git-tar-tree: no more void pointer arithmetic
[git.git] / git-svnimport.perl
index 45d77c5..38ac732 100755 (executable)
 # The head revision is on branch "origin" by default.
 # You can change that with the '-o' option.
 
-require 5.008; # for shell-safe open("-|",LIST)
 use strict;
 use warnings;
 use Getopt::Std;
+use File::Copy;
 use File::Spec;
 use File::Temp qw(tempfile);
 use File::Path qw(mkpath);
@@ -30,19 +30,21 @@ die "Need SVN:Core 1.2.1 or better" if $SVN::Core::VERSION lt "1.2.1";
 $SIG{'PIPE'}="IGNORE";
 $ENV{'TZ'}="UTC";
 
-our($opt_h,$opt_o,$opt_v,$opt_u,$opt_C,$opt_i,$opt_m,$opt_M,$opt_t,$opt_T,$opt_b,$opt_s,$opt_l,$opt_d,$opt_D);
+our($opt_h,$opt_o,$opt_v,$opt_u,$opt_C,$opt_i,$opt_m,$opt_M,$opt_t,$opt_T,
+    $opt_b,$opt_r,$opt_I,$opt_A,$opt_s,$opt_l,$opt_d,$opt_D);
 
 sub usage() {
        print STDERR <<END;
 Usage: ${\basename $0}     # fetch/update GIT from SVN
-       [-o branch-for-HEAD] [-h] [-v] [-l max_num_changes]
+       [-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]
+       [-d|-D] [-i] [-u] [-r] [-I ignorefilename] [-s start_chg]
+       [-m] [-M regex] [-A author_file] [SVN_URL]
 END
        exit(1);
 }
 
-getopts("b:C:dDhil:mM:o:s:t:T:uv") or usage();
+getopts("A:b:C:dDhiI:l:mM:o:rs:t:T:uv") or usage();
 usage if $opt_h;
 
 my $tag_name = $opt_t || "tags";
@@ -61,10 +63,36 @@ my $svn_dir = $ARGV[1];
 
 our @mergerx = ();
 if ($opt_m) {
-       @mergerx = ( qr/\W(?:from|of|merge|merging|merged) (\w+)/i );
+       my $branch_esc = quotemeta ($branch_name);
+       my $trunk_esc  = quotemeta ($trunk_name);
+       @mergerx =
+       (
+               qr!\b(?:merg(?:ed?|ing))\b.*?\b((?:(?<=$branch_esc/)[\w\.\-]+)|(?:$trunk_esc))\b!i,
+               qr!\b(?:from|of)\W+((?:(?<=$branch_esc/)[\w\.\-]+)|(?:$trunk_esc))\b!i,
+               qr!\b(?:from|of)\W+(?:the )?([\w\.\-]+)[-\s]branch\b!i
+       );
 }
 if ($opt_M) {
-       push (@mergerx, qr/$opt_M/);
+       unshift (@mergerx, qr/$opt_M/);
+}
+
+# Absolutize filename now, since we will have chdir'ed by the time we
+# get around to opening it.
+$opt_A = File::Spec->rel2abs($opt_A) if $opt_A;
+
+our %users = ();
+our $users_file = undef;
+sub read_users($) {
+       $users_file = File::Spec->rel2abs(@_);
+       die "Cannot open $users_file\n" unless -f $users_file;
+       open(my $authors,$users_file);
+       while(<$authors>) {
+               chomp;
+               next unless /^(\S+?)\s*=\s*(.+?)\s*<(.+)>\s*$/;
+               (my $user,my $name,my $email) = ($1,$2,$3);
+               $users{$user} = [$name,$email];
+       }
+       close($authors);
 }
 
 select(STDERR); $|=1; select(STDOUT);
@@ -77,6 +105,7 @@ package SVNconn;
 use File::Spec;
 use File::Temp qw(tempfile);
 use POSIX qw(strftime dup2);
+use Fcntl qw(SEEK_SET);
 
 sub new {
        my($what,$repo) = @_;
@@ -96,8 +125,10 @@ 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;
@@ -111,23 +142,63 @@ sub file {
                    DIR => File::Spec->tmpdir(), UNLINK => 1);
 
