cherry-pick uri security fix

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

diff --git a/IkiWiki/Rcs/git.pm b/IkiWiki/Rcs/git.pm
index 05eaa8d78eba8f408bdaa4645a3eca28d31ea213..26a6f42661a652249c8a4c7475dfb45578b8fc03 100644 (file)
--- a/IkiWiki/Rcs/git.pm
+++ b/IkiWiki/Rcs/git.pm
@@ -8,8 +8,6 @@ use open qw{:utf8 :std};
 
 package IkiWiki;
 
 
 package IkiWiki;
 

-my $origin_branch    = 'origin';            # Git ref for main repository
-my $master_branch    = 'master';            # working branch

 my $sha1_pattern     = qr/[0-9a-fA-F]{40}/; # pattern to validate Git sha1sums
 my $dummy_commit_msg = 'dummy commit';      # message to skip in recent changes
 
 my $sha1_pattern     = qr/[0-9a-fA-F]{40}/; # pattern to validate Git sha1sums
 my $dummy_commit_msg = 'dummy commit';      # message to skip in recent changes
 


@@ -25,8 +23,6 @@ sub _safe_git (&@) { #{{{
 
        if (!$pid) {
                # In child.
 
        if (!$pid) {
                # In child.

-               open STDERR, ">&STDOUT"
-                   or error("Cannot dup STDOUT: $!");

                # Git commands want to be in wc.
                chdir $config{srcdir}
                    or error("Cannot chdir to $config{srcdir}: $!");
                # Git commands want to be in wc.
                chdir $config{srcdir}
                    or error("Cannot chdir to $config{srcdir}: $!");


@@ -42,7 +38,7 @@ sub _safe_git (&@) { #{{{
 
        close $OUT;
 
 
        close $OUT;
 

-       ($error_handler || sub { })->("'@cmdline' failed: $!") if $?;
+       $error_handler->("'@cmdline' failed: $!") if $? && $error_handler;

 
        return wantarray ? @lines : ($? == 0);
 }
 
        return wantarray ? @lines : ($? == 0);
 }


@@ -55,7 +51,7 @@ sub run_or_non ($@) { _safe_git(undef,            @_) }
 sub _merge_past ($$$) { #{{{
        # Unlike with Subversion, Git cannot make a 'svn merge -rN:M file'.
        # Git merge commands work with the committed changes, except in the
 sub _merge_past ($$$) { #{{{
        # Unlike with Subversion, Git cannot make a 'svn merge -rN:M file'.
        # Git merge commands work with the committed changes, except in the

-       # implicit case of '-m' of git-checkout(1).  So we should invent a
+       # implicit case of '-m' of git checkout(1).  So we should invent a

        # kludge here.  In principle, we need to create a throw-away branch
        # in preparing for the merge itself.  Since branches are cheap (and
        # branching is fast), this shouldn't cost high.
        # kludge here.  In principle, we need to create a throw-away branch
        # in preparing for the merge itself.  Since branches are cheap (and
        # branching is fast), this shouldn't cost high.


@@ -64,7 +60,7 @@ sub _merge_past ($$$) { #{{{
        # possible approach to get rid of this situation could be that we first
        # make a temporary commit in the master branch and later restore the
        # initial state (this is possible since Git has the ability to undo a
        # possible approach to get rid of this situation could be that we first
        # make a temporary commit in the master branch and later restore the
        # initial state (this is possible since Git has the ability to undo a

-       # commit, i.e. 'git-reset --soft HEAD^').  The method can be summarized
+       # commit, i.e. 'git reset --soft HEAD^').  The method can be summarized

        # as follows:
        #
        #       - create a diff of HEAD:current-sha1
        # as follows:
        #
        #       - create a diff of HEAD:current-sha1


@@ -107,30 +103,30 @@ sub _merge_past ($$$) { #{{{
                my $branch = "throw_away_${sha1}"; # supposed to be unique
 
                # Create a throw-away branch and rewind backward.
                my $branch = "throw_away_${sha1}"; # supposed to be unique
 
                # Create a throw-away branch and rewind backward.

-               push @undo, sub { run_or_cry('git-branch', '-D', $branch) };
-               run_or_die('git-branch', $branch, $sha1);
+               push @undo, sub { run_or_cry('git', 'branch', '-D', $branch) };
+               run_or_die('git', 'branch', $branch, $sha1);

