* 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

1 files changed, 23 insertions, 14 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;
           }
         }