TLPSRC.pm (from_file),

TeXCatalogue.pm (beautify): collapse multiple whitespace.
(Still on the quest for identical longdesc results from our various
sources.)


git-svn-id: svn://tug.org/texlive/trunk@54167 c570f23f-e606-0410-a88d-b1316a301751

3 files changed, 11 insertions, 10 deletions

diff --git a/Master/tlpkg/TeXLive/TLPSRC.pm b/Master/tlpkg/TeXLive/TLPSRC.pm
index 4a461a8fa13..66c3aed24fc 100644
--- a/Master/tlpkg/TeXLive/TLPSRC.pm
+++ b/Master/tlpkg/TeXLive/TLPSRC.pm
@@ -232,14 +232,14 @@ sub from_file {
     die "Cannot deduce name from file argument and name tag not found";
   }
   #
-  # We should probably call TeXCatalogue::beautify(), but trailing
-  # whitespace seems to be the only thing that comes up in practice. We
-  # want the parsing from .tlpsrc to result in exactly the same string,
-  # including spaces, as parsing from texlive.tlpdb. Otherwise
-  # tl-update-tlpdb's tlpdb_catalogue_compare will think the strings
-  # are always different.
-  $shortdesc =~ s/\s+$//g;
-  $longdesc =~ s/\s+$//g;
+  # We should call TeXCatalogue::beautify(), but let's be lazy since not
+  # everything comes up in practice. We want the parsing from .tlpsrc to
+  # result in exactly the same string, including spaces, as parsing from
+  # texlive.tlpdb. Otherwise tl-update-tlpdb's tlpdb_catalogue_compare
+  # will think the strings are always different.
+  $shortdesc =~ s/\s+$//g;  # rm trailing whitespace (shortdesc)
+  $longdesc =~ s/\s+$//g;   # rm trailing whitespace (longdesc)
+  $longdesc =~ s/\s\s+/ /g; # collapse multiple whitespace characters to one
   #
   $self->name($name);
   $self->category($category);
diff --git a/Master/tlpkg/TeXLive/TeXCatalogue.pm b/Master/tlpkg/TeXLive/TeXCatalogue.pm
index da0b3640d99..9f422ceaaed 100644
--- a/Master/tlpkg/TeXLive/TeXCatalogue.pm
+++ b/Master/tlpkg/TeXLive/TeXCatalogue.pm
@@ -140,7 +140,8 @@ sub beautify {
   $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
+  $txt =~ s/\s\s+/ /g; # collapse multiple whitespace characters to one
+  $txt =~ s/\t/ /g;    # tabs to spaces
   # transliterate to ascii: it allows the final tlpdb to be pure ascii,
   # avoiding problems since we don't control the user's terminal encoding
   return Text::Unidecode::unidecode($txt);
diff --git a/Master/tlpkg/doc/releng.txt b/Master/tlpkg/doc/releng.txt
index 4c75377be80..14f60e6b2b6 100644
--- a/Master/tlpkg/doc/releng.txt
+++ b/Master/tlpkg/doc/releng.txt
@@ -164,7 +164,7 @@ B. Specifics for tlperl update:
   unzip -q newperl.zip  # replace All files
   comm -13 /tmp/now /tmp/new >/tmp/c2
   comm -23 /tmp/now /tmp/new >/tmp/c1
-  wc -l /tmp/c1 /tmp/c2 # should not be everything
+  wc -l /tmp/c1 /tmp/c2 # should be about 100-200, not 1500+
   svn rm `cat /tmp/c1`
   svn add --parents --no-auto-props `cat /tmp/c2`
   svn status | fgrep \? # shouldn't be any besides newperl.zip