X-Git-Url: http://git.vanrenterghem.biz/git.ikiwiki.info.git/blobdiff_plain/57d10171e342046cc6062e5053186920ea04d486..35bc35660cafad2e8661077a9196bbf9b6fd3663:/IkiWiki/Rcs/git.pm?ds=inline

diff --git a/IkiWiki/Rcs/git.pm b/IkiWiki/Rcs/git.pm
index 3fedd64f1..26a6f4266 100644
--- a/IkiWiki/Rcs/git.pm
+++ b/IkiWiki/Rcs/git.pm
@@ -1,20 +1,15 @@
 #!/usr/bin/perl
-# Git backend for IkiWiki.
-# Copyright 2006  Recai Oktaş <roktas@debian.org>
-#
-# Licensed under the same terms as IkiWiki.
 
 use warnings;
 use strict;
 use IkiWiki;
+use Encode;
+use open qw{:utf8 :std};
 
 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 $web_commit_msg   = qr/^web commit by (\w+):?(.*)/; # pattern for web commits
 
 sub _safe_git (&@) { #{{{
 	# Start a child process safely without resorting /bin/sh.
@@ -28,8 +23,6 @@ sub _safe_git (&@) { #{{{
 
 	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}: $!");
@@ -45,12 +38,12 @@ sub _safe_git (&@) { #{{{
 
 	close $OUT;
 
-	($error_handler || sub { })->("'@cmdline' failed: $!") if $?;
+	$error_handler->("'@cmdline' failed: $!") if $? && $error_handler;
 
 	return wantarray ? @lines : ($? == 0);
 }
 # Convenient wrappers.
-sub run_or_die ($@) { _safe_git(\&IkiWiki::error, @_) }
+sub run_or_die ($@) { _safe_git(\&error, @_) }
 sub run_or_cry ($@) { _safe_git(sub { warn @_ },  @_) }
 sub run_or_non ($@) { _safe_git(undef,            @_) }
 #}}}
@@ -58,7 +51,7 @@ sub run_or_non ($@) { _safe_git(undef,            @_) }
 sub _merge_past ($$$) { #{{{
 	# Unlike with Subversion, Git cannot make a 'svn merge -rN:M file'.
 	# Git merge commands work with the committed changes, except in the
-	# implicit case of '-m' of git-checkout(1).  So we should invent a
+	# implicit case of '-m' of git checkout(1).  So we should invent a
 	# kludge here.  In principle, we need to create a throw-away branch
 	# in preparing for the merge itself.  Since branches are cheap (and
 	# branching is fast), this shouldn't cost high.
@@ -67,7 +60,7 @@ sub _merge_past ($$$) { #{{{
 	# 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
-	# 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
@@ -110,30 +103,30 @@ sub _merge_past ($$$) { #{{{
 		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 {
-			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.
-		run_or_non('git-commit', '-m', $message, '-a');
+		run_or_non('git', 'commit', '-m', $message, '-a');
 		# ... 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.
-		if (!run_or_non('git-pull', '--no-commit', '.', $branch)) {
+		if (!run_or_non('git', 'pull', '--no-commit', '.', $branch)) {
 			$conflict = readfile($target);
-			run_or_die('git-reset', '--hard');
+			run_or_die('git', 'reset', '--hard');
 		}
 	};
 	my $failure = $@;
@@ -149,34 +142,37 @@ sub _merge_past ($$$) { #{{{
 	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.
 
-	my ($dt_ref) = @_;
+	my ($prefix, $dt_ref) = @_;
 
 	# 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.
-	HEADER: while (my $line = shift @{ $dt_ref }) {
-		return if $line !~ m/^diff-tree ($sha1_pattern)/;
+	while (my $line = shift @{ $dt_ref }) {
+		return if $line !~ m/^(.+) ($sha1_pattern)/;
 
-		my $sha1 = $1;
+		my $sha1 = $2;
 		$ci{'sha1'} = $sha1;
-		last HEADER;
+		last;
 	}
 
 	# 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'} = $1;
-		} elsif ($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'} }, $1;
-		} elsif ($line =~ m/^(author|committer) (.*) ([0-9]+) (.*)$/) {
+		}
+		elsif ($line =~ m/^(author|committer) (.*) ([0-9]+) (.*)$/) {
 			my ($who, $name, $epoch, $tz) =
 			   ($1,   $2,    $3,     $4 );
 
