sample update-hook: sanely handle a new branch head.
[gitweb.git] / git-svnimport.perl
index 20a85724cb1660a3411fa34717cd5f1e98ebd731..6e3a44a2bc9fbf170db45a7e390d6ac8ffbb6f02 100755 (executable)
@@ -10,7 +10,7 @@
 # The head revision is on branch "origin" by default.
 # You can change that with the '-o' option.
 
-require v5.8.0; # for shell-safe open("-|",LIST)
+require 5.008; # for shell-safe open("-|",LIST)
 use strict;
 use warnings;
 use Getopt::Std;
@@ -25,7 +25,7 @@
 use SVN::Core;
 use SVN::Ra;
 
-die "Need CVN:Core 1.2.1 or better" if $SVN::Core::VERSION lt "1.2.1";
+die "Need SVN:Core 1.2.1 or better" if $SVN::Core::VERSION lt "1.2.1";
 
 $SIG{'PIPE'}="IGNORE";
 $ENV{'TZ'}="UTC";
@@ -34,8 +34,8 @@
 
 sub usage() {
        print STDERR <<END;
-Usage: ${\basename $0}     # fetch/update GIT from CVS
-       [-o branch-for-HEAD] [-h] [-v] [-l max_num_changes]
+Usage: ${\basename $0}     # fetch/update GIT from SVN
+       [-o branch-for-HEAD] [-h] [-v] [-l max_rev]
        [-C GIT_repository] [-t tagname] [-T trunkname] [-b branchname]
        [-d|-D] [-i] [-u] [-s start_chg] [-m] [-M regex] [SVN_URL]
 END
@@ -53,7 +53,6 @@ END
 
 $opt_o ||= "origin";
 $opt_s ||= 1;
-$opt_l = 100 unless defined $opt_l;
 my $git_tree = $opt_C;
 $git_tree ||= ".";
 
@@ -97,8 +96,10 @@ sub new {
 sub conn {
        my $self = shift;
        my $repo = $self->{'fullrep'};
-       my $s = SVN::Ra->new($repo);
-
+       my $auth = SVN::Core::auth_open ([SVN::Client::get_simple_provider,
+                         SVN::Client::get_ssl_server_trust_file_provider,
+                         SVN::Client::get_username_provider]);
+       my $s = SVN::Ra->new(url => $repo, auth => $auth);
        die "SVN connection to $repo: $!\n" unless defined $s;
        $self->{'svn'} = $s;
        $self->{'repo'} = $repo;
@@ -112,7 +113,9 @@ sub file {
                    DIR => File::Spec->tmpdir(), UNLINK => 1);
 
        print "... $rev $path ...\n" if $opt_v;
-       eval { $self->{'svn'}->get_file($path,$rev,$fh); };
+       my $pool = SVN::Pool->new();
+       eval { $self->{'svn'}->get_file($path,$rev,$fh,$pool); };
+       $pool->clear;
        if($@) {
                return undef if $@ =~ /Attempted to get checksum/;
                die $@;
@@ -125,8 +128,9 @@ sub file {
 package main;
 use URI;
 
-my $svn = $svn_url;
+our $svn = $svn_url;
 $svn .= "/$svn_dir" if defined $svn_dir;
+my $svn2 = SVNconn->new($svn);
 $svn = SVNconn->new($svn);
 
 my $lwp_ua;
@@ -197,7 +201,7 @@ ($$)
 my $maxnum = 0;
 my $last_rev = "";
 my $last_branch;
-my $current_rev = $opt_s-1;
+my $current_rev = $opt_s || 1;
 unless(-d $git_dir) {
        system("git-init-db");
        die "Cannot init the GIT db at $git_tree: $?\n" if $?;
@@ -215,7 +219,11 @@ ($$)
        -f "$git_dir/svn2git"
                or die "'$git_dir/svn2git' does not exist.\n".
                       "You need that file for incremental imports.\n";
-       $last_branch = basename(readlink("$git_dir/HEAD"));
+       open(F, "git-symbolic-ref HEAD |") or
+               die "Cannot run git-symbolic-ref: $!\n";
+       chomp ($last_branch = <F>);
+       $last_branch = basename($last_branch);
+       close(F);
        unless($last_branch) {
                warn "Cannot read the last branch name: $! -- assuming 'master'\n";
                $last_branch = "master";
@@ -249,7 +257,7 @@ ($$)
                my($num,$branch,$ref) = split;
                $branches{$branch}{$num} = $ref;
                $branches{$branch}{"LAST"} = $ref;
-               $current_rev = $num if $current_rev < $num;
+               $current_rev = $num+1 if $current_rev <= $num;
        }
        close($B);
 }
@@ -258,10 +266,17 @@ ($$)
 
 open BRANCHES,">>", "$git_dir/svn2git";
 
