| 1 | n/a | /* Copyright (c) 1998, 1999 Thai Open Source Software Center Ltd |
|---|
| 2 | n/a | See the file COPYING for copying permission. |
|---|
| 3 | n/a | */ |
|---|
| 4 | n/a | |
|---|
| 5 | n/a | /* This file is included! */ |
|---|
| 6 | n/a | #ifdef XML_TOK_IMPL_C |
|---|
| 7 | n/a | |
|---|
| 8 | n/a | #ifndef IS_INVALID_CHAR |
|---|
| 9 | n/a | #define IS_INVALID_CHAR(enc, ptr, n) (0) |
|---|
| 10 | n/a | #endif |
|---|
| 11 | n/a | |
|---|
| 12 | n/a | #define INVALID_LEAD_CASE(n, ptr, nextTokPtr) \ |
|---|
| 13 | n/a | case BT_LEAD ## n: \ |
|---|
| 14 | n/a | if (end - ptr < n) \ |
|---|
| 15 | n/a | return XML_TOK_PARTIAL_CHAR; \ |
|---|
| 16 | n/a | if (IS_INVALID_CHAR(enc, ptr, n)) { \ |
|---|
| 17 | n/a | *(nextTokPtr) = (ptr); \ |
|---|
| 18 | n/a | return XML_TOK_INVALID; \ |
|---|
| 19 | n/a | } \ |
|---|
| 20 | n/a | ptr += n; \ |
|---|
| 21 | n/a | break; |
|---|
| 22 | n/a | |
|---|
| 23 | n/a | #define INVALID_CASES(ptr, nextTokPtr) \ |
|---|
| 24 | n/a | INVALID_LEAD_CASE(2, ptr, nextTokPtr) \ |
|---|
| 25 | n/a | INVALID_LEAD_CASE(3, ptr, nextTokPtr) \ |
|---|
| 26 | n/a | INVALID_LEAD_CASE(4, ptr, nextTokPtr) \ |
|---|
| 27 | n/a | case BT_NONXML: \ |
|---|
| 28 | n/a | case BT_MALFORM: \ |
|---|
| 29 | n/a | case BT_TRAIL: \ |
|---|
| 30 | n/a | *(nextTokPtr) = (ptr); \ |
|---|
| 31 | n/a | return XML_TOK_INVALID; |
|---|
| 32 | n/a | |
|---|
| 33 | n/a | #define CHECK_NAME_CASE(n, enc, ptr, end, nextTokPtr) \ |
|---|
| 34 | n/a | case BT_LEAD ## n: \ |
|---|
| 35 | n/a | if (end - ptr < n) \ |
|---|
| 36 | n/a | return XML_TOK_PARTIAL_CHAR; \ |
|---|
| 37 | n/a | if (!IS_NAME_CHAR(enc, ptr, n)) { \ |
|---|
| 38 | n/a | *nextTokPtr = ptr; \ |
|---|
| 39 | n/a | return XML_TOK_INVALID; \ |
|---|
| 40 | n/a | } \ |
|---|
| 41 | n/a | ptr += n; \ |
|---|
| 42 | n/a | break; |
|---|
| 43 | n/a | |
|---|
| 44 | n/a | #define CHECK_NAME_CASES(enc, ptr, end, nextTokPtr) \ |
|---|
| 45 | n/a | case BT_NONASCII: \ |
|---|
| 46 | n/a | if (!IS_NAME_CHAR_MINBPC(enc, ptr)) { \ |
|---|
| 47 | n/a | *nextTokPtr = ptr; \ |
|---|
| 48 | n/a | return XML_TOK_INVALID; \ |
|---|
| 49 | n/a | } \ |
|---|
| 50 | n/a | case BT_NMSTRT: \ |
|---|
| 51 | n/a | case BT_HEX: \ |
|---|
| 52 | n/a | case BT_DIGIT: \ |
|---|
| 53 | n/a | case BT_NAME: \ |
|---|
| 54 | n/a | case BT_MINUS: \ |
|---|
| 55 | n/a | ptr += MINBPC(enc); \ |
|---|
| 56 | n/a | break; \ |
|---|
| 57 | n/a | CHECK_NAME_CASE(2, enc, ptr, end, nextTokPtr) \ |
|---|
| 58 | n/a | CHECK_NAME_CASE(3, enc, ptr, end, nextTokPtr) \ |
|---|
| 59 | n/a | CHECK_NAME_CASE(4, enc, ptr, end, nextTokPtr) |
|---|
| 60 | n/a | |
|---|
| 61 | n/a | #define CHECK_NMSTRT_CASE(n, enc, ptr, end, nextTokPtr) \ |
|---|
| 62 | n/a | case BT_LEAD ## n: \ |
|---|
| 63 | n/a | if (end - ptr < n) \ |
|---|
| 64 | n/a | return XML_TOK_PARTIAL_CHAR; \ |
|---|
| 65 | n/a | if (!IS_NMSTRT_CHAR(enc, ptr, n)) { \ |
|---|
| 66 | n/a | *nextTokPtr = ptr; \ |
|---|
| 67 | n/a | return XML_TOK_INVALID; \ |
|---|
| 68 | n/a | } \ |
|---|
| 69 | n/a | ptr += n; \ |
|---|
| 70 | n/a | break; |
|---|
| 71 | n/a | |
|---|
| 72 | n/a | #define CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr) \ |
|---|
| 73 | n/a | case BT_NONASCII: \ |
|---|
| 74 | n/a | if (!IS_NMSTRT_CHAR_MINBPC(enc, ptr)) { \ |
|---|
| 75 | n/a | *nextTokPtr = ptr; \ |
|---|
| 76 | n/a | return XML_TOK_INVALID; \ |
|---|
| 77 | n/a | } \ |
|---|
| 78 | n/a | case BT_NMSTRT: \ |
|---|
| 79 | n/a | case BT_HEX: \ |
|---|
| 80 | n/a | ptr += MINBPC(enc); \ |
|---|
| 81 | n/a | break; \ |
|---|
| 82 | n/a | CHECK_NMSTRT_CASE(2, enc, ptr, end, nextTokPtr) \ |
|---|
| 83 | n/a | CHECK_NMSTRT_CASE(3, enc, ptr, end, nextTokPtr) \ |
|---|
| 84 | n/a | CHECK_NMSTRT_CASE(4, enc, ptr, end, nextTokPtr) |
|---|
| 85 | n/a | |
|---|
| 86 | n/a | #ifndef PREFIX |
|---|
| 87 | n/a | #define PREFIX(ident) ident |
|---|
| 88 | n/a | #endif |
|---|
| 89 | n/a | |
|---|
| 90 | n/a | /* ptr points to character following "<!-" */ |
|---|
| 91 | n/a | |
|---|
| 92 | n/a | static int PTRCALL |
|---|
| 93 | n/a | PREFIX(scanComment)(const ENCODING *enc, const char *ptr, |
|---|
| 94 | n/a | const char *end, const char **nextTokPtr) |
|---|
| 95 | n/a | { |
|---|
| 96 | n/a | if (ptr != end) { |
|---|
| 97 | n/a | if (!CHAR_MATCHES(enc, ptr, ASCII_MINUS)) { |
|---|
| 98 | n/a | *nextTokPtr = ptr; |
|---|
| 99 | n/a | return XML_TOK_INVALID; |
|---|
| 100 | n/a | } |
|---|
| 101 | n/a | ptr += MINBPC(enc); |
|---|
| 102 | n/a | while (ptr != end) { |
|---|
| 103 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 104 | n/a | INVALID_CASES(ptr, nextTokPtr) |
|---|
| 105 | n/a | case BT_MINUS: |
|---|
| 106 | n/a | if ((ptr += MINBPC(enc)) == end) |
|---|
| 107 | n/a | return XML_TOK_PARTIAL; |
|---|
| 108 | n/a | if (CHAR_MATCHES(enc, ptr, ASCII_MINUS)) { |
|---|
| 109 | n/a | if ((ptr += MINBPC(enc)) == end) |
|---|
| 110 | n/a | return XML_TOK_PARTIAL; |
|---|
| 111 | n/a | if (!CHAR_MATCHES(enc, ptr, ASCII_GT)) { |
|---|
| 112 | n/a | *nextTokPtr = ptr; |
|---|
| 113 | n/a | return XML_TOK_INVALID; |
|---|
| 114 | n/a | } |
|---|
| 115 | n/a | *nextTokPtr = ptr + MINBPC(enc); |
|---|
| 116 | n/a | return XML_TOK_COMMENT; |
|---|
| 117 | n/a | } |
|---|
| 118 | n/a | break; |
|---|
| 119 | n/a | default: |
|---|
| 120 | n/a | ptr += MINBPC(enc); |
|---|
| 121 | n/a | break; |
|---|
| 122 | n/a | } |
|---|
| 123 | n/a | } |
|---|
| 124 | n/a | } |
|---|
| 125 | n/a | return XML_TOK_PARTIAL; |
|---|
| 126 | n/a | } |
|---|
| 127 | n/a | |
|---|
| 128 | n/a | /* ptr points to character following "<!" */ |
|---|
| 129 | n/a | |
|---|
| 130 | n/a | static int PTRCALL |
|---|
| 131 | n/a | PREFIX(scanDecl)(const ENCODING *enc, const char *ptr, |
|---|
| 132 | n/a | const char *end, const char **nextTokPtr) |
|---|
| 133 | n/a | { |
|---|
| 134 | n/a | if (ptr == end) |
|---|
| 135 | n/a | return XML_TOK_PARTIAL; |
|---|
| 136 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 137 | n/a | case BT_MINUS: |
|---|
| 138 | n/a | return PREFIX(scanComment)(enc, ptr + MINBPC(enc), end, nextTokPtr); |
|---|
| 139 | n/a | case BT_LSQB: |
|---|
| 140 | n/a | *nextTokPtr = ptr + MINBPC(enc); |
|---|
| 141 | n/a | return XML_TOK_COND_SECT_OPEN; |
|---|
| 142 | n/a | case BT_NMSTRT: |
|---|
| 143 | n/a | case BT_HEX: |
|---|
| 144 | n/a | ptr += MINBPC(enc); |
|---|
| 145 | n/a | break; |
|---|
| 146 | n/a | default: |
|---|
| 147 | n/a | *nextTokPtr = ptr; |
|---|
| 148 | n/a | return XML_TOK_INVALID; |
|---|
| 149 | n/a | } |
|---|
| 150 | n/a | while (ptr != end) { |
|---|
| 151 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 152 | n/a | case BT_PERCNT: |
|---|
| 153 | n/a | if (ptr + MINBPC(enc) == end) |
|---|
| 154 | n/a | return XML_TOK_PARTIAL; |
|---|
| 155 | n/a | /* don't allow <!ENTITY% foo "whatever"> */ |
|---|
| 156 | n/a | switch (BYTE_TYPE(enc, ptr + MINBPC(enc))) { |
|---|
| 157 | n/a | case BT_S: case BT_CR: case BT_LF: case BT_PERCNT: |
|---|
| 158 | n/a | *nextTokPtr = ptr; |
|---|
| 159 | n/a | return XML_TOK_INVALID; |
|---|
| 160 | n/a | } |
|---|
| 161 | n/a | /* fall through */ |
|---|
| 162 | n/a | case BT_S: case BT_CR: case BT_LF: |
|---|
| 163 | n/a | *nextTokPtr = ptr; |
|---|
| 164 | n/a | return XML_TOK_DECL_OPEN; |
|---|
| 165 | n/a | case BT_NMSTRT: |
|---|
| 166 | n/a | case BT_HEX: |
|---|
| 167 | n/a | ptr += MINBPC(enc); |
|---|
| 168 | n/a | break; |
|---|
| 169 | n/a | default: |
|---|
| 170 | n/a | *nextTokPtr = ptr; |
|---|
| 171 | n/a | return XML_TOK_INVALID; |
|---|
| 172 | n/a | } |
|---|
| 173 | n/a | } |
|---|
| 174 | n/a | return XML_TOK_PARTIAL; |
|---|
| 175 | n/a | } |
|---|
| 176 | n/a | |
|---|
| 177 | n/a | static int PTRCALL |
|---|
| 178 | n/a | PREFIX(checkPiTarget)(const ENCODING *enc, const char *ptr, |
|---|
| 179 | n/a | const char *end, int *tokPtr) |
|---|
| 180 | n/a | { |
|---|
| 181 | n/a | int upper = 0; |
|---|
| 182 | n/a | *tokPtr = XML_TOK_PI; |
|---|
| 183 | n/a | if (end - ptr != MINBPC(enc)*3) |
|---|
| 184 | n/a | return 1; |
|---|
| 185 | n/a | switch (BYTE_TO_ASCII(enc, ptr)) { |
|---|
| 186 | n/a | case ASCII_x: |
|---|
| 187 | n/a | break; |
|---|
| 188 | n/a | case ASCII_X: |
|---|
| 189 | n/a | upper = 1; |
|---|
| 190 | n/a | break; |
|---|
| 191 | n/a | default: |
|---|
| 192 | n/a | return 1; |
|---|
| 193 | n/a | } |
|---|
| 194 | n/a | ptr += MINBPC(enc); |
|---|
| 195 | n/a | switch (BYTE_TO_ASCII(enc, ptr)) { |
|---|
| 196 | n/a | case ASCII_m: |
|---|
| 197 | n/a | break; |
|---|
| 198 | n/a | case ASCII_M: |
|---|
| 199 | n/a | upper = 1; |
|---|
| 200 | n/a | break; |
|---|
| 201 | n/a | default: |
|---|
| 202 | n/a | return 1; |
|---|
| 203 | n/a | } |
|---|
| 204 | n/a | ptr += MINBPC(enc); |
|---|
| 205 | n/a | switch (BYTE_TO_ASCII(enc, ptr)) { |
|---|
| 206 | n/a | case ASCII_l: |
|---|
| 207 | n/a | break; |
|---|
| 208 | n/a | case ASCII_L: |
|---|
| 209 | n/a | upper = 1; |
|---|
| 210 | n/a | break; |
|---|
| 211 | n/a | default: |
|---|
| 212 | n/a | return 1; |