        print "... $rev $path ...\n" if $opt_v;
+       my (undef, $properties);
        my $pool = SVN::Pool->new();
-       eval { $self->{'svn'}->get_file($path,$rev,$fh,$pool); };
+       eval { (undef, $properties)
+                  = $self->{'svn'}->get_file($path,$rev,$fh,$pool); };
        $pool->clear;
        if($@) {
                return undef if $@ =~ /Attempted to get checksum/;
                die $@;
        }
+       my $mode;
+       if (exists $properties->{'svn:executable'}) {
+               $mode = '100755';
+       } elsif (exists $properties->{'svn:special'}) {
+               my ($special_content, $filesize);
+               $filesize = tell $fh;
+               seek $fh, 0, SEEK_SET;
+               read $fh, $special_content, $filesize;
+               if ($special_content =~ s/^link //) {
+                       $mode = '120000';
+                       seek $fh, 0, SEEK_SET;
+                       truncate $fh, 0;
+                       print $fh $special_content;
+               } else {
+                       die "unexpected svn:special file encountered";
+               }
+       } else {
+               $mode = '100644';
+       }
        close ($fh);
 
-       return $name;
+       return ($name, $mode);
+}
+
+sub ignore {
+       my($self,$path,$rev) = @_;
+
+       print "... $rev $path ...\n" if $opt_v;
+       my (undef,undef,$properties)
+           = $self->{'svn'}->get_dir($path,$rev,undef);
+       if (exists $properties->{'svn:ignore'}) {
+               my ($fh, $name) = tempfile('gitsvn.XXXXXX',
+                                          DIR => File::Spec->tmpdir(),
+                                          UNLINK => 1);
+               print $fh $properties->{'svn:ignore'};
+               close($fh);
+               return $name;
+       } else {
+               return undef;
+       }
 }
 
 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;
@@ -198,7 +269,7 @@ $ENV{GIT_INDEX_FILE} = $git_index;
 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 $?;
@@ -254,13 +325,21 @@ EOM
                my($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);
 }
 -d $git_dir
        or die "Could not create git subdir ($git_dir).\n";
 
+my $default_authors = "$git_dir/svn-authors";
+if ($opt_A) {
+       read_users($opt_A);
+       copy($opt_A,$default_authors) or die "Copy failed: $!";
+} else {
+       read_users($default_authors) if -f $default_authors;
+}
+
 open BRANCHES,">>", "$git_dir/svn2git";
 
 sub node_kind($$$) {
@@ -294,7 +373,7 @@ sub get_file($$$) {
        my $svnpath = revert_split_path($branch,$path);
 
        # now get it
-       my $name;
+       my ($name,$mode);
        if($opt_d) {
                my($req,$res);
 
@@ -314,21 +393,53 @@ sub get_file($$$) {
                        return undef if $res->code == 301; # directory?
                        die $res->status_line." at $url\n";
                }
+               $mode = '0644'; # can't obtain mode via direct http request?
        } else {
-               $name = $svn->file("/$svnpath",$rev);
+               ($name,$mode) = $svn->file("$svnpath",$rev);
                return undef unless defined $name;
        }
 
-       open my $F, '-|', "git-hash-object", "-w", $name
+       my $pid = open(my $F, '-|');
+       die $! unless defined $pid;
+       if (!$pid) {
+           exec("git-hash-object", "-w", $name)
                or die "Cannot create object: $!\n";
+       }
        my $sha = <$F>;
        chomp $sha;
        close $F;
        unlink $name;
-       my $mode = "0644"; # SV does not seem to store any file modes
        return [$mode, $sha, $path];
 }
 
+sub get_ignore($$$$$) {
+       my($new,$old,$rev,$branch,$path) = @_;
+
+       return unless $opt_I;
+       my $svnpath = revert_split_path($branch,$path);
+       my $name = $svn->ignore("$svnpath",$rev);
+       if ($path eq '/') {
+               $path = $opt_I;
+       } else {
+               $path = File::Spec->catfile($path,$opt_I);
+       }
+       if (defined $name) {
+               my $pid = open(my $F, '-|');
+               die $! unless defined $pid;
+               if (!$pid) {
+                       exec("git-hash-object", "-w", $name)
+                           or die "Cannot create object: $!\n";
+               }
+               my $sha = <$F>;
+               chomp $sha;
+               close $F;
+               unlink $name;
+               push(@$new,['0644',$sha,$path]);
+       } else {
+               push(@$old,$path);
+       }
+}
+
 sub split_path($$) {
        my($rev,$path) = @_;
        my $branch;
@@ -395,7 +506,12 @@ sub copy_path($$$$$$$$) {
                        $srcpath =~ s#/*$#/#;
        }
        
