From 03669d59ee98d609a039b7bace7bcf2a6e09cb13 Mon Sep 17 00:00:00 2001
From: Stefan Metzmacher <metze@samba.org>
Date: Thu, 20 Dec 2007 19:15:53 +0100
Subject: [PATCH] add code based on SVN and SVK perl bindings to get the full
 diffs out of subversion even for deleted subdirs

metze
---
 SVN2GitEditor.pm  | 316 ++++++++++++++++++++++++++++++++++++++++++++++
 SVN2GitPatch.pm   | 200 +++++++++++++++++++++++++++++
 sync-v4-0-test.pl |  69 ++++++++++
 3 files changed, 585 insertions(+)
 create mode 100644 SVN2GitEditor.pm
 create mode 100644 SVN2GitPatch.pm
 create mode 100755 sync-v4-0-test.pl

diff --git a/SVN2GitEditor.pm b/SVN2GitEditor.pm
new file mode 100644
index 0000000..c3416c4
--- /dev/null
+++ b/SVN2GitEditor.pm
@@ -0,0 +1,316 @@
+
+package SVN2GitEditor;
+use strict;
+
+require SVN::Delta;
+our @ISA = qw(SVN::Delta::Editor);
+
+use SVK::I18N;
+use SVK::XD;
+use autouse 'SVK::Util' => qw( slurp_fh tmpfile mimetype_is_text catfile );
+
+sub set_target_revision {
+    my ($self, $revision) = @_;
+}
+
+sub open_root {
+    my ($self, $baserev) = @_;
+    return '';
+}
+
+sub add_file {
+    my ($self, $path, $pdir, $from_path, $from_rev, $pool) = @_;
+    $self->{info}{$path}{fpool} = $pool;
+    if (defined $from_path) {
+	$self->{info}{$path}{from_path} = $from_path;
+	$self->{info}{$path}{copied} = 1;
+    } else {
+	$self->{info}{$path}{added} = 1;
+    }
+    return $path;
+}
+
+sub open_file {
+    my ($self, $path, $pdir, $rev, $pool) = @_;
+    $self->{info}{$path}{from_path} = $path;
+    $self->{info}{$path}{fpool} = $pool;
+
+    return $path;
+}
+
+sub apply_textdelta {
+    my ($self, $path, $checksum, $pool) = @_;
+    return unless $path;
+    my $info = $self->{info}{$path};
+    $info->{base} = $self->{cb_basecontent}($info->{from_path}, $info->{fpool})
+	unless $info->{added};
+
+    unless ($self->{external}) {
+	my $newtype = $info->{prop} && $info->{prop}{'svn:mime-type'};
+	my $is_text = !$newtype || mimetype_is_text ($newtype);
+	if ($is_text && !$info->{added}) {
+	    my $basetype = $self->{cb_baseprop}->($info->{from_path}, 'svn:mime-type', $pool);
+	    $is_text = !$basetype || mimetype_is_text ($basetype);
+	}
+	unless ($is_text) {
+            confess("Cannot display: file marked as a binary type.\n");
+	}
+    }
+    my $new;
+    if ($self->{external}) {
+	my $tmp = tmpfile ('diff');
+	slurp_fh ($info->{base}, $tmp)
+	    if $info->{base};
+	seek $tmp, 0, 0;
+	$info->{base} = $tmp;
+	$info->{new} = $new = tmpfile ('diff');
+    }
+    else {
+	$info->{new} = '';
+	open $new, '>', \$info->{new};
+    }
+
+    return [SVN::TxDelta::apply ($info->{base}, $new,
+				 undef, undef, $pool)];
+}
+
+sub close_file {
+    my ($self, $path, $checksum, $pool) = @_;
+    return unless $path;
+    my $info = $self->{info}{$path};
+
+    if (exists $info->{new}) {
+	no warnings 'uninitialized';
+	my $rpath = $self->{report} ? catfile($self->{report}, $path) : $path;
+	my @label = map { $self->{$_} || $self->{"cb_$_"}->($path) } qw/llabel rlabel/;
