sl@0
|
1 |
#
|
sl@0
|
2 |
# Copyright (c) 2009 Nokia Corporation and/or its subsidiary(-ies).
|
sl@0
|
3 |
# All rights reserved.
|
sl@0
|
4 |
# This component and the accompanying materials are made available
|
sl@0
|
5 |
# under the terms of the License "Eclipse Public License v1.0"
|
sl@0
|
6 |
# which accompanies this distribution, and is available
|
sl@0
|
7 |
# at the URL "http://www.eclipse.org/legal/epl-v10.html".
|
sl@0
|
8 |
#
|
sl@0
|
9 |
# Initial Contributors:
|
sl@0
|
10 |
# Nokia Corporation - initial contribution.
|
sl@0
|
11 |
#
|
sl@0
|
12 |
# Contributors:
|
sl@0
|
13 |
#
|
sl@0
|
14 |
# Description:
|
sl@0
|
15 |
#
|
sl@0
|
16 |
# UnicodeMaxDecompose.pl
|
sl@0
|
17 |
#
|
sl@0
|
18 |
# Adds maximal decompositions of the character and maximal decompositions of
|
sl@0
|
19 |
# its folded varient to the Unicode data.
|
sl@0
|
20 |
#
|
sl@0
|
21 |
# Added as the fourth field after the 'Symbain:' marker in the following format:
|
sl@0
|
22 |
#
|
sl@0
|
23 |
# Symbian:<grapheme-role>;<excluded>;<folded>;<max-decomposition>;<folded-decomposition>
|
sl@0
|
24 |
# where each of <max-decomposition> and <folded-decomposition> are strings
|
sl@0
|
25 |
# of hex numbers separated by spaces, representing the complete decomposition
|
sl@0
|
26 |
# of the character and its folded equivalent respectively.
|
sl@0
|
27 |
#
|
sl@0
|
28 |
# Usage:
|
sl@0
|
29 |
# perl -w UnicodeMaxDecompose.pl < <output-of-UnicodeAddFolded>
|
sl@0
|
30 |
|
sl@0
|
31 |
use strict;
|
sl@0
|
32 |
|
sl@0
|
33 |
if (scalar(@ARGV) != 0)
|
sl@0
|
34 |
{
|
sl@0
|
35 |
print (STDERR "Usage:\nperl -w UnicodeMaxDecompose.pl < <output-of-UnicodeAddFolded>\n");
|
sl@0
|
36 |
exit 1;
|
sl@0
|
37 |
}
|
sl@0
|
38 |
|
sl@0
|
39 |
my %StatedDecomposition = ();
|
sl@0
|
40 |
my %CompleteDecomposition = ();
|
sl@0
|
41 |
|
sl@0
|
42 |
sub Decompose
|
sl@0
|
43 |
{
|
sl@0
|
44 |
my ($code) = @_;
|
sl@0
|
45 |
return unless exists $StatedDecomposition{$code};
|
sl@0
|
46 |
my $stated = $StatedDecomposition{$code};
|
sl@0
|
47 |
delete $StatedDecomposition{$code};
|
sl@0
|
48 |
my @complete = ();
|
sl@0
|
49 |
foreach my $hexelt ( split(' ', $stated) )
|
sl@0
|
50 |
{
|
sl@0
|
51 |
if ($hexelt)
|
sl@0
|
52 |
{
|
sl@0
|
53 |
Decompose($hexelt);
|
sl@0
|
54 |
if (exists $CompleteDecomposition{$hexelt})
|
sl@0
|
55 |
{
|
sl@0
|
56 |
push @complete, $CompleteDecomposition{$hexelt};
|
sl@0
|
57 |
}
|
sl@0
|
58 |
else
|
sl@0
|
59 |
{
|
sl@0
|
60 |
push @complete, $hexelt;
|
sl@0
|
61 |
}
|
sl@0
|
62 |
}
|
sl@0
|
63 |
}
|
sl@0
|
64 |
$CompleteDecomposition{$code} = join(' ', @complete);
|
sl@0
|
65 |
}
|
sl@0
|
66 |
|
sl@0
|
67 |
my %Folded = ();
|
sl@0
|
68 |
my %LineToCode = ();
|
sl@0
|
69 |
my @RawLine = ();
|
sl@0
|
70 |
|
sl@0
|
71 |
my $lineNo = 0;
|
sl@0
|
72 |
while (my $line = <STDIN>)
|
sl@0
|
73 |
{
|
sl@0
|
74 |
chomp $line;
|
sl@0
|
75 |
$lineNo++;
|
sl@0
|
76 |
# Split into fields: make sure trailing null strings are not
|
sl@0
|
77 |
# deleted by adding a dummy final field
|
sl@0
|
78 |
my @attribute = split(/;/, $line.';dummy');
|
sl@0
|
79 |
# Delete the dummy field
|
sl@0
|
80 |
pop @attribute;
|
sl@0
|
81 |
die ("Line $lineNo is missing 'Symbian:' entries. Has UnicodeAddFolded been run?")
