Commit | Line | Data |
---|---|---|
61dad979 KW |
1 | use v5.16.0; |
2 | use strict; | |
3 | use warnings; | |
4 | require 'regen/regen_lib.pl'; | |
ad88cddb | 5 | require 'regen/charset_translations.pl'; |
61dad979 KW |
6 | use charnames qw(:loose); |
7 | ||
1b0f46bf | 8 | my $out_fh = open_new('unicode_constants.h', '>', |
ad88cddb | 9 | {style => '*', by => $0, |
61dad979 KW |
10 | from => "Unicode data"}); |
11 | ||
12 | print $out_fh <<END; | |
d10c72f2 | 13 | |
1b0f46bf KW |
14 | #ifndef H_UNICODE_CONSTANTS /* Guard against nested #includes */ |
15 | #define H_UNICODE_CONSTANTS 1 | |
d10c72f2 | 16 | |
61dad979 | 17 | /* This file contains #defines for various Unicode code points. The values |
525b6419 KW |
18 | * the macros expand to are the native Unicode code point, or all or portions |
19 | * of the UTF-8 encoding for the code point. In the former case, the macro | |
20 | * name has the suffix "_NATIVE"; otherwise, the suffix "_UTF8". | |
61dad979 | 21 | * |
525b6419 KW |
22 | * The macros that have the suffix "_UTF8" may have further suffixes, as |
23 | * follows: | |
24 | * "_FIRST_BYTE" if the value is just the first byte of the UTF-8 | |
25 | * representation; the value will be a numeric constant. | |
26 | * "_TAIL" if instead it represents all but the first byte. This, and | |
27 | * with no additional suffix are both string constants */ | |
61dad979 KW |
28 | |
29 | END | |
30 | ||
76837d21 | 31 | # The data are at the end of this file. A blank line is output as-is. |
5a731a17 KW |
32 | # Comments (lines whose first non-blank is a '#') are converted to C-style, |
33 | # though empty comments are converted to blank lines. Otherwise, each line | |
34 | # represents one #define, and begins with either a Unicode character name with | |
35 | # the blanks and dashes in it squeezed out or replaced by underscores; or it | |
1dfa4f52 | 36 | # may be a hexadecimal Unicode code point of the form U+xxxx. In the latter |
76837d21 | 37 | # case, the name will be looked-up to use as the name of the macro. In either |
e9cddfae KW |
38 | # case, the macro name will have suffixes as listed above, and all blanks and |
39 | # dashes will be replaced by underscores. | |
61dad979 KW |
40 | # |
41 | # Each line may optionally have one of the following flags on it, separated by | |
42 | # white space from the initial token. | |
5f1720e9 | 43 | # string indicates that the output is to be of the string form |
61dad979 | 44 | # described in the comments above that are placed in the file. |
632c9f80 KW |
45 | # string_skip_ifundef is the same as 'string', but instead of dying if the |
46 | # code point doesn't exist, the line is just skipped: no output is | |
47 | # generated for it | |
5f1720e9 | 48 | # first indicates that the output is to be of the FIRST_BYTE form. |
61dad979 | 49 | # tail indicates that the output is of the _TAIL form. |
525b6419 KW |
50 | # native indicates that the output is the code point, converted to the |
51 | # platform's native character set if applicable | |
61dad979 | 52 | # |
765ec46c KW |
53 | # If the code point has no official name, the desired name may be appended |
54 | # after the flag, which will be ignored if there is an official name. | |
55 | # | |
61dad979 KW |
56 | # This program is used to make it convenient to create compile time constants |
57 | # of UTF-8, and to generate proper EBCDIC as well as ASCII without manually | |
58 | # having to figure things out. | |
59 | ||
ad88cddb KW |
60 | my @data = <DATA>; |
61 | ||
62 | foreach my $charset (get_supported_code_pages()) { | |
63 | print $out_fh "\n" . get_conditional_compile_line_start($charset); | |
64 | ||
65 | my @a2n = get_a2n($charset); | |
66 | ||
4a4b1311 KW |
67 | for ( @data ) { |
68 | chomp; | |
69 | ||
70 | # Convert any '#' comments to /* ... */; empty lines and comments are | |
71 | # output as blank lines | |