|---|
| 213 | n/a | } |
|---|
| 214 | n/a | if (upper) |
|---|
| 215 | n/a | return 0; |
|---|
| 216 | n/a | *tokPtr = XML_TOK_XML_DECL; |
|---|
| 217 | n/a | return 1; |
|---|
| 218 | n/a | } |
|---|
| 219 | n/a | |
|---|
| 220 | n/a | /* ptr points to character following "<?" */ |
|---|
| 221 | n/a | |
|---|
| 222 | n/a | static int PTRCALL |
|---|
| 223 | n/a | PREFIX(scanPi)(const ENCODING *enc, const char *ptr, |
|---|
| 224 | n/a | const char *end, const char **nextTokPtr) |
|---|
| 225 | n/a | { |
|---|
| 226 | n/a | int tok; |
|---|
| 227 | n/a | const char *target = ptr; |
|---|
| 228 | n/a | if (ptr == end) |
|---|
| 229 | n/a | return XML_TOK_PARTIAL; |
|---|
| 230 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 231 | n/a | CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr) |
|---|
| 232 | n/a | default: |
|---|
| 233 | n/a | *nextTokPtr = ptr; |
|---|
| 234 | n/a | return XML_TOK_INVALID; |
|---|
| 235 | n/a | } |
|---|
| 236 | n/a | while (ptr != end) { |
|---|
| 237 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 238 | n/a | CHECK_NAME_CASES(enc, ptr, end, nextTokPtr) |
|---|
| 239 | n/a | case BT_S: case BT_CR: case BT_LF: |
|---|
| 240 | n/a | if (!PREFIX(checkPiTarget)(enc, target, ptr, &tok)) { |
|---|
| 241 | n/a | *nextTokPtr = ptr; |
|---|
| 242 | n/a | return XML_TOK_INVALID; |
|---|
| 243 | n/a | } |
|---|
| 244 | n/a | ptr += MINBPC(enc); |
|---|
| 245 | n/a | while (ptr != end) { |
|---|
| 246 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 247 | n/a | INVALID_CASES(ptr, nextTokPtr) |
|---|
| 248 | n/a | case BT_QUEST: |
|---|
| 249 | n/a | ptr += MINBPC(enc); |
|---|
| 250 | n/a | if (ptr == end) |
|---|
| 251 | n/a | return XML_TOK_PARTIAL; |
|---|
| 252 | n/a | if (CHAR_MATCHES(enc, ptr, ASCII_GT)) { |
|---|
| 253 | n/a | *nextTokPtr = ptr + MINBPC(enc); |
|---|
| 254 | n/a | return tok; |
|---|
| 255 | n/a | } |
|---|
| 256 | n/a | break; |
|---|
| 257 | n/a | default: |
|---|
| 258 | n/a | ptr += MINBPC(enc); |
|---|
| 259 | n/a | break; |
|---|
| 260 | n/a | } |
|---|
| 261 | n/a | } |
|---|
| 262 | n/a | return XML_TOK_PARTIAL; |
|---|
| 263 | n/a | case BT_QUEST: |
|---|
| 264 | n/a | if (!PREFIX(checkPiTarget)(enc, target, ptr, &tok)) { |
|---|
| 265 | n/a | *nextTokPtr = ptr; |
|---|
| 266 | n/a | return XML_TOK_INVALID; |
|---|
| 267 | n/a | } |
|---|
| 268 | n/a | ptr += MINBPC(enc); |
|---|
| 269 | n/a | if (ptr == end) |
|---|
| 270 | n/a | return XML_TOK_PARTIAL; |
|---|
| 271 | n/a | if (CHAR_MATCHES(enc, ptr, ASCII_GT)) { |
|---|
| 272 | n/a | *nextTokPtr = ptr + MINBPC(enc); |
|---|
| 273 | n/a | return tok; |
|---|
| 274 | n/a | } |
|---|
| 275 | n/a | /* fall through */ |
|---|
| 276 | n/a | default: |
|---|
| 277 | n/a | *nextTokPtr = ptr; |
|---|
| 278 | n/a | return XML_TOK_INVALID; |
|---|
| 279 | n/a | } |
|---|
| 280 | n/a | } |
|---|
| 281 | n/a | return XML_TOK_PARTIAL; |
|---|
| 282 | n/a | } |
|---|
| 283 | n/a | |
|---|
| 284 | n/a | static int PTRCALL |
|---|
| 285 | n/a | PREFIX(scanCdataSection)(const ENCODING *enc, const char *ptr, |
|---|
| 286 | n/a | const char *end, const char **nextTokPtr) |
|---|
| 287 | n/a | { |
|---|
| 288 | n/a | static const char CDATA_LSQB[] = { ASCII_C, ASCII_D, ASCII_A, |
|---|
| 289 | n/a | ASCII_T, ASCII_A, ASCII_LSQB }; |
|---|
| 290 | n/a | int i; |
|---|
| 291 | n/a | /* CDATA[ */ |
|---|
| 292 | n/a | if (end - ptr < 6 * MINBPC(enc)) |
|---|
| 293 | n/a | return XML_TOK_PARTIAL; |
|---|
| 294 | n/a | for (i = 0; i < 6; i++, ptr += MINBPC(enc)) { |
|---|
| 295 | n/a | if (!CHAR_MATCHES(enc, ptr, CDATA_LSQB[i])) { |
|---|
| 296 | n/a | *nextTokPtr = ptr; |
|---|
| 297 | n/a | return XML_TOK_INVALID; |
|---|
| 298 | n/a | } |
|---|
| 299 | n/a | } |
|---|
| 300 | n/a | *nextTokPtr = ptr; |
|---|
| 301 | n/a | return XML_TOK_CDATA_SECT_OPEN; |
|---|
| 302 | n/a | } |
|---|
| 303 | n/a | |
|---|
| 304 | n/a | static int PTRCALL |
|---|
| 305 | n/a | PREFIX(cdataSectionTok)(const ENCODING *enc, const char *ptr, |
|---|
| 306 | n/a | const char *end, const char **nextTokPtr) |
|---|
| 307 | n/a | { |
|---|
| 308 | n/a | if (ptr == end) |
|---|
| 309 | n/a | return XML_TOK_NONE; |
|---|
| 310 | n/a | if (MINBPC(enc) > 1) { |
|---|
| 311 | n/a | size_t n = end - ptr; |
|---|
| 312 | n/a | if (n & (MINBPC(enc) - 1)) { |
|---|
| 313 | n/a | n &= ~(MINBPC(enc) - 1); |
|---|
| 314 | n/a | if (n == 0) |
|---|
| 315 | n/a | return XML_TOK_PARTIAL; |
|---|
| 316 | n/a | end = ptr + n; |
|---|
| 317 | n/a | } |
|---|
| 318 | n/a | } |
|---|
| 319 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 320 | n/a | case BT_RSQB: |
|---|
| 321 | n/a | ptr += MINBPC(enc); |
|---|
| 322 | n/a | if (ptr == end) |
|---|
| 323 | n/a | return XML_TOK_PARTIAL; |
|---|
| 324 | n/a | if (!CHAR_MATCHES(enc, ptr, ASCII_RSQB)) |
|---|
| 325 | n/a | break; |
|---|
| 326 | n/a | ptr += MINBPC(enc); |
|---|
| 327 | n/a | if (ptr == end) |
|---|
| 328 | n/a | return XML_TOK_PARTIAL; |
|---|
| 329 | n/a | if (!CHAR_MATCHES(enc, ptr, ASCII_GT)) { |
|---|
| 330 | n/a | ptr -= MINBPC(enc); |
|---|
| 331 | n/a | break; |
|---|
| 332 | n/a | } |
|---|
| 333 | n/a | *nextTokPtr = ptr + MINBPC(enc); |
|---|
| 334 | n/a | return XML_TOK_CDATA_SECT_CLOSE; |
|---|
| 335 | n/a | case BT_CR: |
|---|
| 336 | n/a | ptr += MINBPC(enc); |
|---|
| 337 | n/a | if (ptr == end) |
|---|
| 338 | n/a | return XML_TOK_PARTIAL; |
|---|
| 339 | n/a | if (BYTE_TYPE(enc, ptr) == BT_LF) |
|---|
| 340 | n/a | ptr += MINBPC(enc); |
|---|
| 341 | n/a | *nextTokPtr = ptr; |
|---|
| 342 | n/a | return XML_TOK_DATA_NEWLINE; |
|---|
| 343 | n/a | case BT_LF: |
|---|
| 344 | n/a | *nextTokPtr = ptr + MINBPC(enc); |
|---|
| 345 | n/a | return XML_TOK_DATA_NEWLINE; |
|---|
| 346 | n/a | INVALID_CASES(ptr, nextTokPtr) |
|---|
| 347 | n/a | default: |
|---|
| 348 | n/a | ptr += MINBPC(enc); |
|---|
| 349 | n/a | break; |
|---|
| 350 | n/a | } |
|---|
| 351 | n/a | while (ptr != end) { |
|---|
| 352 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 353 | n/a | #define LEAD_CASE(n) \ |
|---|
| 354 | n/a | case BT_LEAD ## n: \ |
|---|
| 355 | n/a | if (end - ptr < n || IS_INVALID_CHAR(enc, ptr, n)) { \ |
|---|
| 356 | n/a | *nextTokPtr = ptr; \ |
|---|
| 357 | n/a | return XML_TOK_DATA_CHARS; \ |
|---|
| 358 | n/a | } \ |
|---|
| 359 | n/a | ptr += n; \ |
|---|
| 360 | n/a | break; |
|---|
| 361 | n/a | LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4) |
|---|
| 362 | n/a | #undef LEAD_CASE |
|---|
| 363 | n/a | case BT_NONXML: |
|---|
| 364 | n/a | case BT_MALFORM: |
|---|
| 365 | n/a | case BT_TRAIL: |
|---|
| 366 | n/a | case BT_CR: |
|---|
| 367 | n/a | case BT_LF: |
|---|
| 368 | n/a | case BT_RSQB: |
|---|
| 369 | n/a | *nextTokPtr = ptr; |
|---|
| 370 | n/a | return XML_TOK_DATA_CHARS; |
|---|
| 371 | n/a | default: |
|---|
| 372 | n/a | ptr += MINBPC(enc); |
|---|
| 373 | n/a | break; |
|---|
| 374 | n/a | } |
|---|
| 375 | n/a | } |
|---|
| 376 | n/a | *nextTokPtr = ptr; |
|---|
| 377 | n/a | return XML_TOK_DATA_CHARS; |
|---|
| 378 | n/a | } |
|---|
| 379 | n/a | |
|---|
| 380 | n/a | /* ptr points to character following "</" */ |
|---|
| 381 | n/a | |
|---|
| 382 | n/a | static int PTRCALL |
|---|
| 383 | n/a | PREFIX(scanEndTag)(const ENCODING *enc, const char *ptr, |
|---|
| 384 | n/a | const char *end, const char **nextTokPtr) |
|---|
| 385 | n/a | { |
|---|
| 386 | n/a | if (ptr == end) |
|---|
| 387 | n/a | return XML_TOK_PARTIAL; |
|---|
| 388 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 389 | n/a | CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr) |
|---|
| 390 | n/a | default: |
|---|
| 391 | n/a | *nextTokPtr = ptr; |
|---|
| 392 | n/a | return XML_TOK_INVALID; |
|---|
| 393 | n/a | } |
|---|
| 394 | n/a | while (ptr != end) { |
|---|
| 395 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 396 | n/a | CHECK_NAME_CASES(enc, ptr, end, nextTokPtr) |
|---|
| 397 | n/a | case BT_S: case BT_CR: case BT_LF: |
|---|
| 398 | n/a | for (ptr += MINBPC(enc); ptr != end; ptr += MINBPC(enc)) { |
|---|
| 399 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 400 | n/a | case BT_S: case BT_CR: case BT_LF: |
|---|
| 401 | n/a | break; |
|---|
| 402 | n/a | case BT_GT: |
|---|
| 403 | n/a | *nextTokPtr = ptr + MINBPC(enc); |
|---|
| 404 | n/a | return XML_TOK_END_TAG; |
|---|
| 405 | n/a | default: |
|---|
| 406 | n/a | *nextTokPtr = ptr; |
|---|
| 407 | n/a | return XML_TOK_INVALID; |
|---|
| 408 | n/a | } |
|---|
| 409 | n/a | } |
|---|
| 410 | n/a | return XML_TOK_PARTIAL; |
|---|
| 411 | n/a | #ifdef XML_NS |
|---|
| 412 | n/a | case BT_COLON: |
|---|
| 413 | n/a | /* no need to check qname syntax here, |
|---|
| 414 | n/a | since end-tag must match exactly */ |
|---|
| 415 | n/a | ptr += MINBPC(enc); |
|---|
| 416 | n/a | break; |
|---|
| 417 | n/a | #endif |
|---|
| 418 | n/a | case BT_GT: |
|---|
| 419 | n/a | *nextTokPtr = ptr + MINBPC(enc); |
|---|
| 420 | n/a | return XML_TOK_END_TAG; |
|---|
| 421 | n/a | default: |
|---|
