unknown order parameter");
        }
 
        my @list = git_get_projects_list($project);
        if (!@list) {
-               die_error(undef, "No forks found");
+               die_error(404, "No forks found");
        }
 
        git_header_html();
@@ -4000,7 +4480,7 @@ sub git_summary {
        my @taglist  = git_get_tags_list(16);
        my @headlist = git_get_heads_list(16);
        my @forklist;
-       my ($check_forks) = gitweb_check_feature('forks');
+       my $check_forks = gitweb_check_feature('forks');
 
        if ($check_forks) {
                @forklist = git_get_projects_list($project);
@@ -4011,10 +4491,10 @@ sub git_summary {
 
        print "<div class=\"title\">&nbsp;</div>\n";
        print "<table class=\"projects_list\">\n" .
-             "<tr><td>description</td><td>" . esc_html($descr) . "</td></tr>\n" .
-             "<tr><td>owner</td><td>" . esc_html($owner) . "</td></tr>\n";
+             "<tr id=\"metadata_desc\"><td>description</td><td>" . esc_html($descr) . "</td></tr>\n" .
+             "<tr id=\"metadata_owner\"><td>owner</td><td>" . esc_html($owner) . "</td></tr>\n";
        if (defined $cd{'rfc2822'}) {
-               print "<tr><td>last change</td><td>$cd{'rfc2822'}</td></tr>\n";
+               print "<tr id=\"metadata_lchange\"><td>last change</td><td>$cd{'rfc2822'}</td></tr>\n";
        }
 
        # use per project git URL list in $projectroot/$project/cloneurl
@@ -4024,19 +4504,32 @@ sub git_summary {
        @url_list = map { "$_/$project" } @git_base_url_list unless @url_list;
        foreach my $git_url (@url_list) {
                next unless $git_url;
-               print "<tr><td>$url_tag</td><td>$git_url</td></tr>\n";
+               print "<tr class=\"metadata_url\"><td>$url_tag</td><td>$git_url</td></tr>\n";
                $url_tag = "";
        }
+
+       # Tag cloud
+       my $show_ctags = gitweb_check_feature('ctags');
+       if ($show_ctags) {
+               my $ctags = git_get_project_ctags($project);
+               my $cloud = git_populate_project_tagcloud($ctags);
+               print "<tr id=\"metadata_ctags\"><td>Content tags:<br />";
+               print "</td>\n<td>" unless %$ctags;
+               print "<form action=\"$show_ctags\" method=\"post\"><input type=\"hidden\" name=\"p\" value=\"$project\" />Add: <input type=\"text\" name=\"t\" size=\"8\" /></form>";
+               print "</td>\n<td>" if %$ctags;
+               print git_show_project_tagcloud($cloud, 48);
+               print "</td></tr>";
+       }
+
        print "</table>\n";
 
-       if (-s "$projectroot/$project/README.html") {
-               if (open my $fd, "$projectroot/$project/README.html") {
-                       print "<div class=\"title\">readme</div>\n" .
-                             "<div class=\"readme\">\n";
-                       print $_ while (<$fd>);
-                       print "\n</div>\n"; # class="readme"
-                       close $fd;
-               }
+       # If XSS prevention is on, we don't include README.html.
+       # TODO: Allow a readme in some safe format.
+       if (!$prevent_xss && -s "$projectroot/$project/README.html") {
+               print "<div class=\"title\">readme</div>\n" .
+                     "<div class=\"readme\">\n";
+               insert_file("$projectroot/$project/README.html");
+               print "\n</div>\n"; # class="readme"
        }
 
