* TLPOBJ.pm (update_from_catalogue): compare true basename of ctan file,

resulting in some ~350 additional Catalogue details strings being
found.  Also prefer first match from Catalogue, for the sake of
top-level README files.

* TeXCatalogue.pm (initialize): show error when parse fails.

* tl-update-tlpdb: doc fixes.


git-svn-id: svn://tug.org/texlive/trunk@35744 c570f23f-e606-0410-a88d-b1316a301751

3 files changed, 42 insertions, 30 deletions

diff --git a/Master/tlpkg/TeXLive/TLPOBJ.pm b/Master/tlpkg/TeXLive/TLPOBJ.pm
index 5715a6e71e2..7bd98607626 100644
--- a/Master/tlpkg/TeXLive/TLPOBJ.pm
+++ b/Master/tlpkg/TeXLive/TLPOBJ.pm
@@ -720,8 +720,7 @@ sub total_size {
 # Update the current TLPOBJ object with the information from the
 # corresponding entry in C<$tlc->entries>.
 #
-sub update_from_catalogue
-{
+sub update_from_catalogue {
   my ($self, $tlc) = @_;
   my $tlcname = $self->name;
   if (defined($self->catalogue)) {
@@ -782,28 +781,38 @@ sub update_from_catalogue
     # <documentation details='Package documentation' language='de'
     #   href='ctan:/macros/latex/contrib/juramisc/doc/jmgerdoc.pdf'/>
     # <ctan path='/macros/latex/contrib/juramisc'/>
-    my @tcdocfiles = keys %{$entry->docs};
-    my @tlpdocfiles = $self->docfiles;
-    foreach my $tcdocfile (@tcdocfiles) {
-      # basename also kills the ctan: prefix!
+    my @tcdocfiles = keys %{$entry->docs};  # Catalogue doc files.
+    my %tcdocfilebasenames;                 # basenames of those, as we go.
+    my @tlpdocfiles = $self->docfiles;      # TL doc files.
+    foreach my $tcdocfile (sort @tcdocfiles) {  # sort so shortest first
+      #warn "looking at tcdocfile $tcdocfile\n";
       my $tcdocfilebasename = $tcdocfile;
-      # remove the ctan: prefix
-      $tcdocfilebasename =~ s/^ctan://;
-      # remove the ctan path if present
-      my $tcctanpath = $entry->ctan ? $entry->ctan : "";
-      $tcdocfilebasename =~ s/^$tcctanpath//;
+      $tcdocfilebasename =~ s/^ctan://;  # remove ctan: prefix
+      $tcdocfilebasename =~ s,.*/,,;     # remove all but the base file name
+      #warn "  got basename $tcdocfilebasename\n";
+      #
+      # If we've already seen this basename, skip.  This is for the sake
+      # of README files, which can exist in different directories but
+      # get renamed into different files in TL for various annoying reasons;
+      # e.g., ibygrk, rsfs, songbook.  In these cases, it turns out we
+      # always prefer the first entry (top-level README).
+      next if exists $tcdocfilebasenames{$tcdocfilebasename};
+      $tcdocfilebasenames{$tcdocfilebasename} = 1;
+      #
       foreach my $tlpdocfile (@tlpdocfiles) {
-        if ($tlpdocfile =~ m/$tcdocfilebasename$/) {
-          # update the language/detail tags if present!
+        warn "considering merge into tlpdocfile $tlpdocfile\n";
+        if ($tlpdocfile =~ m,/$tcdocfilebasename$,) {
+          # update the language/detail tags from Catalogue if present.
           if (defined($entry->docs->{$tcdocfile}{'details'})) {
             my $tmp = $entry->docs->{$tcdocfile}{'details'};
+            #warn "merging details for $tcdocfile: $tmp\n";
             # remove all embedded quotes, they are just a pain
             $tmp =~ s/"//g;
             $self->{'docfiledata'}{$tlpdocfile}{'details'} = $tmp;
           }
           if (defined($entry->docs->{$tcdocfile}{'language'})) {
             my $tmp = $entry->docs->{$tcdocfile}{'language'};
-            $tmp =~ s/"//g;
+            #warn "merging lang for $tcdocfile: $tmp\n";
             $self->{'docfiledata'}{$tlpdocfile}{'language'} = $tmp;
           }
         }
diff --git a/Master/tlpkg/TeXLive/TeXCatalogue.pm b/Master/tlpkg/TeXLive/TeXCatalogue.pm
index cdea8db9b47..2f5cdeb9d16 100644
--- a/Master/tlpkg/TeXLive/TeXCatalogue.pm
+++ b/Master/tlpkg/TeXLive/TeXCatalogue.pm
@@ -56,8 +56,9 @@ sub new {
 sub initialize {
   my $self = shift;
   # parse all the files
-  my $parser = new XML::XPath->new(ioref => $self->{'ioref'}, parser => $_parser) ||
-    die "Failed to parse the given ioref";
+  my $parser
+    = new XML::XPath->new(ioref => $self->{'ioref'}, parser => $_parser)
+      || die "Failed to parse given ioref";
   $self->{'entry'}{'id'} = $parser->findvalue('/entry/@id');
   $self->{'entry'}{'date'} = $parser->findvalue('/entry/@datestamp');
   $self->{'entry'}{'modder'} = $parser->findvalue('/entry/@modifier');
@@ -79,8 +80,8 @@ sub initialize {
   foreach my $node ($docset->get_nodelist) {
     my $docfile = $parser->find('./@href',$node);
     # see comments at end of beautify()
-    my $details = Text::Unidecode::unidecode(
-        $parser->find('./@details',$node));
+    my $details
+      = Text::Unidecode::unidecode($parser->find('./@details',$node));
     my $language = $parser->find('./@language',$node);
     $self->{'docs'}{$docfile}{'available'} = 1;
     if ($details) { $self->{'docs'}{$docfile}{'details'} = $details; }
@@ -186,13 +187,14 @@ sub initialize {
   || die "chdir($self->{location}/entries failed: $!";
   # parse all the files
   foreach (glob("?/*.xml")) {
+    # for debugging, nice to skip everything but: next unless /pst-node/;
     open(my $io,"<$_") or die "open($_) failed: $!";
     our $tce;
     # the XML parser die's on malformed xml entries, so we catch
     # that and continue, simply skipping the entry
     eval { $tce = TeXLive::TeXCatalogue::Entry->new( 'ioref' => $io ); };
     if ($@) {
-      printf STDERR "TeXCatalogue.pm: cannot parse $_, skipping it!\n";
+      warn "TeXCatalogue.pm:$_: cannot parse, skipping: $@\n";
       close($io);
       next;
     }
diff --git a/Master/tlpkg/bin/tl-update-tlpdb b/Master/tlpkg/bin/tl-update-tlpdb
index 98dc06e519f..562da3af3e8 100755
--- a/Master/tlpkg/bin/tl-update-tlpdb
+++ b/Master/tlpkg/bin/tl-update-tlpdb
@@ -2,7 +2,7 @@
 # $Id$
 # convert a set of tlpsrc files (given as arguments) into tlpdb
 # 
-# Copyright 2007-2012 Norbert Preining
+# Copyright 2007-2014 Norbert Preining
 # This file is licensed under the GNU General Public License version 2
 # or any later version.
 