| 422 | n/a | *nextTokPtr = ptr; |
|---|
| 423 | n/a | return XML_TOK_INVALID; |
|---|
| 424 | n/a | } |
|---|
| 425 | n/a | } |
|---|
| 426 | n/a | return XML_TOK_PARTIAL; |
|---|
| 427 | n/a | } |
|---|
| 428 | n/a | |
|---|
| 429 | n/a | /* ptr points to character following "&#X" */ |
|---|
| 430 | n/a | |
|---|
| 431 | n/a | static int PTRCALL |
|---|
| 432 | n/a | PREFIX(scanHexCharRef)(const ENCODING *enc, const char *ptr, |
|---|
| 433 | n/a | const char *end, const char **nextTokPtr) |
|---|
| 434 | n/a | { |
|---|
| 435 | n/a | if (ptr != end) { |
|---|
| 436 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 437 | n/a | case BT_DIGIT: |
|---|
| 438 | n/a | case BT_HEX: |
|---|
| 439 | n/a | break; |
|---|
| 440 | n/a | default: |
|---|
| 441 | n/a | *nextTokPtr = ptr; |
|---|
| 442 | n/a | return XML_TOK_INVALID; |
|---|
| 443 | n/a | } |
|---|
| 444 | n/a | for (ptr += MINBPC(enc); ptr != end; ptr += MINBPC(enc)) { |
|---|
| 445 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 446 | n/a | case BT_DIGIT: |
|---|
| 447 | n/a | case BT_HEX: |
|---|
| 448 | n/a | break; |
|---|
| 449 | n/a | case BT_SEMI: |
|---|
| 450 | n/a | *nextTokPtr = ptr + MINBPC(enc); |
|---|
| 451 | n/a | return XML_TOK_CHAR_REF; |
|---|
| 452 | n/a | default: |
|---|
| 453 | n/a | *nextTokPtr = ptr; |
|---|
| 454 | n/a | return XML_TOK_INVALID; |
|---|
| 455 | n/a | } |
|---|
| 456 | n/a | } |
|---|
| 457 | n/a | } |
|---|
| 458 | n/a | return XML_TOK_PARTIAL; |
|---|
| 459 | n/a | } |
|---|
| 460 | n/a | |
|---|
| 461 | n/a | /* ptr points to character following "&#" */ |
|---|
| 462 | n/a | |
|---|
| 463 | n/a | static int PTRCALL |
|---|
| 464 | n/a | PREFIX(scanCharRef)(const ENCODING *enc, const char *ptr, |
|---|
| 465 | n/a | const char *end, const char **nextTokPtr) |
|---|
| 466 | n/a | { |
|---|
| 467 | n/a | if (ptr != end) { |
|---|
| 468 | n/a | if (CHAR_MATCHES(enc, ptr, ASCII_x)) |
|---|
| 469 | n/a | return PREFIX(scanHexCharRef)(enc, ptr + MINBPC(enc), end, nextTokPtr); |
|---|
| 470 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 471 | n/a | case BT_DIGIT: |
|---|
| 472 | n/a | break; |
|---|
| 473 | n/a | default: |
|---|
| 474 | n/a | *nextTokPtr = ptr; |
|---|
| 475 | n/a | return XML_TOK_INVALID; |
|---|
| 476 | n/a | } |
|---|
| 477 | n/a | for (ptr += MINBPC(enc); ptr != end; ptr += MINBPC(enc)) { |
|---|
| 478 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 479 | n/a | case BT_DIGIT: |
|---|
| 480 | n/a | break; |
|---|
| 481 | n/a | case BT_SEMI: |
|---|
| 482 | n/a | *nextTokPtr = ptr + MINBPC(enc); |
|---|
| 483 | n/a | return XML_TOK_CHAR_REF; |
|---|
| 484 | n/a | default: |
|---|
| 485 | n/a | *nextTokPtr = ptr; |
|---|
| 486 | n/a | return XML_TOK_INVALID; |
|---|
| 487 | n/a | } |
|---|
| 488 | n/a | } |
|---|
| 489 | n/a | } |
|---|
| 490 | n/a | return XML_TOK_PARTIAL; |
|---|
| 491 | n/a | } |
|---|
| 492 | n/a | |
|---|
| 493 | n/a | /* ptr points to character following "&" */ |
|---|
| 494 | n/a | |
|---|
| 495 | n/a | static int PTRCALL |
|---|
| 496 | n/a | PREFIX(scanRef)(const ENCODING *enc, const char *ptr, const char *end, |
|---|
| 497 | n/a | const char **nextTokPtr) |
|---|
| 498 | n/a | { |
|---|
| 499 | n/a | if (ptr == end) |
|---|
| 500 | n/a | return XML_TOK_PARTIAL; |
|---|
| 501 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 502 | n/a | CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr) |
|---|
| 503 | n/a | case BT_NUM: |
|---|
| 504 | n/a | return PREFIX(scanCharRef)(enc, ptr + MINBPC(enc), end, nextTokPtr); |
|---|
| 505 | n/a | default: |
|---|
| 506 | n/a | *nextTokPtr = ptr; |
|---|
| 507 | n/a | return XML_TOK_INVALID; |
|---|
| 508 | n/a | } |
|---|
| 509 | n/a | while (ptr != end) { |
|---|
| 510 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 511 | n/a | CHECK_NAME_CASES(enc, ptr, end, nextTokPtr) |
|---|
| 512 | n/a | case BT_SEMI: |
|---|
| 513 | n/a | *nextTokPtr = ptr + MINBPC(enc); |
|---|
| 514 | n/a | return XML_TOK_ENTITY_REF; |
|---|
| 515 | n/a | default: |
|---|
| 516 | n/a | *nextTokPtr = ptr; |
|---|
| 517 | n/a | return XML_TOK_INVALID; |
|---|
| 518 | n/a | } |
|---|
| 519 | n/a | } |
|---|
| 520 | n/a | return XML_TOK_PARTIAL; |
|---|
| 521 | n/a | } |
|---|
| 522 | n/a | |
|---|
| 523 | n/a | /* ptr points to character following first character of attribute name */ |
|---|
| 524 | n/a | |
|---|
| 525 | n/a | static int PTRCALL |
|---|
| 526 | n/a | PREFIX(scanAtts)(const ENCODING *enc, const char *ptr, const char *end, |
|---|
| 527 | n/a | const char **nextTokPtr) |
|---|
| 528 | n/a | { |
|---|
| 529 | n/a | #ifdef XML_NS |
|---|
| 530 | n/a | int hadColon = 0; |
|---|
| 531 | n/a | #endif |
|---|
| 532 | n/a | while (ptr != end) { |
|---|
| 533 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 534 | n/a | CHECK_NAME_CASES(enc, ptr, end, nextTokPtr) |
|---|
| 535 | n/a | #ifdef XML_NS |
|---|
| 536 | n/a | case BT_COLON: |
|---|
| 537 | n/a | if (hadColon) { |
|---|
| 538 | n/a | *nextTokPtr = ptr; |
|---|
| 539 | n/a | return XML_TOK_INVALID; |
|---|
| 540 | n/a | } |
|---|
| 541 | n/a | hadColon = 1; |
|---|
| 542 | n/a | ptr += MINBPC(enc); |
|---|
| 543 | n/a | if (ptr == end) |
|---|
| 544 | n/a | return XML_TOK_PARTIAL; |
|---|
| 545 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 546 | n/a | CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr) |
|---|
| 547 | n/a | default: |
|---|
| 548 | n/a | *nextTokPtr = ptr; |
|---|
| 549 | n/a | return XML_TOK_INVALID; |
|---|
| 550 | n/a | } |
|---|
| 551 | n/a | break; |
|---|
| 552 | n/a | #endif |
|---|
| 553 | n/a | case BT_S: case BT_CR: case BT_LF: |
|---|
| 554 | n/a | for (;;) { |
|---|
| 555 | n/a | int t; |
|---|
| 556 | n/a | |
|---|
| 557 | n/a | ptr += MINBPC(enc); |
|---|
| 558 | n/a | if (ptr == end) |
|---|
| 559 | n/a | return XML_TOK_PARTIAL; |
|---|
| 560 | n/a | t = BYTE_TYPE(enc, ptr); |
|---|
| 561 | n/a | if (t == BT_EQUALS) |
|---|
| 562 | n/a | break; |
|---|
| 563 | n/a | switch (t) { |
|---|
| 564 | n/a | case BT_S: |
|---|
| 565 | n/a | case BT_LF: |
|---|
| 566 | n/a | case BT_CR: |
|---|
| 567 | n/a | break; |
|---|
| 568 | n/a | default: |
|---|
| 569 | n/a | *nextTokPtr = ptr; |
|---|
| 570 | n/a | return XML_TOK_INVALID; |
|---|
| 571 | n/a | } |
|---|
| 572 | n/a | } |
|---|
| 573 | n/a | /* fall through */ |
|---|
| 574 | n/a | case BT_EQUALS: |
|---|
| 575 | n/a | { |
|---|
| 576 | n/a | int open; |
|---|
| 577 | n/a | #ifdef XML_NS |
|---|
| 578 | n/a | hadColon = 0; |
|---|
| 579 | n/a | #endif |
|---|
| 580 | n/a | for (;;) { |
|---|
| 581 | n/a | ptr += MINBPC(enc); |
|---|
| 582 | n/a | if (ptr == end) |
|---|
| 583 | n/a | return XML_TOK_PARTIAL; |
|---|
| 584 | n/a | open = BYTE_TYPE(enc, ptr); |
|---|
| 585 | n/a | if (open == BT_QUOT || open == BT_APOS) |
|---|
| 586 | n/a | break; |
|---|
| 587 | n/a | switch (open) { |
|---|
| 588 | n/a | case BT_S: |
|---|
| 589 | n/a | case BT_LF: |
|---|
| 590 | n/a | case BT_CR: |
|---|
| 591 | n/a | break; |
|---|
| 592 | n/a | default: |
|---|
| 593 | n/a | *nextTokPtr = ptr; |
|---|
| 594 | n/a | return XML_TOK_INVALID; |
|---|
| 595 | n/a | } |
|---|
| 596 | n/a | } |
|---|
| 597 | n/a | ptr += MINBPC(enc); |
|---|
| 598 | n/a | /* in attribute value */ |
|---|
| 599 | n/a | for (;;) { |
|---|
| 600 | n/a | int t; |
|---|
| 601 | n/a | if (ptr == end) |
|---|
| 602 | n/a | return XML_TOK_PARTIAL; |
|---|
| 603 | n/a | t = BYTE_TYPE(enc, ptr); |
|---|
| 604 | n/a | if (t == open) |
|---|
| 605 | n/a | break; |
|---|
| 606 | n/a | switch (t) { |
|---|
| 607 | n/a | INVALID_CASES(ptr, nextTokPtr) |
|---|
| 608 | n/a | case BT_AMP: |
|---|
| 609 | n/a | { |
|---|
| 610 | n/a | int tok = PREFIX(scanRef)(enc, ptr + MINBPC(enc), end, &ptr); |
|---|
| 611 | n/a | if (tok <= 0) { |
|---|
| 612 | n/a | if (tok == XML_TOK_INVALID) |
|---|
| 613 | n/a | *nextTokPtr = ptr; |
|---|
| 614 | n/a | return tok; |
|---|
| 615 | n/a | } |
|---|
| 616 | n/a | break; |
|---|
| 617 | n/a | } |
|---|
| 618 | n/a | case BT_LT: |
|---|
| 619 | n/a | *nextTokPtr = ptr; |
|---|
| 620 | n/a | return XML_TOK_INVALID; |
|---|
| 621 | n/a | default: |
|---|
| 622 | n/a | ptr += MINBPC(enc); |
|---|
| 623 | n/a | break; |
|---|
| 624 | n/a | } |
|---|
| 625 | n/a | } |
|---|
| 626 | n/a | ptr += MINBPC(enc); |
|---|
| 627 | n/a | if (ptr == end) |
|---|
| 628 | n/a | return XML_TOK_PARTIAL; |
|---|
| 629 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 630 | n/a | case BT_S: |
|---|
| 631 | n/a | case BT_CR: |
|---|
| 632 | n/a | case BT_LF: |
|---|
| 633 | n/a | break; |
|---|
| 634 | n/a | case BT_SOL: |
|---|
| 635 | n/a | goto sol; |
|---|
| 636 | n/a | case BT_GT: |
|---|
| 637 | n/a | goto gt; |
|---|
| 638 | n/a | default: |
|---|
| 639 | n/a | *nextTokPtr = ptr; |
|---|
| 640 | n/a | return XML_TOK_INVALID; |
|---|
| 641 | n/a | } |
|---|
| 642 | n/a | /* ptr points to closing quote */ |
|---|
| 643 | n/a | for (;;) { |
|---|
| 644 | n/a | ptr += MINBPC(enc); |
|---|
| 645 | n/a | if (ptr == end) |
|---|