 
                # Switch to throw-away branch for the merge operation.
                push @undo, sub {
 
                # Switch to throw-away branch for the merge operation.
                push @undo, sub {

-                       if (!run_or_cry('git-checkout', $master_branch)) {
-                               run_or_cry('git-checkout','-f',$master_branch);
+                       if (!run_or_cry('git', 'checkout', $config{gitmaster_branch})) {
+                               run_or_cry('git', 'checkout','-f',$config{gitmaster_branch});

                        }
                };
                        }
                };

-               run_or_die('git-checkout', $branch);
+               run_or_die('git', 'checkout', $branch);

 
                # Put the modified file in _this_ branch.
                rename($hidden, $target)
                    or error("rename '$hidden' to '$target' failed: $!");
 
                # _Silently_ commit all modifications in the current branch.
 
                # Put the modified file in _this_ branch.
                rename($hidden, $target)
                    or error("rename '$hidden' to '$target' failed: $!");
 
                # _Silently_ commit all modifications in the current branch.

-               run_or_non('git-commit', '-m', $message, '-a');
+               run_or_non('git', 'commit', '-m', $message, '-a');

                # ... and re-switch to master.
                # ... and re-switch to master.

-               run_or_die('git-checkout', $master_branch);
+               run_or_die('git', 'checkout', $config{gitmaster_branch});

 
                # Attempt to merge without complaining.
 
                # Attempt to merge without complaining.

-               if (!run_or_non('git-pull', '--no-commit', '.', $branch)) {
+               if (!run_or_non('git', 'pull', '--no-commit', '.', $branch)) {

                        $conflict = readfile($target);
                        $conflict = readfile($target);

-                       run_or_die('git-reset', '--hard');
+                       run_or_die('git', 'reset', '--hard');

                }
        };
        my $failure = $@;
                }
        };
        my $failure = $@;


@@ -146,34 +142,37 @@ sub _merge_past ($$$) { #{{{
        return $conflict;
 } #}}}
 
        return $conflict;
 } #}}}
 

-sub _parse_diff_tree (@) { #{{{
+sub _parse_diff_tree ($@) { #{{{

        # Parse the raw diff tree chunk and return the info hash.
        # See git-diff-tree(1) for the syntax.
 
        # Parse the raw diff tree chunk and return the info hash.
        # See git-diff-tree(1) for the syntax.
 

-       my ($dt_ref) = @_;
+       my ($prefix, $dt_ref) = @_;

 
        # End of stream?
 
        # End of stream?

-       return if !defined @{ $dt_ref } || !length @{ $dt_ref }[0];
+       return if !defined @{ $dt_ref } ||
+                 !defined @{ $dt_ref }[0] || !length @{ $dt_ref }[0];

 
        my %ci;
        # Header line.
 
        my %ci;
        # Header line.

-       HEADER: while (my $line = shift @{ $dt_ref }) {
+       while (my $line = shift @{ $dt_ref }) {

                return if $line !~ m/^(.+) ($sha1_pattern)/;
 
                return if $line !~ m/^(.+) ($sha1_pattern)/;
 

-               my $sha1 = $1;
+               my $sha1 = $2;

                $ci{'sha1'} = $sha1;
                $ci{'sha1'} = $sha1;

-               last HEADER;
+               last;

        }
 
        # Identification lines for the commit.
        }
 
        # Identification lines for the commit.

-       IDENT: while (my $line = shift @{ $dt_ref }) {
+       while (my $line = shift @{ $dt_ref }) {

                # Regexps are semi-stolen from gitweb.cgi.
                if ($line =~ m/^tree ([0-9a-fA-F]{40})$/) {
                        $ci{'tree'} = $1;
                # Regexps are semi-stolen from gitweb.cgi.
                if ($line =~ m/^tree ([0-9a-fA-F]{40})$/) {
                        $ci{'tree'} = $1;

-               } elsif ($line =~ m/^parent ([0-9a-fA-F]{40})$/) {
+               }
+               elsif ($line =~ m/^parent ([0-9a-fA-F]{40})$/) {

                        # XXX: collecting in reverse order
                        push @{ $ci{'parents'} }, $1;
                        # XXX: collecting in reverse order
                        push @{ $ci{'parents'} }, $1;

