1 /**********************************************************************
2 regenc.c - Oniguruma (regular expression library)
3 **********************************************************************/
4 /*-
5 * Copyright (c) 2002-2007 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
6 * All rights reserved.
7 *
8 * (C) Copyright 2015 Hewlett Packard Enterprise Development LP<BR>
9 *
10 * Redistribution and use in source and binary forms, with or without
11 * modification, are permitted provided that the following conditions
12 * are met:
13 * 1. Redistributions of source code must retain the above copyright
14 * notice, this list of conditions and the following disclaimer.
15 * 2. Redistributions in binary form must reproduce the above copyright
16 * notice, this list of conditions and the following disclaimer in the
17 * documentation and/or other materials provided with the distribution.
18 *
19 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29 * SUCH DAMAGE.
30 */
31
32 #include "regint.h"
33
34 OnigEncoding OnigEncDefaultCharEncoding = ONIG_ENCODING_INIT_DEFAULT;
35
36 extern int
onigenc_init(void)37 onigenc_init(void)
38 {
39 return 0;
40 }
41
42 extern OnigEncoding
onigenc_get_default_encoding(void)43 onigenc_get_default_encoding(void)
44 {
45 return OnigEncDefaultCharEncoding;
46 }
47
48 extern int
onigenc_set_default_encoding(OnigEncoding enc)49 onigenc_set_default_encoding(OnigEncoding enc)
50 {
51 OnigEncDefaultCharEncoding = enc;
52 return 0;
53 }
54
55 extern UChar*
onigenc_get_right_adjust_char_head(OnigEncoding enc,const UChar * start,const UChar * s)56 onigenc_get_right_adjust_char_head(OnigEncoding enc, const UChar* start, const UChar* s)
57 {
58 UChar* p = ONIGENC_LEFT_ADJUST_CHAR_HEAD(enc, start, s);
59 if (p < s) {
60 p += enclen(enc, p);
61 }
62 return p;
63 }
64
65 extern UChar*
onigenc_get_right_adjust_char_head_with_prev(OnigEncoding enc,const UChar * start,const UChar * s,const UChar ** prev)66 onigenc_get_right_adjust_char_head_with_prev(OnigEncoding enc,
67 const UChar* start, const UChar* s, const UChar** prev)
68 {
69 UChar* p = ONIGENC_LEFT_ADJUST_CHAR_HEAD(enc, start, s);
70
71 if (p < s) {
72 if (prev) *prev = (const UChar* )p;
73 p += enclen(enc, p);
74 }
75 else {
76 if (prev) *prev = (const UChar* )NULL; /* Sorry */
77 }
78 return p;
79 }
80
81 extern UChar*
onigenc_get_prev_char_head(OnigEncoding enc,const UChar * start,const UChar * s)82 onigenc_get_prev_char_head(OnigEncoding enc, const UChar* start, const UChar* s)
83 {
84 if (s <= start)
85 return (UChar* )NULL;
86
87 return ONIGENC_LEFT_ADJUST_CHAR_HEAD(enc, start, s - 1);
88 }
89
90 extern UChar*
onigenc_step_back(OnigEncoding enc,const UChar * start,const UChar * s,int n)91 onigenc_step_back(OnigEncoding enc, const UChar* start, const UChar* s, int n)
92 {
93 while (ONIG_IS_NOT_NULL(s) && n-- > 0) {
94 if (s <= start)
95 return (UChar* )NULL;
96
97 s = ONIGENC_LEFT_ADJUST_CHAR_HEAD(enc, start, s - 1);
98 }
99 return (UChar* )s;
100 }
101
102 extern UChar*
onigenc_step(OnigEncoding enc,const UChar * p,const UChar * end,int n)103 onigenc_step(OnigEncoding enc, const UChar* p, const UChar* end, int n)
104 {
105 UChar* q = (UChar* )p;
106 while (n-- > 0) {
107 q += ONIGENC_MBC_ENC_LEN(enc, q);
108 }
109 return (q <= end ? q : NULL);
110 }
111
112 extern int
onigenc_strlen(OnigEncoding enc,const UChar * p,const UChar * end)113 onigenc_strlen(OnigEncoding enc, const UChar* p, const UChar* end)
114 {
115 int n = 0;
116 UChar* q = (UChar* )p;
117
118 while (q < end) {
119 q += ONIGENC_MBC_ENC_LEN(enc, q);
120 n++;
121 }
122 return n;
123 }
124
125 extern int
onigenc_strlen_null(OnigEncoding enc,const UChar * s)126 onigenc_strlen_null(OnigEncoding enc, const UChar* s)
127 {
128 int n = 0;
