TLPOBJ.pm (from_fh): do not remove trailing whitespace; turns out this

takes consequential time, and the creation programs should not ever
produce it.  Also added trailing-whitespace check to cron job for master
tlpdb creation.  With this change, tlpdb-read-test time is reduced from
~2.4sec to ~1.2sec.

TeXCatalogue.pm (beautify): remove trailing whitespace.  Looks like this
can matter in <caption>s which we turn into shortdescs, but nowhere else.

dev/tlpdb-read-test: trivial test script to read tlpdb and nothing else.


git-svn-id: svn://tug.org/texlive/trunk@22498 c570f23f-e606-0410-a88d-b1316a301751

4 files changed, 45 insertions, 22 deletions

diff --git a/Master/tlpkg/TeXLive/TLPOBJ.pm b/Master/tlpkg/TeXLive/TLPOBJ.pm
index b73ee0e3cff..9713c87b90a 100644
--- a/Master/tlpkg/TeXLive/TLPOBJ.pm
+++ b/Master/tlpkg/TeXLive/TLPOBJ.pm
@@ -83,7 +83,11 @@ sub from_fh {
   my $size;
 
   while (my $line = <$fh>) {
+    # we do not worry about whitespace at the end of a line;
+    # that would be a bug in the db creation, and it takes some
+    # noticeable time to get rid of it.  So just chomp.
     chomp($line);
+    
     # we call tllog only when something will be logged, to speed things up.
     # this is the inner loop bounding the time to read tlpdb.
     dddebug("reading line: >>>$line<<<\n") if ($::opt_verbosity >= 3);
@@ -111,8 +115,7 @@ sub from_fh {
         die("Continuation of $lastcmd not allowed, please fix tlpobj: line = $line!\n");
       }
     }
-    # remove white space at the end of a line
-    $line =~ s/\s*$//;
+    
     if ($line =~ /^name\s*([-.\w]+)/o) {
       $name = "$1";
       $lastcmd = "name";
diff --git a/Master/tlpkg/TeXLive/TeXCatalogue.pm b/Master/tlpkg/TeXLive/TeXCatalogue.pm
index 95ba60442a1..34f7a7f512b 100644
--- a/Master/tlpkg/TeXLive/TeXCatalogue.pm
+++ b/Master/tlpkg/TeXLive/TeXCatalogue.pm
@@ -1,6 +1,6 @@
 # $Id$
 # TeXLive::TeXCatalogue - module for accessing the TeX Catalogue
-# Copyright 2007, 2008, 2009, 2010 Norbert Preining
+# Copyright 2007, 2008, 2009, 2010, 2011 Norbert Preining
 # This file is licensed under the GNU General Public License version 2
 # or any later version.
 # 
@@ -88,15 +88,15 @@ sub initialize {
   }
 }
 
-sub beautify
-{
+sub beautify {
   my ($txt) = @_;
-  $txt =~ s/\n/ /g;
-  $txt =~ s/^[[:space:]]*//g;
-  $txt =~ s/[[:space:]][[:space:]]*/ /g;
+  $txt =~ s/\n/ /g;  # make one line
+  $txt =~ s/^\s+//g; # rm leading whitespace
+  $txt =~ s/\s+$//g; # rm trailing whitespace
+  $txt =~ s/\s\s+/ /g; # multiple spaces to one
   # transliterate to ascii: it allows the final tlpdb to be pure ascii,
   # avoiding problems since we don't control the user's terminal encoding
-  return(Text::Unidecode::unidecode($txt));
+  return Text::Unidecode::unidecode($txt);
 }
 
 sub name {
@@ -197,8 +197,7 @@ sub initialize {
 # Copy every catalogue $entry under the name $entry->{'texlive'}
 # if it makes sense.
 # 
-sub quest4texlive
-{
+sub quest4texlive {
   my $self = shift;
 
   # The catalogue has a partial mapping from catalogue entries to
diff --git a/Master/tlpkg/dev/README b/Master/tlpkg/dev/README
index 62ce79ed5e2..bee33f3e2c1 100644
--- a/Master/tlpkg/dev/README
+++ b/Master/tlpkg/dev/README
@@ -1,20 +1,20 @@
 $Id$
 (Public domain.)
 
-lib
-	tlpkg libraries in other languages, none usable
+check-tk-used-font.pl
+	?
 
-dev.tempfile-for-tlpdb.patch
-	trial to make symlink attacks harder by using tmpfiles instead
-	of the always same filename when downloading the tlpdb.xz
+dev.multi-source-support
+	Multiple repositories.  Needs serious testing.
+
+dev.update-tlpdb-option-on-the-fly
+	?
 
 makeself-runscript.sh
 	a run.sh for a makeself (tlmgr-update-latest.sh) that even
 	allows installation of texlive by installing the bare bone
 	minimal (texlive.infra) and then bootstrapping into some
-	minimal env
-
-	Needs work, both on the script and tlmgr itself
+	minimal env. Needs work, both on the script and tlmgr itself.
 
 mktextex.pl
 	Simple mktextex program that can be used to query the tlpdb
@@ -22,12 +22,15 @@ mktextex.pl
 	Probably not useable in real life, only for testing.
 
 mktexupd.texlua
-	rewrite of mktexupd in texlua
+	Rewrite of mktexupd in texlua.
 
 srclist.txt
 	List of duplicated files to keep in sync.  Read by cron.tl.
 
 tlnet-disabled-packages.txt
-	this file specifies packages that should NOT be updated in 
-	tlnet.  It is read by tl-update-containers.
+	This file specifies packages that should NOT be updated in 
+	tlnet.  Read by tl-update-containers.
 
+tlpdb-read-test
+	Minimal script to exercise tlpdb reading and nothing else,
+	for testing/optimization.
diff --git a/Master/tlpkg/dev/tlpdb-read-test b/Master/tlpkg/dev/tlpdb-read-test
new file mode 100755
index 00000000000..30ac0d05b15
--- /dev/null
+++ b/Master/tlpkg/dev/tlpdb-read-test
@@ -0,0 +1,18 @@
+#!/usr/bin/env perl
+# Public domain.  (Originally written by Karl Berry, 2011.)
+# This script does nothing but read TLPDB so we can work 
+# on optimizing the inner loop, TLPOBJ::from_fh.
+
+our $mydir;
+
+BEGIN {
+  $^W = 1;
+  ($mydir = $0) =~ s,/[^/]*$,,;
+  unshift (@INC, "$mydir/..");
+}
+
+use strict;
+use TeXLive::TLPDB;
+
+chomp (my $Master = `cd $mydir/../.. && pwd`);
+my $tlpdb = TeXLive::TLPDB->new ("root" => $Master);