        # we need to request one more than 16 (0..15) to check if
@@ -4065,10 +4558,10 @@ sub git_summary {
 
        if (@forklist) {
                git_print_header_div('forks');
-               git_project_list_body(\@forklist, undef, 0, 15,
+               git_project_list_body(\@forklist, 'age', 0, 15,
                                      $#forklist <= 15 ? undef :
                                      $cgi->a({-href => href(action=>"forks")}, "..."),
-                                     'noheader');
+                                     'no_header');
        }
 
        git_footer_html();
@@ -4081,7 +4574,7 @@ sub git_tag {
        my %tag = parse_tag($hash);
 
        if (! %tag) {
-               die_error(undef, "Unknown tag object");
+               die_error(404, "Unknown tag object");
        }
 
        git_print_header_div('commit', esc_html($tag{'name'}), $hash);
@@ -4113,30 +4606,34 @@ sub git_tag {
        git_footer_html();
 }
 
-sub git_blame2 {
-       my $fd;
-       my $ftype;
+sub git_blame {
+       # permissions
+       gitweb_check_feature('blame')
+               or die_error(403, "Blame view not allowed");
 
-       my ($have_blame) = gitweb_check_feature('blame');
-       if (!$have_blame) {
-               die_error('403 Permission denied', "Permission denied");
-       }
-       die_error('404 Not Found', "File name not defined") if (!$file_name);
+       # error checking
+       die_error(400, "No file name given") unless $file_name;
        $hash_base ||= git_get_head_hash($project);
-       die_error(undef, "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(undef, "Reading commit failed");
+               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(undef, "Error looking up file");
-       }
-       $ftype = git_get_type($hash);
-       if ($ftype !~ "blob") {
-               die_error('400 Bad Request', "Object is not a blob");
+                       or die_error(404, "Error looking up file");
+       } else {
+               $ftype = git_get_type($hash);
+               if ($ftype !~ "blob") {
+                       die_error(400, "Object is not a blob");
+               }
        }
-       open ($fd, "-|", git_cmd(), "blame", '-p', '--',
-             $file_name, $hash_base)
-               or die_error(undef, "Open git-blame failed");
+
+       # 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)},
@@ -4150,42 +4647,46 @@ sub git_blame2 {
        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\"";
@@ -4194,20 +4695,25 @@ HTML
                        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(undef, "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>";
@@ -4218,103 +4724,8 @@ HTML
        print "</div>";
        close $fd
                or print "Reading blob failed\n";
-       git_footer_html();
-}
 
-sub git_blame {
-       my $fd;
-
-       my ($have_blame) = gitweb_check_feature('blame');
-       if (!$have_blame) {
-               die_error('403 Permission denied', "Permission denied");
-       }
-       die_error('404 Not Found', "File name not defined") if (!$file_name);
-       $hash_base ||= git_get_head_hash($project);
-       die_error(undef, "Couldn't find base commit") unless ($hash_base);
-       my %co = parse_commit($hash_base)
-               or die_error(undef, "Reading commit failed");
-       if (!defined $hash) {
-               $hash = git_get_hash_by_path($hash_base, $file_name, "blob")
-                       or die_error(undef, "Error lookup file");
-       }
-       open ($fd, "-|", git_cmd(), "annotate", '-l', '-t', '-r', $file_name, $hash_base)
-               or die_error(undef, "Open git-annotate failed");
-       git_header_html();
-       my $formats_nav =
-               $cgi->a({-href => href(action=>"blob", hash=>$hash, hash_base=>$hash_base, file_name=>$file_name)},
-                       "blob") .
-               " | " .
-               $cgi->a({-href => href(action=>"history", hash=>$hash, hash_base=>$hash_base, file_name=>$file_name)},
-                       "history") .
-               " | " .
-               $cgi->a({-href => href(action=>"blame", file_name=>$file_name)},
-                       "HEAD");
-       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, 'blob', $hash_base);
-       print "<div class=\"page_body\">\n";
-       print <<HTML;
-<table class="blame">
-  <tr>
-    <th>Commit</th>
-    <th>Age</th>
-    <th>Author</th>
-    <th>Line</th>
-    <th>Data</th>
-  </tr>
-HTML
-       my @line_class = (qw(light dark));
-       my $line_class_len = scalar (@line_class);
-       my $line_class_num = $#line_class;
-       while (my $line = <$fd>) {
-               my $long_rev;
-               my $short_rev;
-               my $author;
-               my $time;
-               my $lineno;
-               my $data;
-               my $age;
-               my $age_str;
-               my $age_class;
-
-               chomp $line;
-               $line_class_num = ($line_class_num + 1) % $line_class_len;
-
-               if ($line =~ m/^([0-9a-fA-F]{40})\t\(\s*([^\t]+)\t(\d+) [+-]\d\d\d\d\t(\d+)\)(.*)$/) {
-                       $long_rev = $1;
-                       $author   = $2;
-                       $time     = $3;
-                       $lineno   = $4;
-                       $data     = $5;
-               } else {
-                       print qq(  <tr><td colspan="5" class="error">Unable to parse: $line</td></tr>\n);
-                       next;
-               }
-               $short_rev  = substr ($long_rev, 0, 8);
-               $age        = time () - $time;
-               $age_str    = age_string ($age);
-               $age_str    =~ s/ /&nbsp;/g;
-               $age_class  = age_class($age);
-               $author     = esc_html ($author);
-               $author     =~ s/ /&nbsp;/g;
-
-               $data = untabify($data);
-               $data = esc_html ($data);
-
-               print <<HTML;
-  <tr class="$line_class[$line_class_num]">
-    <td class="sha1"><a href="${\href (action=>"commit", hash=>$long_rev)}" class="text">$short_rev..</a></td>
-    <td class="$age_class">$age_str</td>
-    <td>$author</td>
-    <td class="linenr"><a id="$lineno" href="#$lineno" class="linenr">$lineno</a></td>
-    <td class="pre">$data</td>
-  </tr>
-HTML
-       } # while (my $line = <$fd>)
-       print "</table>\n\n";
-       close $fd
-               or print "Reading blob failed.\n";
-       print "</div>";
+       # page footer
        git_footer_html();
 }
 
