replace the run time code in lib/utf8_pva.pl with data generated

at build by mktables, stored in lib/unicore/PVA.pl

p4raw-id: //depot/perl@22881

diff --git a/MANIFEST b/MANIFEST
index a74806f..609919d 100644 (file)
--- a/MANIFEST
+++ b/MANIFEST
@@ -1846,7 +1846,6 @@ lib/User/grent.t          See if User::grwent works
 lib/User/pwent.pm              By-name interface to Perl's builtin getpw*
 lib/User/pwent.t               See if User::pwent works
 lib/utf8_heavy.pl              Support routines for utf8 pragma
-lib/utf8_pva.pl                        Support routines for utf8 pragma
 lib/utf8.pm                    Pragma to control Unicode support
 lib/utf8.t                     See if utf8 operations work
 lib/validate.pl                        Perl library supporting wholesale file mode validation

diff --git a/Makefile.SH b/Makefile.SH
index a72a372..b502fac 100644 (file)
--- a/Makefile.SH
+++ b/Makefile.SH
@@ -320,7 +320,7 @@ addedbyconf = UU $(shextract) $(plextract) lib/lib.pm pstruct
 # Unicode data files generated by mktables
 unidatafiles = lib/unicore/Canonical.pl lib/unicore/Exact.pl \
        lib/unicore/Properties lib/unicore/Decomposition.pl \
-       lib/unicore/CombiningClass.pl lib/unicore/Name.pl
+       lib/unicore/CombiningClass.pl lib/unicore/Name.pl lib/unicore/PVA.pl
 
 # Directories of Unicode data files generated by mktables
 unidatadirs = lib/unicore/To lib/unicore/lib

diff --git a/lib/unicore/mktables b/lib/unicore/mktables
index ae83de8..aa92643 100644 (file)
--- a/lib/unicore/mktables
+++ b/lib/unicore/mktables
@@ -22,7 +22,6 @@ use File::Spec;
 ## Base names already used in lib/gc_sc (for avoiding 8.3 conflicts)
 my %BaseNames;
 
-
 ##
 ## Process any args.
 ##
@@ -154,10 +153,15 @@ sub Build_Aliases()
 {
     ##
     ## Most of the work with aliases doesn't occur here,
-    ## but rather in utf8_heavy.pl, which uses utf8_pva.pl,
-    ## which contains just this function.  However, this one
-    ## 
-    ##   -- japhy (2004/04/13)
+    ## but rather in utf8_heavy.pl, which uses PVA.pl,
+
+    # Placate the warnings about used only once. (They are used again, but
+    # via a typeglob lookup)
+    %utf8::PropertyAlias = ();
+    %utf8::PA_reverse = ();
+    %utf8::PropValueAlias = ();
+    %utf8::PVA_reverse = ();
+    %utf8::PVA_abbr_map = ();
 
     open PA, "< PropertyAliases.txt"
        or confess "Can't open PropertyAliases.txt: $!";
@@ -170,6 +174,12 @@ sub Build_Aliases()
         next if $abbrev eq "n/a";
        $PropertyAlias{$abbrev} = $name;
         $PA_reverse{$name} = $abbrev;
+
+       # The %utf8::... versions use japhy's code originally from utf8_pva.pl
+       # However, it's moved here so that we build the tables at runtime.
+       tr/ _-//d for $abbrev, $name;
+       $utf8::PropertyAlias{lc $abbrev} = $name;
+       $utf8::PA_reverse{lc $name} = $abbrev;
     }
     close PA;
 
