annotate lib/unictype/scripts.c @ 40057:b06060465f09

maint: Run 'make update-copyright'
author Paul Eggert <eggert@cs.ucla.edu>
date Tue, 01 Jan 2019 00:25:11 +0100
parents 10eb9086bea0
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
9471
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
1 /* Scripts of Unicode characters.
40057
b06060465f09 maint: Run 'make update-copyright'
Paul Eggert <eggert@cs.ucla.edu>
parents: 19484
diff changeset
2 Copyright (C) 2007, 2009-2019 Free Software Foundation, Inc.
9471
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
3 Written by Bruno Haible <bruno@clisp.org>, 2007.
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
4
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
5 This program is free software: you can redistribute it and/or modify it
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
6 under the terms of the GNU Lesser General Public License as published
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
7 by the Free Software Foundation; either version 3 of the License, or
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
8 (at your option) any later version.
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
9
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
10 This program is distributed in the hope that it will be useful,
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
11 but WITHOUT ANY WARRANTY; without even the implied warranty of
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
12 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
13 Lesser General Public License for more details.
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
14
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
15 You should have received a copy of the GNU Lesser General Public License
19190
9759915b2aca all: prefer https: URLs
Paul Eggert <eggert@cs.ucla.edu>
parents: 18626
diff changeset
16 along with this program. If not, see <https://www.gnu.org/licenses/>. */
9471
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
17
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
18 #include <config.h>
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
19
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
20 /* Specification. */
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
21 #include "unictype.h"
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
22
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
23 #include <string.h>
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
24
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
25 #include "scripts.h"
9499
5d136ebd8610 Fix a build error when someone uses "make maintainer-clean".
Bruno Haible <bruno@clisp.org>
parents: 9471
diff changeset
26 #include "unictype/scripts_byname.h"
9471
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
27
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
28 const uc_script_t *
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
29 uc_script (ucs4_t uc)
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
30 {
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
31 unsigned int index1 = uc >> script_header_0;
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
32 if (index1 < script_header_1)
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
33 {
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
34 int lookup1 = u_script.level1[index1];
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
35 if (lookup1 >= 0)
12421
e8d2c6fc33ad Use spaces for indentation, not tabs.
Bruno Haible <bruno@clisp.org>
parents: 9499
diff changeset
36 {
e8d2c6fc33ad Use spaces for indentation, not tabs.
Bruno Haible <bruno@clisp.org>
parents: 9499
diff changeset
37 unsigned int index2 = (uc >> script_header_2) & script_header_3;
e8d2c6fc33ad Use spaces for indentation, not tabs.
Bruno Haible <bruno@clisp.org>
parents: 9499
diff changeset
38 int lookup2 = u_script.level2[lookup1 + index2];
e8d2c6fc33ad Use spaces for indentation, not tabs.
Bruno Haible <bruno@clisp.org>
parents: 9499
diff changeset
39 if (lookup2 >= 0)
e8d2c6fc33ad Use spaces for indentation, not tabs.
Bruno Haible <bruno@clisp.org>
parents: 9499
diff changeset
40 {
e8d2c6fc33ad Use spaces for indentation, not tabs.
Bruno Haible <bruno@clisp.org>
parents: 9499
diff changeset
41 unsigned int index3 = (uc & script_header_4);
e8d2c6fc33ad Use spaces for indentation, not tabs.
Bruno Haible <bruno@clisp.org>
parents: 9499
diff changeset
42 unsigned char lookup3 = u_script.level3[lookup2 + index3];
9471
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
43
12421
e8d2c6fc33ad Use spaces for indentation, not tabs.
Bruno Haible <bruno@clisp.org>
parents: 9499
diff changeset
44 if (lookup3 != 0xff)
e8d2c6fc33ad Use spaces for indentation, not tabs.
Bruno Haible <bruno@clisp.org>
parents: 9499
diff changeset
45 return &scripts[lookup3];
e8d2c6fc33ad Use spaces for indentation, not tabs.
Bruno Haible <bruno@clisp.org>
parents: 9499
diff changeset
46 }
e8d2c6fc33ad Use spaces for indentation, not tabs.
Bruno Haible <bruno@clisp.org>
parents: 9499
diff changeset
47 }
9471
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
48 }
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
49 return NULL;
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
50 }
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
51
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
52 const uc_script_t *
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
53 uc_script_byname (const char *script_name)
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
54 {
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
55 const struct named_script *found;
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
56
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
57 found = uc_script_lookup (script_name, strlen (script_name));
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
58 if (found != NULL)
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
59 return &scripts[found->index];
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
60 else
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
61 return NULL;
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
62 }
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
63
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
64 bool
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
65 uc_is_script (ucs4_t uc, const uc_script_t *script)
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
66 {
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
67 return uc_script (uc) == script;
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
68 }
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
69
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
70 void
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
71 uc_all_scripts (const uc_script_t **scriptsp, size_t *countp)
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
72 {
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
73 *scriptsp = scripts;
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
74 *countp = sizeof (scripts) / sizeof (scripts[0]);
6dc73c76eced Unicode character classification functions.
Bruno Haible <bruno@clisp.org>
parents:
diff changeset
75 }