@@ -4352,9 +4763,9 @@ sub git_blob_plain {
                if (defined $file_name) {
                        my $base = $hash_base || git_get_head_hash($project);
                        $hash = git_get_hash_by_path($base, $file_name, "blob")
-                               or die_error(undef, "Error lookup file");
+                               or die_error(404, "Cannot find file");
                } else {
-                       die_error(undef, "No file name defined");
+                       die_error(400, "No file name defined");
                }
        } elsif ($hash =~ m/^[0-9a-fA-F]{40}$/) {
                # blobs defined by non-textual hash id's can be cached
@@ -4362,7 +4773,7 @@ sub git_blob_plain {
        }
 
        open my $fd, "-|", git_cmd(), "cat-file", "blob", $hash
-               or die_error(undef, "Open git-cat-file blob '$hash' failed");
+               or die_error(500, "Open git-cat-file blob '$hash' failed");
 
        # content-type (can include charset)
        $type = blob_contenttype($fd, $file_name, $type);
@@ -4375,10 +4786,21 @@ 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>;
@@ -4394,18 +4816,18 @@ sub git_blob {
                if (defined $file_name) {
                        my $base = $hash_base || git_get_head_hash($project);
                        $hash = git_get_hash_by_path($base, $file_name, "blob")
-                               or die_error(undef, "Error lookup file");
+                               or die_error(404, "Cannot find file");
                } else {
-                       die_error(undef, "No file name defined");
+                       die_error(400, "No file name defined");
                }
        } elsif ($hash =~ m/^[0-9a-fA-F]{40}$/) {
                # blobs defined by non-textual hash id's can be cached
                $expires = "+1d";
        }
 