129 UChar* p = (UChar* )s;
130
131 while (1) {
132 if (*p == '\0') {
133 UChar* q;
134 int len = ONIGENC_MBC_MINLEN(enc);
135
136 if (len == 1) return n;
137 q = p + 1;
138 while (len > 1) {
139 if (*q != '\0') break;
140 q++;
141 len--;
142 }
143 if (len == 1) return n;
144 }
145 p += ONIGENC_MBC_ENC_LEN(enc, p);
146 n++;
147 }
148 }
149
150 extern int
onigenc_str_bytelen_null(OnigEncoding enc,const UChar * s)151 onigenc_str_bytelen_null(OnigEncoding enc, const UChar* s)
152 {
153 UChar* start = (UChar* )s;
154 UChar* p = (UChar* )s;
155
156 while (1) {
157 if (*p == '\0') {
158 UChar* q;
159 int len = ONIGENC_MBC_MINLEN(enc);
160
161 if (len == 1) return (int )(p - start);
162 q = p + 1;
163 while (len > 1) {
164 if (*q != '\0') break;
165 q++;
166 len--;
167 }
168 if (len == 1) return (int )(p - start);
169 }
170 p += ONIGENC_MBC_ENC_LEN(enc, p);
171 }
172 }
173
174 const UChar OnigEncAsciiToLowerCaseTable[] = {
175 0000, 0001, 0002, 0003, 0004, 0005, 0006, 0007,
176 0010, 0011, 0012, 0013, 0014, 0015, 0016, 0017,
177 0020, 0021, 0022, 0023, 0024, 0025, 0026, 0027,
178 0030, 0031, 0032, 0033, 0034, 0035, 0036, 0037,
179 0040, 0041, 0042, 0043, 0044, 0045, 0046, 0047,
180 0050, 0051, 0052, 0053, 0054, 0055, 0056, 0057,
181 0060, 0061, 0062, 0063, 0064, 0065, 0066, 0067,
182 0070, 0071, 0072, 0073, 0074, 0075, 0076, 0077,
183 0100, 0141, 0142, 0143, 0144, 0145, 0146, 0147,
184 0150, 0151, 0152, 0153, 0154, 0155, 0156, 0157,
185 0160, 0161, 0162, 0163, 0164, 0165, 0166, 0167,
186 0170, 0171, 0172, 0133, 0134, 0135, 0136, 0137,
187 0140, 0141, 0142, 0143, 0144, 0145, 0146, 0147,
188 0150, 0151, 0152, 0153, 0154, 0155, 0156, 0157,
189 0160, 0161, 0162, 0163, 0164, 0165, 0166, 0167,
190 0170, 0171, 0172, 0173, 0174, 0175, 0176, 0177,
191 0200, 0201, 0202, 0203, 0204, 0205, 0206, 0207,
192 0210, 0211, 0212, 0213, 0214, 0215, 0216, 0217,
193 0220, 0221, 0222, 0223, 0224, 0225, 0226, 0227,
194 0230, 0231, 0232, 0233, 0234, 0235, 0236, 0237,
195 0240, 0241, 0242, 0243, 0244, 0245, 0246, 0247,
196 0250, 0251, 0252, 0253, 0254, 0255, 0256, 0257,
197 0260, 0261, 0262, 0263, 0264, 0265, 0266, 0267,
198 0270, 0271, 0272, 0273, 0274, 0275, 0276, 0277,
199 0300, 0301, 0302, 0303, 0304, 0305, 0306, 0307,
200 0310, 0311, 0312, 0313, 0314, 0315, 0316, 0317,
201 0320, 0321, 0322, 0323, 0324, 0325, 0326, 0327,
202 0330, 0331, 0332, 0333, 0334, 0335, 0336, 0337,
203 0340, 0341, 0342, 0343, 0344, 0345, 0346, 0347,
204 0350, 0351, 0352, 0353, 0354, 0355, 0356, 0357,
205 0360, 0361, 0362, 0363, 0364, 0365, 0366, 0367,
206 0370, 0371, 0372, 0373, 0374, 0375, 0376, 0377,
207 };
208
209 #ifdef USE_UPPER_CASE_TABLE
210 const UChar OnigEncAsciiToUpperCaseTable[256] = {
211 0000, 0001, 0002, 0003, 0004, 0005, 0006, 0007,
212 0010, 0011, 0012, 0013, 0014, 0015, 0016, 0017,
213 0020, 0021, 0022, 0023, 0024, 0025, 0026, 0027,
214 0030, 0031, 0032, 0033, 0034, 0035, 0036, 0037,
215 0040, 0041, 0042, 0043, 0044, 0045, 0046, 0047,
216 0050, 0051, 0052, 0053, 0054, 0055, 0056, 0057,
217 0060, 0061, 0062, 0063, 0064, 0065, 0066, 0067,
218 0070, 0071, 0072, 0073, 0074, 0075, 0076, 0077,
219 0100, 0101, 0102, 0103, 0104, 0105, 0106, 0107,
220 0110, 0111, 0112, 0113, 0114, 0115, 0116, 0117,
221 0120, 0121, 0122, 0123, 0124, 0125, 0126, 0127,
222 0130, 0131, 0132, 0133, 0134, 0135, 0136, 0137,
223 0140, 0101, 0102, 0103, 0104, 0105, 0106, 0107,
224 0110, 0111, 0112, 0113, 0114, 0115, 0116, 0117,
225 0120, 0121, 0122, 0123, 0124, 0125, 0126, 0127,
226 0130, 0131, 0132, 0173, 0174, 0175, 0176, 0177,
227 0200, 0201, 0202, 0203, 0204, 0205, 0206, 0207,
228 0210, 0211, 0212, 0213, 0214, 0215, 0216, 0217,
229 0220, 0221, 0222, 0223, 0224, 0225, 0226, 0227,
230 0230, 0231, 0232, 0233, 0234, 0235, 0236, 0237,
231 0240, 0241, 0242, 0243, 0244, 0245, 0246, 0247,