@@ -184,86 +180,96 @@ sub _parse_diff_tree (@) { #{{{
 			$ci{ "${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;
-			} else {
+			}
+			else {
 				$ci{"${who}_fullname"} =
 					$ci{"${who}_username"} = $name;
 			}
-		} elsif ($line =~ m/^$/) {
+		}
+		elsif ($line =~ m/^$/) {
 			# 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.
-	COMMENT: while (my $line = shift @{ $dt_ref }) {
+	while (my $line = shift @{ $dt_ref }) {
 		if ($line =~ m/^$/) {
 			# Trailing empty line signals next section.
-			last COMMENT;
+			last;
 		};
 		$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
 		$}xo) {
-			my ($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;
 			}
+			$file =~ s/^\Q$prefix\E//;
 			if (length $file) {
 				push @{ $ci{'details'} }, {
-					'file'      => $file,
-					'sha1_from' => $sha1_from,
+					'file'      => decode_utf8($file),
+					'sha1_from' => $sha1_from[0],
 					'sha1_to'   => $sha1_to,
 				};
 			}
-			next FILE;
+			next;
 		};
-		last FILE;
+		last;
 	}
 
-	warn "No detail in diff-tree output" if !defined $ci{'details'};
-
 	return \%ci;
 } #}}}
 
-sub git_commit_info (;$$) { #{{{
+sub git_commit_info ($;$) { #{{{
 	# 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 @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;
-	while (my $parsed = _parse_diff_tree(\@raw_lines)) {
+	while (my $parsed = _parse_diff_tree(($prefix or ""), \@raw_lines)) {
 		push @ci, $parsed;
 	}
 
+	warn "Cannot parse commit info for '$sha1' commit" if !@ci;
+
 	return wantarray ? @ci : $ci[0];
 } #}}}
 
@@ -273,7 +279,7 @@ sub git_sha1 (;$) { #{{{
 	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 '$file'.") }
@@ -283,7 +289,9 @@ sub git_sha1 (;$) { #{{{
 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 ($) { #{{{
@@ -295,18 +303,27 @@ sub rcs_prepedit ($) { #{{{
 	return git_sha1($file);
 } #}}}
 
-sub rcs_commit ($$$) { #{{{
+sub rcs_commit ($$$;$$) { #{{{
 	# Try to commit the page; returns undef on _success_ and
 	# a version of the page with the rcs's conflict markers on
 	# failure.
 
-	my ($file, $message, $rcstoken) = @_;
+	my ($file, $message, $rcstoken, $user, $ipaddr) = @_;
+
+	if (defined $user) {
+		$message = "web commit by $user" .
+		    (length $message ? ": $message" : "");
+	}
+	elsif (defined $ipaddr) {
+		$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
-	# (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
@@ -319,12 +336,14 @@ sub rcs_commit ($$$) { #{{{
 		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);
-	if (run_or_non('git-commit', '-m', $message, '-i', $file)) {
+	if (run_or_non('git', 'commit', '-q', '-m', $message, '-i', $file)) {
 		unlockwiki();
-		run_or_cry('git-push', $origin_branch);
+		if (length $config{gitorigin_branch}) {
+			run_or_cry('git', 'push', $config{gitorigin_branch});
+		}
 	}
 
 	return undef; # success
@@ -335,7 +354,7 @@ sub rcs_add ($) { # {{{
 
 	my ($file) = @_;
 
-	run_or_cry('git-add', $file);
+	run_or_cry('git', 'add', $file);
 } #}}}
 
 sub rcs_recentchanges ($) { #{{{
@@ -343,147 +362,72 @@ sub rcs_recentchanges ($) { #{{{
 
 	my ($num) = @_;
 
-	eval q{use CGI 'escapeHTML'};
 	eval q{use Date::Parse};
-	eval q{use Time::Duration};
-
-	my ($sha1, $type, $when, $diffurl, $user, @pages, @message, @rets);
-	INFO: foreach my $ci (git_commit_info('HEAD', $num)) {
-		my $title = @{ $ci->{'comment'} }[0];
+	error($@) if $@;
 
+	my @rets;
+	foreach my $ci (git_commit_info('HEAD', $num)) {
 		# Skip redundant commits.
-		next INFO if ($title eq $dummy_commit_msg);
+		next if (@{$ci->{'comment'}}[0] eq $dummy_commit_msg);
+
+		my ($sha1, $when) = (
+			$ci->{'sha1'},
+			$ci->{'author_epoch'}
+		);
 
