1 /*
2 * Copyright (C) 2009 The Android Open Source Project
3 *
4 * Licensed under the Apache License, Version 2.0 (the "License");
5 * you may not use this file except in compliance with the License.
6 * You may obtain a copy of the License at
7 *
8 * http://www.apache.org/licenses/LICENSE-2.0
9 *
10 * Unless required by applicable law or agreed to in writing, software
11 * distributed under the License is distributed on an "AS IS" BASIS,
12 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13 * See the License for the specific language governing permissions and
14 * limitations under the License.
15 */
16
17 #include <stdio.h>
18 #include <stdlib.h>
19
20 #include "PhoneticStringUtils.h"
21 #include <utils/Unicode.h>
22
23 // We'd like 0 length string last of sorted list. So when input string is NULL
24 // or 0 length string, we use these instead.
25 #define CODEPOINT_FOR_NULL_STR 0xFFFD
26 #define STR_FOR_NULL_STR "\xEF\xBF\xBD"
27
28 // We assume that users will not notice strings not sorted properly when the
29 // first 128 characters are the same.
30 #define MAX_CODEPOINTS 128
31
32 namespace android {
33
34 // Get hiragana from halfwidth katakana.
GetHiraganaFromHalfwidthKatakana(char32_t codepoint,char32_t next_codepoint,bool * next_is_consumed)35 static int GetHiraganaFromHalfwidthKatakana(char32_t codepoint,
36 char32_t next_codepoint,
37 bool *next_is_consumed) {
38 if (codepoint < 0xFF66 || 0xFF9F < codepoint) {
39 return codepoint;
40 }
41
42 switch (codepoint) {
43 case 0xFF66: // wo
44 return 0x3092;
45 case 0xFF67: // xa
46 return 0x3041;
47 case 0xFF68: // xi
48 return 0x3043;
49 case 0xFF69: // xu
50 return 0x3045;
51 case 0xFF6A: // xe
52 return 0x3047;
53 case 0xFF6B: // xo
54 return 0x3049;
55 case 0xFF6C: // xya
56 return 0x3083;
57 case 0xFF6D: // xyu
58 return 0x3085;
59 case 0xFF6E: // xyo
60 return 0x3087;
61 case 0xFF6F: // xtsu
62 return 0x3063;
63 case 0xFF70: // -
64 return 0x30FC;
65 case 0xFF9C: // wa
66 return 0x308F;
67 case 0xFF9D: // n
68 return 0x3093;
69 break;
70 default: {
71 if (0xFF71 <= codepoint && codepoint <= 0xFF75) {
72 // a, i, u, e, o
73 if (codepoint == 0xFF73 && next_codepoint == 0xFF9E) {
74 if (next_is_consumed != NULL) {
75 *next_is_consumed = true;
76 }
77 return 0x3094; // vu
78 } else {
79 return 0x3042 + (codepoint - 0xFF71) * 2;
80 }
81 } else if (0xFF76 <= codepoint && codepoint <= 0xFF81) {
82 // ka - chi
83 if (next_codepoint == 0xFF9E) {
84 // "dakuten" (voiced mark)
85 if (next_is_consumed != NULL) {
86 *next_is_consumed = true;
87 }
88 return 0x304B + (codepoint - 0xFF76) * 2 + 1;
89 } else {
90 return 0x304B + (codepoint - 0xFF76) * 2;
91 }
92 } else if (0xFF82 <= codepoint && codepoint <= 0xFF84) {
93 // tsu, te, to (skip xtsu)
94 if (next_codepoint == 0xFF9E) {
95 // "dakuten" (voiced mark)
96 if (next_is_consumed != NULL) {
97 *next_is_consumed = true;
98 }
99 return 0x3064 + (codepoint - 0xFF82) * 2 + 1;
100 } else {
101 return 0x3064 + (codepoint - 0xFF82) * 2;