232 0250, 0251, 0252, 0253, 0254, 0255, 0256, 0257,
233 0260, 0261, 0262, 0263, 0264, 0265, 0266, 0267,
234 0270, 0271, 0272, 0273, 0274, 0275, 0276, 0277,
235 0300, 0301, 0302, 0303, 0304, 0305, 0306, 0307,
236 0310, 0311, 0312, 0313, 0314, 0315, 0316, 0317,
237 0320, 0321, 0322, 0323, 0324, 0325, 0326, 0327,
238 0330, 0331, 0332, 0333, 0334, 0335, 0336, 0337,
239 0340, 0341, 0342, 0343, 0344, 0345, 0346, 0347,
240 0350, 0351, 0352, 0353, 0354, 0355, 0356, 0357,
241 0360, 0361, 0362, 0363, 0364, 0365, 0366, 0367,
242 0370, 0371, 0372, 0373, 0374, 0375, 0376, 0377,
243 };
244 #endif
245
246 const unsigned short OnigEncAsciiCtypeTable[256] = {
247 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008,
248 0x4008, 0x420c, 0x4209, 0x4208, 0x4208, 0x4208, 0x4008, 0x4008,
249 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008,
250 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008, 0x4008,
251 0x4284, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0,
252 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0,
253 0x78b0, 0x78b0, 0x78b0, 0x78b0, 0x78b0, 0x78b0, 0x78b0, 0x78b0,
254 0x78b0, 0x78b0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x41a0,
255 0x41a0, 0x7ca2, 0x7ca2, 0x7ca2, 0x7ca2, 0x7ca2, 0x7ca2, 0x74a2,
256 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2,
257 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2, 0x74a2,
258 0x74a2, 0x74a2, 0x74a2, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x51a0,
259 0x41a0, 0x78e2, 0x78e2, 0x78e2, 0x78e2, 0x78e2, 0x78e2, 0x70e2,
260 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2,
261 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2, 0x70e2,
262 0x70e2, 0x70e2, 0x70e2, 0x41a0, 0x41a0, 0x41a0, 0x41a0, 0x4008,
263 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
264 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
265 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
266 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
267 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
268 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
269 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
270 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
271 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
272 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
273 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
274 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
275 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
276 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
277 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
278 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000
279 };
280
281 const UChar OnigEncISO_8859_1_ToLowerCaseTable[256] = {
282 0000, 0001, 0002, 0003, 0004, 0005, 0006, 0007,
283 0010, 0011, 0012, 0013, 0014, 0015, 0016, 0017,
284 0020, 0021, 0022, 0023, 0024, 0025, 0026, 0027,
285 0030, 0031, 0032, 0033, 0034, 0035, 0036, 0037,
286 0040, 0041, 0042, 0043, 0044, 0045, 0046, 0047,
287 0050, 0051, 0052, 0053, 0054, 0055, 0056, 0057,
288 0060, 0061, 0062, 0063, 0064, 0065, 0066, 0067,
289 0070, 0071, 0072, 0073, 0074, 0075, 0076, 0077,
290 0100, 0141, 0142, 0143, 0144, 0145, 0146, 0147,
291 0150, 0151, 0152, 0153, 0154, 0155, 0156, 0157,
292 0160, 0161, 0162, 0163, 0164, 0165, 0166, 0167,
293 0170, 0171, 0172, 0133, 0134, 0135, 0136, 0137,
294 0140, 0141, 0142, 0143, 0144, 0145, 0146, 0147,
295 0150, 0151, 0152, 0153, 0154, 0155, 0156, 0157,
296 0160, 0161, 0162, 0163, 0164, 0165, 0166, 0167,
297 0170, 0171, 0172, 0173, 0174, 0175, 0176, 0177,
298 0200, 0201, 0202, 0203, 0204, 0205, 0206, 0207,
299 0210, 0211, 0212, 0213, 0214, 0215, 0216, 0217,
300 0220, 0221, 0222, 0223, 0224, 0225, 0226, 0227,