| 646 | n/a | return XML_TOK_PARTIAL; |
|---|
| 647 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 648 | n/a | CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr) |
|---|
| 649 | n/a | case BT_S: case BT_CR: case BT_LF: |
|---|
| 650 | n/a | continue; |
|---|
| 651 | n/a | case BT_GT: |
|---|
| 652 | n/a | gt: |
|---|
| 653 | n/a | *nextTokPtr = ptr + MINBPC(enc); |
|---|
| 654 | n/a | return XML_TOK_START_TAG_WITH_ATTS; |
|---|
| 655 | n/a | case BT_SOL: |
|---|
| 656 | n/a | sol: |
|---|
| 657 | n/a | ptr += MINBPC(enc); |
|---|
| 658 | n/a | if (ptr == end) |
|---|
| 659 | n/a | return XML_TOK_PARTIAL; |
|---|
| 660 | n/a | if (!CHAR_MATCHES(enc, ptr, ASCII_GT)) { |
|---|
| 661 | n/a | *nextTokPtr = ptr; |
|---|
| 662 | n/a | return XML_TOK_INVALID; |
|---|
| 663 | n/a | } |
|---|
| 664 | n/a | *nextTokPtr = ptr + MINBPC(enc); |
|---|
| 665 | n/a | return XML_TOK_EMPTY_ELEMENT_WITH_ATTS; |
|---|
| 666 | n/a | default: |
|---|
| 667 | n/a | *nextTokPtr = ptr; |
|---|
| 668 | n/a | return XML_TOK_INVALID; |
|---|
| 669 | n/a | } |
|---|
| 670 | n/a | break; |
|---|
| 671 | n/a | } |
|---|
| 672 | n/a | break; |
|---|
| 673 | n/a | } |
|---|
| 674 | n/a | default: |
|---|
| 675 | n/a | *nextTokPtr = ptr; |
|---|
| 676 | n/a | return XML_TOK_INVALID; |
|---|
| 677 | n/a | } |
|---|
| 678 | n/a | } |
|---|
| 679 | n/a | return XML_TOK_PARTIAL; |
|---|
| 680 | n/a | } |
|---|
| 681 | n/a | |
|---|
| 682 | n/a | /* ptr points to character following "<" */ |
|---|
| 683 | n/a | |
|---|
| 684 | n/a | static int PTRCALL |
|---|
| 685 | n/a | PREFIX(scanLt)(const ENCODING *enc, const char *ptr, const char *end, |
|---|
| 686 | n/a | const char **nextTokPtr) |
|---|
| 687 | n/a | { |
|---|
| 688 | n/a | #ifdef XML_NS |
|---|
| 689 | n/a | int hadColon; |
|---|
| 690 | n/a | #endif |
|---|
| 691 | n/a | if (ptr == end) |
|---|
| 692 | n/a | return XML_TOK_PARTIAL; |
|---|
| 693 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 694 | n/a | CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr) |
|---|
| 695 | n/a | case BT_EXCL: |
|---|
| 696 | n/a | if ((ptr += MINBPC(enc)) == end) |
|---|
| 697 | n/a | return XML_TOK_PARTIAL; |
|---|
| 698 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 699 | n/a | case BT_MINUS: |
|---|
| 700 | n/a | return PREFIX(scanComment)(enc, ptr + MINBPC(enc), end, nextTokPtr); |
|---|
| 701 | n/a | case BT_LSQB: |
|---|
| 702 | n/a | return PREFIX(scanCdataSection)(enc, ptr + MINBPC(enc), |
|---|
| 703 | n/a | end, nextTokPtr); |
|---|
| 704 | n/a | } |
|---|
| 705 | n/a | *nextTokPtr = ptr; |
|---|
| 706 | n/a | return XML_TOK_INVALID; |
|---|
| 707 | n/a | case BT_QUEST: |
|---|
| 708 | n/a | return PREFIX(scanPi)(enc, ptr + MINBPC(enc), end, nextTokPtr); |
|---|
| 709 | n/a | case BT_SOL: |
|---|
| 710 | n/a | return PREFIX(scanEndTag)(enc, ptr + MINBPC(enc), end, nextTokPtr); |
|---|
| 711 | n/a | default: |
|---|
| 712 | n/a | *nextTokPtr = ptr; |
|---|
| 713 | n/a | return XML_TOK_INVALID; |
|---|
| 714 | n/a | } |
|---|
| 715 | n/a | #ifdef XML_NS |
|---|
| 716 | n/a | hadColon = 0; |
|---|
| 717 | n/a | #endif |
|---|
| 718 | n/a | /* we have a start-tag */ |
|---|
| 719 | n/a | while (ptr != end) { |
|---|
| 720 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 721 | n/a | CHECK_NAME_CASES(enc, ptr, end, nextTokPtr) |
|---|
| 722 | n/a | #ifdef XML_NS |
|---|
| 723 | n/a | case BT_COLON: |
|---|
| 724 | n/a | if (hadColon) { |
|---|
| 725 | n/a | *nextTokPtr = ptr; |
|---|
| 726 | n/a | return XML_TOK_INVALID; |
|---|
| 727 | n/a | } |
|---|
| 728 | n/a | hadColon = 1; |
|---|
| 729 | n/a | ptr += MINBPC(enc); |
|---|
| 730 | n/a | if (ptr == end) |
|---|
| 731 | n/a | return XML_TOK_PARTIAL; |
|---|
| 732 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 733 | n/a | CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr) |
|---|
| 734 | n/a | default: |
|---|
| 735 | n/a | *nextTokPtr = ptr; |
|---|
| 736 | n/a | return XML_TOK_INVALID; |
|---|
| 737 | n/a | } |
|---|
| 738 | n/a | break; |
|---|
| 739 | n/a | #endif |
|---|
| 740 | n/a | case BT_S: case BT_CR: case BT_LF: |
|---|
| 741 | n/a | { |
|---|
| 742 | n/a | ptr += MINBPC(enc); |
|---|
| 743 | n/a | while (ptr != end) { |
|---|
| 744 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 745 | n/a | CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr) |
|---|
| 746 | n/a | case BT_GT: |
|---|
| 747 | n/a | goto gt; |
|---|
| 748 | n/a | case BT_SOL: |
|---|
| 749 | n/a | goto sol; |
|---|
| 750 | n/a | case BT_S: case BT_CR: case BT_LF: |
|---|
| 751 | n/a | ptr += MINBPC(enc); |
|---|
| 752 | n/a | continue; |
|---|
| 753 | n/a | default: |
|---|
| 754 | n/a | *nextTokPtr = ptr; |
|---|
| 755 | n/a | return XML_TOK_INVALID; |
|---|
| 756 | n/a | } |
|---|
| 757 | n/a | return PREFIX(scanAtts)(enc, ptr, end, nextTokPtr); |
|---|
| 758 | n/a | } |
|---|
| 759 | n/a | return XML_TOK_PARTIAL; |
|---|
| 760 | n/a | } |
|---|
| 761 | n/a | case BT_GT: |
|---|
| 762 | n/a | gt: |
|---|
| 763 | n/a | *nextTokPtr = ptr + MINBPC(enc); |
|---|
| 764 | n/a | return XML_TOK_START_TAG_NO_ATTS; |
|---|
| 765 | n/a | case BT_SOL: |
|---|
| 766 | n/a | sol: |
|---|
| 767 | n/a | ptr += MINBPC(enc); |
|---|
| 768 | n/a | if (ptr == end) |
|---|
| 769 | n/a | return XML_TOK_PARTIAL; |
|---|
| 770 | n/a | if (!CHAR_MATCHES(enc, ptr, ASCII_GT)) { |
|---|
| 771 | n/a | *nextTokPtr = ptr; |
|---|
| 772 | n/a | return XML_TOK_INVALID; |
|---|
| 773 | n/a | } |
|---|
| 774 | n/a | *nextTokPtr = ptr + MINBPC(enc); |
|---|
| 775 | n/a | return XML_TOK_EMPTY_ELEMENT_NO_ATTS; |
|---|
| 776 | n/a | default: |
|---|
| 777 | n/a | *nextTokPtr = ptr; |
|---|
| 778 | n/a | return XML_TOK_INVALID; |
|---|
| 779 | n/a | } |
|---|
| 780 | n/a | } |
|---|
| 781 | n/a | return XML_TOK_PARTIAL; |
|---|
| 782 | n/a | } |
|---|
| 783 | n/a | |
|---|
| 784 | n/a | static int PTRCALL |
|---|
| 785 | n/a | PREFIX(contentTok)(const ENCODING *enc, const char *ptr, const char *end, |
|---|
| 786 | n/a | const char **nextTokPtr) |
|---|
| 787 | n/a | { |
|---|
| 788 | n/a | if (ptr == end) |
|---|
| 789 | n/a | return XML_TOK_NONE; |
|---|
| 790 | n/a | if (MINBPC(enc) > 1) { |
|---|
| 791 | n/a | size_t n = end - ptr; |
|---|
| 792 | n/a | if (n & (MINBPC(enc) - 1)) { |
|---|
| 793 | n/a | n &= ~(MINBPC(enc) - 1); |
|---|
| 794 | n/a | if (n == 0) |
|---|
| 795 | n/a | return XML_TOK_PARTIAL; |
|---|
| 796 | n/a | end = ptr + n; |
|---|
| 797 | n/a | } |
|---|
| 798 | n/a | } |
|---|
| 799 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 800 | n/a | case BT_LT: |
|---|
| 801 | n/a | return PREFIX(scanLt)(enc, ptr + MINBPC(enc), end, nextTokPtr); |
|---|
| 802 | n/a | case BT_AMP: |
|---|
| 803 | n/a | return PREFIX(scanRef)(enc, ptr + MINBPC(enc), end, nextTokPtr); |
|---|
| 804 | n/a | case BT_CR: |
|---|
| 805 | n/a | ptr += MINBPC(enc); |
|---|
| 806 | n/a | if (ptr == end) |
|---|
| 807 | n/a | return XML_TOK_TRAILING_CR; |
|---|
| 808 | n/a | if (BYTE_TYPE(enc, ptr) == BT_LF) |
|---|
| 809 | n/a | ptr += MINBPC(enc); |
|---|
| 810 | n/a | *nextTokPtr = ptr; |
|---|
| 811 | n/a | return XML_TOK_DATA_NEWLINE; |
|---|
| 812 | n/a | case BT_LF: |
|---|
| 813 | n/a | *nextTokPtr = ptr + MINBPC(enc); |
|---|
| 814 | n/a | return XML_TOK_DATA_NEWLINE; |
|---|
| 815 | n/a | case BT_RSQB: |
|---|
| 816 | n/a | ptr += MINBPC(enc); |
|---|
| 817 | n/a | if (ptr == end) |
|---|
| 818 | n/a | return XML_TOK_TRAILING_RSQB; |
|---|
| 819 | n/a | if (!CHAR_MATCHES(enc, ptr, ASCII_RSQB)) |
|---|
| 820 | n/a | break; |
|---|
| 821 | n/a | ptr += MINBPC(enc); |
|---|
| 822 | n/a | if (ptr == end) |
|---|
| 823 | n/a | return XML_TOK_TRAILING_RSQB; |
|---|
| 824 | n/a | if (!CHAR_MATCHES(enc, ptr, ASCII_GT)) { |
|---|
| 825 | n/a | ptr -= MINBPC(enc); |
|---|
| 826 | n/a | break; |
|---|
| 827 | n/a | } |
|---|
| 828 | n/a | *nextTokPtr = ptr; |
|---|
| 829 | n/a | return XML_TOK_INVALID; |
|---|
| 830 | n/a | INVALID_CASES(ptr, nextTokPtr) |
|---|
| 831 | n/a | default: |
|---|
| 832 | n/a | ptr += MINBPC(enc); |
|---|
| 833 | n/a | break; |
|---|
| 834 | n/a | } |
|---|
| 835 | n/a | while (ptr != end) { |
|---|
| 836 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 837 | n/a | #define LEAD_CASE(n) \ |
|---|
| 838 | n/a | case BT_LEAD ## n: \ |
|---|
| 839 | n/a | if (end - ptr < n || IS_INVALID_CHAR(enc, ptr, n)) { \ |
|---|
| 840 | n/a | *nextTokPtr = ptr; \ |
|---|
| 841 | n/a | return XML_TOK_DATA_CHARS; \ |
|---|
| 842 | n/a | } \ |
|---|
| 843 | n/a | ptr += n; \ |
|---|
| 844 | n/a | break; |
|---|
| 845 | n/a | LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4) |
|---|
| 846 | n/a | #undef LEAD_CASE |
|---|
| 847 | n/a | case BT_RSQB: |
|---|
| 848 | n/a | if (ptr + MINBPC(enc) != end) { |
|---|
| 849 | n/a | if (!CHAR_MATCHES(enc, ptr + MINBPC(enc), ASCII_RSQB)) { |
|---|
| 850 | n/a | ptr += MINBPC(enc); |
|---|
| 851 | n/a | break; |
|---|
| 852 | n/a | } |
|---|
| 853 | n/a | if (ptr + 2*MINBPC(enc) != end) { |
|---|
