Faster generation of bzip2 deltas: Do not repeatedly decompress input file.

diff --git a/debian/changelog b/debian/changelog
index ddf2697..f74f1cc 100644 (file)
--- a/debian/changelog
+++ b/debian/changelog
@@ -1,3 +1,10 @@
+pristine-tar (1.02) UNRELEASED; urgency=low
+
+  * Faster generation of bzip2 deltas: Do not repeatedly decompress
+    input file.
+
+ -- Joey Hess <joeyh@debian.org>  Wed, 12 May 2010 21:12:58 -0400
+
 pristine-tar (1.01) unstable; urgency=low
 
   * Use italics in man pages to highlight which parts of command

diff --git a/pristine-bz2 b/pristine-bz2
index cca9377..83f818f 100755 (executable)
--- a/pristine-bz2
+++ b/pristine-bz2
@@ -81,6 +81,7 @@ use strict;
 use File::Temp;
 use Getopt::Long;
 use File::Basename qw/basename/;
+use IPC::Open2;
 
 use constant BZIP2_DEBUG        => 1;
 
@@ -117,6 +118,16 @@ sub doit {
        }
 }
 
+sub doit_redir {
+       no warnings 'once';
+       my ($in, $out, @args) = @_;
+       vprint(@args, "<", $in, ">", $out);
+       open INFILE, "<", $in or die("Could not open '$in' for reading: $!\n");
+       open OUTFILE, ">", $out or die("Could not open '$out' for reading: $!\n");
+       my $pid = open2(">&OUTFILE", "<&INFILE", @args);
+       waitpid $pid, 0;
+}
+
 sub tempdir {
        return File::Temp::tempdir("pristine-bz2.XXXXXXXXXX",
                TMPDIR => 1, CLEANUP => !$keep);
@@ -169,36 +180,27 @@ sub comparefiles {
 }
 
 sub testvariant {
-       my ($old, $new, $bzip2_program, @args) = @_;
-
-       # Unzip attempt of the previous run (or the original on the first run)
-       doit('bunzip2', $new);
-       if (-e $new) {
-               die "bunzip2 failed, aborting";
-       }
-       else {
-               $new =~ s/\.bz2$//;
-               unless (-e $new) {
-                       die("bunzip2 succeeded but I can't find the new file");
-               }
-       }
+       my ($old, $tmpin, $bzip2_program, @args) = @_;
 
        # try bzip2'ing with the arguments passed
-       doit($bzip2_program, @args, $new);
-       $new .= '.bz2';
-       unless (-e $new) {
+       doit($bzip2_program, @args, $tmpin);
+       $tmpin .= '.bz2';
+       unless (-e $tmpin) {
                die("$bzip2_program failed, aborting");
        }
 
        # and compare the generated with the original
-       return !comparefiles($old, $new);
+       return !comparefiles($old, $tmpin);
 }
 
 sub reproducebzip2 {
-       my ($wd, $orig, $new) = (shift, shift, shift);
+       my ($wd, $orig) = (shift, shift);
+       
+       my $tmpin="$wd/test";
+       doit_redir($orig, "$tmpin.bak", "bzip2", "-dc");
 
        # read fields from bzip2 headers
-       my ($level) = readbzip2($new);
+       my ($level) = readbzip2($orig);
        debug("level: $level");
 
        # try to guess the bzip2 arguments that are needed by the
@@ -206,7 +208,14 @@ sub reproducebzip2 {
        my @args = predictbzip2args($level);
 
        foreach my $program (@supported_bzip2_programs) {
-               testvariant($orig, $new, $program, @args)
+               # some compressors eat the uncompressed file, some
+               # do not; restore as needed
+               if (! -e $tmpin) {
+                       doit("cp", "-a", "$tmpin.bak", "$tmpin");
+
+               }
+
+               testvariant($orig, $tmpin, $program, @args)
                        && return $program, @args;
        }
        # More to come?
@@ -291,9 +300,8 @@ sub gendelta {
 
        my @files=qw(version type params program);
 
-       doit("cp", $bzip2file, "$tempdir/test.bz2");
        my ($program, @params)=
-               reproducebzip2($tempdir, $bzip2file, "$tempdir/test.bz2");
+               reproducebzip2($tempdir, $bzip2file);
 
        open(OUT, ">", "$tempdir/version") || die "$!";
        print OUT "2.0\n";