This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
regexec.c: More efficient Korean \X processing
[perl5.git] / regen / unicode_constants.pl
CommitLineData
61dad979
KW
1use v5.16.0;
2use strict;
3use warnings;
4require 'regen/regen_lib.pl';
5use charnames qw(:loose);
6
1b0f46bf 7my $out_fh = open_new('unicode_constants.h', '>',
61dad979
KW
8 {style => '*', by => $0,
9 from => "Unicode data"});
10
11print $out_fh <<END;
d10c72f2 12
1b0f46bf
KW
13#ifndef H_UNICODE_CONSTANTS /* Guard against nested #includes */
14#define H_UNICODE_CONSTANTS 1
d10c72f2 15
61dad979 16/* This file contains #defines for various Unicode code points. The values
525b6419
KW
17 * the macros expand to are the native Unicode code point, or all or portions
18 * of the UTF-8 encoding for the code point. In the former case, the macro
19 * name has the suffix "_NATIVE"; otherwise, the suffix "_UTF8".
61dad979 20 *
525b6419
KW
21 * The macros that have the suffix "_UTF8" may have further suffixes, as
22 * follows:
23 * "_FIRST_BYTE" if the value is just the first byte of the UTF-8
24 * representation; the value will be a numeric constant.
25 * "_TAIL" if instead it represents all but the first byte. This, and
26 * with no additional suffix are both string constants */
61dad979
KW
27
28END
29
76837d21
KW
30# The data are at the end of this file. A blank line is output as-is.
31# Otherwise, each line represents one #define, and begins with either a
32# Unicode character name with the blanks in it squeezed out or replaced by
33# underscores; or it may be a hexadecimal Unicode code point. In the latter
34# case, the name will be looked-up to use as the name of the macro. In either
35# case, the macro name will have suffixes as listed above, and all blanks will
36# be replaced by underscores.
61dad979
KW
37#
38# Each line may optionally have one of the following flags on it, separated by
39# white space from the initial token.
5f1720e9 40# string indicates that the output is to be of the string form
61dad979 41# described in the comments above that are placed in the file.
5f1720e9 42# first indicates that the output is to be of the FIRST_BYTE form.
61dad979 43# tail indicates that the output is of the _TAIL form.
525b6419
KW
44# native indicates that the output is the code point, converted to the
45# platform's native character set if applicable
61dad979 46#
765ec46c
KW
47# If the code point has no official name, the desired name may be appended
48# after the flag, which will be ignored if there is an official name.
49#
61dad979
KW
50# This program is used to make it convenient to create compile time constants
51# of UTF-8, and to generate proper EBCDIC as well as ASCII without manually
52# having to figure things out.
53
54while ( <DATA> ) {
76837d21
KW
55 if ($_ !~ /\S/) {
56 print $out_fh "\n";
57 next;
58 }
59
61dad979
KW
60 chomp;
61 unless ($_ =~ m/ ^ ( [^\ ]* ) # Name or code point token
765ec46c
KW
62 (?: [\ ]+ ( [^ ]* ) )? # optional flag
63 (?: [\ ]+ ( .* ) )? # name if unnamed; flag is required
61dad979
KW
64 /x)
65 {
66 die "Unexpected syntax at line $.: $_\n";
67 }
68
69 my $name_or_cp = $1;
70 my $flag = $2;
765ec46c 71 my $desired_name = $3;
61dad979
KW
72
73 my $name;
74 my $cp;
75
76 if ($name_or_cp =~ /[^[:xdigit:]]/) {
77
78 # Anything that isn't a hex value must be a name.
79 $name = $name_or_cp;
80 $cp = charnames::vianame($name =~ s/_/ /gr);
81 die "Unknown name '$name' at line $.: $_\n" unless defined $name;
82 }
83 else {
84 $cp = $name_or_cp;
765ec46c
KW
85 $name = charnames::viacode("0$cp") // ""; # viacode requires a leading
86 # zero to be sure that the
87 # argument is hex
61dad979
KW
88 die "Unknown code point '$cp' at line $.: $_\n" unless defined $cp;
89 }
90
765ec46c 91 $name = $desired_name if $name eq "";
61dad979
KW
92 $name =~ s/ /_/g; # The macro name can have no blanks in it
93
94 my $str = join "", map { sprintf "\\x%02X", $_ }
95 unpack("U0C*", pack("U", hex $cp));
96
97 my $suffix = '_UTF8';
5f1720e9 98 if (! defined $flag || $flag eq 'string') {
61dad979
KW
99 $str = "\"$str\""; # Will be a string constant
100 } elsif ($flag eq 'tail') {
101 $str =~ s/\\x..//; # Remove the first byte
102 $suffix .= '_TAIL';
103 $str = "\"$str\""; # Will be a string constant
104 }
105 elsif ($flag eq 'first') {
106 $str =~ s/ \\x ( .. ) .* /$1/x; # Get the two nibbles of the 1st byte
107 $suffix .= '_FIRST_BYTE';
108 $str = "0x$str"; # Is a numeric constant
109 }
525b6419
KW
110 elsif ($flag eq 'native') {
111 die "Are you sure you want to run this on an above-Latin1 code point?" if hex $cp > 0xff;
112 $suffix = '_NATIVE';
113 $str = utf8::unicode_to_native(hex $cp);
114 $str = "0x$cp"; # Is a numeric constant
115 }
61dad979
KW
116 else {
117 die "Unknown flag at line $.: $_\n";
118 }
119 print $out_fh "#define ${name}$suffix $str /* U+$cp */\n";
120}
121
1b0f46bf 122print $out_fh "\n#endif /* H_UNICODE_CONSTANTS */\n";
d10c72f2 123
61dad979
KW
124read_only_bottom_close_and_rename($out_fh);
125
126__DATA__
5f1720e9
KW
1270300 string
1280301 string
1290308 string
76837d21 130
0a982f06 13103B9 string
76837d21 132
0a982f06 13303C5 string
76837d21 134
5f1720e9 1352010 string
765ec46c 136D800 first FIRST_SURROGATE
525b6419
KW
137
138007F native
13900DF native
14000E5 native
14100C5 native
14200FF native
14300B5 native
1440085 native