-sub get_file($$$) {
-       my($rev,$branch,$path) = @_;
+sub node_kind($$$) {
+       my ($branch, $path, $revision) = @_;
+       my $pool=SVN::Pool->new;
+       my $kind = $svn->{'svn'}->check_path(revert_split_path($branch,$path),$revision,$pool);
+       $pool->clear;
+       return $kind;
+}
+
+sub revert_split_path($$) {
+       my($branch,$path) = @_;
 
-       # revert split_path(), below
        my $svnpath;
        $path = "" if $path eq "/"; # this should not happen, but ...
        if($branch eq "/") {
@@ -272,6 +287,15 @@ ($$$)
                $svnpath = "$branch_name/$branch/$path";
        }
 
+       $svnpath =~ s#/+$##;
+       return $svnpath;
+}
+
+sub get_file($$$) {
+       my($rev,$branch,$path) = @_;
+
+       my $svnpath = revert_split_path($branch,$path);
+
        # now get it
        my $name;
        if($opt_d) {
@@ -319,28 +343,61 @@ ($$)
        } elsif($path =~ s#^/\Q$branch_name\E/([^/]+)/?##) {
                $branch = $1;
        } else {
-               print STDERR "$rev: Unrecognized path: $path\n";
+               my %no_error = (
+                       "/" => 1,
+                       "/$tag_name" => 1,
+                       "/$branch_name" => 1
+               );
+               print STDERR "$rev: Unrecognized path: $path\n" unless (defined $no_error{$path});
                return ()
        }
        $path = "/" if $path eq "";
        return ($branch,$path);
 }
 
-sub copy_subdir($$$$$$) {
+sub branch_rev($$) {
+
+       my ($srcbranch,$uptorev) = @_;
+
+       my $bbranches = $branches{$srcbranch};
+       my @revs = reverse sort { ($a eq 'LAST' ? 0 : $a) <=> ($b eq 'LAST' ? 0 : $b) } keys %$bbranches;
+       my $therev;
+       foreach my $arev(@revs) {
+               next if  ($arev eq 'LAST');
+               if ($arev <= $uptorev) {
+                       $therev = $arev;
+                       last;
+               }
+       }
+       return $therev;
+}
+
+sub copy_path($$$$$$$$) {
        # Somebody copied a whole subdirectory.
        # We need to find the index entries from the old version which the
        # SVN log entry points to, and add them to the new place.
 
-       my($newrev,$newbranch,$path,$oldpath,$rev,$new) = @_;
-       my($branch,$srcpath) = split_path($rev,$oldpath);
+       my($newrev,$newbranch,$path,$oldpath,$rev,$node_kind,$new,$parents) = @_;
 
-       my $gitrev = $branches{$branch}{$rev};
+       my($srcbranch,$srcpath) = split_path($rev,$oldpath);
+       unless(defined $srcbranch) {
+               print "Path not found when copying from $oldpath @ $rev\n";
+               return;
+       }
+       my $therev = branch_rev($srcbranch, $rev);
+       my $gitrev = $branches{$srcbranch}{$therev};
        unless($gitrev) {
                print STDERR "$newrev:$newbranch: could not find $oldpath \@ $rev\n";
                return;
        }
-       print "$newrev:$newbranch:$path: copying from $branch:$srcpath @ $rev\n" if $opt_v;
-       $srcpath =~ s#/*$#/#;
+       if ($srcbranch ne $newbranch) {
+               push(@$parents, $branches{$srcbranch}{'LAST'});
+       }
+       print "$newrev:$newbranch:$path: copying from $srcbranch:$srcpath @ $rev\n" if $opt_v;
+       if ($node_kind eq $SVN::Node::dir) {
+                       $srcpath =~ s#/*$#/#;
+       }
+       
        open my $f,"-|","git-ls-tree","-r","-z",$gitrev,$srcpath;
        local $/ = "\0";
        while(<$f>) {
@@ -348,9 +405,12 @@ ($$$$$$)
                my($m,$p) = split(/\t/,$_,2);
                my($mode,$type,$sha1) = split(/ /,$m);
                next if $type ne "blob";
-               $p = substr($p,length($srcpath)-1);
-               print "... found $path$p ...\n" if $opt_v;
-               push(@$new,[$mode,$sha1,$path.$p]);
+               if ($node_kind eq $SVN::Node::dir) {
+                       $p = $path . substr($p,length($srcpath)-1);
+               } else {
+                       $p = $path;
+               }
+               push(@$new,[$mode,$sha1,$p]);   
        }
        close($f) or
                print STDERR "$newrev:$newbranch: could not list files in $oldpath \@ $rev\n";
@@ -359,7 +419,7 @@ ($$$$$$)
 sub commit {
        my($branch, $changed_paths, $revision, $author, $date, $message) = @_;
        my($author_name,$author_email,$dest);
-       my(@old,@new);
+       my(@old,@new,@parents);
 