-       my ($have_blame) = gitweb_check_feature('blame');
+       my $have_blame = gitweb_check_feature('blame');
        open my $fd, "-|", git_cmd(), "cat-file", "blob", $hash
-               or die_error(undef, "Couldn't cat $file_name, $hash");
+               or die_error(500, "Couldn't cat $file_name, $hash");
        my $mimetype = blob_mimetype($fd, $file_name);
        if ($mimetype !~ m!^(?:text/|image/(?:gif|png|jpeg)$)! && -B $fd) {
                close $fd;
@@ -4484,18 +4906,19 @@ sub git_tree {
                        $hash = $hash_base;
                }
        }
+       die_error(404, "No such tree") unless defined($hash);
        $/ = "\0";
        open my $fd, "-|", git_cmd(), "ls-tree", '-z', $hash
-               or die_error(undef, "Open git-ls-tree failed");
+               or die_error(500, "Open git-ls-tree failed");
        my @entries = map { chomp; $_ } <$fd>;
-       close $fd or die_error(undef, "Reading tree failed");
+       close $fd or die_error(404, "Reading tree failed");
        $/ = "\n";
 
        my $refs = git_get_references();
        my $ref = format_ref_marker($refs, $hash_base);
        git_header_html();
        my $basedir = '';
-       my ($have_blame) = gitweb_check_feature('blame');
+       my $have_blame = gitweb_check_feature('blame');
        if (defined $hash_base && (my %co = parse_commit($hash_base))) {
                my @views_nav = ();
                if (defined $file_name) {
@@ -4524,8 +4947,8 @@ sub git_tree {
                if ($basedir ne '' && substr($basedir, -1) ne '/') {
                        $basedir .= '/';
                }
+               git_print_page_path($file_name, 'tree', $hash_base);
        }
-       git_print_page_path($file_name, 'tree', $hash_base);
        print "<div class=\"page_body\">\n";
        print "<table class=\"tree\">\n";
        my $alternate = 1;
@@ -4573,21 +4996,18 @@ sub git_tree {
 }
 
 sub git_snapshot {
-       my @supported_fmts = gitweb_check_feature('snapshot');
-       @supported_fmts = filter_snapshot_fmts(@supported_fmts);
-
-       my $format = $cgi->param('sf');
-       if (!@supported_fmts) {
-               die_error('403 Permission denied', "Permission denied");
+       my $format = $input_params{'snapshot_format'};
+       if (!@snapshot_fmts) {
+               die_error(403, "Snapshots not allowed");
        }
        # default to first supported snapshot format
-       $format ||= $supported_fmts[0];
+       $format ||= $snapshot_fmts[0];
        if ($format !~ m/^[a-z0-9]+$/) {
-               die_error(undef, "Invalid snapshot format parameter");
+               die_error(400, "Invalid snapshot format parameter");
        } elsif (!exists($known_snapshot_formats{$format})) {
-               die_error(undef, "Unknown snapshot format");
-       } elsif (!grep($_ eq $format, @supported_fmts)) {
-               die_error(undef, "Unsupported snapshot format");
+               die_error(400, "Unknown snapshot format");
+       } elsif (!grep($_ eq $format, @snapshot_fmts)) {
+               die_error(403, "Unsupported snapshot format");
        }
 
        if (!defined $hash) {
@@ -4615,7 +5035,7 @@ sub git_snapshot {
                -status => '200 OK');
 
        open my $fd, "-|", $cmd
-               or die_error(undef, "Execute git-archive failed");
+               or die_error(500, "Execute git-archive failed");
        binmode STDOUT, ':raw';
        print <$fd>;
        binmode STDOUT, ':utf8'; # as set at the beginning of gitweb.cgi
@@ -4636,6 +5056,15 @@ 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);
 
@@ -4683,10 +5112,8 @@ sub git_log {
 
 sub git_commit {
        $hash ||= $hash_base || "HEAD";
-       my %co = parse_commit($hash);
-       if (!%co) {
-               die_error(undef, "Unknown commit object");
-       }
+       my %co = parse_commit($hash)
+           or die_error(404, "Unknown commit object");
        my %ad = parse_date($co{'author_epoch'}, $co{'author_tz'});
        my %cd = parse_date($co{'committer_epoch'}, $co{'committer_tz'});
 
