adding support for md5 sums

git-svn-id: svn://tug.org/texlive/trunk@9115 c570f23f-e606-0410-a88d-b1316a301751

2 files changed, 96 insertions, 13 deletions

diff --git a/Master/tlpkg/TeXLive/TLPOBJ.pm b/Master/tlpkg/TeXLive/TLPOBJ.pm
index f911f2a6cff..61f766d8e19 100644
--- a/Master/tlpkg/TeXLive/TLPOBJ.pm
+++ b/Master/tlpkg/TeXLive/TLPOBJ.pm
@@ -179,6 +179,18 @@ sub from_fh {
         $self->doccontainersize("$1");
         $lastcmd = "doccontainersize";
         next;
+      } elsif ($line =~ /^containermd5\s+([a-f0-9]+)\s*/o) {
+        $self->containermd5("$1");
+        $lastcmd = "containermd5";
+        next;
+      } elsif ($line =~ /^srccontainermd5\s+([a-f0-9]+)\s*/o) {
+        $self->srccontainermd5("$1");
+        $lastcmd = "srccontainermd5";
+        next;
+      } elsif ($line =~ /^doccontainermd5\s+([a-f0-9]+)\s*/o) {
+        $self->doccontainermd5("$1");
+        $lastcmd = "doccontainermd5";
+        next;
       } elsif ($line =~ /^catalogue\s+(.*)\s*/o) {
         $self->catalogue("$1");
         $lastcmd = "catalogue";
@@ -305,9 +317,15 @@ sub writeout {
   if (defined($self->{'containersize'})) {
     print $fd "containersize $self->{'containersize'}\n";
   }
+  if (defined($self->{'containermd5'})) {
+    print $fd "containermd5 $self->{'containermd5'}\n";
+  }
   if (defined($self->{'doccontainersize'})) {
     print $fd "doccontainersize $self->{'doccontainersize'}\n";
   }
+  if (defined($self->{'doccontainermd5'})) {
+    print $fd "doccontainermd5 $self->{'doccontainermd5'}\n";
+  }
   if (defined($self->{'docfiles'}) && (@{$self->{'docfiles'}})) {
     print $fd "docfiles size=$self->{'docsize'}\n";
     foreach my $f (sort @{$self->{'docfiles'}}) {
@@ -324,6 +342,9 @@ sub writeout {
   if (defined($self->{'srccontainersize'})) {
     print $fd "srccontainersize $self->{'srccontainersize'}\n";
   }
+  if (defined($self->{'srccontainermd5'})) {
+    print $fd "srccontainermd5 $self->{'srccontainermd5'}\n";
+  }
   if (defined($self->{'srcfiles'}) && (@{$self->{'srcfiles'}})) {
     print $fd "srcfiles size=$self->{'srcsize'}\n";
     foreach (sort @{$self->{'srcfiles'}}) {
@@ -587,12 +608,13 @@ sub make_container {
   }
   # compute the size
   my $size = (stat "$destdir/$tarname.lzma") [7];
+  my $m = TeXLive::TLUtils::tlmd5("$destdir/$tarname.lzma");
   # cleaning up
   unlink("$tlpobjdir/$self->{'name'}.tlpobj");
   rmdir("$tlpobjdir") if $removetlpobjdir;
   chdir($cwd);
   ddebug("Done $containername\n");
-  return($size);
+  return($size,$m);
 }
 
 
@@ -907,6 +929,21 @@ sub doccontainersize {
   if (@_) { $self->{'doccontainersize'} = shift }
   return ( defined($self->{'doccontainersize'}) ? $self->{'doccontainersize'} : -1 );
 }
+sub containermd5 {
+  my $self = shift;
+  if (@_) { $self->{'containermd5'} = shift }
+  return ( defined($self->{'containermd5'}) ? $self->{'containermd5'} : "" );
+}
+sub srccontainermd5 {
+  my $self = shift;
+  if (@_) { $self->{'srccontainermd5'} = shift }
+  return ( defined($self->{'srccontainermd5'}) ? $self->{'srccontainermd5'} : "" );
+}
+sub doccontainermd5 {
+  my $self = shift;
+  if (@_) { $self->{'doccontainermd5'} = shift }
+  return ( defined($self->{'doccontainermd5'}) ? $self->{'doccontainermd5'} : "" );
+}
 sub srcsize {
   my $self = shift;
   if (@_) { $self->{'srcsize'} = shift }
@@ -1165,13 +1202,17 @@ architectures as keys, similar to the C<runfiles> functions (see above).
 
 Futhermore, if the tlpobj is contained ina tlpdb which describes a media
 where the files are distributed in packed format (usually as .tar.lzma), 
-there are 3 more possible keys:
+there are 6 more possible keys:
 
   $tlpobj->containersize
   $tlpobj->doccontainersize
   $tlpobj->srccontainersize
+  $tlpobj->containermd5
+  $tlpobj->doccontainermd5
+  $tlpobj->srccontainermd5
 
-describing the respective sizes in bytes. The latter two are only present
+describing the respective sizes and md5sums in bytes and as hex string, resp. 
+The latter two are only present
 if src/doc file container splitting is activated for that install medium.
 