-               } elsif ($line =~ m/^(author|committer) (.*) ([0-9]+) (.*)$/) {
+               }
+               elsif ($line =~ m/^(author|committer) (.*) ([0-9]+) (.*)$/) {

                        my ($who, $name, $epoch, $tz) =
                           ($1,   $2,    $3,     $4 );
 
                        my ($who, $name, $epoch, $tz) =
                           ($1,   $2,    $3,     $4 );
 


@@ -181,83 +180,91 @@ sub _parse_diff_tree (@) { #{{{
                        $ci{ "${who}_epoch" } = $epoch;
                        $ci{ "${who}_tz"    } = $tz;
 
                        $ci{ "${who}_epoch" } = $epoch;
                        $ci{ "${who}_tz"    } = $tz;
 

-                       if ($name =~ m/^([^<]+) <([^@]+)/) {
+                       if ($name =~ m/^([^<]+) <([^@>]+)/) {

                                my ($fullname, $username) = ($1, $2);
                                $ci{"${who}_fullname"}    = $fullname;
                                $ci{"${who}_username"}    = $username;
                                my ($fullname, $username) = ($1, $2);
                                $ci{"${who}_fullname"}    = $fullname;
                                $ci{"${who}_username"}    = $username;

-                       } else {
+                       }
+                       else {

                                $ci{"${who}_fullname"} =
                                        $ci{"${who}_username"} = $name;
                        }
                                $ci{"${who}_fullname"} =
                                        $ci{"${who}_username"} = $name;
                        }

-               } elsif ($line =~ m/^$/) {
+               }
+               elsif ($line =~ m/^$/) {

                        # Trailing empty line signals next section.
                        # Trailing empty line signals next section.

-                       last IDENT;
+                       last;

                }
        }
 
                }
        }
 

-       error("No 'tree' or 'parents' seen in diff-tree output")
-           if !defined $ci{'tree'} || !defined $ci{'parents'};
+       debug("No 'tree' seen in diff-tree output") if !defined $ci{'tree'};

 
 

-       $ci{'parent'} = @{ $ci{'parents'} }[0];
+       if (defined $ci{'parents'}) {
+               $ci{'parent'} = @{ $ci{'parents'} }[0];
+       }
+       else {
+               $ci{'parent'} = 0 x 40;
+       }

 
        # Commit message.
 
        # Commit message.

-       COMMENT: while (my $line = shift @{ $dt_ref }) {
+       while (my $line = shift @{ $dt_ref }) {

                if ($line =~ m/^$/) {
                        # Trailing empty line signals next section.
                if ($line =~ m/^$/) {
                        # Trailing empty line signals next section.

-                       last COMMENT;
+                       last;

                };
                $line =~ s/^    //;
                push @{ $ci{'comment'} }, $line;
        }
 
        # Modified files.
                };
                $line =~ s/^    //;
                push @{ $ci{'comment'} }, $line;
        }
 
        # Modified files.

-       FILE: while (my $line = shift @{ $dt_ref }) {
-               if ($line =~ m{^:
-                       ([0-7]{6})[ ]      # from mode
-                       ([0-7]{6})[ ]      # to mode
-                       ($sha1_pattern)[ ] # from sha1
-                       ($sha1_pattern)[ ] # to sha1
-                       (.)                # status
-                       ([0-9]{0,3})\t     # similarity
-                       (.*)               # file
+       while (my $line = shift @{ $dt_ref }) {
+               if ($line =~ m{^
+                       (:+)       # number of parents
+                       ([^\t]+)\t # modes, sha1, status
+                       (.*)       # file names

                $}xo) {
                $}xo) {

-                       my ($sha1_from, $sha1_to, $file) =
-                          ($3,         $4,       $7   );
+                       my $num_parents = length $1;
+                       my @tmp = split(" ", $2);
+                       my ($file, $file_to) = split("\t", $3);
+                       my @mode_from = splice(@tmp, 0, $num_parents);
+                       my $mode_to = shift(@tmp);
+                       my @sha1_from = splice(@tmp, 0, $num_parents);
+                       my $sha1_to = shift(@tmp);
+                       my $status = shift(@tmp);