@@ -4717,6 +5144,11 @@ sub git_commit {
                        } @$parents ) .
                        ')';
        }
+       if (gitweb_check_feature('patches')) {
+               $formats_nav .= " | " .
+                       $cgi->a({-href => href(action=>"patch", -replay=>1)},
+                               "patch");
+       }
 
        if (!defined $parent) {
                $parent = "--root";
@@ -4726,9 +5158,9 @@ sub git_commit {
                @diff_opts,
                (@$parents <= 1 ? $parent : '-c'),
                $hash, "--"
-               or die_error(undef, "Open git-diff-tree failed");
+               or die_error(500, "Open git-diff-tree failed");
        @difftree = map { chomp; $_ } <$fd>;
-       close $fd or die_error(undef, "Reading git-diff-tree failed");
+       close $fd or die_error(404, "Reading git-diff-tree failed");
 
        # non-textual hash id's can be cached
        my $expires;
@@ -4821,33 +5253,33 @@ sub git_object {
 
                open my $fd, "-|", quote_command(
                        git_cmd(), 'cat-file', '-t', $object_id) . ' 2> /dev/null'
-                       or die_error('404 Not Found', "Object does not exist");
+                       or die_error(404, "Object does not exist");
                $type = <$fd>;
                chomp $type;
                close $fd
-                       or die_error('404 Not Found', "Object does not exist");
+                       or die_error(404, "Object does not exist");
 
        # - hash_base and file_name
        } elsif ($hash_base && defined $file_name) {
                $file_name =~ s,/+$,,;
 
                system(git_cmd(), "cat-file", '-e', $hash_base) == 0
-                       or die_error('404 Not Found', "Base object does not exist");
+                       or die_error(404, "Base object does not exist");
 
                # here errors should not hapen
                open my $fd, "-|", git_cmd(), "ls-tree", $hash_base, "--", $file_name
-                       or die_error(undef, "Open git-ls-tree failed");
+                       or die_error(500, "Open git-ls-tree failed");
                my $line = <$fd>;
                close $fd;
 
                #'100644 blob 0fa3f3a66fb6a137f6ec2c19351ed4d807070ffa  panic.c'
                unless ($line && $line =~ m/^([0-9]+) (.+) ([0-9a-fA-F]{40})\t/) {
-                       die_error('404 Not Found', "File or directory for given base does not exist");
+                       die_error(404, "File or directory for given base does not exist");
                }
                $type = $2;
                $hash = $3;
        } else {
-               die_error('404 Not Found', "Not enough information to find object");
+               die_error(400, "Not enough information to find object");
        }
 
        print $cgi->redirect(-uri => href(action=>$type, -full=>1,
@@ -4872,12 +5304,12 @@ sub git_blobdiff {
                        open $fd, "-|", git_cmd(), "diff-tree", '-r', @diff_opts,
                                $hash_parent_base, $hash_base,
                                "--", (defined $file_parent ? $file_parent : ()), $file_name
-                               or die_error(undef, "Open git-diff-tree failed");
+                               or die_error(500, "Open git-diff-tree failed");
                        @difftree = map { chomp; $_ } <$fd>;
                        close $fd
-                               or die_error(undef, "Reading git-diff-tree failed");
+                               or die_error(404, "Reading git-diff-tree failed");
                        @difftree
-                               or die_error('404 Not Found', "Blob diff not found");
+                               or die_error(404, "Blob diff not found");
 
                } elsif (defined $hash &&
                         $hash =~ /[0-9a-fA-F]{40}/) {
@@ -4886,23 +5318,23 @@ sub git_blobdiff {
                        # read filtered raw output
                        open $fd, "-|", git_cmd(), "diff-tree", '-r', @diff_opts,
                                $hash_parent_base, $hash_base, "--"
-                               or die_error(undef, "Open git-diff-tree failed");
+                               or die_error(500, "Open git-diff-tree failed");
                        @difftree =
                                # ':100644 100644 03b21826... 3b93d5e7... M     ls-files.c'
                                # $hash == to_id
                                grep { /^:[0-7]{6} [0-7]{6} [0-9a-fA-F]{40} $hash/ }
                                map { chomp; $_ } <$fd>;
                        close $fd
-                               or die_error(undef, "Reading git-diff-tree failed");
+                               or die_error(404, "Reading git-diff-tree failed");
                        @difftree
-                               or die_error('404 Not Found', "Blob diff not found");
+                               or die_error(404, "Blob diff not found");
 