 =head1 OTHER FUNCTIONS
diff --git a/Master/tlpkg/TeXLive/TLUtils.pm b/Master/tlpkg/TeXLive/TLUtils.pm
index 3968c18bcae..59ab8bd92d8 100644
--- a/Master/tlpkg/TeXLive/TLUtils.pm
+++ b/Master/tlpkg/TeXLive/TLUtils.pm
@@ -66,6 +66,7 @@ C<TeXLive::TLUtils> -- utilities used in the TeX Live infrastructure
   TeXLive::TLUtils::kpsewhich;
   TeXLive::TLUtils::conv_to_win_path($path);
   TeXLive::TLUtils::give_ctan_mirror($path);
+  TeXLive::TLUtils::tlmd5($path);
 
 =head1 DESCRIPTION
 
@@ -115,6 +116,7 @@ BEGIN {
     &welcome
     &welcome_paths
     &give_ctan_mirror
+    &tlmd5
   );
   @EXPORT = qw(setup_programs download_file info log debug ddebug tlwarn process_logging_options win32);
 }
@@ -122,6 +124,7 @@ BEGIN {
 
 use TeXLive::TLConfig;
 use Getopt::Long;
+use Digest::MD5;
 
 $::opt_verbosity = 0;
 
@@ -622,7 +625,7 @@ sub install_packages {
     } elsif (&media eq 'NET') {
       $container = "$TeXLiveURL/$NetArchive/$package.$DefaultContainerExtension";
     }
-    install_package($container, $tlpobj->containersize,\@installfiles,$totlpdb->root,$vars{'this_platform'});
+    install_package($container, $tlpobj->containersize, $tlpobj->containermd5, \@installfiles,$totlpdb->root,$vars{'this_platform'});
     # if we are installing from CD or NET we have to fetch the respective
     # source and doc packages $pkg.source and $pkg.doc and install them, too
     if ((&media eq 'NET') || (&media eq 'CD')) {
@@ -635,13 +638,13 @@ sub install_packages {
           ($package !~ m/\./) && $tlpobj->srcfiles) {
         my $srccontainer = $container;
         $srccontainer =~ s/(\.tar\.lzma|\.zip)$/.source$1/;
-        install_package($srccontainer, $tlpobj->srccontainersize,\@installfiles,$totlpdb->root,$vars{'this_platform'});
+        install_package($srccontainer, $tlpobj->srccontainersize, $tlpobj->srccontainermd5, \@installfiles,$totlpdb->root,$vars{'this_platform'});
       }
       if ($container_doc_split && $real_opt_doc &&
           ($package !~ m/\./) && $tlpobj->docfiles) {
         my $doccontainer = $container;
         $doccontainer =~ s/(\.tar\.lzma|\.zip)$/.doc$1/;
-        install_package($doccontainer, $tlpobj->doccontainersize, \@installfiles,$totlpdb->root,$vars{'this_platform'});
+        install_package($doccontainer, $tlpobj->doccontainersize, $tlpobj->doccontainermd5, \@installfiles,$totlpdb->root,$vars{'this_platform'});
       }
     }
     # we don't want to have wrong information in the tlpdb, so remove the
@@ -693,7 +696,7 @@ is not taken into account (should be fixed!).
 =cut
 
 sub install_package {
-  my ($what, $whatsize, $filelistref, $target, $platform) = @_;
+  my ($what, $whatsize, $whatmd5, $filelistref, $target, $platform) = @_;
 
   my @filelist = @$filelistref;
 
@@ -737,14 +740,41 @@ sub install_package {
     my $gotfiledone = 0;
     if (-r $lzmafile) {
       # check that the downloaded file is not partial
-      my $size = (stat $lzmafile)[7];
-      if ($size == $whatsize) {
-        debug("Reusing already downloaded container $lzmafile\n");
-        $gotfiledone = 1;
+      if ($whatsize >= 0) {
+        # we have the size given, so check that first
+        my $size = (stat $lzmafile)[7];
+        if ($size == $whatsize) {
+          # we want to check also the md5sum if we have it present
+          if ($whatmd5) {
+            if (tlmd5($what) eq $whatmd5) {
+              $gotfiledone = 1;
+            } else {
+              tlwarn("Downloaded file present, size equal, but md5sum differs\nredownloading $what!\n");
+            }
+          } else {
+            # size ok, no md5sum
+            tlwarn("Downloaded container present, size equal, but no md5sum available. Hope that is fine!\n");
+            $gotfiledone = 1;
+          }
+        } else {
+          tlwarn("Partial download of $what found, removing it!\n");
+          unlink($tarfile, $lzmafile);
+        }
       } else {
-        tlwarn("Partial download of $what found, removing it!\n");
-        unlink($tarfile, $lzmafile);
+        # ok no size information, hopefully we have md5 sums
+        if ($whatmd5) {
+          if (tlmd5($what) eq $whatmd5) {
+            $gotfiledone = 1;
+          } else {
+            tlwarn("Downloaded file present, but md5sum differs, removing it!\n");
+          }
+        } else {
+          tlwarn("container found, but we cannot verify size of md5sum!\nContinuing with it and hope that it is ok!\n");
+          $gotfiledone = 1;
+        }
       }
+      debug("Reusing already downloaded container $lzmafile\n")
+        if ($gotfiledone);
     }
     if (!$gotfiledone) {
       if ($what =~ m,http://|ftp://,) {
@@ -1448,6 +1478,18 @@ sub give_ctan_mirror {
   return($mirror);
 }
 
+sub tlmd5 {
+  my $file = shift;
+  if (-r $file) {
+    open(FILE, $file) or die "Can't open '$file': $!";
+    binmode(FILE);
+    return Digest::MD5->new->addfile(*FILE)->hexdigest;
+    close(FILE);
+  } else {
+    return "";
+  }
+}
+
 #############################################
 #
 # Taken from Text::ParseWords