-       open my $f,"-|","git-ls-tree","-r","-z",$gitrev,$srcpath;
+       my $pid = open my $f,'-|';
+       die $! unless defined $pid;
+       if (!$pid) {
+               exec("git-ls-tree","-r","-z",$gitrev,$srcpath)
+                       or die $!;
+       }
        local $/ = "\0";
        while(<$f>) {
                chomp;
@@ -420,6 +536,10 @@ sub commit {
 
        if (not defined $author) {
                $author_name = $author_email = "unknown";
+       } elsif (defined $users_file) {
+               die "User $author is not listed in $users_file\n"
+                   unless exists $users{$author};
+               ($author_name,$author_email) = @{$users{$author}};
        } elsif ($author =~ /^(.*?)\s+<(.*)>$/) {
                ($author_name, $author_email) = ($1, $2);
        } else {
@@ -519,9 +639,7 @@ sub commit {
                        }
                        if(($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 $SVN::Node::file) {
+                               if ($node_kind eq $SVN::Node::file) {
                                        my $f = get_file($revision,$branch,$path);
                                        if ($f) {
                                                push(@new,$f) if $f;
@@ -529,6 +647,16 @@ sub commit {
                                                my $opath = $action->[3];
                                                print STDERR "$revision: $branch: could not fetch '$opath'\n";
                                        }
+                               } elsif ($node_kind eq $SVN::Node::dir) {
+                                       if($action->[1]) {
+                                               copy_path($revision, $branch,
+                                                         $path, $action->[1],
+                                                         $action->[2], $node_kind,
+                                                         \@new, \@parents);
+                                       } else {
+                                               get_ignore(\@new, \@old, $revision,
+                                                          $branch, $path);
+                                       }
                                }
                        } elsif ($action->[0] eq "D") {
                                push(@old,$path);
@@ -537,29 +665,43 @@ sub commit {
                                if ($node_kind eq $SVN::Node::file) {
                                        my $f = get_file($revision,$branch,$path);
                                        push(@new,$f) if $f;
+                               } elsif ($node_kind eq $SVN::Node::dir) {
+                                       get_ignore(\@new, \@old, $revision,
+                                                  $branch,$path);
                                }
                        } else {
                                die "$revision: unknown action '".$action->[0]."' for $path\n";
                        }
                }
 
-               if(@old) {
-                       open my $F, "-|", "git-ls-files", "-z", @old or die $!;
-                       @old = ();
+               while(@old) {
+                       my @o1;
+                       if(@old > 55) {
+                               @o1 = splice(@old,0,50);
+                       } else {
+                               @o1 = @old;
+                               @old = ();
+                       }
+                       my $pid = open my $F, "-|";
+                       die "$!" unless defined $pid;
+                       if (!$pid) {
+                               exec("git-ls-files", "-z", @o1) or die $!;
+                       }
+                       @o1 = ();
                        local $/ = "\0";
                        while(<$F>) {
                                chomp;
-                               push(@old,$_);
+                               push(@o1,$_);
                        }
                        close($F);
 
-                       while(@old) {
+                       while(@o1) {
                                my @o2;
-                               if(@old > 55) {
-                                       @o2 = splice(@old,0,50);
+                               if(@o1 > 55) {
+                                       @o2 = splice(@o1,0,50);
                                } else {
-                                       @o2 = @old;
-                                       @old = ();
+                                       @o2 = @o1;
+                                       @o1 = ();
                                }
                                system("git-update-index","--force-remove","--",@o2);
                                die "Cannot remove files: $?\n" if $?;
@@ -640,6 +782,7 @@ sub commit {
                $pr->reader();
 
                $message =~ s/[\s\n]+\z//;
+               $message = "r$revision: $message" if $opt_r;
 
                print $pw "$message\n"
                        or die "Error writing to git-commit-tree: $!\n";
@@ -708,17 +851,17 @@ 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;
@@ -734,19 +877,20 @@ sub commit_all {
        }
 }
 
-while(++$current_rev <= $svn->{'maxrev'}) {
-       if (defined $opt_l) {
-               $opt_l--;
-               if ($opt_l < 0) {
-                       last;
-               }
-       }
-       my $pool=SVN::Pool->new;
-       $svn->{'svn'}->get_log("/",$current_rev,$current_rev,1,1,1,\&_commit_all,$pool);
-       $pool->clear;
-       commit_all();
+$opt_l = $svn->{'maxrev'} if not defined $opt_l or $opt_l > $svn->{'maxrev'};
+
+if ($opt_l < $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);