• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
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/String8.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 0x308E: // xwa
157             return codepoint + 1;
158         case 0x3095: // xka
159             return 0x304B;
160         case 0x3096: // xku
161             return 0x304F;
162         default:
163             return codepoint;
164     }
165 }
166 
GetNormalizedKana(char32_t codepoint,char32_t next_codepoint,bool * next_is_consumed)167 static int GetNormalizedKana(char32_t codepoint,
168                              char32_t next_codepoint,
169                              bool *next_is_consumed) {
170     // First, convert fullwidth katakana and halfwidth katakana to hiragana.
171     if (0x30A1 <= codepoint && codepoint <= 0x30F6) {
172         // Make fullwidth katakana same as hiragana.
173         // 96 == 0x30A1 - 0x3041c
174         codepoint = codepoint - 96;
175     } else {
176         codepoint = GetHiraganaFromHalfwidthKatakana(
177                 codepoint, next_codepoint, next_is_consumed);
178     }
179 
180     // Normalize Hiragana.
181     return GetNormalizedHiragana(codepoint);
182 }
183 
GetPhoneticallySortableCodePoint(char32_t codepoint,char32_t next_codepoint,bool * next_is_consumed)184 int GetPhoneticallySortableCodePoint(char32_t codepoint,
185                                      char32_t next_codepoint,
186                                      bool *next_is_consumed) {
187     if (next_is_consumed != NULL) {
188         *next_is_consumed = false;
189     }
190 
191     if (codepoint <= 0x0020 || codepoint == 0x3000) {
192         // Whitespace should be ignored.
193         // Note: Formally, more "whitespace" exist. This block only
194         // handles part of them
195         return -1;
196     } else if ((0x0021 <= codepoint && codepoint <= 0x007E) ||
197                (0xFF01 <= codepoint && codepoint <= 0xFF5E)) {
198         // Ascii and fullwidth ascii
199 
200         if (0x0021 <= codepoint && codepoint <= 0x007E) {
201             // Convert ascii to fullwidth ascii so that they become
202             // behind hiragana.
203             // 65248 = 0xFF01 - 0x0021
204             codepoint += 65248;
205         }
206 
207         // Now, there is only fullwidth ascii.
208         if (0xFF10 <= codepoint && codepoint <= 0xFF19) {
209             // Numbers should be after alphabets but before symbols.
210             // 86 = 0xFF66
211             // (the beginning of halfwidth-katakankana space) - 0xFF10
212             return codepoint + 86;
213         } else if (0xFF41 <= codepoint && codepoint <= 0xFF5A) {
214             // Make lower alphabets same as capital alphabets.
215             // 32 = 0xFF41 - 0xFF21
216             return codepoint - 32;
217         } else if (0xFF01 <= codepoint && codepoint <= 0xFF0F) {
218             // Symbols (Ascii except alphabet nor number)
219             // These should be at the end of sorting, just after numebers
220             // (see below)
221             //
222             // We use halfwidth-katakana space for storing those symbols.
223             // 111 = 0xFF70 (0xFF19 + 86 + 1) - 0xFF01
224             return codepoint + 111;
225         } else if (0xFF1A <= codepoint && codepoint <= 0xFF20) {
226             // Symbols (cont.)
227             // 101 = 0xFF7F (0xFF0F + 111 + 1) - 0xFF1A
228             return codepoint + 101;
229         } else if (0xFF3B <= codepoint && codepoint <= 0xFF40) {
230             // Symbols (cont.)
231             // 75 = 0xFF86 (0xFF20 + 101 + 1) - 0xFF3B (= 101 - 26)
232             return codepoint + 75;
233         } else if (0xFF5B <= codepoint && codepoint <= 0xFF5E) {
234             // Symbols (cont.)
235             // 49 = 0xFF8C (0xFF40 + 75 + 1) - 0xFF5B (= 75 - 26)
236             return codepoint + 49;
237         } else {
238             return codepoint;
239         }
240     } else if (codepoint == 0x02DC || codepoint == 0x223C) {
241         // tilde
242         return 0xFF5E;
243     } else if (codepoint <= 0x3040 ||
244                (0x3100 <= codepoint && codepoint < 0xFF00) ||
245                codepoint == CODEPOINT_FOR_NULL_STR) {
246         // Move Kanji and other non-Japanese characters behind symbols.