| 854 | n/a | if (!CHAR_MATCHES(enc, ptr + 2*MINBPC(enc), ASCII_GT)) { |
|---|
| 855 | n/a | ptr += MINBPC(enc); |
|---|
| 856 | n/a | break; |
|---|
| 857 | n/a | } |
|---|
| 858 | n/a | *nextTokPtr = ptr + 2*MINBPC(enc); |
|---|
| 859 | n/a | return XML_TOK_INVALID; |
|---|
| 860 | n/a | } |
|---|
| 861 | n/a | } |
|---|
| 862 | n/a | /* fall through */ |
|---|
| 863 | n/a | case BT_AMP: |
|---|
| 864 | n/a | case BT_LT: |
|---|
| 865 | n/a | case BT_NONXML: |
|---|
| 866 | n/a | case BT_MALFORM: |
|---|
| 867 | n/a | case BT_TRAIL: |
|---|
| 868 | n/a | case BT_CR: |
|---|
| 869 | n/a | case BT_LF: |
|---|
| 870 | n/a | *nextTokPtr = ptr; |
|---|
| 871 | n/a | return XML_TOK_DATA_CHARS; |
|---|
| 872 | n/a | default: |
|---|
| 873 | n/a | ptr += MINBPC(enc); |
|---|
| 874 | n/a | break; |
|---|
| 875 | n/a | } |
|---|
| 876 | n/a | } |
|---|
| 877 | n/a | *nextTokPtr = ptr; |
|---|
| 878 | n/a | return XML_TOK_DATA_CHARS; |
|---|
| 879 | n/a | } |
|---|
| 880 | n/a | |
|---|
| 881 | n/a | /* ptr points to character following "%" */ |
|---|
| 882 | n/a | |
|---|
| 883 | n/a | static int PTRCALL |
|---|
| 884 | n/a | PREFIX(scanPercent)(const ENCODING *enc, const char *ptr, const char *end, |
|---|
| 885 | n/a | const char **nextTokPtr) |
|---|
| 886 | n/a | { |
|---|
| 887 | n/a | if (ptr == end) |
|---|
| 888 | n/a | return XML_TOK_PARTIAL; |
|---|
| 889 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 890 | n/a | CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr) |
|---|
| 891 | n/a | case BT_S: case BT_LF: case BT_CR: case BT_PERCNT: |
|---|
| 892 | n/a | *nextTokPtr = ptr; |
|---|
| 893 | n/a | return XML_TOK_PERCENT; |
|---|
| 894 | n/a | default: |
|---|
| 895 | n/a | *nextTokPtr = ptr; |
|---|
| 896 | n/a | return XML_TOK_INVALID; |
|---|
| 897 | n/a | } |
|---|
| 898 | n/a | while (ptr != end) { |
|---|
| 899 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 900 | n/a | CHECK_NAME_CASES(enc, ptr, end, nextTokPtr) |
|---|
| 901 | n/a | case BT_SEMI: |
|---|
| 902 | n/a | *nextTokPtr = ptr + MINBPC(enc); |
|---|
| 903 | n/a | return XML_TOK_PARAM_ENTITY_REF; |
|---|
| 904 | n/a | default: |
|---|
| 905 | n/a | *nextTokPtr = ptr; |
|---|
| 906 | n/a | return XML_TOK_INVALID; |
|---|
| 907 | n/a | } |
|---|
| 908 | n/a | } |
|---|
| 909 | n/a | return XML_TOK_PARTIAL; |
|---|
| 910 | n/a | } |
|---|
| 911 | n/a | |
|---|
| 912 | n/a | static int PTRCALL |
|---|
| 913 | n/a | PREFIX(scanPoundName)(const ENCODING *enc, const char *ptr, const char *end, |
|---|
| 914 | n/a | const char **nextTokPtr) |
|---|
| 915 | n/a | { |
|---|
| 916 | n/a | if (ptr == end) |
|---|
| 917 | n/a | return XML_TOK_PARTIAL; |
|---|
| 918 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 919 | n/a | CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr) |
|---|
| 920 | n/a | default: |
|---|
| 921 | n/a | *nextTokPtr = ptr; |
|---|
| 922 | n/a | return XML_TOK_INVALID; |
|---|
| 923 | n/a | } |
|---|
| 924 | n/a | while (ptr != end) { |
|---|
| 925 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 926 | n/a | CHECK_NAME_CASES(enc, ptr, end, nextTokPtr) |
|---|
| 927 | n/a | case BT_CR: case BT_LF: case BT_S: |
|---|
| 928 | n/a | case BT_RPAR: case BT_GT: case BT_PERCNT: case BT_VERBAR: |
|---|
| 929 | n/a | *nextTokPtr = ptr; |
|---|
| 930 | n/a | return XML_TOK_POUND_NAME; |
|---|
| 931 | n/a | default: |
|---|
| 932 | n/a | *nextTokPtr = ptr; |
|---|
| 933 | n/a | return XML_TOK_INVALID; |
|---|
| 934 | n/a | } |
|---|
| 935 | n/a | } |
|---|
| 936 | n/a | return -XML_TOK_POUND_NAME; |
|---|
| 937 | n/a | } |
|---|
| 938 | n/a | |
|---|
| 939 | n/a | static int PTRCALL |
|---|
| 940 | n/a | PREFIX(scanLit)(int open, const ENCODING *enc, |
|---|
| 941 | n/a | const char *ptr, const char *end, |
|---|
| 942 | n/a | const char **nextTokPtr) |
|---|
| 943 | n/a | { |
|---|
| 944 | n/a | while (ptr != end) { |
|---|
| 945 | n/a | int t = BYTE_TYPE(enc, ptr); |
|---|
| 946 | n/a | switch (t) { |
|---|
| 947 | n/a | INVALID_CASES(ptr, nextTokPtr) |
|---|
| 948 | n/a | case BT_QUOT: |
|---|
| 949 | n/a | case BT_APOS: |
|---|
| 950 | n/a | ptr += MINBPC(enc); |
|---|
| 951 | n/a | if (t != open) |
|---|
| 952 | n/a | break; |
|---|
| 953 | n/a | if (ptr == end) |
|---|
| 954 | n/a | return -XML_TOK_LITERAL; |
|---|
| 955 | n/a | *nextTokPtr = ptr; |
|---|
| 956 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 957 | n/a | case BT_S: case BT_CR: case BT_LF: |
|---|
| 958 | n/a | case BT_GT: case BT_PERCNT: case BT_LSQB: |
|---|
| 959 | n/a | return XML_TOK_LITERAL; |
|---|
| 960 | n/a | default: |
|---|
| 961 | n/a | return XML_TOK_INVALID; |
|---|
| 962 | n/a | } |
|---|
| 963 | n/a | default: |
|---|
| 964 | n/a | ptr += MINBPC(enc); |
|---|
| 965 | n/a | break; |
|---|
| 966 | n/a | } |
|---|
| 967 | n/a | } |
|---|
| 968 | n/a | return XML_TOK_PARTIAL; |
|---|
| 969 | n/a | } |
|---|
| 970 | n/a | |
|---|
| 971 | n/a | static int PTRCALL |
|---|
| 972 | n/a | PREFIX(prologTok)(const ENCODING *enc, const char *ptr, const char *end, |
|---|
| 973 | n/a | const char **nextTokPtr) |
|---|
| 974 | n/a | { |
|---|
| 975 | n/a | int tok; |
|---|
| 976 | n/a | if (ptr == end) |
|---|
| 977 | n/a | return XML_TOK_NONE; |
|---|
| 978 | n/a | if (MINBPC(enc) > 1) { |
|---|
| 979 | n/a | size_t n = end - ptr; |
|---|
| 980 | n/a | if (n & (MINBPC(enc) - 1)) { |
|---|
| 981 | n/a | n &= ~(MINBPC(enc) - 1); |
|---|
| 982 | n/a | if (n == 0) |
|---|
| 983 | n/a | return XML_TOK_PARTIAL; |
|---|
| 984 | n/a | end = ptr + n; |
|---|
| 985 | n/a | } |
|---|
| 986 | n/a | } |
|---|
| 987 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 988 | n/a | case BT_QUOT: |
|---|
| 989 | n/a | return PREFIX(scanLit)(BT_QUOT, enc, ptr + MINBPC(enc), end, nextTokPtr); |
|---|
| 990 | n/a | case BT_APOS: |
|---|
| 991 | n/a | return PREFIX(scanLit)(BT_APOS, enc, ptr + MINBPC(enc), end, nextTokPtr); |
|---|
| 992 | n/a | case BT_LT: |
|---|
| 993 | n/a | { |
|---|
| 994 | n/a | ptr += MINBPC(enc); |
|---|
| 995 | n/a | if (ptr == end) |
|---|
| 996 | n/a | return XML_TOK_PARTIAL; |
|---|
| 997 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 998 | n/a | case BT_EXCL: |
|---|
| 999 | n/a | return PREFIX(scanDecl)(enc, ptr + MINBPC(enc), end, nextTokPtr); |
|---|
| 1000 | n/a | case BT_QUEST: |
|---|
| 1001 | n/a | return PREFIX(scanPi)(enc, ptr + MINBPC(enc), end, nextTokPtr); |
|---|
| 1002 | n/a | case BT_NMSTRT: |
|---|
| 1003 | n/a | case BT_HEX: |
|---|
| 1004 | n/a | case BT_NONASCII: |
|---|
| 1005 | n/a | case BT_LEAD2: |
|---|
| 1006 | n/a | case BT_LEAD3: |
|---|
| 1007 | n/a | case BT_LEAD4: |
|---|
| 1008 | n/a | *nextTokPtr = ptr - MINBPC(enc); |
|---|
| 1009 | n/a | return XML_TOK_INSTANCE_START; |
|---|
| 1010 | n/a | } |
|---|
| 1011 | n/a | *nextTokPtr = ptr; |
|---|
| 1012 | n/a | return XML_TOK_INVALID; |
|---|
| 1013 | n/a | } |
|---|
| 1014 | n/a | case BT_CR: |
|---|
| 1015 | n/a | if (ptr + MINBPC(enc) == end) { |
|---|
| 1016 | n/a | *nextTokPtr = end; |
|---|
| 1017 | n/a | /* indicate that this might be part of a CR/LF pair */ |
|---|
| 1018 | n/a | return -XML_TOK_PROLOG_S; |
|---|
| 1019 | n/a | } |
|---|
| 1020 | n/a | /* fall through */ |
|---|
| 1021 | n/a | case BT_S: case BT_LF: |
|---|
| 1022 | n/a | for (;;) { |
|---|
| 1023 | n/a | ptr += MINBPC(enc); |
|---|
| 1024 | n/a | if (ptr == end) |
|---|
| 1025 | n/a | break; |
|---|
| 1026 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 1027 | n/a | case BT_S: case BT_LF: |
|---|
| 1028 | n/a | break; |
|---|
| 1029 | n/a | case BT_CR: |
|---|
| 1030 | n/a | /* don't split CR/LF pair */ |
|---|
| 1031 | n/a | if (ptr + MINBPC(enc) != end) |
|---|
| 1032 | n/a | break; |
|---|
| 1033 | n/a | /* fall through */ |
|---|
| 1034 | n/a | default: |
|---|
| 1035 | n/a | *nextTokPtr = ptr; |
|---|
| 1036 | n/a | return XML_TOK_PROLOG_S; |
|---|
| 1037 | n/a | } |
|---|
| 1038 | n/a | } |
|---|
| 1039 | n/a | *nextTokPtr = ptr; |
|---|
| 1040 | n/a | return XML_TOK_PROLOG_S; |
|---|
| 1041 | n/a | case BT_PERCNT: |
|---|
| 1042 | n/a | return PREFIX(scanPercent)(enc, ptr + MINBPC(enc), end, nextTokPtr); |
|---|
| 1043 | n/a | case BT_COMMA: |
|---|
| 1044 | n/a | *nextTokPtr = ptr + MINBPC(enc); |
|---|
| 1045 | n/a | return XML_TOK_COMMA; |
|---|
| 1046 | n/a | case BT_LSQB: |
|---|
| 1047 | n/a | *nextTokPtr = ptr + MINBPC(enc); |
|---|
| 1048 | n/a | return XML_TOK_OPEN_BRACKET; |
|---|
| 1049 | n/a | case BT_RSQB: |
|---|
| 1050 | n/a | ptr += MINBPC(enc); |
|---|
| 1051 | n/a | if (ptr == end) |
|---|
| 1052 | n/a | return -XML_TOK_CLOSE_BRACKET; |
|---|
| 1053 | n/a | if (CHAR_MATCHES(enc, ptr, ASCII_RSQB)) { |
|---|
| 1054 | n/a | if (ptr + MINBPC(enc) == end) |
|---|
| 1055 | n/a | return XML_TOK_PARTIAL; |
|---|
| 1056 | n/a | if (CHAR_MATCHES(enc, ptr + MINBPC(enc), ASCII_GT)) { |
|---|
| 1057 | n/a | *nextTokPtr = ptr + 2*MINBPC(enc); |
|---|
| 1058 | n/a | return XML_TOK_COND_SECT_CLOSE; |
|---|
| 1059 | n/a | } |
|---|
| 1060 | n/a | } |
|---|
| 1061 | n/a | *nextTokPtr = ptr; |