|
sl@0
|
82 |
if (scalar(@attribute) == 16);
|
sl@0
|
83 |
if (scalar(@attribute) == 17)
|
sl@0
|
84 |
{
|
sl@0
|
85 |
die ("Line $lineNo is missing 'Symbian:' entries. Has UnicodeAddFolded been run?")
|
sl@0
|
86 |
if ($attribute[15] !~ /^[ \t]*symbian:/i);
|
sl@0
|
87 |
my $code = $attribute[0];
|
sl@0
|
88 |
die("First attribute '$code' not a valid Unicode codepoint at line $lineNo")
|
sl@0
|
89 |
unless ($code =~ /^1?[0-9a-fA-F]{4,5}$/ && hex($code) < 0x110000);
|
sl@0
|
90 |
my $decomposition = $attribute[5];
|
sl@0
|
91 |
die("Decomposition '$decomposition' at line $lineNo is not a valid Unicode decomposition.")
|
sl@0
|
92 |
unless $decomposition =~ /^[ \t]*(<.*>[ \t]*[0-9a-fA-F])?[0-9a-fA-F \t]*$/;
|
sl@0
|
93 |
my $folded = $attribute[16];
|
sl@0
|
94 |
die ("'$folded' not a valid string of hex values at line $lineNo.")
|
sl@0
|
95 |
unless $folded =~ /[0-9a-fA-F \t]*/;
|
sl@0
|
96 |
# Store all decompositions that have no tag and at least one value
|
sl@0
|
97 |
if ($decomposition =~ /^[ \t]*[0-9a-fA-F]/)
|
sl@0
|
98 |
{
|
sl@0
|
99 |
$StatedDecomposition{$code} = $decomposition;
|
sl@0
|
100 |
}
|
sl@0
|
101 |
if ($folded =~ /[0-9a-fA-F]/)
|
sl@0
|
102 |
{
|
sl@0
|
103 |
$Folded{$code} = $folded;
|
sl@0
|
104 |
}
|
sl@0
|
105 |
$LineToCode{$lineNo-1} = $code;
|
sl@0
|
106 |
}
|
sl@0
|
107 |
elsif ($line !~ /^[ \t]*$/)
|
sl@0
|
108 |
{
|
sl@0
|
109 |
die 'Do not understand line '.$lineNo;
|
sl@0
|
110 |
}
|
sl@0
|
111 |
$RawLine[$lineNo-1] = $line;
|
sl@0
|
112 |
}
|
sl@0
|
113 |
|
sl@0
|
114 |
# Completely decompose all strings in the %StatedDecomposition
|
sl@0
|
115 |
foreach my $code (keys %StatedDecomposition)
|
sl@0
|
116 |
{
|
sl@0
|
117 |
Decompose($code);
|
sl@0
|
118 |
}
|
sl@0
|
119 |
|
sl@0
|
120 |
# Now decompose all the folded versions
|
sl@0
|
121 |
foreach my $code (keys %Folded)
|
sl@0
|
122 |
{
|
sl@0
|
123 |
my @result = ();
|
sl@0
|
124 |
foreach my $hexelt (split(' ', $Folded{$code}))
|
sl@0
|
125 |
{
|
sl@0
|
126 |
if (exists $CompleteDecomposition{$hexelt})
|
sl@0
|
127 |
{
|
sl@0
|
128 |
push @result, split(' ', $CompleteDecomposition{$hexelt});
|
sl@0
|
129 |
}
|
sl@0
|
130 |
else
|
sl@0
|
131 |
{
|
sl@0
|
132 |
push @result, $hexelt;
|
sl@0
|
133 |
}
|
sl@0
|
134 |
}
|
sl@0
|
135 |
$Folded{$code} = join(' ', @result);
|
sl@0
|
136 |
}
|
sl@0
|
137 |
|
sl@0
|
138 |
# Now output all the results
|
sl@0
|
139 |
for (my $i = 0; $i != scalar(@RawLine); $i++)
|
sl@0
|
140 |
{
|
sl@0
|
141 |
print $RawLine[$i];
|
sl@0
|
142 |
if (exists $LineToCode{$i})
|
sl@0
|
143 |
{
|
sl@0
|
144 |
my $code = $LineToCode{$i};
|
sl@0
|
145 |
print ';';
|
sl@0
|
146 |
my $decomp = '';
|
sl@0
|
147 |
$decomp = $CompleteDecomposition{$code}
|
sl@0
|
148 |
if exists $CompleteDecomposition{$code};
|
sl@0
|
149 |
print $decomp.';';
|
sl@0
|
150 |
if (exists $Folded{$code})
|
sl@0
|
151 |
{
|
sl@0
|
152 |
print $Folded{$code}
|
sl@0
|
153 |
}
|
sl@0
|
154 |
else
|
sl@0
|
155 |
{
|
sl@0
|
156 |
# If there is no folded value, but there is a decomposition
|
sl@0
|
157 |
# sequence, the character must fold to the decomposition
|
sl@0
|
158 |
# sequence too.
|
sl@0
|
159 |
print $decomp;
|
sl@0
|
160 |
}
|
sl@0
|
161 |
}
|
sl@0
|
162 |
print "\n";
|
sl@0
|
163 |
}
|