301 0230, 0231, 0232, 0233, 0234, 0235, 0236, 0237,
302 0240, 0241, 0242, 0243, 0244, 0245, 0246, 0247,
303 0250, 0251, 0252, 0253, 0254, 0255, 0256, 0257,
304 0260, 0261, 0262, 0263, 0264, 0265, 0266, 0267,
305 0270, 0271, 0272, 0273, 0274, 0275, 0276, 0277,
306 0340, 0341, 0342, 0343, 0344, 0345, 0346, 0347,
307 0350, 0351, 0352, 0353, 0354, 0355, 0356, 0357,
308 0360, 0361, 0362, 0363, 0364, 0365, 0366, 0327,
309 0370, 0371, 0372, 0373, 0374, 0375, 0376, 0337,
310 0340, 0341, 0342, 0343, 0344, 0345, 0346, 0347,
311 0350, 0351, 0352, 0353, 0354, 0355, 0356, 0357,
312 0360, 0361, 0362, 0363, 0364, 0365, 0366, 0367,
313 0370, 0371, 0372, 0373, 0374, 0375, 0376, 0377
314 };
315
316 #ifdef USE_UPPER_CASE_TABLE
317 const UChar OnigEncISO_8859_1_ToUpperCaseTable[256] = {
318 0000, 0001, 0002, 0003, 0004, 0005, 0006, 0007,
319 0010, 0011, 0012, 0013, 0014, 0015, 0016, 0017,
320 0020, 0021, 0022, 0023, 0024, 0025, 0026, 0027,
321 0030, 0031, 0032, 0033, 0034, 0035, 0036, 0037,
322 0040, 0041, 0042, 0043, 0044, 0045, 0046, 0047,
323 0050, 0051, 0052, 0053, 0054, 0055, 0056, 0057,
324 0060, 0061, 0062, 0063, 0064, 0065, 0066, 0067,
325 0070, 0071, 0072, 0073, 0074, 0075, 0076, 0077,
326 0100, 0101, 0102, 0103, 0104, 0105, 0106, 0107,
327 0110, 0111, 0112, 0113, 0114, 0115, 0116, 0117,
328 0120, 0121, 0122, 0123, 0124, 0125, 0126, 0127,
329 0130, 0131, 0132, 0133, 0134, 0135, 0136, 0137,
330 0140, 0101, 0102, 0103, 0104, 0105, 0106, 0107,
331 0110, 0111, 0112, 0113, 0114, 0115, 0116, 0117,
332 0120, 0121, 0122, 0123, 0124, 0125, 0126, 0127,
333 0130, 0131, 0132, 0173, 0174, 0175, 0176, 0177,
334 0200, 0201, 0202, 0203, 0204, 0205, 0206, 0207,
335 0210, 0211, 0212, 0213, 0214, 0215, 0216, 0217,
336 0220, 0221, 0222, 0223, 0224, 0225, 0226, 0227,
337 0230, 0231, 0232, 0233, 0234, 0235, 0236, 0237,
338 0240, 0241, 0242, 0243, 0244, 0245, 0246, 0247,
339 0250, 0251, 0252, 0253, 0254, 0255, 0256, 0257,
340 0260, 0261, 0262, 0263, 0264, 0265, 0266, 0267,
341 0270, 0271, 0272, 0273, 0274, 0275, 0276, 0277,
342 0300, 0301, 0302, 0303, 0304, 0305, 0306, 0307,
343 0310, 0311, 0312, 0313, 0314, 0315, 0316, 0317,
344 0320, 0321, 0322, 0323, 0324, 0325, 0326, 0327,
345 0330, 0331, 0332, 0333, 0334, 0335, 0336, 0337,
346 0300, 0301, 0302, 0303, 0304, 0305, 0306, 0307,
347 0310, 0311, 0312, 0313, 0314, 0315, 0316, 0317,
348 0320, 0321, 0322, 0323, 0324, 0325, 0326, 0367,
349 0330, 0331, 0332, 0333, 0334, 0335, 0336, 0377,
350 };
351 #endif
352
353 extern void
onigenc_set_default_caseconv_table(const UChar * table ARG_UNUSED)354 onigenc_set_default_caseconv_table(const UChar* table ARG_UNUSED)
355 {
356 /* nothing */
357 /* obsoleted. */
358 }
359
360 extern UChar*
onigenc_get_left_adjust_char_head(OnigEncoding enc,const UChar * start,const UChar * s)361 onigenc_get_left_adjust_char_head(OnigEncoding enc, const UChar* start, const UChar* s)
362 {
363 return ONIGENC_LEFT_ADJUST_CHAR_HEAD(enc, start, s);
364 }
365
366 const OnigPairCaseFoldCodes OnigAsciiLowerMap[] = {
367 { 0x41, 0x61 },
368 { 0x42, 0x62 },
369 { 0x43, 0x63 },
370 { 0x44, 0x64 },
371 { 0x45, 0x65 },
372 { 0x46, 0x66 },
373 { 0x47, 0x67 },
374 { 0x48, 0x68 },
375 { 0x49, 0x69 },
376 { 0x4a, 0x6a },
377 { 0x4b, 0x6b },
378 { 0x4c, 0x6c },
379 { 0x4d, 0x6d },
380 { 0x4e, 0x6e },
381 { 0x4f, 0x6f },
382 { 0x50, 0x70 },
383 { 0x51, 0x71 },
384 { 0x52, 0x72 },
385 { 0x53, 0x73 },
386 { 0x54, 0x74 },
387 { 0x55, 0x75 },
388 { 0x56, 0x76 },
389 { 0x57, 0x77 },
390 { 0x58, 0x78 },
391 { 0x59, 0x79 },
392 { 0x5a, 0x7a }
393 };
394
395 extern int
onigenc_ascii_apply_all_case_fold(OnigCaseFoldType flag ARG_UNUSED,OnigApplyAllCaseFoldFunc f,void * arg)396 onigenc_ascii_apply_all_case_fold(OnigCaseFoldType flag ARG_UNUSED,
397 OnigApplyAllCaseFoldFunc f, void* arg)