@@ -446,20 +446,18 @@ tl-update-tlpdb [OPTION]... [TLPSRC]...
 
 =over 4
 
-=item B<--catalogue> I<Catalogue_file>
+=item B<--catalogue> I<Catalogue_dir>
 
-The I<Catalogue_file> must point to a valid TeX Catalogue as a Perl data
-dump; default is C<Master/texmf-dist/scripts/texlive/var/texcatalogue.data>,
-where C<Master> is the master directory described below.  That file is
-update nightly by the C<tl-update-auto> script.  If this file is given
-and readable, the generated tlpobj files will contain information
-gathered from the TeX Catalogue.
+The I<Catalogue_file> must point to a valid TeX Catalogue checkout; the
+default is C</home/httpd/html/catalogue>.  If the directory is readable,
+the generated tlpobj files will contain information gathered from the
+TeX Catalogue.
 
 =item B<--fix-reverse-revisions>
 
 If a package with revisions going backward is found, this option makes
 C<tl-update-tlpdb> try to check in a dummy change to the package's
-tlpsrc file to artificially increase the version number.
+tlpsrc file to artificially increase its version number.
 
 =item B<--from-files>
 
@@ -519,8 +517,11 @@ omits adding anything to the configs set in the tlpsrc file.
 
 By default, C<tl-update-tlpdb> recreates the tlpdb from all the tlpsrc
 files present in the current tree.  If this option is given, the tlpsrc
-file(s) have to be given on the command line.  If you're trying to debug
-the conversion of one particular package, this is the option to use.
+file(s), or package names, are given on the command line.  If you're
+trying to debug the conversion of one particular package, this is the
+option to use (along with C<-n>, most likely).  Indeed, if any
+non-option arguments are given, the program assumes you meant to specify
+this option.
 
 =item B<--with-w32-pattern-warning>