|---|
| 1062 | n/a | return XML_TOK_CLOSE_BRACKET; |
|---|
| 1063 | n/a | case BT_LPAR: |
|---|
| 1064 | n/a | *nextTokPtr = ptr + MINBPC(enc); |
|---|
| 1065 | n/a | return XML_TOK_OPEN_PAREN; |
|---|
| 1066 | n/a | case BT_RPAR: |
|---|
| 1067 | n/a | ptr += MINBPC(enc); |
|---|
| 1068 | n/a | if (ptr == end) |
|---|
| 1069 | n/a | return -XML_TOK_CLOSE_PAREN; |
|---|
| 1070 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 1071 | n/a | case BT_AST: |
|---|
| 1072 | n/a | *nextTokPtr = ptr + MINBPC(enc); |
|---|
| 1073 | n/a | return XML_TOK_CLOSE_PAREN_ASTERISK; |
|---|
| 1074 | n/a | case BT_QUEST: |
|---|
| 1075 | n/a | *nextTokPtr = ptr + MINBPC(enc); |
|---|
| 1076 | n/a | return XML_TOK_CLOSE_PAREN_QUESTION; |
|---|
| 1077 | n/a | case BT_PLUS: |
|---|
| 1078 | n/a | *nextTokPtr = ptr + MINBPC(enc); |
|---|
| 1079 | n/a | return XML_TOK_CLOSE_PAREN_PLUS; |
|---|
| 1080 | n/a | case BT_CR: case BT_LF: case BT_S: |
|---|
| 1081 | n/a | case BT_GT: case BT_COMMA: case BT_VERBAR: |
|---|
| 1082 | n/a | case BT_RPAR: |
|---|
| 1083 | n/a | *nextTokPtr = ptr; |
|---|
| 1084 | n/a | return XML_TOK_CLOSE_PAREN; |
|---|
| 1085 | n/a | } |
|---|
| 1086 | n/a | *nextTokPtr = ptr; |
|---|
| 1087 | n/a | return XML_TOK_INVALID; |
|---|
| 1088 | n/a | case BT_VERBAR: |
|---|
| 1089 | n/a | *nextTokPtr = ptr + MINBPC(enc); |
|---|
| 1090 | n/a | return XML_TOK_OR; |
|---|
| 1091 | n/a | case BT_GT: |
|---|
| 1092 | n/a | *nextTokPtr = ptr + MINBPC(enc); |
|---|
| 1093 | n/a | return XML_TOK_DECL_CLOSE; |
|---|
| 1094 | n/a | case BT_NUM: |
|---|
| 1095 | n/a | return PREFIX(scanPoundName)(enc, ptr + MINBPC(enc), end, nextTokPtr); |
|---|
| 1096 | n/a | #define LEAD_CASE(n) \ |
|---|
| 1097 | n/a | case BT_LEAD ## n: \ |
|---|
| 1098 | n/a | if (end - ptr < n) \ |
|---|
| 1099 | n/a | return XML_TOK_PARTIAL_CHAR; \ |
|---|
| 1100 | n/a | if (IS_NMSTRT_CHAR(enc, ptr, n)) { \ |
|---|
| 1101 | n/a | ptr += n; \ |
|---|
| 1102 | n/a | tok = XML_TOK_NAME; \ |
|---|
| 1103 | n/a | break; \ |
|---|
| 1104 | n/a | } \ |
|---|
| 1105 | n/a | if (IS_NAME_CHAR(enc, ptr, n)) { \ |
|---|
| 1106 | n/a | ptr += n; \ |
|---|
| 1107 | n/a | tok = XML_TOK_NMTOKEN; \ |
|---|
| 1108 | n/a | break; \ |
|---|
| 1109 | n/a | } \ |
|---|
| 1110 | n/a | *nextTokPtr = ptr; \ |
|---|
| 1111 | n/a | return XML_TOK_INVALID; |
|---|
| 1112 | n/a | LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4) |
|---|
| 1113 | n/a | #undef LEAD_CASE |
|---|
| 1114 | n/a | case BT_NMSTRT: |
|---|
| 1115 | n/a | case BT_HEX: |
|---|
| 1116 | n/a | tok = XML_TOK_NAME; |
|---|
| 1117 | n/a | ptr += MINBPC(enc); |
|---|
| 1118 | n/a | break; |
|---|
| 1119 | n/a | case BT_DIGIT: |
|---|
| 1120 | n/a | case BT_NAME: |
|---|
| 1121 | n/a | case BT_MINUS: |
|---|
| 1122 | n/a | #ifdef XML_NS |
|---|
| 1123 | n/a | case BT_COLON: |
|---|
| 1124 | n/a | #endif |
|---|
| 1125 | n/a | tok = XML_TOK_NMTOKEN; |
|---|
| 1126 | n/a | ptr += MINBPC(enc); |
|---|
| 1127 | n/a | break; |
|---|
| 1128 | n/a | case BT_NONASCII: |
|---|
| 1129 | n/a | if (IS_NMSTRT_CHAR_MINBPC(enc, ptr)) { |
|---|
| 1130 | n/a | ptr += MINBPC(enc); |
|---|
| 1131 | n/a | tok = XML_TOK_NAME; |
|---|
| 1132 | n/a | break; |
|---|
| 1133 | n/a | } |
|---|
| 1134 | n/a | if (IS_NAME_CHAR_MINBPC(enc, ptr)) { |
|---|
| 1135 | n/a | ptr += MINBPC(enc); |
|---|
| 1136 | n/a | tok = XML_TOK_NMTOKEN; |
|---|
| 1137 | n/a | break; |
|---|
| 1138 | n/a | } |
|---|
| 1139 | n/a | /* fall through */ |
|---|
| 1140 | n/a | default: |
|---|
| 1141 | n/a | *nextTokPtr = ptr; |
|---|
| 1142 | n/a | return XML_TOK_INVALID; |
|---|
| 1143 | n/a | } |
|---|
| 1144 | n/a | while (ptr != end) { |
|---|
| 1145 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 1146 | n/a | CHECK_NAME_CASES(enc, ptr, end, nextTokPtr) |
|---|
| 1147 | n/a | case BT_GT: case BT_RPAR: case BT_COMMA: |
|---|
| 1148 | n/a | case BT_VERBAR: case BT_LSQB: case BT_PERCNT: |
|---|
| 1149 | n/a | case BT_S: case BT_CR: case BT_LF: |
|---|
| 1150 | n/a | *nextTokPtr = ptr; |
|---|
| 1151 | n/a | return tok; |
|---|
| 1152 | n/a | #ifdef XML_NS |
|---|
| 1153 | n/a | case BT_COLON: |
|---|
| 1154 | n/a | ptr += MINBPC(enc); |
|---|
| 1155 | n/a | switch (tok) { |
|---|
| 1156 | n/a | case XML_TOK_NAME: |
|---|
| 1157 | n/a | if (ptr == end) |
|---|
| 1158 | n/a | return XML_TOK_PARTIAL; |
|---|
| 1159 | n/a | tok = XML_TOK_PREFIXED_NAME; |
|---|
| 1160 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 1161 | n/a | CHECK_NAME_CASES(enc, ptr, end, nextTokPtr) |
|---|
| 1162 | n/a | default: |
|---|
| 1163 | n/a | tok = XML_TOK_NMTOKEN; |
|---|
| 1164 | n/a | break; |
|---|
| 1165 | n/a | } |
|---|
| 1166 | n/a | break; |
|---|
| 1167 | n/a | case XML_TOK_PREFIXED_NAME: |
|---|
| 1168 | n/a | tok = XML_TOK_NMTOKEN; |
|---|
| 1169 | n/a | break; |
|---|
| 1170 | n/a | } |
|---|
| 1171 | n/a | break; |
|---|
| 1172 | n/a | #endif |
|---|
| 1173 | n/a | case BT_PLUS: |
|---|
| 1174 | n/a | if (tok == XML_TOK_NMTOKEN) { |
|---|
| 1175 | n/a | *nextTokPtr = ptr; |
|---|
| 1176 | n/a | return XML_TOK_INVALID; |
|---|
| 1177 | n/a | } |
|---|
| 1178 | n/a | *nextTokPtr = ptr + MINBPC(enc); |
|---|
| 1179 | n/a | return XML_TOK_NAME_PLUS; |
|---|
| 1180 | n/a | case BT_AST: |
|---|
| 1181 | n/a | if (tok == XML_TOK_NMTOKEN) { |
|---|
| 1182 | n/a | *nextTokPtr = ptr; |
|---|
| 1183 | n/a | return XML_TOK_INVALID; |
|---|
| 1184 | n/a | } |
|---|
| 1185 | n/a | *nextTokPtr = ptr + MINBPC(enc); |
|---|
| 1186 | n/a | return XML_TOK_NAME_ASTERISK; |
|---|
| 1187 | n/a | case BT_QUEST: |
|---|
| 1188 | n/a | if (tok == XML_TOK_NMTOKEN) { |
|---|
| 1189 | n/a | *nextTokPtr = ptr; |
|---|
| 1190 | n/a | return XML_TOK_INVALID; |
|---|
| 1191 | n/a | } |
|---|
| 1192 | n/a | *nextTokPtr = ptr + MINBPC(enc); |
|---|
| 1193 | n/a | return XML_TOK_NAME_QUESTION; |
|---|
| 1194 | n/a | default: |
|---|
| 1195 | n/a | *nextTokPtr = ptr; |
|---|
| 1196 | n/a | return XML_TOK_INVALID; |
|---|
| 1197 | n/a | } |
|---|
| 1198 | n/a | } |
|---|
| 1199 | n/a | return -tok; |
|---|
| 1200 | n/a | } |
|---|
| 1201 | n/a | |
|---|
| 1202 | n/a | static int PTRCALL |
|---|
| 1203 | n/a | PREFIX(attributeValueTok)(const ENCODING *enc, const char *ptr, |
|---|
| 1204 | n/a | const char *end, const char **nextTokPtr) |
|---|
| 1205 | n/a | { |
|---|
| 1206 | n/a | const char *start; |
|---|
| 1207 | n/a | if (ptr == end) |
|---|
| 1208 | n/a | return XML_TOK_NONE; |
|---|
| 1209 | n/a | start = ptr; |
|---|
| 1210 | n/a | while (ptr != end) { |
|---|
| 1211 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 1212 | n/a | #define LEAD_CASE(n) \ |
|---|
| 1213 | n/a | case BT_LEAD ## n: ptr += n; break; |
|---|
| 1214 | n/a | LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4) |
|---|
| 1215 | n/a | #undef LEAD_CASE |
|---|
| 1216 | n/a | case BT_AMP: |
|---|
| 1217 | n/a | if (ptr == start) |
|---|
| 1218 | n/a | return PREFIX(scanRef)(enc, ptr + MINBPC(enc), end, nextTokPtr); |
|---|
| 1219 | n/a | *nextTokPtr = ptr; |
|---|
| 1220 | n/a | return XML_TOK_DATA_CHARS; |
|---|
| 1221 | n/a | case BT_LT: |
|---|
| 1222 | n/a | /* this is for inside entity references */ |
|---|
| 1223 | n/a | *nextTokPtr = ptr; |
|---|
| 1224 | n/a | return XML_TOK_INVALID; |
|---|
| 1225 | n/a | case BT_LF: |
|---|
| 1226 | n/a | if (ptr == start) { |
|---|
| 1227 | n/a | *nextTokPtr = ptr + MINBPC(enc); |
|---|
| 1228 | n/a | return XML_TOK_DATA_NEWLINE; |
|---|
| 1229 | n/a | } |
|---|
| 1230 | n/a | *nextTokPtr = ptr; |
|---|
| 1231 | n/a | return XML_TOK_DATA_CHARS; |
|---|
| 1232 | n/a | case BT_CR: |
|---|
| 1233 | n/a | if (ptr == start) { |
|---|
| 1234 | n/a | ptr += MINBPC(enc); |
|---|
| 1235 | n/a | if (ptr == end) |
|---|
| 1236 | n/a | return XML_TOK_TRAILING_CR; |
|---|
| 1237 | n/a | if (BYTE_TYPE(enc, ptr) == BT_LF) |
|---|
| 1238 | n/a | ptr += MINBPC(enc); |
|---|
| 1239 | n/a | *nextTokPtr = ptr; |
|---|
| 1240 | n/a | return XML_TOK_DATA_NEWLINE; |
|---|
| 1241 | n/a | } |
|---|
| 1242 | n/a | *nextTokPtr = ptr; |
|---|
| 1243 | n/a | return XML_TOK_DATA_CHARS; |
|---|
| 1244 | n/a | case BT_S: |
|---|
| 1245 | n/a | if (ptr == start) { |
|---|
| 1246 | n/a | *nextTokPtr = ptr + MINBPC(enc); |
|---|
| 1247 | n/a | return XML_TOK_ATTRIBUTE_VALUE_S; |
|---|
| 1248 | n/a | } |
|---|
| 1249 | n/a | *nextTokPtr = ptr; |
|---|
| 1250 | n/a | return XML_TOK_DATA_CHARS; |
|---|
| 1251 | n/a | default: |
|---|
| 1252 | n/a | ptr += MINBPC(enc); |
|---|
| 1253 | n/a | break; |
|---|
| 1254 | n/a | } |
|---|
| 1255 | n/a | } |
|---|
| 1256 | n/a | *nextTokPtr = ptr; |
|---|
| 1257 | n/a | return XML_TOK_DATA_CHARS; |
|---|
| 1258 | n/a | } |
|---|
| 1259 | n/a | |
|---|
| 1260 | n/a | static int PTRCALL |
|---|
| 1261 | n/a | PREFIX(entityValueTok)(const ENCODING *enc, const char *ptr, |
|---|
| 1262 | n/a | const char *end, const char **nextTokPtr) |
|---|
| 1263 | n/a | { |
|---|
| 1264 | n/a | const char *start; |
|---|
| 1265 | n/a | if (ptr == end) |
|---|
| 1266 | n/a | return XML_TOK_NONE; |