398 {
399 OnigCodePoint code;
400 int i, r;
401
402 for (i = 0;
403 i < (int )(sizeof(OnigAsciiLowerMap)/sizeof(OnigPairCaseFoldCodes));
404 i++) {
405 code = OnigAsciiLowerMap[i].to;
406 r = (*f)(OnigAsciiLowerMap[i].from, &code, 1, arg);
407 if (r != 0) return r;
408
409 code = OnigAsciiLowerMap[i].from;
410 r = (*f)(OnigAsciiLowerMap[i].to, &code, 1, arg);
411 if (r != 0) return r;
412 }
413
414 return 0;
415 }
416
417 extern int
onigenc_ascii_get_case_fold_codes_by_str(OnigCaseFoldType flag ARG_UNUSED,const OnigUChar * p,const OnigUChar * end ARG_UNUSED,OnigCaseFoldCodeItem items[])418 onigenc_ascii_get_case_fold_codes_by_str(OnigCaseFoldType flag ARG_UNUSED,
419 const OnigUChar* p, const OnigUChar* end ARG_UNUSED,
420 OnigCaseFoldCodeItem items[])
421 {
422 if (0x41 <= *p && *p <= 0x5a) {
423 items[0].byte_len = 1;
424 items[0].code_len = 1;
425 items[0].code[0] = (OnigCodePoint )(*p + 0x20);
426 return 1;
427 }
428 else if (0x61 <= *p && *p <= 0x7a) {
429 items[0].byte_len = 1;
430 items[0].code_len = 1;
431 items[0].code[0] = (OnigCodePoint )(*p - 0x20);
432 return 1;
433 }
434 else
435 return 0;
436 }
437
438 static int
ss_apply_all_case_fold(OnigCaseFoldType flag ARG_UNUSED,OnigApplyAllCaseFoldFunc f,void * arg)439 ss_apply_all_case_fold(OnigCaseFoldType flag ARG_UNUSED,
440 OnigApplyAllCaseFoldFunc f, void* arg)
441 {
442 static OnigCodePoint ss[] = { 0x73, 0x73 };
443
444 return (*f)((OnigCodePoint )0xdf, ss, 2, arg);
445 }
446
447 extern int
onigenc_apply_all_case_fold_with_map(int map_size,const OnigPairCaseFoldCodes map[],int ess_tsett_flag,OnigCaseFoldType flag,OnigApplyAllCaseFoldFunc f,void * arg)448 onigenc_apply_all_case_fold_with_map(int map_size,
449 const OnigPairCaseFoldCodes map[],
450 int ess_tsett_flag, OnigCaseFoldType flag,
451 OnigApplyAllCaseFoldFunc f, void* arg)
452 {
453 OnigCodePoint code;
454 int i, r;
455
456 r = onigenc_ascii_apply_all_case_fold(flag, f, arg);
457 if (r != 0) return r;
458
459 for (i = 0; i < map_size; i++) {
460 code = map[i].to;
461 r = (*f)(map[i].from, &code, 1, arg);
462 if (r != 0) return r;
463
464 code = map[i].from;
465 r = (*f)(map[i].to, &code, 1, arg);
466 if (r != 0) return r;
467 }
468
469 if (ess_tsett_flag != 0)
470 return ss_apply_all_case_fold(flag, f, arg);
471
472 return 0;
473 }
474
475 extern int
onigenc_get_case_fold_codes_by_str_with_map(int map_size,const OnigPairCaseFoldCodes map[],int ess_tsett_flag,OnigCaseFoldType flag ARG_UNUSED,const OnigUChar * p,const OnigUChar * end,OnigCaseFoldCodeItem items[])476 onigenc_get_case_fold_codes_by_str_with_map(int map_size,
477 const OnigPairCaseFoldCodes map[],
478 int ess_tsett_flag, OnigCaseFoldType flag ARG_UNUSED,
479 const OnigUChar* p, const OnigUChar* end, OnigCaseFoldCodeItem items[])
480 {
481 if (0x41 <= *p && *p <= 0x5a) {
482 items[0].byte_len = 1;
483 items[0].code_len = 1;
484 items[0].code[0] = (OnigCodePoint )(*p + 0x20);
485 if (*p == 0x53 && ess_tsett_flag != 0 && end > p + 1
486 && (*(p+1) == 0x53 || *(p+1) == 0x73)) {
487 /* SS */
488 items[1].byte_len = 2;
489 items[1].code_len = 1;
490 items[1].code[0] = (OnigCodePoint )0xdf;
491 return 2;
492 }
493 else
494 return 1;
495 }
496 else if (0x61 <= *p && *p <= 0x7a) {
497 items[0].byte_len = 1;
498 items[0].code_len = 1;
499 items[0].code[0] = (OnigCodePoint )(*p - 0x20);
500 if (*p == 0x73 && ess_tsett_flag != 0 && end > p + 1
501 && (*(p+1) == 0x73 || *(p+1) == 0x53)) {
502 /* ss */
503 items[1].byte_len = 2;
504 items[1].code_len = 1;
505 items[1].code[0] = (OnigCodePoint )0xdf;
506 return 2;
507 }
508 else
509 return 1;
510 }
511 else if (*p == 0xdf && ess_tsett_flag != 0) {
512 items[0].byte_len = 1;
513 items[0].code_len = 2;
514 items[0].code[0] = (OnigCodePoint )'s';
