badcommit-fixup: reorganise arg parsing

[dgit-junk.git] / badcommit-fixup

diff --git a/badcommit-fixup b/badcommit-fixup
index 15b1720cf4d265028b5d9bfa73a5001b1e6fdf2c..ca2e0dfbdd28a8f02db50d38cd45244c06464d1b 100755 (executable)
--- a/badcommit-fixup
+++ b/badcommit-fixup
@@ -1,59 +1,179 @@
 #!/usr/bin/perl -w
 
+# usage:
+#   .../badcommit-fixup [-- --test
+#   .../badcommit-fixup --real
+
 use strict;
 
-set -e
-set -o pipefail
-
-tmp=.git/dgit-badcommit-fixup-tmp
-rm -rf $tmp
-mkdir $tmp
-
-LC_MESSAGES=C git fsck --no-dangling >$tmp/gfo 2>&1 || test $? = 1
-
-perl -ne '
-       print $1, "\n" or die $! if
- m/^error in commit (\w+):.*invalid format - expected '\''committer/;
-' <$tmp/gfo >$tmp/bad
-
-case `wc -l <$tmp/bad` in
-    0)
-       echo >&2 'nothing bad found - is git-fsck doing as we expect?' ;
-       exit 8 ;;
-    1)
-       read <$tmp/bad bads
-       nots="^$bads^0 ^$bads^1"
-       ;;
-    *)
-       bads="cat $tmp/bad"
-       ;;
-esac
-
-args="$nots"
-
-refs=`git-for-each-ref --format='%(refname)'`
-
-if git-symbolic-ref HEAD >/dev/null 2>&1; then
-    refs+=' HEAD'
-fi
-
-for head in $refs; do
-    exec <$tmp/bad
-    needed=false
-    for bad in $bads; do
-       if git merge-base --is-ancestor $bad $head; then
-           needed=true
-           break
-       fi
-    done 
-    if ! $needed; then continue; fi
-    args+=" $head"
-done
-
-git filter-branch --original dgit-badcommit --commit-filter '
-echo >&2 "FOO $*"
-cat >&2
-echo >&2 ====
-false
-    sed -e '\''1,/^$/ s/^commiter /committer /'\''
-' $args
+use POSIX;
+use IPC::Open2;
+use Data::Dumper;
+
+my $real;
+
+foreach my $a (@ARGV) {
+    if ($a eq '--test') {
+       $real = 0;
+    } elsif ($a eq '--real') {
+       $real = 1;
+    } else {
+       die "$a ?";
+    }
+}
+
+die unless defined $real;
+
+my $gcfpid = open2 \*GCFO, \*GCFI, 'git cat-file --batch' or die $!;
+
+our %count;
+
+no warnings qw(recursion);
+
+sub getobj ($$) {
+    my ($obj, $type) = @_;
+    print GCFI $obj, "\n" or die $!;
+    my $x = <GCFO>;
+    my ($gtype, $gsize) = $x =~ m/^\w+ (\w+) (\d+)\n/ or die "$obj ?";
+    $gtype eq $type or die "$obj $gtype != $type ?";
+    my $gdata;
+    (read GCFO, $gdata, $gsize) == $gsize or die "$obj $!";
+#print STDERR ">$obj|$x|$gdata|$gsize<\n";
+    $x = <GCFO>;
+    $x eq "\n" or die "$obj ($_) $!";
+    $count{inspected}++;
+    return $gdata;
+}
+
+sub hashobj ($$) {
+    my ($data,$type) = @_;
+    my $gwopid = open2 \*GWO, \*GWI,
+       "git hash-object -w -t $type --stdin"
+       or die $!;
+    print GWI $data or die $!;
+    close GWI or die $!;
+    $_ = <GWO>;
+    close GWO or die $!;
+    waitpid $gwopid,0 == $gwopid or die $!;
+    die $? if $?;
+    m/^(\w+)\n/ or die "$_ ?";
+    $count{"rewritten $type"}++;
+    return $1;
+}
+
+our %memo;
+
+sub rewrite_commit ($);
+sub rewrite_commit ($) {