|---|
| 1267 | n/a | start = ptr; |
|---|
| 1268 | n/a | while (ptr != end) { |
|---|
| 1269 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 1270 | n/a | #define LEAD_CASE(n) \ |
|---|
| 1271 | n/a | case BT_LEAD ## n: ptr += n; break; |
|---|
| 1272 | n/a | LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4) |
|---|
| 1273 | n/a | #undef LEAD_CASE |
|---|
| 1274 | n/a | case BT_AMP: |
|---|
| 1275 | n/a | if (ptr == start) |
|---|
| 1276 | n/a | return PREFIX(scanRef)(enc, ptr + MINBPC(enc), end, nextTokPtr); |
|---|
| 1277 | n/a | *nextTokPtr = ptr; |
|---|
| 1278 | n/a | return XML_TOK_DATA_CHARS; |
|---|
| 1279 | n/a | case BT_PERCNT: |
|---|
| 1280 | n/a | if (ptr == start) { |
|---|
| 1281 | n/a | int tok = PREFIX(scanPercent)(enc, ptr + MINBPC(enc), |
|---|
| 1282 | n/a | end, nextTokPtr); |
|---|
| 1283 | n/a | return (tok == XML_TOK_PERCENT) ? XML_TOK_INVALID : tok; |
|---|
| 1284 | n/a | } |
|---|
| 1285 | n/a | *nextTokPtr = ptr; |
|---|
| 1286 | n/a | return XML_TOK_DATA_CHARS; |
|---|
| 1287 | n/a | case BT_LF: |
|---|
| 1288 | n/a | if (ptr == start) { |
|---|
| 1289 | n/a | *nextTokPtr = ptr + MINBPC(enc); |
|---|
| 1290 | n/a | return XML_TOK_DATA_NEWLINE; |
|---|
| 1291 | n/a | } |
|---|
| 1292 | n/a | *nextTokPtr = ptr; |
|---|
| 1293 | n/a | return XML_TOK_DATA_CHARS; |
|---|
| 1294 | n/a | case BT_CR: |
|---|
| 1295 | n/a | if (ptr == start) { |
|---|
| 1296 | n/a | ptr += MINBPC(enc); |
|---|
| 1297 | n/a | if (ptr == end) |
|---|
| 1298 | n/a | return XML_TOK_TRAILING_CR; |
|---|
| 1299 | n/a | if (BYTE_TYPE(enc, ptr) == BT_LF) |
|---|
| 1300 | n/a | ptr += MINBPC(enc); |
|---|
| 1301 | n/a | *nextTokPtr = ptr; |
|---|
| 1302 | n/a | return XML_TOK_DATA_NEWLINE; |
|---|
| 1303 | n/a | } |
|---|
| 1304 | n/a | *nextTokPtr = ptr; |
|---|
| 1305 | n/a | return XML_TOK_DATA_CHARS; |
|---|
| 1306 | n/a | default: |
|---|
| 1307 | n/a | ptr += MINBPC(enc); |
|---|
| 1308 | n/a | break; |
|---|
| 1309 | n/a | } |
|---|
| 1310 | n/a | } |
|---|
| 1311 | n/a | *nextTokPtr = ptr; |
|---|
| 1312 | n/a | return XML_TOK_DATA_CHARS; |
|---|
| 1313 | n/a | } |
|---|
| 1314 | n/a | |
|---|
| 1315 | n/a | #ifdef XML_DTD |
|---|
| 1316 | n/a | |
|---|
| 1317 | n/a | static int PTRCALL |
|---|
| 1318 | n/a | PREFIX(ignoreSectionTok)(const ENCODING *enc, const char *ptr, |
|---|
| 1319 | n/a | const char *end, const char **nextTokPtr) |
|---|
| 1320 | n/a | { |
|---|
| 1321 | n/a | int level = 0; |
|---|
| 1322 | n/a | if (MINBPC(enc) > 1) { |
|---|
| 1323 | n/a | size_t n = end - ptr; |
|---|
| 1324 | n/a | if (n & (MINBPC(enc) - 1)) { |
|---|
| 1325 | n/a | n &= ~(MINBPC(enc) - 1); |
|---|
| 1326 | n/a | end = ptr + n; |
|---|
| 1327 | n/a | } |
|---|
| 1328 | n/a | } |
|---|
| 1329 | n/a | while (ptr != end) { |
|---|
| 1330 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 1331 | n/a | INVALID_CASES(ptr, nextTokPtr) |
|---|
| 1332 | n/a | case BT_LT: |
|---|
| 1333 | n/a | if ((ptr += MINBPC(enc)) == end) |
|---|
| 1334 | n/a | return XML_TOK_PARTIAL; |
|---|
| 1335 | n/a | if (CHAR_MATCHES(enc, ptr, ASCII_EXCL)) { |
|---|
| 1336 | n/a | if ((ptr += MINBPC(enc)) == end) |
|---|
| 1337 | n/a | return XML_TOK_PARTIAL; |
|---|
| 1338 | n/a | if (CHAR_MATCHES(enc, ptr, ASCII_LSQB)) { |
|---|
| 1339 | n/a | ++level; |
|---|
| 1340 | n/a | ptr += MINBPC(enc); |
|---|
| 1341 | n/a | } |
|---|
| 1342 | n/a | } |
|---|
| 1343 | n/a | break; |
|---|
| 1344 | n/a | case BT_RSQB: |
|---|
| 1345 | n/a | if ((ptr += MINBPC(enc)) == end) |
|---|
| 1346 | n/a | return XML_TOK_PARTIAL; |
|---|
| 1347 | n/a | if (CHAR_MATCHES(enc, ptr, ASCII_RSQB)) { |
|---|
| 1348 | n/a | if ((ptr += MINBPC(enc)) == end) |
|---|
| 1349 | n/a | return XML_TOK_PARTIAL; |
|---|
| 1350 | n/a | if (CHAR_MATCHES(enc, ptr, ASCII_GT)) { |
|---|
| 1351 | n/a | ptr += MINBPC(enc); |
|---|
| 1352 | n/a | if (level == 0) { |
|---|
| 1353 | n/a | *nextTokPtr = ptr; |
|---|
| 1354 | n/a | return XML_TOK_IGNORE_SECT; |
|---|
| 1355 | n/a | } |
|---|
| 1356 | n/a | --level; |
|---|
| 1357 | n/a | } |
|---|
| 1358 | n/a | } |
|---|
| 1359 | n/a | break; |
|---|
| 1360 | n/a | default: |
|---|
| 1361 | n/a | ptr += MINBPC(enc); |
|---|
| 1362 | n/a | break; |
|---|
| 1363 | n/a | } |
|---|
| 1364 | n/a | } |
|---|
| 1365 | n/a | return XML_TOK_PARTIAL; |
|---|
| 1366 | n/a | } |
|---|
| 1367 | n/a | |
|---|
| 1368 | n/a | #endif /* XML_DTD */ |
|---|
| 1369 | n/a | |
|---|
| 1370 | n/a | static int PTRCALL |
|---|
| 1371 | n/a | PREFIX(isPublicId)(const ENCODING *enc, const char *ptr, const char *end, |
|---|
| 1372 | n/a | const char **badPtr) |
|---|
| 1373 | n/a | { |
|---|
| 1374 | n/a | ptr += MINBPC(enc); |
|---|
| 1375 | n/a | end -= MINBPC(enc); |
|---|
| 1376 | n/a | for (; ptr != end; ptr += MINBPC(enc)) { |
|---|
| 1377 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 1378 | n/a | case BT_DIGIT: |
|---|
| 1379 | n/a | case BT_HEX: |
|---|
| 1380 | n/a | case BT_MINUS: |
|---|
| 1381 | n/a | case BT_APOS: |
|---|
| 1382 | n/a | case BT_LPAR: |
|---|
| 1383 | n/a | case BT_RPAR: |
|---|
| 1384 | n/a | case BT_PLUS: |
|---|
| 1385 | n/a | case BT_COMMA: |
|---|
| 1386 | n/a | case BT_SOL: |
|---|
| 1387 | n/a | case BT_EQUALS: |
|---|
| 1388 | n/a | case BT_QUEST: |
|---|
| 1389 | n/a | case BT_CR: |
|---|
| 1390 | n/a | case BT_LF: |
|---|
| 1391 | n/a | case BT_SEMI: |
|---|
| 1392 | n/a | case BT_EXCL: |
|---|
| 1393 | n/a | case BT_AST: |
|---|
| 1394 | n/a | case BT_PERCNT: |
|---|
| 1395 | n/a | case BT_NUM: |
|---|
| 1396 | n/a | #ifdef XML_NS |
|---|
| 1397 | n/a | case BT_COLON: |
|---|
| 1398 | n/a | #endif |
|---|
| 1399 | n/a | break; |
|---|
| 1400 | n/a | case BT_S: |
|---|
| 1401 | n/a | if (CHAR_MATCHES(enc, ptr, ASCII_TAB)) { |
|---|
| 1402 | n/a | *badPtr = ptr; |
|---|
| 1403 | n/a | return 0; |
|---|
| 1404 | n/a | } |
|---|
| 1405 | n/a | break; |
|---|
| 1406 | n/a | case BT_NAME: |
|---|
| 1407 | n/a | case BT_NMSTRT: |
|---|
| 1408 | n/a | if (!(BYTE_TO_ASCII(enc, ptr) & ~0x7f)) |
|---|
| 1409 | n/a | break; |
|---|
| 1410 | n/a | default: |
|---|
| 1411 | n/a | switch (BYTE_TO_ASCII(enc, ptr)) { |
|---|
| 1412 | n/a | case 0x24: /* $ */ |
|---|
| 1413 | n/a | case 0x40: /* @ */ |
|---|
| 1414 | n/a | break; |
|---|
| 1415 | n/a | default: |
|---|
| 1416 | n/a | *badPtr = ptr; |
|---|
| 1417 | n/a | return 0; |
|---|
| 1418 | n/a | } |
|---|
| 1419 | n/a | break; |
|---|
| 1420 | n/a | } |
|---|
| 1421 | n/a | } |
|---|
| 1422 | n/a | return 1; |
|---|
| 1423 | n/a | } |
|---|
| 1424 | n/a | |
|---|
| 1425 | n/a | /* This must only be called for a well-formed start-tag or empty |
|---|
| 1426 | n/a | element tag. Returns the number of attributes. Pointers to the |
|---|
| 1427 | n/a | first attsMax attributes are stored in atts. |
|---|
| 1428 | n/a | */ |
|---|
| 1429 | n/a | |
|---|
| 1430 | n/a | static int PTRCALL |
|---|
| 1431 | n/a | PREFIX(getAtts)(const ENCODING *enc, const char *ptr, |
|---|
| 1432 | n/a | int attsMax, ATTRIBUTE *atts) |
|---|
| 1433 | n/a | { |
|---|
| 1434 | n/a | enum { other, inName, inValue } state = inName; |
|---|
| 1435 | n/a | int nAtts = 0; |
|---|
| 1436 | n/a | int open = 0; /* defined when state == inValue; |
|---|
| 1437 | n/a | initialization just to shut up compilers */ |
|---|
| 1438 | n/a | |
|---|
| 1439 | n/a | for (ptr += MINBPC(enc);; ptr += MINBPC(enc)) { |
|---|
| 1440 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 1441 | n/a | #define START_NAME \ |
|---|
| 1442 | n/a | if (state == other) { \ |
|---|
| 1443 | n/a | if (nAtts < attsMax) { \ |
|---|
| 1444 | n/a | atts[nAtts].name = ptr; \ |
|---|
| 1445 | n/a | atts[nAtts].normalized = 1; \ |
|---|
| 1446 | n/a | } \ |
|---|
| 1447 | n/a | state = inName; \ |
|---|
| 1448 | n/a | } |
|---|
| 1449 | n/a | #define LEAD_CASE(n) \ |
|---|
| 1450 | n/a | case BT_LEAD ## n: START_NAME ptr += (n - MINBPC(enc)); break; |
|---|
| 1451 | n/a | LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4) |
|---|
| 1452 | n/a | #undef LEAD_CASE |
|---|
| 1453 | n/a | case BT_NONASCII: |
|---|
| 1454 | n/a | case BT_NMSTRT: |
|---|
| 1455 | n/a | case BT_HEX: |
|---|
| 1456 | n/a | START_NAME |
|---|
| 1457 | n/a | break; |
|---|
| 1458 | n/a | #undef START_NAME |
|---|
| 1459 | n/a | case BT_QUOT: |
|---|
| 1460 | n/a | if (state != inValue) { |
|---|
| 1461 | n/a | if (nAtts < attsMax) |
|---|
| 1462 | n/a | atts[nAtts].valuePtr = ptr + MINBPC(enc); |
|---|
| 1463 | n/a | state = inValue; |
|---|
| 1464 | n/a | open = BT_QUOT; |
|---|
| 1465 | n/a | } |
|---|
| 1466 | n/a | else if (open == BT_QUOT) { |
|---|
| 1467 | n/a | state = other; |
|---|
| 1468 | n/a | if (nAtts < attsMax) |
|---|
| 1469 | n/a | atts[nAtts].valueEnd = ptr; |
|---|
| 1470 | n/a | nAtts++; |
|---|
| 1471 | n/a | } |
|---|
| 1472 | n/a | break; |
|---|
| 1473 | n/a | case BT_APOS: |
|---|
| 1474 | n/a | if (state != inValue) { |
|---|
| 1475 | n/a | if (nAtts < attsMax) |
|---|
| 1476 | n/a | atts[nAtts].valuePtr = ptr + MINBPC(enc); |