        if (not defined $author) {
                $author_name = $author_email = "unknown";
@@ -446,6 +506,8 @@ sub commit {
                $last_rev = $rev;
        }
 
+       push (@parents, $rev) if defined $rev;
+
        my $cid;
        if($tag and not %$changed_paths) {
                $cid = $rev;
@@ -454,39 +516,31 @@ sub commit {
                foreach my $path(@paths) {
                        my $action = $changed_paths->{$path};
 
-                       if ($action->[0] eq "A") {
-                               my $f = get_file($revision,$branch,$path);
-                               if($f) {
-                                       push(@new,$f) if $f;
-                               } elsif($action->[1]) {
-                                       copy_subdir($revision,$branch,$path,$action->[1],$action->[2],\@new);
-                               } else {
-                                       my $opath = $action->[3];
-                                       print STDERR "$revision: $branch: could not fetch '$opath'\n";
+                       if ($action->[0] eq "R") {
+                               # refer to a file/tree in an earlier commit
+                               push(@old,$path); # remove any old stuff
+                       }
+                       if(($action->[0] eq "A") || ($action->[0] eq "R")) {
+                               my $node_kind = node_kind($branch,$path,$revision);
+                               if($action->[1]) {
+                                       copy_path($revision,$branch,$path,$action->[1],$action->[2],$node_kind,\@new,\@parents);
+                               } elsif ($node_kind eq $SVN::Node::file) {
+                                       my $f = get_file($revision,$branch,$path);
+                                       if ($f) {
+                                               push(@new,$f) if $f;
+                                       } else {
+                                               my $opath = $action->[3];
+                                               print STDERR "$revision: $branch: could not fetch '$opath'\n";
+                                       }
                                }
                        } elsif ($action->[0] eq "D") {
                                push(@old,$path);
                        } elsif ($action->[0] eq "M") {
-                               my $f = get_file($revision,$branch,$path);
-                               push(@new,$f) if $f;
-                       } elsif ($action->[0] eq "R") {
-                               # refer to a file/tree in an earlier commit
-                               push(@old,$path); # remove any old stuff
-
-                               # ... and add any new stuff
-                               my($b,$srcpath) = split_path($revision,$action->[1]);
-                               $srcpath =~ s#/*$#/#;
-                               open my $F,"-|","git-ls-tree","-r","-z", $branches{$b}{$action->[2]}, $srcpath;
-                               local $/ = "\0";
-                               while(<$F>) {
-                                       chomp;
-                                       my($m,$p) = split(/\t/,$_,2);
-                                       my($mode,$type,$sha1) = split(/ /,$m);
-                                       next if $type ne "blob";
-                                       $p = substr($p,length($srcpath)-1);
-                                       push(@new,[$mode,$sha1,$path.$p]);
+                               my $node_kind = node_kind($branch,$path,$revision);
+                               if ($node_kind eq $SVN::Node::file) {
+                                       my $f = get_file($revision,$branch,$path);
+                                       push(@new,$f) if $f;
                                }
-                               close($F);
                        } else {
                                die "$revision: unknown action '".$action->[0]."' for $path\n";
                        }
@@ -554,7 +608,6 @@ sub commit {
                        $pw->close();
 