                } else {
-                       die_error('404 Not Found', "Missing one of the blob diff parameters");
+                       die_error(400, "Missing one of the blob diff parameters");
                }
 
                if (@difftree > 1) {
-                       die_error('404 Not Found', "Ambiguous blob diff specification");
+                       die_error(400, "Ambiguous blob diff specification");
                }
 
                %diffinfo = parse_difftree_raw_line($difftree[0]);
@@ -4923,46 +5355,12 @@ sub git_blobdiff {
                        '-p', ($format eq 'html' ? "--full-index" : ()),
                        $hash_parent_base, $hash_base,
                        "--", (defined $file_parent ? $file_parent : ()), $file_name
-                       or die_error(undef, "Open git-diff-tree failed");
+                       or die_error(500, "Open git-diff-tree failed");
        }
 
-       # old/legacy style URI
-       if (!%diffinfo && # if new style URI failed
-           defined $hash && defined $hash_parent) {
-               # fake git-diff-tree raw output
-               $diffinfo{'from_mode'} = $diffinfo{'to_mode'} = "blob";
-               $diffinfo{'from_id'} = $hash_parent;
-               $diffinfo{'to_id'}   = $hash;
-               if (defined $file_name) {
-                       if (defined $file_parent) {
-                               $diffinfo{'status'} = '2';
-                               $diffinfo{'from_file'} = $file_parent;
-                               $diffinfo{'to_file'}   = $file_name;
-                       } else { # assume not renamed
-                               $diffinfo{'status'} = '1';
-                               $diffinfo{'from_file'} = $file_name;
-                               $diffinfo{'to_file'}   = $file_name;
-                       }
-               } else { # no filename given
-                       $diffinfo{'status'} = '2';
-                       $diffinfo{'from_file'} = $hash_parent;
-                       $diffinfo{'to_file'}   = $hash;
-               }
-
-               # non-textual hash id's can be cached
-               if ($hash =~ m/^[0-9a-fA-F]{40}$/ &&
-                   $hash_parent =~ m/^[0-9a-fA-F]{40}$/) {
-                       $expires = '+1d';
-               }
-
-               # open patch output
-               open $fd, "-|", git_cmd(), "diff", @diff_opts,
-                       '-p', ($format eq 'html' ? "--full-index" : ()),
-                       $hash_parent, $hash, "--"
-                       or die_error(undef, "Open git-diff failed");
-       } else  {
+       # old/legacy style URI -- not generated anymore since 1.4.3.
+       if (!%diffinfo) {
                die_error('404 Not Found', "Missing one of the blob diff parameters")
-                       unless %diffinfo;
        }
 
        # header
@@ -4994,7 +5392,7 @@ sub git_blobdiff {
                print "X-Git-Url: " . $cgi->self_url() . "\n\n";
 
        } else {
-               die_error(undef, "Unknown blobdiff format");
+               die_error(400, "Unknown blobdiff format");
        }
 
        # patch
@@ -5027,13 +5425,18 @@ sub git_blobdiff_plain {
 }
 
 sub git_commitdiff {
-       my $format = shift || 'html';
-       $hash ||= $hash_base || "HEAD";
-       my %co = parse_commit($hash);
-       if (!%co) {
-               die_error(undef, "Unknown commit object");
+       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");
+
        # choose format for commitdiff for merge
        if (! defined $hash_parent && @{$co{'parents'}} > 1) {
                $hash_parent = '--cc';
@@ -5044,6 +5447,11 @@ 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') {
@@ -5114,7 +5522,7 @@ sub git_commitdiff {
                open $fd, "-|", git_cmd(), "diff-tree", '-r', @diff_opts,
                        "--no-commit-id", "--patch-with-raw", "--full-index",
                        $hash_parent_param, $hash, "--"
-                       or die_error(undef, "Open git-diff-tree failed");
+                       or die_error(500, "Open git-diff-tree failed");
 