 
                        if ($file =~ m/^"(.*)"$/) {
                                ($file=$1) =~ s/\\([0-7]{1,3})/chr(oct($1))/eg;
                        }
 
                        if ($file =~ m/^"(.*)"$/) {
                                ($file=$1) =~ s/\\([0-7]{1,3})/chr(oct($1))/eg;
                        }

+                       $file =~ s/^\Q$prefix\E//;

                        if (length $file) {
                                push @{ $ci{'details'} }, {
                                        'file'      => decode_utf8($file),
                        if (length $file) {
                                push @{ $ci{'details'} }, {
                                        'file'      => decode_utf8($file),

-                                       'sha1_from' => $sha1_from,
+                                       'sha1_from' => $sha1_from[0],

                                        'sha1_to'   => $sha1_to,
                                };
                        }
                                        'sha1_to'   => $sha1_to,
                                };
                        }

-                       next FILE;
+                       next;

                };
                };

-               last FILE;
+               last;

        }
 
        }
 

-       warn "No detail in diff-tree output" if !defined $ci{'details'};
-

        return \%ci;
 } #}}}
 
        return \%ci;
 } #}}}
 

-sub git_commit_info (;$$) { #{{{
+sub git_commit_info ($;$) { #{{{

        # Return an array of commit info hashes of num commits (default: 1)
        # Return an array of commit info hashes of num commits (default: 1)

-       # starting from the given sha1sum (default: HEAD).
+       # starting from the given sha1sum.

 
        my ($sha1, $num) = @_;
 
        $num ||= 1;
 
 
        my ($sha1, $num) = @_;
 
        $num ||= 1;
 

-       my @raw_lines =
-           run_or_die(qq{git-rev-list --max-count=$num $sha1 |
-                         git-diff-tree --stdin --pretty=raw -M -r});
+       my @raw_lines = run_or_die('git', 'log', "--max-count=$num", 
+               '--pretty=raw', '--raw', '--abbrev=40', '--always', '-c',
+               '-r', $sha1, '--', '.');
+       my ($prefix) = run_or_die('git', 'rev-parse', '--show-prefix');

 
        my @ci;
 
        my @ci;

-       while (my $parsed = _parse_diff_tree(\@raw_lines)) {
+       while (my $parsed = _parse_diff_tree(($prefix or ""), \@raw_lines)) {

                push @ci, $parsed;
        }
 
                push @ci, $parsed;
        }
 


@@ -272,7 +279,7 @@ sub git_sha1 (;$) { #{{{
        my $file = shift || q{--};
 
        # Ignore error since a non-existing file might be given.
        my $file = shift || q{--};
 
        # Ignore error since a non-existing file might be given.

-       my ($sha1) = run_or_non('git-rev-list', '--max-count=1', 'HEAD', $file);
+       my ($sha1) = run_or_non('git', 'rev-list', '--max-count=1', 'HEAD', $file);

        if ($sha1) {
                ($sha1) = $sha1 =~ m/($sha1_pattern)/; # sha1 is untainted now
        } else { debug("Empty sha1sum for '$file'.") }
        if ($sha1) {
                ($sha1) = $sha1 =~ m/($sha1_pattern)/; # sha1 is untainted now
        } else { debug("Empty sha1sum for '$file'.") }


@@ -282,7 +289,9 @@ sub git_sha1 (;$) { #{{{
 sub rcs_update () { #{{{
        # Update working directory.
 
 sub rcs_update () { #{{{
        # Update working directory.
 

-       run_or_cry('git-pull', $origin_branch);
+       if (length $config{gitorigin_branch}) {
+               run_or_cry('git', 'pull', $config{gitorigin_branch});
+       }

 } #}}}
 
 sub rcs_prepedit ($) { #{{{
 } #}}}
 
 sub rcs_prepedit ($) { #{{{


@@ -302,17 +311,19 @@ sub rcs_commit ($$$;$$) { #{{{
        my ($file, $message, $rcstoken, $user, $ipaddr) = @_;
 
        if (defined $user) {
        my ($file, $message, $rcstoken, $user, $ipaddr) = @_;
 
        if (defined $user) {

-               $message="web commit by $user".(length $message ? ": $message" : "");
+               $message = "web commit by $user" .
+                   (length $message ? ": $message" : "");