+	my $showpath = ($self->{lpath} ne $self->{rpath});
+	my @showpath = map { $showpath ? $self->{$_} : undef } qw/lpath rpath/;
+	if ($self->{external}) {
+	    # XXX: the 2nd file could be - and save some disk IO
+	    my @content = map { ($info->{$_}->filename) } qw/base new/;
+	    @content = reverse @content if $self->{reverse};
+	    (system (split (/ /, $self->{external}),
+		    '-L', _full_label ($rpath, $showpath[0], $label[0]),
+		    $content[0],
+		    '-L', _full_label ($rpath, $showpath[1], $label[1]),
+		    $content[1]) >= 0) or die loc("Could not run %1: %2", $self->{external}, $?);
+	}
+	else {
+	    $info->{base} = '';
+	    $info->{base} = $self->{cb_basecontent}($info->{from_path}, $info->{fpool})
+		unless $info->{added};
+	    my @content = ($info->{base}, \$self->{info}{$path}{new});
+	    @content = reverse @content if $self->{reverse};
+	    $self->output_diff ($rpath, @label, @showpath, @content);
+	}
+    }
+
+#    $self->output_prop_diff ($path, $pool);
+    delete $self->{info}{$path};
+}
+
+sub oldfilemode
+{
+	my ($self, $path) = @_;
+	my $pool = $self->{info}{$path}{fpool};
+	my $name = "svn:executable";
+
+	return undef if $self->{info}{$path}{added};
+
+	my $prop = $self->{cb_baseprop}->($path, $name, $pool);
+
+	return "100644" unless defined($prop);
+
+	return "100755" if (length($prop) > 0);
+
+	return "100644";
+}
+
+sub newfilemode
+{
+	my ($self, $path, $oldmode) = @_;
+	my $name = "svn:executable";
+
+	return undef if $self->{info}{$path}{deleted};
+
+	my $prop = $self->{info}{$path}{prop}{$name};
+
+	my $changed = 1;
+	$changed = 0 unless defined($prop);
+	$changed = 1 if $self->{info}{$path}{added};
+
+	return $oldmode unless $changed;
+
+	return "100755" if (length($prop) > 0);
+
+	return "100644";
+}
+
+sub output_diff {
+    my ($self, $path, $llabel, $rlabel, $lpath, $rpath) = splice(@_, 0, 6);
+    my $fh = $self->_output_fh;
+
+    my $ofile = $self->{info}{$path}{added} ? "/dev/null": "a/$path";
+    my $nfile = $self->{info}{$path}{deleted} ? "/dev/null": "b/$path";
+
+    my $osha1 = $self->{info}{$path}{added} ? "0000000": "1234567";
+    my $nsha1 = $self->{info}{$path}{deleted} ? "0000000": "7654321";
+
+    my $omode = $self->oldfilemode($path);
+    my $nmode = $self->newfilemode($path, $omode);
+
+    my $name = "";
+    my $mode = "";
+
+    if (defined($omode) and defined($nmode)) {
+	if ($omode ne $nmode) {
+	    $mode = "old mode $omode\nnew mode $nmode\n";
+	}
+    } elsif (defined($omode)) {
+	$mode = "deleted file mode $omode\n";
+    } elsif (defined($nmode)) {
+	$mode = "new file mode $nmode\n";
+    }
+
+    print $fh (
+	"diff --git $ofile $nfile\n",
+	$name,
+	$mode,
+	"index $osha1..$nsha1\n"
+    );
+
+    unshift @_, $self->_output_fh;
+    push @_, $ofile, $nfile;
+
+    goto &{$self->can('_output_diff_content')};
+}
+
+# _output_diff_content($fh, $ltext, $rtext, $llabel, $rlabel)
+sub _output_diff_content {
+    my ($fh, $ltext, $rtext, $llabel, $rlabel) = @_;
+
+    my ($lfh, $lfn) = tmpfile ('diff');
+    my ($rfh, $rfn) = tmpfile ('diff');