@@ -191,8 +201,23 @@ sub Build_Aliases()
            $PropValueAlias{$prop}{$data[0]} = $data[1];
             $PVA_reverse{$prop}{$data[1]} = $data[0];
        }
+
+       shift @data if $prop eq 'ccc';
+       next if $data[0] eq "n/a";
+
+       $data[1] =~ tr/ _-//d;
+       $utf8::PropValueAlias{$prop}{lc $data[0]} = $data[1];
+       $utf8::PVA_reverse{$prop}{lc $data[1]} = $data[0];
+
+       my $abbr_class = ($prop eq 'gc' or $prop eq 'sc') ? 'gc_sc' : $prop;
+       $utf8::PVA_abbr_map{$abbr_class}{lc $data[0]} = $data[0];
     }
     close PVA;
+
+    # backwards compatibility for L& -> LC
+    $utf8::PropValueAlias{gc}{'l&'} = $utf8::PropValueAlias{gc}{lc};
+    $utf8::PVA_abbr_map{gc_sc}{'l&'} = $utf8::PVA_abbr_map{gc_sc}{lc};
+
 }
 
 
@@ -623,6 +648,33 @@ my $General  = Table->New(); ## all characters, grouped by category
 my %General;
 my %Cat;
 
+## Simple Data::Dumper alike. Good enough for our needs. We can't use the real
+## thing as we have to run under miniperl
+sub simple_dumper {
+    my @lines;
+    my $item;
+    foreach $item (@_) {
+       if (ref $item) {
+           if (ref $item eq 'ARRAY') {
+               push @lines, "[\n", simple_dumper (@$item), "],\n";
+           } elsif (ref $item eq 'HASH') {
+               push @lines, "{\n", simple_dumper (%$item), "},\n";
+           } else {
+               die "Can't cope with $item";
+           }
+       } else {
+           if (defined $item) {
+               my $copy = $item;
+               $copy =~ s/([\'\\])/\\$1/gs;
+               push @lines, "'$copy',\n";
+           } else {
+               push @lines, "undef,\n";
+           }
+       }
+    }
+    @lines;
+}
+
 ##
 ## Process UnicodeData.txt (Categories, etc.)
 ##
@@ -959,6 +1011,18 @@ sub UnicodeData_Txt()
     ##
     $Name->Write("Name.pl");
 
+    {
+       my @PVA = $HEADER;
+       foreach my $name (qw (PropertyAlias PA_reverse PropValueAlias
+                             PVA_reverse PVA_abbr_map)) {
+           # Should I really jump through typeglob hoops just to avoid a
+           # symbolic reference? (%{"utf8::$name})
+           push @PVA, "\n", "\%utf8::$name = (\n",
+               simple_dumper (%{$utf8::{$name}}), ");\n";
+       }
+       WriteIfChanged("PVA.pl", @PVA);
+    }
+
     # $Bidi->Write("Bidirectional.pl");
     for (keys %Bidi) {
        $Bidi{$_}->Write(

diff --git a/lib/utf8_heavy.pl b/lib/utf8_heavy.pl
index 309cbef..b6fdeb9 100644 (file)
--- a/lib/utf8_heavy.pl
+++ b/lib/utf8_heavy.pl
@@ -77,7 +77,7 @@ sub SWASHNEW {
            ##
            ## See if it's in some enumeration.
            ##
-           require "utf8_pva.pl";
+           require "unicore/PVA.pl";
            if ($type =~ /^([\w\s]+)[:=]\s*(.*)/) {
                my ($enum, $val) = (lc $1, lc $2);
                $enum =~ tr/ _-//d;