                        my @par = ();
-                       @par = ("-p",$rev) if defined $rev;
 
                        # loose detection of merges
                        # based on the commit msg
@@ -564,11 +617,17 @@ sub commit {
                                        if ($mparent eq 'HEAD') { $mparent = $opt_o };
                                        if ( -e "$git_dir/refs/heads/$mparent") {
                                                $mparent = get_headref($mparent, $git_dir);
-                                               push @par, '-p', $mparent;
+                                               push (@parents, $mparent);
                                                print OUT "Merge parent branch: $mparent\n" if $opt_v;
                                        }
                                }
                        }
+                       my %seen_parents = ();
+                       my @unique_parents = grep { ! $seen_parents{$_} ++ } @parents;
+                       foreach my $bparent (@unique_parents) {
+                               push @par, '-p', $bparent;
+                               print OUT "Merge parent branch: $bparent\n" if $opt_v;
+                       }
 
                        exec("env",
                                "GIT_AUTHOR_NAME=$author_name",
@@ -600,6 +659,10 @@ sub commit {
                die "Error running git-commit-tree: $?\n" if $?;
        }
 
+       if (not defined $cid) {
+               $cid = $branches{"/"}{"LAST"};
+       }
+
        if(not defined $dest) {
                print "... no known parent\n" if $opt_v;
        } elsif(not $tag) {
@@ -616,6 +679,7 @@ sub commit {
                # the tag was 'complex', i.e. did not refer to a "real" revision
 
                $dest =~ tr/_/\./ if $opt_u;
+               $branch = $dest;
 
                my $pid = open2($in, $out, 'git-mktag');
                print $out ("object $cid\n".
@@ -647,17 +711,17 @@ sub commit {
        print "DONE: $revision $dest $cid\n" if $opt_v;
 }
 
-my ($changed_paths, $revision, $author, $date, $message, $pool) = @_;
-sub _commit_all {
-       ($changed_paths, $revision, $author, $date, $message, $pool) = @_;
+sub commit_all {
+       # Recursive use of the SVN connection does not work
+       local $svn = $svn2;
+
+       my ($changed_paths, $revision, $author, $date, $message, $pool) = @_;
        my %p;
        while(my($path,$action) = each %$changed_paths) {
                $p{$path} = [ $action->action,$action->copyfrom_path, $action->copyfrom_rev, $path ];
        }
        $changed_paths = \%p;
-}
 
-sub commit_all {
        my %done;
        my @col;
        my $pref;
@@ -673,16 +737,20 @@ sub commit_all {
        }
 }
 
-while(++$current_rev <= $svn->{'maxrev'}) {
-       $svn->{'svn'}->get_log("/",$current_rev,$current_rev,$current_rev,1,1,\&_commit_all,"");
-       commit_all();
-       if($opt_l and not --$opt_l) {
-               print STDERR "Stopping, because there is a memory leak (in the SVN library).\n";
-               print STDERR "Please repeat this command; it will continue safely\n";
-               last;
-       }
+$opt_l = $svn->{'maxrev'} if not defined $opt_l or $opt_l > $svn->{'maxrev'};
+
+if ($svn->{'maxrev'} < $current_rev) {
+    print "Up to date: no new revisions to fetch!\n" if $opt_v;
+    unlink("$git_dir/SVN2GIT_HEAD");
+    exit;
 }
 
+print "Fetching from $current_rev to $opt_l ...\n" if $opt_v;
+
+my $pool=SVN::Pool->new;
+$svn->{'svn'}->get_log("/",$current_rev,$opt_l,0,1,1,\&commit_all,$pool);
+$pool->clear;
+
 
 unlink($git_index);
 
@@ -706,8 +774,7 @@ sub commit_all {
        print "DONE; creating $orig_branch branch\n" if $opt_v and (not defined $opt_l or $opt_l > 0);
        system("cp","$git_dir/refs/heads/$opt_o","$git_dir/refs/heads/master")
                unless -f "$git_dir/refs/heads/master";
-       unlink("$git_dir/HEAD");
-       symlink("refs/heads/$orig_branch","$git_dir/HEAD");
+       system('git-update-ref', 'HEAD', "$orig_branch");
        unless ($opt_i) {
                system('git checkout');
                die "checkout failed: $?\n" if $?;