+
+    slurp_fh ($ltext => $lfh); close ($lfh);
+    slurp_fh ($rtext => $rfh); close ($rfh);
+
+    my $diff = SVN::Core::diff_file_diff( $lfn, $rfn );
+
+    SVN::Core::diff_file_output_unified(
+        $fh, $diff, $lfn, $rfn, $llabel, $rlabel
+    );
+
+    unlink ($lfn, $rfn);
+}
+
+sub output_prop_diff {
+    my ($self, $path, $pool) = @_;
+    if ($self->{info}{$path}{prop}) {
+	my $rpath = $self->{report} ? catfile($self->{report}, $path) : $path;
+	$self->_print("\n", loc("Property changes on: %1\n", $rpath), ('_' x 67), "\n");
+	for (sort keys %{$self->{info}{$path}{prop}}) {
+	    $self->_print(loc("Name: %1\n", $_));
+	    my $baseprop;
+	    $baseprop = $self->{cb_baseprop}->($path, $_, $pool)
+		unless $self->{info}{$path}{added};
+            my @args =
+                map \$_,
+                map { (length || /\n$/) ? "$_\n" : $_ }
+                    ($baseprop||''), ($self->{info}{$path}{prop}{$_}||'');
+            @args = reverse @args if $self->{reverse};
+
+            my $diff = '';
+            open my $fh, '>', \$diff;
+            _output_diff_content($fh, @args, '', '');
+            $diff =~ s/.*\n.*\n//;
+            $diff =~ s/^\@.*\n//mg;
+            $diff =~ s/^/ /mg;
+            $self->_print($diff);
+	}
+	$self->_print("\n");
+    }
+}
+
+sub add_directory {
+    my ($self, $path, $pdir, @arg) = @_;
+#    $self->{info}{$path}{added} = 1;
+    return $path;
+}
+
+sub open_directory {
+    my ($self, $path, $pdir, $rev, @arg) = @_;
+    return $path;
+}
+
+sub close_directory {
+    my ($self, $path, $pool) = @_;
+#    $self->output_prop_diff ($path, $pool);
+    delete $self->{info}{$path};
+}
+
+sub delete_entry {
+    my ($self, $path, $revision, $pdir, $pool) = @_;
+
+    my $fullpath = "$self->{branch}/$path";
+
+    if ($self->{oldroot}->is_file($fullpath)) {
+	$self->{info}{$path}{from_path} = $path;
+	$self->{info}{$path}{deleted} = 1;
+	$self->{info}{$path}{fpool} = $pool;
+	$self->{info}{$path}{new} = '';
+	$self->close_file($path, undef, $pool);
+	return;
+    }
+
+    my $entries = $self->{oldroot}->dir_entries($fullpath);
+    foreach my $c (keys %{$entries}) {
+	$self->delete_entry("$path/$c", $revision, $path, $pool);
+    }
+}
+
+sub change_file_prop {
+    my ($self, $path, $name, $value) = @_;
+    $self->{info}{$path}{prop}{$name} = $value;
+}
+
+sub change_dir_prop {
+    my ($self, $path, $name, $value) = @_;
+}
+
+sub close_edit {
+    my ($self, @arg) = @_;
+}
+
+sub _print {
+    my $self = shift;
+    $self->{output} or return print @_;
+    ${ $self->{output} } .= $_ for @_;
+}
+
+sub _output_fh {
+    my $self = shift;
+
+    no strict 'refs';
+    $self->{output} or return \*{select()};
+
+    open my $fh, '>>', $self->{output};
+    return $fh;
+}
+
+=head1 AUTHORS
+
+Chia-liang Kao E<lt>clkao@clkao.orgE<gt>
+
+=head1 COPYRIGHT
+
+Copyright 2003-2005 by Chia-liang Kao E<lt>clkao@clkao.orgE<gt>.
+
+This program is free software; you can redistribute it and/or modify it
+under the same terms as Perl itself.