        }
        elsif (defined $ipaddr) {
        }
        elsif (defined $ipaddr) {

-               $message="web commit from $ipaddr".(length $message ? ": $message" : "");
+               $message = "web commit from $ipaddr" .
+                   (length $message ? ": $message" : "");

        }
 
        # XXX: Wiki directory is in the unlocked state when starting this
        # action.  But it takes time for a Git process to finish its job
        # (especially if a merge required), so we must re-lock to prevent
        # race conditions.  Only when the time of the real commit action
        }
 
        # XXX: Wiki directory is in the unlocked state when starting this
        # action.  But it takes time for a Git process to finish its job
        # (especially if a merge required), so we must re-lock to prevent
        # race conditions.  Only when the time of the real commit action

-       # (i.e. git-push(1)) comes, we'll unlock the directory.
+       # (i.e. git push) comes, we'll unlock the directory.

        lockwiki();
 
        # Check to see if the page has been changed by someone else since
        lockwiki();
 
        # Check to see if the page has been changed by someone else since


@@ -325,12 +336,14 @@ sub rcs_commit ($$$;$$) { #{{{
                return $conflict if defined $conflict;
        }
 
                return $conflict if defined $conflict;
        }
 

-       # git-commit(1) returns non-zero if file has not been really changed.
+       # git commit returns non-zero if file has not been really changed.

        # so we should ignore its exit status (hence run_or_non).
        $message = possibly_foolish_untaint($message);
        # so we should ignore its exit status (hence run_or_non).
        $message = possibly_foolish_untaint($message);

-       if (run_or_non('git-commit', '-m', $message, '-i', $file)) {
+       if (run_or_non('git', 'commit', '-q', '-m', $message, '-i', $file)) {

                unlockwiki();
                unlockwiki();

-               run_or_cry('git-push', $origin_branch);
+               if (length $config{gitorigin_branch}) {
+                       run_or_cry('git', 'push', $config{gitorigin_branch});
+               }

        }
 
        return undef; # success
        }
 
        return undef; # success


@@ -341,7 +354,7 @@ sub rcs_add ($) { # {{{
 
        my ($file) = @_;
 
 
        my ($file) = @_;
 

-       run_or_cry('git-add', $file);
+       run_or_cry('git', 'add', $file);

 } #}}}
 
 sub rcs_recentchanges ($) { #{{{
 } #}}}
 
 sub rcs_recentchanges ($) { #{{{


@@ -352,21 +365,21 @@ sub rcs_recentchanges ($) { #{{{
        eval q{use Date::Parse};
        error($@) if $@;
 
        eval q{use Date::Parse};
        error($@) if $@;
 

-       my ($sha1, $type, $when, $diffurl, $user, @pages, @message, @rets);
-       INFO: foreach my $ci (git_commit_info('HEAD', $num)) {
-               my $title = @{ $ci->{'comment'} }[0];
-
+       my @rets;
+       foreach my $ci (git_commit_info('HEAD', $num)) {

                # Skip redundant commits.
                # Skip redundant commits.

-               next INFO if ($title eq $dummy_commit_msg);
+               next if (@{$ci->{'comment'}}[0] eq $dummy_commit_msg);

 
 

-               $sha1 = $ci->{'sha1'};
-               $type = "web";
-               $when = time - $ci->{'author_epoch'};
+               my ($sha1, $when) = (
+                       $ci->{'sha1'},
+                       $ci->{'author_epoch'}
+               );

 
 

-               DETAIL: foreach my $detail (@{ $ci->{'details'} }) {
-                       my $diffurl = $config{'diffurl'};
-                       my $file    = $detail->{'file'};
+               my (@pages, @messages);
+               foreach my $detail (@{ $ci->{'details'} }) {
+                       my $file = $detail->{'file'};

 
 

+                       my $diffurl = $config{'diffurl'};

                        $diffurl =~ s/\[\[file\]\]/$file/go;
                        $diffurl =~ s/\[\[sha1_parent\]\]/$ci->{'parent'}/go;
                        $diffurl =~ s/\[\[sha1_from\]\]/$detail->{'sha1_from'}/go;
                        $diffurl =~ s/\[\[file\]\]/$file/go;
                        $diffurl =~ s/\[\[sha1_parent\]\]/$ci->{'parent'}/go;
                        $diffurl =~ s/\[\[sha1_from\]\]/$detail->{'sha1_from'}/go;