247         return codepoint + 0x10000;
248     }
249 
250     // Below is Kana-related handling.
251 
252     return GetNormalizedKana(codepoint, next_codepoint, next_is_consumed);
253 }
254 
GetNormalizedCodePoint(char32_t codepoint,char32_t next_codepoint,bool * next_is_consumed)255 int GetNormalizedCodePoint(char32_t codepoint,
256                            char32_t next_codepoint,
257                            bool *next_is_consumed) {
258     if (next_is_consumed != NULL) {
259         *next_is_consumed = false;
260     }
261 
262     if (codepoint <= 0x0020 || codepoint == 0x3000) {
263         // Whitespaces. Keep it as is.
264         return codepoint;
265     } else if ((0x0021 <= codepoint && codepoint <= 0x007E) ||
266                (0xFF01 <= codepoint && codepoint <= 0xFF5E)) {
267         // Ascii and fullwidth ascii. Keep it as is
268         return codepoint;
269     } else if (codepoint == 0x02DC || codepoint == 0x223C) {
270         // tilde
271         return 0xFF5E;
272     } else if (codepoint <= 0x3040 ||
273                (0x3100 <= codepoint && codepoint < 0xFF00) ||
274                codepoint == CODEPOINT_FOR_NULL_STR) {
275         // Keep it as is.
276         return codepoint;
277     }
278 
279     // Below is Kana-related handling.
280 
281     return GetNormalizedKana(codepoint, next_codepoint, next_is_consumed);
282 }
283 
GetExpectedString(const char * src,char ** dst,size_t * dst_len,int (* get_codepoint_function)(char32_t,char32_t,bool *))284 static bool GetExpectedString(
285     const char *src, char **dst, size_t *dst_len,
286     int (*get_codepoint_function)(char32_t, char32_t, bool*)) {
287     if (dst == NULL || dst_len == NULL) {
288         return false;
289     }
290 
291     if (src == NULL || *src == '\0') {
292         src = STR_FOR_NULL_STR;
293     }
294 
295     char32_t codepoints[MAX_CODEPOINTS]; // if array size is changed the for loop needs to be changed
296 
297     size_t src_len = utf8_length(src);
298     if (src_len == 0) {
299         return false;
300     }
301     bool next_is_consumed;
302     size_t j = 0;
303     for (size_t i = 0; i < src_len && j < MAX_CODEPOINTS;) {
304         int32_t ret = utf32_at(src, src_len, i, &i);
305         if (ret < 0) {
306             // failed to parse UTF-8
307             return false;
308         }
309         ret = get_codepoint_function(
310                 static_cast<char32_t>(ret),
311                 i + 1 < src_len ? src[i + 1] : 0,
312                 &next_is_consumed);
313         if (ret > 0) {
314             codepoints[j] = static_cast<char32_t>(ret);
315             j++;
316         }
317         if (next_is_consumed) {
318             i++;
319         }
320     }
321     size_t length = j;
322 
323     if (length == 0) {
324         // If all of codepoints are invalid, we place the string at the end of
325         // the list.
326         codepoints[0] = 0x10000 + CODEPOINT_FOR_NULL_STR;
327         length = 1;
328     }
329 
330     size_t new_len = utf8_length_from_utf32(codepoints, length);
331     *dst = static_cast<char *>(malloc(new_len + 1));
332     if (*dst == NULL) {
333         return false;
334     }
335 
336     if (utf32_to_utf8(codepoints, length, *dst, new_len + 1) != new_len) {
337         free(*dst);
338         *dst = NULL;
339         return false;
340     }
341 
342     *dst_len = new_len;
343     return true;
344 }
345 
GetPhoneticallySortableString(const char * src,char ** dst,size_t * len)346 bool GetPhoneticallySortableString(const char *src, char **dst, size_t *len) {
347     return GetExpectedString(src, dst, len, GetPhoneticallySortableCodePoint);
348 }
349 
GetNormalizedString(const char * src,char ** dst,size_t * len)350 bool GetNormalizedString(const char *src, char **dst, size_t *len) {
351     return GetExpectedString(src, dst, len, GetNormalizedCodePoint);
352 }
353 
354 }  // namespace android
355