102 }
103 } else if (0xFF85 <= codepoint && codepoint <= 0xFF89) {
104 // na, ni, nu, ne, no
105 return 0x306A + (codepoint - 0xFF85);
106 } else if (0xFF8A <= codepoint && codepoint <= 0xFF8E) {
107 // ha, hi, hu, he, ho
108 if (next_codepoint == 0xFF9E) {
109 // "dakuten" (voiced mark)
110 if (next_is_consumed != NULL) {
111 *next_is_consumed = true;
112 }
113 return 0x306F + (codepoint - 0xFF8A) * 3 + 1;
114 } else if (next_codepoint == 0xFF9F) {
115 // "han-dakuten" (half voiced mark)
116 if (next_is_consumed != NULL) {
117 *next_is_consumed = true;
118 }
119 return 0x306F + (codepoint - 0xFF8A) * 3 + 2;
120 } else {
121 return 0x306F + (codepoint - 0xFF8A) * 3;
122 }
123 } else if (0xFF8F <= codepoint && codepoint <= 0xFF93) {
124 // ma, mi, mu, me, mo
125 return 0x307E + (codepoint - 0xFF8F);
126 } else if (0xFF94 <= codepoint && codepoint <= 0xFF96) {
127 // ya, yu, yo
128 return 0x3084 + (codepoint - 0xFF94) * 2;
129 } else if (0xFF97 <= codepoint && codepoint <= 0xFF9B) {
130 // ra, ri, ru, re, ro
131 return 0x3089 + (codepoint - 0xFF97);
132 }
133 // Note: 0xFF9C, 0xFF9D are handled above
134 } // end of default
135 }
136
137 return codepoint;
138 }
139
140 // Assuming input is hiragana, convert the hiragana to "normalized" hiragana.
GetNormalizedHiragana(int codepoint)141 static int GetNormalizedHiragana(int codepoint) {
142 if (codepoint < 0x3040 || 0x309F < codepoint) {
143 return codepoint;
144 }
145
146 // TODO: should care (semi-)voiced mark (0x3099, 0x309A).
147
148 // Trivial kana conversions.
149 // e.g. xa => a
150 switch (codepoint) {
151 case 0x3041:
152 case 0x3043:
153 case 0x3045:
154 case 0x3047:
155 case 0x3049:
156 case 0x3063:
157 case 0x3083:
158 case 0x3085:
159 case 0x3087:
160 case 0x308E: // xwa
161 return codepoint + 1;
162 case 0x3095: // xka
163 return 0x304B;
164 case 0x3096: // xke
165 return 0x3051;
166 case 0x31F0: // xku
167 return 0x304F;
168 case 0x31F1: // xsi
169 return 0x3057;
170 case 0x31F2: // xsu
171 return 0x3059;
172 case 0x31F3: // xto
173 return 0x3068;
174 case 0x31F4: // xnu
175 return 0x306C;
176 case 0x31F5: // xha
177 return 0x306F;
178 case 0x31F6: // xhi
179 return 0x3072;
180 case 0x31F7: // xhu
181 return 0x3075;
182 case 0x31F8: // xhe
183 return 0x3078;
184 case 0x31F9: // xho
185 return 0x307B;
186 case 0x31FA: // xmu
187 return 0x3080;
188 case 0x31FB: // xra
189 case 0x31FC: // xri
190 case 0x31FD: // xru
191 case 0x31FE: // xre
192 case 0x31FF: // xro
193 // ra: 0x3089
194 return 0x3089 + (codepoint - 0x31FB);
195 default:
196 return codepoint;
197 }
198 }
199
GetNormalizedKana(char32_t codepoint,char32_t next_codepoint,bool * next_is_consumed)200 static int GetNormalizedKana(char32_t codepoint,
201 char32_t next_codepoint,
202 bool *next_is_consumed) {
203 // First, convert fullwidth katakana and halfwidth katakana to hiragana.
204 if (0x30A1 <= codepoint && codepoint <= 0x30F6) {
205 // Make fullwidth katakana same as hiragana.