515 items[0].code[1] = (OnigCodePoint )'s';
516
517 items[1].byte_len = 1;
518 items[1].code_len = 2;
519 items[1].code[0] = (OnigCodePoint )'S';
520 items[1].code[1] = (OnigCodePoint )'S';
521
522 items[2].byte_len = 1;
523 items[2].code_len = 2;
524 items[2].code[0] = (OnigCodePoint )'s';
525 items[2].code[1] = (OnigCodePoint )'S';
526
527 items[3].byte_len = 1;
528 items[3].code_len = 2;
529 items[3].code[0] = (OnigCodePoint )'S';
530 items[3].code[1] = (OnigCodePoint )'s';
531
532 return 4;
533 }
534 else {
535 int i;
536
537 for (i = 0; i < map_size; i++) {
538 if (*p == map[i].from) {
539 items[0].byte_len = 1;
540 items[0].code_len = 1;
541 items[0].code[0] = map[i].to;
542 return 1;
543 }
544 else if (*p == map[i].to) {
545 items[0].byte_len = 1;
546 items[0].code_len = 1;
547 items[0].code[0] = map[i].from;
548 return 1;
549 }
550 }
551 }
552
553 return 0;
554 }
555
556
557 extern int
onigenc_not_support_get_ctype_code_range(OnigCtype ctype ARG_UNUSED,OnigCodePoint * sb_out ARG_UNUSED,const OnigCodePoint * ranges[]ARG_UNUSED)558 onigenc_not_support_get_ctype_code_range(OnigCtype ctype ARG_UNUSED,
559 OnigCodePoint* sb_out ARG_UNUSED,
560 const OnigCodePoint* ranges[] ARG_UNUSED)
561 {
562 return ONIG_NO_SUPPORT_CONFIG;
563 }
564
565 extern int
onigenc_is_mbc_newline_0x0a(const UChar * p,const UChar * end)566 onigenc_is_mbc_newline_0x0a(const UChar* p, const UChar* end)
567 {
568 if (p < end) {
569 if (*p == 0x0a) return 1;
570 }
571 return 0;
572 }
573
574 /* for single byte encodings */
575 extern int
onigenc_ascii_mbc_case_fold(OnigCaseFoldType flag ARG_UNUSED,const UChar ** p,const UChar * end ARG_UNUSED,UChar * lower)576 onigenc_ascii_mbc_case_fold(OnigCaseFoldType flag ARG_UNUSED, const UChar** p,
577 const UChar*end ARG_UNUSED, UChar* lower)
578 {
579 *lower = ONIGENC_ASCII_CODE_TO_LOWER_CASE(**p);
580
581 (*p)++;
582 return 1; /* return byte length of converted char to lower */
583 }
584
585 #if 0
586 extern int
587 onigenc_ascii_is_mbc_ambiguous(OnigCaseFoldType flag,
588 const UChar** pp, const UChar* end)
589 {
590 const UChar* p = *pp;
591
592 (*pp)++;
593 return ONIGENC_IS_ASCII_CODE_CASE_AMBIG(*p);
594 }
595 #endif
596
597 extern int
onigenc_single_byte_mbc_enc_len(const UChar * p ARG_UNUSED)598 onigenc_single_byte_mbc_enc_len(const UChar* p ARG_UNUSED)
599 {
600 return 1;
601 }
602
603 extern OnigCodePoint
onigenc_single_byte_mbc_to_code(const UChar * p,const UChar * end ARG_UNUSED)604 onigenc_single_byte_mbc_to_code(const UChar* p, const UChar* end ARG_UNUSED)
605 {
606 return (OnigCodePoint )(*p);
607 }
608
609 extern int
onigenc_single_byte_code_to_mbclen(OnigCodePoint code ARG_UNUSED)610 onigenc_single_byte_code_to_mbclen(OnigCodePoint code ARG_UNUSED)
611 {
612 return (code < 0x100 ? 1 : ONIGERR_INVALID_CODE_POINT_VALUE);
613 }
614
615 extern int
onigenc_single_byte_code_to_mbc(OnigCodePoint code,UChar * buf)616 onigenc_single_byte_code_to_mbc(OnigCodePoint code, UChar *buf)
617 {
618 *buf = (UChar )(code & 0xff);
619 return 1;
620 }
621
622 extern UChar*
onigenc_single_byte_left_adjust_char_head(const UChar * start ARG_UNUSED,const UChar * s)623 onigenc_single_byte_left_adjust_char_head(const UChar* start ARG_UNUSED,
624 const UChar* s)
625 {
626 return (UChar* )s;
627 }
628
629 extern int
onigenc_always_true_is_allowed_reverse_match(const UChar * s ARG_UNUSED,const UChar * end ARG_UNUSED)630 onigenc_always_true_is_allowed_reverse_match(const UChar* s ARG_UNUSED,
631 const UChar* end ARG_UNUSED)
632 {
633 return TRUE;
634 }
635
636 extern int
onigenc_always_false_is_allowed_reverse_match(const UChar * s ARG_UNUSED,const UChar * end ARG_UNUSED)637 onigenc_always_false_is_allowed_reverse_match(const UChar* s ARG_UNUSED,
638 const UChar* end ARG_UNUSED)
639 {
640 return FALSE;
641 }