+
+See L<http://www.perl.com/perl/misc/Artistic.html>
+
+=cut
+
+1;
diff --git a/SVN2GitPatch.pm b/SVN2GitPatch.pm
new file mode 100644
index 0000000..d3f668a
--- /dev/null
+++ b/SVN2GitPatch.pm
@@ -0,0 +1,200 @@
+#!/usr/bin/perl
+#
+
+use strict;
+
+use util;
+
+use Time::Local;
+use Time::gmtime;
+
+use SVN::Core;
+use SVN::Repos;
+use SVN::Fs;
+use SVN::Delta;
+use SVN::Ra;
+use SVK::Editor::Diff;
+use SVN2GitEditor;
+
+package SVN2GitPatch;
+
+sub confess($)
+{
+	# TODO use confess from 'use Carp'
+	my ($str) = @_;
+	die($str);
+}
+
+sub new($$;$) {
+	my ($this, $repo_path, $authors_file) = @_;
+
+	my $self = undef;
+
+	$self->{repos} = SVN::Repos::open($repo_path);
+	$self->{authors_file} = $authors_file;
+	$self->{authors} = undef;
+
+	if (defined($self->{authors_file})) {
+		my $f = util::FileLoad($self->{authors_file});
+		my @lines = split("\n", $f);
+
+		foreach my $l (@lines) {
+			if ($l =~ /^([\w\-]+) = (.*)$/) {
+				$self->{authors}->{$1} = $2;
+				next;
+			}
+
+			confess "line: $l: invalid";
+		}
+	}
+
+	bless $self;
+	return $self;
+}
+
+sub get_last_svn_rev($$)
+{
+	my ($self, $branch) = @_;
+
+	return $self->{repos}->fs->youngest_rev();
+}
+
+sub get_missing_svn_revs($$$)
+{
+	my ($self, $branch, $start_rev) = @_;
+	my $last_rev = $self->get_last_svn_rev($branch);
+	my @ret = ();
+
+	my $nroot = $self->{repos}->fs->revision_root($last_rev);
+	my $hist = $nroot->node_history($branch);
+
+	while ($hist = $hist->prev(0)) {
+		my $crev = $hist->location();
+		last unless defined($crev);
+		last unless $crev > 0;
+		last unless $crev > $start_rev;
+
+		unshift(@ret, $crev);
+	}
+
+	return @ret;
+}
+
+sub svn2git_author($$)
+{
+	my ($self, $in) = @_;
+
+	my $out = $self->{authors}->{$in};
+
+	confess "author: $in:not found" unless defined($out);
+
+	return $out;
+}
+
+sub svn2git_date($$)
+{
+	my ($self, $in) = @_;
+
+
+	my $tmp = $in;
+	my $out = undef;
+
+	if ($tmp =~ /^(\d{4})-(\d{2})-(\d{2})T(\d{2}):(\d{2}):(\d{2})\./) {
+		my $year = $1;
+		my $month = $2-1;
+		my $day = $3;
+		my $hour = $4;
+		my $min = $5;
+		my $sec = $6;
+
+		use Time::Local;
+		use Time::gmtime;
+		my $time = timegm($sec, $min, $hour, $day, $month, $year);
+		$out = gmctime($time);
+	}
+
+	confess "time: $in: invalid" unless defined($out);
+
+	return $out;
+}
+
+sub svn2git_log($$)
+{
+	my ($self, $in) = @_;
+
+	my $out = $in;
+
+	return $out;
+}
+
+sub get_git_patch($$$)
+{
+	my ($self, $branch, $rev) = @_;
+	my $orev = $rev-1;
+	my $nrev = $rev;
+	my $oroot = $self->{repos}->fs->revision_root($orev);
+	my $nroot = $self->{repos}->fs->revision_root($nrev);
+
+	my $p = undef;
+
+	$p->{svn_author} = $self->{repos}->fs->revision_prop($nrev, "svn:author");
+	$p->{svn_date} = $self->{repos}->fs->revision_prop($nrev, "svn:date");
+	$p->{svn_log} = $self->{repos}->fs->revision_prop($nrev, "svn:log");
+	$p->{svn_rev} = $rev;
+
+	$p->{git_diff} = "";
+	my $editor = SVN2GitEditor->new(
+		cb_basecontent => sub {
+			my ($path, $pool) = @_;
+			my $base = $oroot->file_contents("$branch/$path", $pool);
+			return $base;
+		},
+		cb_baseprop => sub {