                while (my $line = <$fd>) {
                        chomp $line;
@@ -5126,10 +5534,34 @@ sub git_commitdiff {
        } elsif ($format eq 'plain') {
                open $fd, "-|", git_cmd(), "diff-tree", '-r', @diff_opts,
                        '-p', $hash_parent_param, $hash, "--"
-                       or die_error(undef, "Open git-diff-tree failed");
-
+                       or die_error(500, "Open git-diff-tree failed");
+       } elsif ($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(undef, "Unknown commitdiff format");
+               die_error(400, "Unknown commitdiff format");
        }
 
        # non-textual hash id's can be cached
@@ -5176,6 +5608,14 @@ 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
@@ -5197,11 +5637,25 @@ 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 {
@@ -5212,19 +5666,15 @@ sub git_history {
                $page = 0;
        }
        my $ftype;
-       my %co = parse_commit($hash_base);
-       if (!%co) {
-               die_error(undef, "Unknown commit object");
-       }
+       my %co = parse_commit($hash_base)
+           or die_error(404, "Unknown commit object");
 
        my $refs = git_get_references();
        my $limit = sprintf("--max-count=%i", (100 * ($page+1)));
 
        my @commitlist = parse_commits($hash_base, 101, (100 * $page),
-                                      $file_name, "--full-history");
-       if (!@commitlist) {
-               die_error('404 Not Found', "No such file or directory on given branch");
-       }
+                                      $file_name, "--full-history")
+           or die_error(404, "No such file or directory on given branch");
 
        if (!defined $hash && defined $file_name) {
                # some commits could have deleted file in question,
@@ -5238,7 +5688,7 @@ sub git_history {
                $ftype = git_get_type($hash);
        }
        if (!defined $ftype) {
-               die_error(undef, "Unknown type of object");
+               die_error(500, "Unknown type of object");
        }
 
        my $paging_nav = '';
@@ -5276,19 +5726,16 @@ sub git_history {
 }
 
 sub git_search {
-       my ($have_search) = gitweb_check_feature('search');
-       if (!$have_search) {
-               die_error('403 Permission denied', "Permission denied");
-       }
+       gitweb_check_feature('search') or die_error(403, "Search is disabled");
        if (!defined $searchtext) {
-               die_error(undef, "Text field empty");
+               die_error(400, "Text field is empty");
        }
        if (!defined $hash) {
                $hash = git_get_head_hash($project);
        }
        my %co = parse_commit($hash);
        if (!%co) {
-               die_error(undef, "Unknown commit object");
+               die_error(404, "Unknown commit object");
        }
        if (!defined $page) {
                $page = 0;
@@ -5298,16 +5745,12 @@ sub git_search {
        if ($searchtype eq 'pickaxe') {
                # pickaxe may take all resources of your box and run for several minutes
                # with every query - so decide by yourself how public you make this feature
-               my ($have_pickaxe) = gitweb_check_feature('pickaxe');
-               if (!$have_pickaxe) {
-                       die_error('403 Permission denied', "Permission denied");
-               }
+               gitweb_check_feature('pickaxe')
+                   or die_error(403, "Pickaxe is disabled");
        }
        if ($searchtype eq 'grep') {
-               my ($have_grep) = gitweb_check_feature('grep');
-               if (!$have_grep) {
-                       die_error('403 Permission denied', "Permission denied");
-               }
+               gitweb_check_feature('grep')
+                   or die_error(403, "Grep is disabled");
        }
 