72 | if ($_ =~ m/ ^ \s* (?: \# ( .* ) )? $ /x) { | |
73 | my $comment_body = $1 // ""; | |
74 | if ($comment_body ne "") { | |
75 | print $out_fh "/* $comment_body */\n"; | |
76 | } | |
77 | else { | |
78 | print $out_fh "\n"; | |
79 | } | |
80 | next; | |
5a731a17 | 81 | } |
76837d21 | 82 | |
4a4b1311 KW |
83 | unless ($_ =~ m/ ^ ( [^\ ]* ) # Name or code point token |
84 | (?: [\ ]+ ( [^ ]* ) )? # optional flag | |
85 | (?: [\ ]+ ( .* ) )? # name if unnamed; flag is required | |
86 | /x) | |
87 | { | |
88 | die "Unexpected syntax at line $.: $_\n"; | |
89 | } | |
61dad979 | 90 | |
4a4b1311 KW |
91 | my $name_or_cp = $1; |
92 | my $flag = $2; | |
93 | my $desired_name = $3; | |
94 | ||
95 | my $name; | |
96 | my $cp; | |
97 | my $U_cp; # code point in Unicode (not-native) terms | |
98 | my $undef_ok = $desired_name || $flag =~ /skip_if_undef/; | |
99 | ||
100 | if ($name_or_cp =~ /^U\+(.*)/) { | |
101 | $U_cp = hex $1; | |
102 | $name = charnames::viacode($name_or_cp); | |
103 | if (! defined $name) { | |
104 | die "Unknown code point '$name_or_cp' at line $.: $_\n" unless $undef_ok; | |
105 | $name = ""; | |
106 | } | |
107 | } | |
108 | else { | |
109 | $name = $name_or_cp; | |
110 | die "Unknown name '$name' at line $.: $_\n" unless defined $name; | |
111 | $U_cp = charnames::vianame($name =~ s/_/ /gr); | |
632c9f80 | 112 | } |
61dad979 | 113 | |
4a4b1311 KW |
114 | $cp = ($U_cp < 256) |
115 | ? $a2n[$U_cp] | |
116 | : $U_cp; | |
ad88cddb | 117 | |
4a4b1311 KW |
118 | $name = $desired_name if $name eq "" && $desired_name; |
119 | $name =~ s/[- ]/_/g; # The macro name can have no blanks nor dashes | |
61dad979 | 120 | |
4a4b1311 KW |
121 | my $str; |
122 | my $suffix; | |
123 | if (defined $flag && $flag eq 'native') { | |
124 | die "Are you sure you want to run this on an above-Latin1 code point?" if $cp > 0xff; | |
125 | $suffix = '_NATIVE'; | |
126 | $str = sprintf "0x%02X", $cp; # Is a numeric constant | |
81a2a11f KW |
127 | } |
128 | else { | |
4a4b1311 KW |
129 | $str = join "", map { sprintf "\\x%02X", ord $_ } split //, cp_2_utfbytes($U_cp, $charset); |
130 | ||
131 | $suffix = '_UTF8'; | |
132 | if (! defined $flag || $flag =~ /^ string (_skip_if_undef)? $/x) { | |
133 | $str = "\"$str\""; # Will be a string constant | |
134 | } elsif ($flag eq 'tail') { | |
135 | $str =~ s/\\x..//; # Remove the first byte | |
136 | $suffix .= '_TAIL'; | |
137 | $str = "\"$str\""; # Will be a string constant | |
138 | } | |
139 | elsif ($flag eq 'first') { | |
140 | $str =~ s/ \\x ( .. ) .* /$1/x; # Get the two nibbles of the 1st byte | |
141 | $suffix .= '_FIRST_BYTE'; | |
142 | $str = "0x$str"; # Is a numeric constant | |
143 | } | |
144 | else { | |
145 | die "Unknown flag at line $.: $_\n"; | |
146 | } | |
81a2a11f | 147 | } |
4a4b1311 | 148 | printf $out_fh "# define %s%s %s /* U+%04X */\n", $name, $suffix, $str, $U_cp; |
a1beba5b | 149 | } |
ad88cddb | 150 | print $out_fh "\n" . get_conditional_compile_line_end(); |
61dad979 KW |
151 | } |
152 | ||
1b0f46bf | 153 | print $out_fh "\n#endif /* H_UNICODE_CONSTANTS */\n"; |
d10c72f2 | 154 | |
61dad979 KW |
155 | read_only_bottom_close_and_rename($out_fh); |
156 | ||
157 | __DATA__ | |
f2e06375 | 158 | U+017F string |
76837d21 | 159 | |
1dfa4f52 | 160 | U+0300 string |
1dfa4f52 | 161 | |
a78bc3c6 KW |
162 | U+0399 string |
163 | U+03BC string | |
164 | ||
f2e06375 KW |
165 | U+1E9E string |
166 | ||
a9f50d33 KW |
167 | U+FB05 string |
168 | U+FB06 string | |
169 | ||
1dfa4f52 KW |
170 | U+2010 string |
171 | U+D800 first FIRST_SURROGATE | |
5f0aa340 KW |
172 | BOM first |
173 | BOM tail | |
525b6419 | 174 | |
df758df2 KW |
175 | NBSP native |
176 | NBSP string | |
177 | ||
05016631 | 178 | DEL native |
c5eda08a KW |
179 | CR native |
180 | LF native | |
1dfa4f52 KW |
181 | U+00DF native |
182 | U+00E5 native | |
183 | U+00C5 native | |
184 | U+00FF native | |
185 | U+00B5 native |