diff --git a/lib/utf8_pva.pl b/lib/utf8_pva.pl
deleted file mode 100644 (file)
index 6db881b..0000000
--- a/lib/utf8_pva.pl
+++ /dev/null
@@ -1,56 +0,0 @@
-package utf8;
-
-##
-## Store the alias definitions for later use.
-##
-
-my $dir;
-for (@INC) {
-  $dir = $_, last if -e "$_/unicore/PropertyAliases.txt";
-}
-
-use Carp 'confess';
-
-local *_;
-local $.; # localizes Pl_last_in_gv
-
-open PA, "< $dir/unicore/PropertyAliases.txt"
-    or confess "Can't open PropertyAliases.txt: $!";
-while (<PA>) {
-    s/#.*//;
-    s/\s+$//;
-    next if /^$/;
-
-    my ($abbrev, $name) = split /\s*;\s*/;
-    next if $abbrev eq "n/a";
-    tr/ _-//d for $abbrev, $name;
-    $PropertyAlias{lc $abbrev} = $name;
-    $PA_reverse{lc $name} = $abbrev;
-}
-close PA;
-
-open PVA, "< $dir/unicore/PropValueAliases.txt"
-    or confess "Can't open PropValueAliases.txt: $!";
-while (<PVA>) {
-    s/#.*//;
-    s/\s+$//;
-    next if /^$/;
-
-    my ($prop, @data) = split /\s*;\s*/;
-    shift @data if $prop eq 'ccc';
-    next if $data[0] eq "n/a";
-
-    $data[1] =~ tr/ _-//d;
-    $PropValueAlias{$prop}{lc $data[0]} = $data[1];
-    $PVA_reverse{$prop}{lc $data[1]} = $data[0];
-
-    my $abbr_class = ($prop eq 'gc' or $prop eq 'sc') ? 'gc_sc' : $prop;
-    $PVA_abbr_map{$abbr_class}{lc $data[0]} = $data[0];
-}
-close PVA;
-
-# backwards compatibility for L& -> LC
-$PropValueAlias{gc}{'l&'} = $PropValueAlias{gc}{lc};
-$PVA_abbr_map{gc_sc}{'l&'} = $PVA_abbr_map{gc_sc}{lc};
-
-1;

diff --git a/vms/descrip_mms.template b/vms/descrip_mms.template
index 8aa9e0d..3cd8827 100644 (file)
--- a/vms/descrip_mms.template
+++ b/vms/descrip_mms.template
@@ -332,7 +332,8 @@ CRTLOPTS =,$(CRTL)/Options
 unidatafiles1 = [.lib.unicore]Canonical.pl [.lib.unicore]Exact.pl
 unidatafiles2 = [.lib.unicore]Properties [.lib.unicore]Decomposition.pl
 unidatafiles3 = [.lib.unicore]CombiningClass.pl [.lib.unicore]Name.pl
-unidatafiles = $(unidatafiles1) $(unidatafiles2) $(unidatafiles3)
+unidatafiles4 = [.lib.unicore]PVA.pl
+unidatafiles = $(unidatafiles1) $(unidatafiles2) $(unidatafiles3) $(unidatafiles4)
 
 # Directories of Unicode data files generated by mktables
 unidatadirs = lib/unicore/To lib/unicore/lib

diff --git a/win32/Makefile b/win32/Makefile
index 308099f..6f3e6ce 100644 (file)
--- a/win32/Makefile
+++ b/win32/Makefile
@@ -454,7 +454,8 @@ X2P         = ..\x2p\a2p.exe
 # Unicode data files generated by mktables
 UNIDATAFILES    = ..\lib\unicore\Canonical.pl ..\lib\unicore\Exact.pl \
                   ..\lib\unicore\Properties ..\lib\unicore\Decomposition.pl \
-                  ..\lib\unicore\CombiningClass.pl ..\lib\unicore\Name.pl
+                  ..\lib\unicore\CombiningClass.pl ..\lib\unicore\Name.pl \
+                  ..\lib\unicore\PVA.pl
 
 # Directories of Unicode data files generated by mktables
 UNIDATADIRS     = ..\lib\unicore\To ..\lib\unicore\lib

diff --git a/win32/makefile.mk b/win32/makefile.mk
index 8aca918..f61b7b5 100644 (file)
--- a/win32/makefile.mk
+++ b/win32/makefile.mk
@@ -577,7 +577,8 @@ X2P         = ..\x2p\a2p.exe
 # Unicode data files generated by mktables
 UNIDATAFILES    = ..\lib\unicore\Canonical.pl ..\lib\unicore\Exact.pl \
                   ..\lib\unicore\Properties ..\lib\unicore\Decomposition.pl \
-                  ..\lib\unicore\CombiningClass.pl ..\lib\unicore\Name.pl
+                  ..\lib\unicore\CombiningClass.pl ..\lib\unicore\Name.pl \
+                  ..\lib\unicore\PVA.pl
 
 # Directories of Unicode data files generated by mktables
 UNIDATADIRS     = ..\lib\unicore\To ..\lib\unicore\lib