@@ -377,14 +390,16 @@ sub rcs_recentchanges ($) { #{{{
                                diffurl => $diffurl,
                        };
                }
                                diffurl => $diffurl,
                        };
                }

+               push @messages, { line => $_ } foreach @{$ci->{'comment'}};

 
 

-               push @message, { line => $title };
+               my ($user, $type) = (q{}, "web");

 
 

-               if (defined $message[0] &&
-                   $message[0]->{line} =~ m/$config{web_commit_regexp}/) {
-                       $user=defined $2 ? "$2" : "$3";
-                       $message[0]->{line}=$4;
-               } else {
+               if (defined $messages[0] &&
+                   $messages[0]->{line} =~ m/$config{web_commit_regexp}/) {
+                       $user = defined $2 ? "$2" : "$3";
+                       $messages[0]->{line} = $4;
+               }
+               else {

                        $type ="git";
                        $user = $ci->{'author_username'};
                }
                        $type ="git";
                        $user = $ci->{'author_username'};
                }


@@ -394,69 +409,25 @@ sub rcs_recentchanges ($) { #{{{
                        user       => $user,
                        committype => $type,
                        when       => $when,
                        user       => $user,
                        committype => $type,
                        when       => $when,

-                       message    => [@message],
+                       message    => [@messages],

                        pages      => [@pages],
                } if @pages;
 
                        pages      => [@pages],
                } if @pages;
 

-               $sha1 = $type = $when = $diffurl = $user = undef;
-               @pages = @message = ();
+               last if @rets >= $num;

        }
 
        return @rets;
 } #}}}
 
        }
 
        return @rets;
 } #}}}
 

-sub rcs_notify () { #{{{
-       # Send notification mail to subscribed users.
-       #
-       # In usual Git usage, hooks/update script is presumed to send
-       # notification mails (see git-receive-pack(1)).  But we prefer
-       # hooks/post-update to support IkiWiki commits coming from a
-       # cloned repository (through command line) because post-update
-       # is called _after_ each ref in repository is updated (update
-       # hook is called _before_ the repository is updated).  Since
-       # post-update hook does not accept command line arguments, we
-       # don't have an $ENV variable in this function.
-       #
-       # Here, we rely on a simple fact: we can extract all parts of the
-       # notification content by parsing the "HEAD" commit (which also
-       # triggers a refresh of IkiWiki pages) and we can obtain the diff
-       # by comparing HEAD and HEAD^ (the previous commit).
-
-       my $sha1 = 'HEAD'; # the commit which triggers this action
-
-       my $ci = git_commit_info($sha1);
-       return if !defined $ci;
-
-       my @changed_pages = map { $_->{'file'} } @{ $ci->{'details'} };
-
-       my ($user, $message);
-       if (@{ $ci->{'comment'} }[0] =~ m/$config{web_commit_regexp}/) {
-               $user    = defined $2 ? "$2" : "$3";
-               $message = $4;
-       } else {
-               $user    = $ci->{'author_username'};
-               $message = join "\n", @{ $ci->{'comment'} };
-       }
-
-       require IkiWiki::UserInfo;
-       send_commit_mails(
-               sub {
-                       $message;
-               },
-               sub {
-                       join "\n", run_or_die('git-diff', "${sha1}^", $sha1);
-               }, $user, @changed_pages);
-} #}}}
-

 sub rcs_getctime ($) { #{{{
 sub rcs_getctime ($) { #{{{

-       # Get the ctime of file.
-
-       my ($file) = @_;
+       my $file=shift;
+       # Remove srcdir prefix
+       $file =~ s/^\Q$config{srcdir}\E\/?//;

 
        my $sha1  = git_sha1($file);
        my $ci    = git_commit_info($sha1);
        my $ctime = $ci->{'author_epoch'};
 
        my $sha1  = git_sha1($file);
        my $ci    = git_commit_info($sha1);
        my $ctime = $ci->{'author_epoch'};

-       debug("ctime for '$file': ". localtime($ctime) . "\n");
+       debug("ctime for '$file': ". localtime($ctime));

 
        return $ctime;
 } #}}}
 
        return $ctime;
 } #}}}