        git_header_html();
@@ -5511,7 +5954,7 @@ insensitive).</p>
 <dt><b>commit</b></dt>
 <dd>The commit messages and authorship information will be scanned for the given pattern.</dd>
 EOT
-       my ($have_grep) = gitweb_check_feature('grep');
+       my $have_grep = gitweb_check_feature('grep');
        if ($have_grep) {
                print <<EOT;
 <dt><b>grep</b></dt>
@@ -5528,7 +5971,7 @@ EOT
 <dt><b>committer</b></dt>
 <dd>Name and e-mail of the committer and date of commit will be scanned for the given pattern.</dd>
 EOT
-       my ($have_pickaxe) = gitweb_check_feature('pickaxe');
+       my $have_pickaxe = gitweb_check_feature('pickaxe');
        if ($have_pickaxe) {
                print <<EOT;
 <dt><b>pickaxe</b></dt>
@@ -5552,7 +5995,11 @@ sub git_shortlog {
        }
        my $refs = git_get_references();
 
-       my @commitlist = parse_commits($hash, 101, (100 * $page));
+       my $commit_hash = $hash;
+       if (defined $hash_parent) {
+               $commit_hash = "$hash_parent..$hash";
+       }
+       my @commitlist = parse_commits($commit_hash, 101, (100 * $page));
 
        my $paging_nav = format_paging_nav('shortlog', $hash, $head, $page, $#commitlist >= 100);
        my $next_link = '';
@@ -5561,6 +6008,14 @@ 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);
@@ -5576,12 +6031,12 @@ sub git_shortlog {
 
 sub git_feed {
        my $format = shift || 'atom';
-       my ($have_blame) = gitweb_check_feature('blame');
+       my $have_blame = gitweb_check_feature('blame');
 
        # Atom: http://www.atomenabled.org/developers/syndication/
        # RSS:  http://www.notestips.com/80256B3A007F2692/1/NAMO5P9UPQ
        if ($format ne 'rss' && $format ne 'atom') {
-               die_error(undef, "Unknown web feed format");
+               die_error(400, "Unknown web feed format");
        }
 
        # log/feed of current (HEAD) branch, log of given branch, history of file/directory
@@ -5598,7 +6053,25 @@ 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 => $content_type,
                        -charset => 'utf-8',
@@ -5657,7 +6130,24 @@ XML
                print "<title>$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 ($format eq 'atom') {
                print <<XML;
 <feed xmlns="http://www.w3.org/2005/Atom">
@@ -5684,6 +6174,7 @@ XML
                } else {
                        print "<updated>$latest_date{'iso-8601'}</updated>\n";
                }
+               print "<generator version='$version/$git_version'>gitweb</generator>\n";
        }
 
        # contents
@@ -5805,7 +6296,11 @@ 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">
@@ -5829,8 +6324,8 @@ XML
                }
 
                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;
index aa0eeca24786dbd5143354fc3bb5e8cdb3ef831f..a01eac814e51edd2115474f4d48a8d7fafa8c0e6 100644 (file)
@@ -435,6 +435,10 @@ div.search {
        right: 12px
 }
 
+p.projsearch {
+       text-align: center;
+}
+
 td.linenr {
        text-align: right;
 }
@@ -481,6 +485,19 @@ span.refs span {
        border-color: #ffccff #ff00ee #ff00ee #ffccff;
 }
 
+span.refs span a {
+       text-decoration: none;
+       color: inherit;
+}
+
+span.refs span a:hover {
+       text-decoration: underline;
+}
+
+span.refs span.indirect {
+       font-style: italic;
+}
+
 span.refs span.ref {
        background-color: #aaaaff;
        border-color: #ccccff #0033cc #0033cc #ccccff;
index 1cfa13b79778003ddc917164a174b9a41cd95add..4d0c44c6987d2d809b361fb5d61b64eb9d9c5654 100755 (executable)
@@ -20,4 +20,5 @@ $feature{'blame'}{'default'} = [1];
 #bernie: not working with rewrite rule
 $feature{'pathinfo'}{'default'} = [1];
 
-$feature{'snapshot'}{'default'} = ['tgz', 'tbz2', 'zip'];
+#bernie: disabled because lame spam harvesters download them all
+#$feature{'snapshot'}{'default'} = ['tgz', 'tbz2', 'zip'];