642
643 extern OnigCodePoint
onigenc_mbn_mbc_to_code(OnigEncoding enc,const UChar * p,const UChar * end)644 onigenc_mbn_mbc_to_code(OnigEncoding enc, const UChar* p, const UChar* end)
645 {
646 int c, i, len;
647 OnigCodePoint n;
648
649 len = enclen(enc, p);
650 n = (OnigCodePoint )(*p++);
651 if (len == 1) return n;
652
653 for (i = 1; i < len; i++) {
654 if (p >= end) break;
655 c = *p++;
656 n <<= 8; n += c;
657 }
658 return n;
659 }
660
661 extern int
onigenc_mbn_mbc_case_fold(OnigEncoding enc,OnigCaseFoldType flag ARG_UNUSED,const UChar ** pp,const UChar * end ARG_UNUSED,UChar * lower)662 onigenc_mbn_mbc_case_fold(OnigEncoding enc, OnigCaseFoldType flag ARG_UNUSED,
663 const UChar** pp, const UChar* end ARG_UNUSED,
664 UChar* lower)
665 {
666 int len;
667 const UChar *p = *pp;
668
669 if (ONIGENC_IS_MBC_ASCII(p)) {
670 *lower = ONIGENC_ASCII_CODE_TO_LOWER_CASE(*p);
671 (*pp)++;
672 return 1;
673 }
674 else {
675 int i;
676
677 len = enclen(enc, p);
678 for (i = 0; i < len; i++) {
679 *lower++ = *p++;
680 }
681 (*pp) += len;
682 return len; /* return byte length of converted to lower char */
683 }
684 }
685
686 #if 0
687 extern int
688 onigenc_mbn_is_mbc_ambiguous(OnigEncoding enc, OnigCaseFoldType flag,
689 const UChar** pp, const UChar* end)
690 {
691 const UChar* p = *pp;
692
693 if (ONIGENC_IS_MBC_ASCII(p)) {
694 (*pp)++;
695 return ONIGENC_IS_ASCII_CODE_CASE_AMBIG(*p);
696 }
697
698 (*pp) += enclen(enc, p);
699 return FALSE;
700 }
701 #endif
702
703 extern int
onigenc_mb2_code_to_mbclen(OnigCodePoint code)704 onigenc_mb2_code_to_mbclen(OnigCodePoint code)
705 {
706 if ((code & 0xff00) != 0) return 2;
707 else return 1;
708 }
709
710 extern int
onigenc_mb4_code_to_mbclen(OnigCodePoint code)711 onigenc_mb4_code_to_mbclen(OnigCodePoint code)
712 {
713 if ((code & 0xff000000) != 0) return 4;
714 else if ((code & 0xff0000) != 0) return 3;
715 else if ((code & 0xff00) != 0) return 2;
716 else return 1;
717 }
718
719 extern int
onigenc_mb2_code_to_mbc(OnigEncoding enc,OnigCodePoint code,UChar * buf)720 onigenc_mb2_code_to_mbc(OnigEncoding enc, OnigCodePoint code, UChar *buf)
721 {
722 UChar *p = buf;
723
724 if ((code & 0xff00) != 0) {
725 *p++ = (UChar )((code >> 8) & 0xff);
726 }
727 *p++ = (UChar )(code & 0xff);
728
729 #if 1
730 if (enclen(enc, buf) != (p - buf))
731 return ONIGERR_INVALID_CODE_POINT_VALUE;
732 #endif
733 return (int)(p - buf);
734 }
735
736 extern int
onigenc_mb4_code_to_mbc(OnigEncoding enc,OnigCodePoint code,UChar * buf)737 onigenc_mb4_code_to_mbc(OnigEncoding enc, OnigCodePoint code, UChar *buf)
738 {
739 UChar *p = buf;
740
741 if ((code & 0xff000000) != 0) {
742 *p++ = (UChar )((code >> 24) & 0xff);
743 }
744 if ((code & 0xff0000) != 0 || p != buf) {
745 *p++ = (UChar )((code >> 16) & 0xff);
746 }
747 if ((code & 0xff00) != 0 || p != buf) {
748 *p++ = (UChar )((code >> 8) & 0xff);
749 }
750 *p++ = (UChar )(code & 0xff);
751
752 #if 1
753 if (enclen(enc, buf) != (p - buf))
754 return ONIGERR_INVALID_CODE_POINT_VALUE;
755 #endif
756 return (int)(p - buf);
757 }
758
759 extern int
onigenc_minimum_property_name_to_ctype(OnigEncoding enc,UChar * p,UChar * end)760 onigenc_minimum_property_name_to_ctype(OnigEncoding enc, UChar* p, UChar* end)
761 {
762 static PosixBracketEntryType PBS[] = {
763 { (UChar* )"Alnum", ONIGENC_CTYPE_ALNUM, 5 },
764 { (UChar* )"Alpha", ONIGENC_CTYPE_ALPHA, 5 },
765 { (UChar* )"Blank", ONIGENC_CTYPE_BLANK, 5 },
766 { (UChar* )"Cntrl", ONIGENC_CTYPE_CNTRL, 5 },
767 { (UChar* )"Digit", ONIGENC_CTYPE_DIGIT, 5 },
768 { (UChar* )"Graph", ONIGENC_CTYPE_GRAPH, 5 },
769 { (UChar* )"Lower", ONIGENC_CTYPE_LOWER, 5 },
770 { (UChar* )"Print", ONIGENC_CTYPE_PRINT, 5 },
771 { (UChar* )"Punct", ONIGENC_CTYPE_PUNCT, 5 },
772 { (UChar* )"Space", ONIGENC_CTYPE_SPACE, 5 },