+    my ($obj) = @_;
+    my $m = \ $memo{$obj};
+    return $$m if defined $$m;
+    my $olddata = getobj $obj, 'commit';
+    $olddata =~ m/(?<=\n)(?=\n)/ or die "$obj ?";
+    my $msg = $';
+    $_ = $`;
+    s{^(parent )(\w+)$}{ $1 . rewrite_commit($2) }gme;
+    $count{'fix overwrite'} += s{^commiter }{committer }gm;
+    if (!m{^author }m && !m{^committer }m) {
+       m{^parent (\w+)}m or die "$obj ?";
+       my $parent = getobj $1, 'commit';
+       $parent =~ m/^(?:.+\n)+(author .*\ncommitter .*\n)/;
+       m/\n$/ or die "$obj ?";
+       $_ .= $1;
+       $count{'fix import'}++;
+    }
+    my $newdata = $_.$msg;
+    my $newobj;
+    if ($newdata eq $olddata) {
+       $newobj = $obj;
+       $count{unchanged}++;
+    } else {
+       $newobj = hashobj $newdata, 'commit';
+    }
+    $$m= $newobj;
+    return $newobj;
+}
+
+sub rewrite_tag ($) {
+    my ($obj) = @_;
+    $_ = getobj $obj, 'tag';
+    m/^type (\w+)\n/m or die "$obj ?";
+    if ($1 ne 'commit') {
+       $count{"oddtags $1"}++;
+       return $obj;
+    }
+    m/^object (\w+)\n/m or die "$obj ?";
+    my $oldref = $1;
+    my $newref = rewrite_commit $oldref;
+    if ($oldref eq $newref) {
+       return $obj;
+    }
+    s/^(object )\w+$/ $1.$newref /me or die "$obj ($_) ?";
+    s/^-----BEGIN PGP SIGNATURE-----\n.*^-----END PGP SIGNATURE-----\n$//sm;
+    return hashobj $_, 'tag';
+}
+
+$!=0; $?=0;
+my $refs=`git for-each-ref`;
+die "$? $!" if $?;
+
+chomp $refs;
+
+our @updates;
+
+foreach my $rline (split /\n/, $refs) {
+    my ($obj, $type, $refname) = 
+       $rline =~ m/^(\w+)\s+(\w+)\s+(\S.*)/
+       or die "$_ ?";
+    my $rewrite;
+    if ($type eq 'commit') {
+       $rewrite = rewrite_commit($obj);
+    } elsif ($type eq 'tag') {
+       $rewrite = rewrite_tag($obj);
+    } else {
+       warn "ref $refname refers to $type\n";
+       next;
+    }
+    next if $rewrite eq $obj;
+    push @updates, [ $refname, $obj, $rewrite ];
+}
+
+open U, "|git update-ref -m 'dgit bad commit fixup' --stdin" or die $!;
+
+if ($real) {
+    $!=0; $?=0;
+    my $bare = `git rev-parse --is-bare-repository`;
+    die "$? $!" if $?;
+    if ($bare eq 'false') {
+       print "detaching your HEAD\n" or die $!;
+       system 'git checkout --detach' and die "$! $?";
+    }
+}
+
+for my $up (@updates) {
+    my ($ref, $old, $new) = @$up;
+    my $otherref = $ref;
+    $otherref =~ s{^refs/}{};
+    if ($real) {
+       print U <<END or die $!;
+create refs/dgit-badcommit/$otherref $old
+update $ref $new $old
+END
+    } else {
+       print U <<END or die $!;
+update refs/dgit-badfixuptest/$otherref $new
+END
+    }
+}
+
+$?=0; $!=0;
+close U or die "$? $!";
+die $? if $?;
+
+print Dumper(\%count);
+
+if ($real) {
+    print "old values saved in refs/dgit-badcommit/\n" or die $!;
+} else {
+    print "testing output saved in refs/dgit-badfixuptest/\n" or die $!;
+}