+			my ($path, $pname, $pool) = @_;
+			my $prop = $oroot->node_prop("$branch/$path", $pname, $pool);
+			return $prop;
+		},
+		oldrepos => $self->{repos},
+		oldroot => $oroot,
+		oldrev => $orev,
+		newrev => $nrev,
+		branch => $branch,
+		llabel => "revision $orev",
+		rlabel => "revision $nrev",
+		external => undef,
+		output => \$p->{git_diff});
+
+	SVN::Repos::dir_delta($oroot,
+			      $branch, '',
+			      $nroot,
+			      $branch,
+			      $editor,
+			      undef, 1, 1, 1, 1);
+
+	$p->{git_author} = $self->svn2git_author($p->{svn_author});
+	$p->{git_date} = $self->svn2git_date($p->{svn_date});
+	$p->{git_log} = $self->svn2git_log($p->{svn_log});
+
+	my @log = split("\n", $p->{git_log});
+	$p->{git_log_subject} = shift @log;
+	$p->{git_log_body} = join("\n", @log);
+
+	if ($p->{git_diff} eq "") {
+		$p->{git_patch} = undef;
+		return $p;
+	}
+
+	$p->{git_patch} = "";
+	$p->{git_patch} .= "From 123456789abcdef\n";
+	$p->{git_patch} .= "From: $p->{git_author}\n";
+	$p->{git_patch} .= "Date: $p->{git_date}\n";
+	$p->{git_patch} .= "Subject: [PATCH] r$p->{svn_rev}: $p->{git_log_subject}\n";
+	$p->{git_patch} .= "$p->{git_log_body}\n";
+	$p->{git_patch} .= "\n";
+	$p->{git_patch} .= $p->{git_diff};
+	$p->{git_patch} .= "\n---\nsvn-sync script\n\n";
+
+	return $p;
+}
+
+1;
diff --git a/sync-v4-0-test.pl b/sync-v4-0-test.pl
new file mode 100755
index 0000000..8aa68de
--- /dev/null
+++ b/sync-v4-0-test.pl
@@ -0,0 +1,69 @@
+#!/usr/bin/perl
+#
+
+use strict;
+
+use SVN2GitPatch;
+
+my $svn_repo_path = "/tmp/svn/samba.repo";
+my $authors_file = "svn-authors";
+
+my $svn_branch = "branches/SAMBA_4_0";
+my $svn_start_rev = 25600;
+my $basepath = "/tmp/svn/v4-0-test";
+
+my $patch_path = "$basepath/patches";
+my $last_svn_rev_file = "$patch_path/latest.svnrev";
+my $git_repo_path = "$basepath/git";
+
+$ENV{LANG} = "en_US.UTF-8";
+
+my $r = SVN2GitPatch->new($svn_repo_path, $authors_file);
+
+sub get_last_svn_rev($;$)
+{
+	my ($file, $default_rev) = @_;
+
+	$default_rev = 25600 if defined($default_rev);
+
+	my $v = util::FileLoad($file);
+
+	$v = $default_rev if $v eq "";
+	$v *= 1;
+	$v = $default_rev if $v == 0;
+
+	print "get_last_svn_rev: $v\n";
+	return $v;
+}
+
+sub set_last_svn_rev($$)
+{
+	my ($file, $rev) = @_;
+	util::FileSave($file, $rev);
+	print "set_last_svn_rev: $rev\n";
+}
+
+my $last_rev = $r->get_last_svn_rev($svn_branch);
+my $start_rev = get_last_svn_rev($last_svn_rev_file, $svn_start_rev);
+my @revs = $r->get_missing_svn_revs($svn_branch, $start_rev);
+
+print "start: $start_rev last: $last_rev\n";
+
+foreach my $rev (@revs) {
+
+	print "Get patch for rev: $rev\n";
+	my $p = $r->get_git_patch($svn_branch, $rev);
+
+	next unless defined($p->{git_patch});
+
+	my $patch_path = "$patch_path/$rev.patch";
+	open(PATCH, ">$patch_path") or die ("failed to $patch_path for write");
+	print PATCH $p->{git_patch}."\n";
+	close PATCH;
+
+	my $applycmd = "cd $git_repo_path && git am --whitespace=nowarn --binary $patch_path";
+
+	my $apply = `$applycmd` or die "$applycmd: failed";
+
+	set_last_svn_rev($last_svn_rev_file, $rev);
+}
-- 
2.34.1