773 { (UChar* )"Upper", ONIGENC_CTYPE_UPPER, 5 },
774 { (UChar* )"XDigit", ONIGENC_CTYPE_XDIGIT, 6 },
775 { (UChar* )"ASCII", ONIGENC_CTYPE_ASCII, 5 },
776 { (UChar* )"Word", ONIGENC_CTYPE_WORD, 4 },
777 { (UChar* )NULL, -1, 0 }
778 };
779
780 PosixBracketEntryType *pb;
781 int len;
782
783 len = onigenc_strlen(enc, p, end);
784 for (pb = PBS; IS_NOT_NULL(pb->name); pb++) {
785 if (len == pb->len &&
786 onigenc_with_ascii_strncmp(enc, p, end, pb->name, pb->len) == 0)
787 return pb->ctype;
788 }
789
790 return ONIGERR_INVALID_CHAR_PROPERTY_NAME;
791 }
792
793 extern int
onigenc_mb2_is_code_ctype(OnigEncoding enc,OnigCodePoint code,unsigned int ctype)794 onigenc_mb2_is_code_ctype(OnigEncoding enc, OnigCodePoint code,
795 unsigned int ctype)
796 {
797 if (code < 128)
798 return ONIGENC_IS_ASCII_CODE_CTYPE(code, ctype);
799 else {
800 if (CTYPE_IS_WORD_GRAPH_PRINT(ctype)) {
801 return (ONIGENC_CODE_TO_MBCLEN(enc, code) > 1 ? TRUE : FALSE);
802 }
803 }
804
805 return FALSE;
806 }
807
808 extern int
onigenc_mb4_is_code_ctype(OnigEncoding enc,OnigCodePoint code,unsigned int ctype)809 onigenc_mb4_is_code_ctype(OnigEncoding enc, OnigCodePoint code,
810 unsigned int ctype)
811 {
812 if (code < 128)
813 return ONIGENC_IS_ASCII_CODE_CTYPE(code, ctype);
814 else {
815 if (CTYPE_IS_WORD_GRAPH_PRINT(ctype)) {
816 return (ONIGENC_CODE_TO_MBCLEN(enc, code) > 1 ? TRUE : FALSE);
817 }
818 }
819
820 return FALSE;
821 }
822
823 extern int
onigenc_with_ascii_strncmp(OnigEncoding enc,const UChar * p,const UChar * end,const UChar * sascii,int n)824 onigenc_with_ascii_strncmp(OnigEncoding enc, const UChar* p, const UChar* end,
825 const UChar* sascii /* ascii */, int n)
826 {
827 int x, c;
828
829 while (n-- > 0) {
830 if (p >= end) return (int )(*sascii);
831
832 c = (int )ONIGENC_MBC_TO_CODE(enc, p, end);
833 x = *sascii - c;
834 if (x) return x;
835
836 sascii++;
837 p += enclen(enc, p);
838 }
839 return 0;
840 }
841
842 /* Property management */
843 static int
resize_property_list(int new_size,const OnigCodePoint *** plist,int * psize)844 resize_property_list(int new_size, const OnigCodePoint*** plist, int* psize)
845 {
846 int size;
847 const OnigCodePoint **list = *plist;
848
849 size = sizeof(OnigCodePoint*) * new_size;
850 if (IS_NULL(list)) {
851 list = (const OnigCodePoint** )xmalloc(size);
852 }
853 else {
854 list = (const OnigCodePoint** )xrealloc((void* )list, size, *psize * sizeof(OnigCodePoint*));
855 }
856
857 if (IS_NULL(list)) return ONIGERR_MEMORY;
858
859 *plist = list;
860 *psize = new_size;
861
862 return 0;
863 }
864
865 extern int
onigenc_property_list_add_property(UChar * name,const OnigCodePoint * prop,hash_table_type ** table,const OnigCodePoint *** plist,int * pnum,int * psize)866 onigenc_property_list_add_property(UChar* name, const OnigCodePoint* prop,
867 hash_table_type **table, const OnigCodePoint*** plist, int *pnum,
868 int *psize)
869 {
870 #define PROP_INIT_SIZE 16
871
872 int r;
873
874 if (*psize <= *pnum) {
875 int new_size = (*psize == 0 ? PROP_INIT_SIZE : *psize * 2);
876 r = resize_property_list(new_size, plist, psize);
877 if (r != 0) return r;
878 }
879
880 (*plist)[*pnum] = prop;
881
882 if (ONIG_IS_NULL(*table)) {
883 *table = onig_st_init_strend_table_with_size(PROP_INIT_SIZE);
884 if (ONIG_IS_NULL(*table)) return ONIGERR_MEMORY;
885 }
886
887 *pnum = *pnum + 1;
888 onig_st_insert_strend(*table, name, name + strlen_s((char* )name, MAX_STRING_SIZE),
889 (hash_data_type )(*pnum + ONIGENC_MAX_STD_CTYPE));
890 return 0;
891 }
892
893 extern int
onigenc_property_list_init(int (* f)(void))894 onigenc_property_list_init(int (*f)(void))
895 {
896 int r;
897
898 THREAD_ATOMIC_START;
899
900 r = f();
901
902 THREAD_ATOMIC_END;
903 return r;
904 }
905