206 // 96 == 0x30A1 - 0x3041c
207 codepoint = codepoint - 96;
208 } else if (codepoint == 0x309F) {
209 // Digraph YORI; Yo
210 codepoint = 0x3088;
211 } else if (codepoint == 0x30FF) {
212 // Digraph KOTO; Ko
213 codepoint = 0x3053;
214 } else {
215 codepoint = GetHiraganaFromHalfwidthKatakana(
216 codepoint, next_codepoint, next_is_consumed);
217 }
218
219 // Normalize Hiragana.
220 return GetNormalizedHiragana(codepoint);
221 }
222
GetNormalizedCodePoint(char32_t codepoint,char32_t next_codepoint,bool * next_is_consumed)223 int GetNormalizedCodePoint(char32_t codepoint,
224 char32_t next_codepoint,
225 bool *next_is_consumed) {
226 if (next_is_consumed != NULL) {
227 *next_is_consumed = false;
228 }
229
230 if (codepoint <= 0x0020 || codepoint == 0x3000) {
231 // Whitespaces. Keep it as is.
232 return codepoint;
233 } else if ((0x0021 <= codepoint && codepoint <= 0x007E) ||
234 (0xFF01 <= codepoint && codepoint <= 0xFF5E)) {
235 // Ascii and fullwidth ascii. Keep it as is
236 return codepoint;
237 } else if (codepoint == 0x02DC || codepoint == 0x223C) {
238 // tilde
239 return 0xFF5E;
240 } else if (codepoint <= 0x3040 ||
241 (0x3100 <= codepoint && codepoint < 0xFF00) ||
242 codepoint == CODEPOINT_FOR_NULL_STR) {
243 // Keep it as is.
244 return codepoint;
245 }
246
247 // Below is Kana-related handling.
248
249 return GetNormalizedKana(codepoint, next_codepoint, next_is_consumed);
250 }
251
GetExpectedString(const char * src,char ** dst,size_t * dst_len,int (* get_codepoint_function)(char32_t,char32_t,bool *))252 static bool GetExpectedString(
253 const char *src, char **dst, size_t *dst_len,
254 int (*get_codepoint_function)(char32_t, char32_t, bool*)) {
255 if (dst == NULL || dst_len == NULL) {
256 return false;
257 }
258
259 if (src == NULL || *src == '\0') {
260 src = STR_FOR_NULL_STR;
261 }
262
263 char32_t codepoints[MAX_CODEPOINTS]; // if array size is changed the for loop needs to be changed
264
265 ssize_t src_len = utf8_length(src);
266 if (src_len <= 0) {
267 return false;
268 }
269
270 bool next_is_consumed;
271 size_t j = 0;
272 for (size_t i = 0; i < (size_t)src_len && j < MAX_CODEPOINTS;) {
273 int32_t ret = utf32_from_utf8_at(src, src_len, i, &i);
274 if (ret < 0) {
275 // failed to parse UTF-8
276 return false;
277 }
278 ret = get_codepoint_function(
279 static_cast<char32_t>(ret),
280 i + 1 < (size_t)src_len ? src[i + 1] : 0,
281 &next_is_consumed);
282 if (ret > 0) {
283 codepoints[j] = static_cast<char32_t>(ret);
284 j++;
285 }
286 if (next_is_consumed) {
287 i++;
288 }
289 }
290 size_t length = j;
291
292 if (length == 0) {
293 // If all of codepoints are invalid, we place the string at the end of
294 // the list.
295 codepoints[0] = 0x10000 + CODEPOINT_FOR_NULL_STR;
296 length = 1;
297 }
298
299 ssize_t new_len = utf32_to_utf8_length(codepoints, length);
300 if (new_len < 0) {
301 return false;
302 }
303
304 *dst = static_cast<char *>(malloc(new_len + 1));
305 if (*dst == NULL) {
306 return false;
307 }
308
309 utf32_to_utf8(codepoints, length, *dst);
310
311 *dst_len = new_len;
312 return true;
313 }
314
GetNormalizedString(const char * src,char ** dst,size_t * len)315 bool GetNormalizedString(const char *src, char **dst, size_t *len) {
316 return GetExpectedString(src, dst, len, GetNormalizedCodePoint);
317 }
318
319 } // namespace android
320