-		$sha1 = $ci->{'sha1'};
-		$type = "web";
-		$when = concise(ago(time - $ci->{'author_epoch'}));
+		my (@pages, @messages);
+		foreach my $detail (@{ $ci->{'details'} }) {
+			my $file = $detail->{'file'};
 
-		foreach my $bit (@{ $ci->{'details'} }) {
 			my $diffurl = $config{'diffurl'};
-			my $file    = $bit->{'file'};
-
 			$diffurl =~ s/\[\[file\]\]/$file/go;
 			$diffurl =~ s/\[\[sha1_parent\]\]/$ci->{'parent'}/go;
-			$diffurl =~ s/\[\[sha1_from\]\]/$bit->{'sha1_from'}/go;
-			$diffurl =~ s/\[\[sha1_to\]\]/$bit->{'sha1_to'}/go;
+			$diffurl =~ s/\[\[sha1_from\]\]/$detail->{'sha1_from'}/go;
+			$diffurl =~ s/\[\[sha1_to\]\]/$detail->{'sha1_to'}/go;
 
 			push @pages, {
-				link => htmllink("", "", pagename($file), 1),
+				page => pagename($file),
 				diffurl => $diffurl,
-			},
+			};
 		}
+		push @messages, { line => $_ } foreach @{$ci->{'comment'}};
 
-		push @message, { line => escapeHTML($title) };
+		my ($user, $type) = (q{}, "web");
 
-		if (defined $message[0] &&
-		    $message[0]->{line} =~ m/$web_commit_msg/) {
-			$user = "$1";
-			$message[0]->{line} = $2;
-		} 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'};
 		}
 
 		push @rets, {
 			rev        => $sha1,
-			user       => htmllink("", "", $user, 1),
+			user       => $user,
 			committype => $type,
 			when       => $when,
-			message    => [@message],
+			message    => [@messages],
 			pages      => [@pages],
 		} if @pages;
 
-		$sha1 = $type = $when = $diffurl = $user = undef;
-		@pages = @message = ();
+		last if @rets >= $num;
 	}
 
 	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);
-	if (!defined $ci) {
-		warn "Cannot parse info for '$sha1' commit";
-		return;
-	}
-
-	my @changed_pages = map { $_->{'file'} } @{ $ci->{'details'} };
-
-	my ($user, $message);
-	if (@{ $ci->{'comment'} }[0] =~ m/$web_commit_msg/) {
-		$user    = "$1";
-		$message = $2;
-	} else {
-		$user    = $ci->{'author_username'};
-		$message = join "\n", @{ $ci->{'comment'} };
-	}
-
-	require IkiWiki::UserInfo;
-	my @email_recipients = commit_notify_list($user, @changed_pages);
-	return if !@email_recipients;
-
-	# TODO: if a commit spans multiple pages, this will send
-	# subscribers a diff that might contain pages they did not
-	# sign up for. Should separate the diff per page and
-	# reassemble into one mail with just the pages subscribed to.
-	my $diff = join "\n", run_or_die('git-diff', "${sha1}^", $sha1);
-
-	my $subject = "$config{wikiname} update of ";
-	if (@changed_pages > 2) {
-		$subject .= "$changed_pages[0] $changed_pages[1] etc";
-	} else {
-		$subject .= join " ", @changed_pages;
-	}
-	$subject .= " by $user";
-
-	my $template = HTML::Template->new(
-		filename => "$config{templatedir}/notifymail.tmpl"
-	);
-	$template->param(
-		wikiname => $config{wikiname},
-		diff     => $diff,
-		user     => $user,
-		message  => $message,
-	);
-
-	eval q{use Mail::Sendmail};
-	foreach my $email (@email_recipients) {
-		sendmail(
-			To      => $email,
-			From    => "$config{wikiname} <$config{adminemail}>",
-			Subject => $subject,
-			Message => $template->output,
-		) or error("Failed to send update notification mail: $!");
-	}
-} #}}}
-
 sub rcs_getctime ($) { #{{{
-	# 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 = $ci->{'author_epoch'};
-	debug("ctime for '$file': ". localtime($ctime) . "\n");
+	debug("ctime for '$file': ". localtime($ctime));
 
 	return $ctime;
 } #}}}