|---|
| 1477 | n/a | state = inValue; |
|---|
| 1478 | n/a | open = BT_APOS; |
|---|
| 1479 | n/a | } |
|---|
| 1480 | n/a | else if (open == BT_APOS) { |
|---|
| 1481 | n/a | state = other; |
|---|
| 1482 | n/a | if (nAtts < attsMax) |
|---|
| 1483 | n/a | atts[nAtts].valueEnd = ptr; |
|---|
| 1484 | n/a | nAtts++; |
|---|
| 1485 | n/a | } |
|---|
| 1486 | n/a | break; |
|---|
| 1487 | n/a | case BT_AMP: |
|---|
| 1488 | n/a | if (nAtts < attsMax) |
|---|
| 1489 | n/a | atts[nAtts].normalized = 0; |
|---|
| 1490 | n/a | break; |
|---|
| 1491 | n/a | case BT_S: |
|---|
| 1492 | n/a | if (state == inName) |
|---|
| 1493 | n/a | state = other; |
|---|
| 1494 | n/a | else if (state == inValue |
|---|
| 1495 | n/a | && nAtts < attsMax |
|---|
| 1496 | n/a | && atts[nAtts].normalized |
|---|
| 1497 | n/a | && (ptr == atts[nAtts].valuePtr |
|---|
| 1498 | n/a | || BYTE_TO_ASCII(enc, ptr) != ASCII_SPACE |
|---|
| 1499 | n/a | || BYTE_TO_ASCII(enc, ptr + MINBPC(enc)) == ASCII_SPACE |
|---|
| 1500 | n/a | || BYTE_TYPE(enc, ptr + MINBPC(enc)) == open)) |
|---|
| 1501 | n/a | atts[nAtts].normalized = 0; |
|---|
| 1502 | n/a | break; |
|---|
| 1503 | n/a | case BT_CR: case BT_LF: |
|---|
| 1504 | n/a | /* This case ensures that the first attribute name is counted |
|---|
| 1505 | n/a | Apart from that we could just change state on the quote. */ |
|---|
| 1506 | n/a | if (state == inName) |
|---|
| 1507 | n/a | state = other; |
|---|
| 1508 | n/a | else if (state == inValue && nAtts < attsMax) |
|---|
| 1509 | n/a | atts[nAtts].normalized = 0; |
|---|
| 1510 | n/a | break; |
|---|
| 1511 | n/a | case BT_GT: |
|---|
| 1512 | n/a | case BT_SOL: |
|---|
| 1513 | n/a | if (state != inValue) |
|---|
| 1514 | n/a | return nAtts; |
|---|
| 1515 | n/a | break; |
|---|
| 1516 | n/a | default: |
|---|
| 1517 | n/a | break; |
|---|
| 1518 | n/a | } |
|---|
| 1519 | n/a | } |
|---|
| 1520 | n/a | /* not reached */ |
|---|
| 1521 | n/a | } |
|---|
| 1522 | n/a | |
|---|
| 1523 | n/a | static int PTRFASTCALL |
|---|
| 1524 | n/a | PREFIX(charRefNumber)(const ENCODING *enc, const char *ptr) |
|---|
| 1525 | n/a | { |
|---|
| 1526 | n/a | int result = 0; |
|---|
| 1527 | n/a | /* skip &# */ |
|---|
| 1528 | n/a | ptr += 2*MINBPC(enc); |
|---|
| 1529 | n/a | if (CHAR_MATCHES(enc, ptr, ASCII_x)) { |
|---|
| 1530 | n/a | for (ptr += MINBPC(enc); |
|---|
| 1531 | n/a | !CHAR_MATCHES(enc, ptr, ASCII_SEMI); |
|---|
| 1532 | n/a | ptr += MINBPC(enc)) { |
|---|
| 1533 | n/a | int c = BYTE_TO_ASCII(enc, ptr); |
|---|
| 1534 | n/a | switch (c) { |
|---|
| 1535 | n/a | case ASCII_0: case ASCII_1: case ASCII_2: case ASCII_3: case ASCII_4: |
|---|
| 1536 | n/a | case ASCII_5: case ASCII_6: case ASCII_7: case ASCII_8: case ASCII_9: |
|---|
| 1537 | n/a | result <<= 4; |
|---|
| 1538 | n/a | result |= (c - ASCII_0); |
|---|
| 1539 | n/a | break; |
|---|
| 1540 | n/a | case ASCII_A: case ASCII_B: case ASCII_C: |
|---|
| 1541 | n/a | case ASCII_D: case ASCII_E: case ASCII_F: |
|---|
| 1542 | n/a | result <<= 4; |
|---|
| 1543 | n/a | result += 10 + (c - ASCII_A); |
|---|
| 1544 | n/a | break; |
|---|
| 1545 | n/a | case ASCII_a: case ASCII_b: case ASCII_c: |
|---|
| 1546 | n/a | case ASCII_d: case ASCII_e: case ASCII_f: |
|---|
| 1547 | n/a | result <<= 4; |
|---|
| 1548 | n/a | result += 10 + (c - ASCII_a); |
|---|
| 1549 | n/a | break; |
|---|
| 1550 | n/a | } |
|---|
| 1551 | n/a | if (result >= 0x110000) |
|---|
| 1552 | n/a | return -1; |
|---|
| 1553 | n/a | } |
|---|
| 1554 | n/a | } |
|---|
| 1555 | n/a | else { |
|---|
| 1556 | n/a | for (; !CHAR_MATCHES(enc, ptr, ASCII_SEMI); ptr += MINBPC(enc)) { |
|---|
| 1557 | n/a | int c = BYTE_TO_ASCII(enc, ptr); |
|---|
| 1558 | n/a | result *= 10; |
|---|
| 1559 | n/a | result += (c - ASCII_0); |
|---|
| 1560 | n/a | if (result >= 0x110000) |
|---|
| 1561 | n/a | return -1; |
|---|
| 1562 | n/a | } |
|---|
| 1563 | n/a | } |
|---|
| 1564 | n/a | return checkCharRefNumber(result); |
|---|
| 1565 | n/a | } |
|---|
| 1566 | n/a | |
|---|
| 1567 | n/a | static int PTRCALL |
|---|
| 1568 | n/a | PREFIX(predefinedEntityName)(const ENCODING *enc, const char *ptr, |
|---|
| 1569 | n/a | const char *end) |
|---|
| 1570 | n/a | { |
|---|
| 1571 | n/a | switch ((end - ptr)/MINBPC(enc)) { |
|---|
| 1572 | n/a | case 2: |
|---|
| 1573 | n/a | if (CHAR_MATCHES(enc, ptr + MINBPC(enc), ASCII_t)) { |
|---|
| 1574 | n/a | switch (BYTE_TO_ASCII(enc, ptr)) { |
|---|
| 1575 | n/a | case ASCII_l: |
|---|
| 1576 | n/a | return ASCII_LT; |
|---|
| 1577 | n/a | case ASCII_g: |
|---|
| 1578 | n/a | return ASCII_GT; |
|---|
| 1579 | n/a | } |
|---|
| 1580 | n/a | } |
|---|
| 1581 | n/a | break; |
|---|
| 1582 | n/a | case 3: |
|---|
| 1583 | n/a | if (CHAR_MATCHES(enc, ptr, ASCII_a)) { |
|---|
| 1584 | n/a | ptr += MINBPC(enc); |
|---|
| 1585 | n/a | if (CHAR_MATCHES(enc, ptr, ASCII_m)) { |
|---|
| 1586 | n/a | ptr += MINBPC(enc); |
|---|
| 1587 | n/a | if (CHAR_MATCHES(enc, ptr, ASCII_p)) |
|---|
| 1588 | n/a | return ASCII_AMP; |
|---|
| 1589 | n/a | } |
|---|
| 1590 | n/a | } |
|---|
| 1591 | n/a | break; |
|---|
| 1592 | n/a | case 4: |
|---|
| 1593 | n/a | switch (BYTE_TO_ASCII(enc, ptr)) { |
|---|
| 1594 | n/a | case ASCII_q: |
|---|
| 1595 | n/a | ptr += MINBPC(enc); |
|---|
| 1596 | n/a | if (CHAR_MATCHES(enc, ptr, ASCII_u)) { |
|---|
| 1597 | n/a | ptr += MINBPC(enc); |
|---|
| 1598 | n/a | if (CHAR_MATCHES(enc, ptr, ASCII_o)) { |
|---|
| 1599 | n/a | ptr += MINBPC(enc); |
|---|
| 1600 | n/a | if (CHAR_MATCHES(enc, ptr, ASCII_t)) |
|---|
| 1601 | n/a | return ASCII_QUOT; |
|---|
| 1602 | n/a | } |
|---|
| 1603 | n/a | } |
|---|
| 1604 | n/a | break; |
|---|
| 1605 | n/a | case ASCII_a: |
|---|
| 1606 | n/a | ptr += MINBPC(enc); |
|---|
| 1607 | n/a | if (CHAR_MATCHES(enc, ptr, ASCII_p)) { |
|---|
| 1608 | n/a | ptr += MINBPC(enc); |
|---|
| 1609 | n/a | if (CHAR_MATCHES(enc, ptr, ASCII_o)) { |
|---|
| 1610 | n/a | ptr += MINBPC(enc); |
|---|
| 1611 | n/a | if (CHAR_MATCHES(enc, ptr, ASCII_s)) |
|---|
| 1612 | n/a | return ASCII_APOS; |
|---|
| 1613 | n/a | } |
|---|
| 1614 | n/a | } |
|---|
| 1615 | n/a | break; |
|---|
| 1616 | n/a | } |
|---|
| 1617 | n/a | } |
|---|
| 1618 | n/a | return 0; |
|---|
| 1619 | n/a | } |
|---|
| 1620 | n/a | |
|---|
| 1621 | n/a | static int PTRCALL |
|---|
| 1622 | n/a | PREFIX(sameName)(const ENCODING *enc, const char *ptr1, const char *ptr2) |
|---|
| 1623 | n/a | { |
|---|
| 1624 | n/a | for (;;) { |
|---|
| 1625 | n/a | switch (BYTE_TYPE(enc, ptr1)) { |
|---|
| 1626 | n/a | #define LEAD_CASE(n) \ |
|---|
| 1627 | n/a | case BT_LEAD ## n: \ |
|---|
| 1628 | n/a | if (*ptr1++ != *ptr2++) \ |
|---|
| 1629 | n/a | return 0; |
|---|
| 1630 | n/a | LEAD_CASE(4) LEAD_CASE(3) LEAD_CASE(2) |
|---|
| 1631 | n/a | #undef LEAD_CASE |
|---|
| 1632 | n/a | /* fall through */ |
|---|
| 1633 | n/a | if (*ptr1++ != *ptr2++) |
|---|
| 1634 | n/a | return 0; |
|---|
| 1635 | n/a | break; |
|---|
| 1636 | n/a | case BT_NONASCII: |
|---|
| 1637 | n/a | case BT_NMSTRT: |
|---|
| 1638 | n/a | #ifdef XML_NS |
|---|
| 1639 | n/a | case BT_COLON: |
|---|
| 1640 | n/a | #endif |
|---|
| 1641 | n/a | case BT_HEX: |
|---|
| 1642 | n/a | case BT_DIGIT: |
|---|
| 1643 | n/a | case BT_NAME: |
|---|
| 1644 | n/a | case BT_MINUS: |
|---|
| 1645 | n/a | if (*ptr2++ != *ptr1++) |
|---|
| 1646 | n/a | return 0; |
|---|
| 1647 | n/a | if (MINBPC(enc) > 1) { |
|---|
| 1648 | n/a | if (*ptr2++ != *ptr1++) |
|---|
| 1649 | n/a | return 0; |
|---|
| 1650 | n/a | if (MINBPC(enc) > 2) { |
|---|
| 1651 | n/a | if (*ptr2++ != *ptr1++) |
|---|
| 1652 | n/a | return 0; |
|---|
| 1653 | n/a | if (MINBPC(enc) > 3) { |
|---|
| 1654 | n/a | if (*ptr2++ != *ptr1++) |
|---|
| 1655 | n/a | return 0; |
|---|
| 1656 | n/a | } |
|---|
| 1657 | n/a | } |
|---|
| 1658 | n/a | } |
|---|
| 1659 | n/a | break; |
|---|
| 1660 | n/a | default: |
|---|
| 1661 | n/a | if (MINBPC(enc) == 1 && *ptr1 == *ptr2) |
|---|
| 1662 | n/a | return 1; |
|---|
| 1663 | n/a | switch (BYTE_TYPE(enc, ptr2)) { |
|---|
| 1664 | n/a | case BT_LEAD2: |
|---|
| 1665 | n/a | case BT_LEAD3: |
|---|
| 1666 | n/a | case BT_LEAD4: |
|---|
| 1667 | n/a | case BT_NONASCII: |
|---|
| 1668 | n/a | case BT_NMSTRT: |
|---|
| 1669 | n/a | #ifdef XML_NS |
|---|
| 1670 | n/a | case BT_COLON: |
|---|
| 1671 | n/a | #endif |
|---|
| 1672 | n/a | case BT_HEX: |
|---|
| 1673 | n/a | case BT_DIGIT: |
|---|
| 1674 | n/a | case BT_NAME: |
|---|
| 1675 | n/a | case BT_MINUS: |
|---|
| 1676 | n/a | return 0; |
|---|
| 1677 | n/a | default: |
|---|
| 1678 | n/a | return 1; |
|---|
| 1679 | n/a | } |
|---|
| 1680 | n/a | } |
|---|
| 1681 | n/a | } |
|---|
| 1682 | n/a | /* not reached */ |
|---|
| 1683 | n/a | } |
|---|
| 1684 | n/a | |
|---|
| 1685 | n/a | static int PTRCALL |
|---|
| 1686 | n/a | PREFIX(nameMatchesAscii)(const ENCODING *enc, const char *ptr1, |
|---|
| 1687 | n/a | const char *end1, const char *ptr2) |
|---|
| 1688 | n/a | { |
|---|
| 1689 | n/a | for (; *ptr2; ptr1 += MINBPC(enc), ptr2++) { |
|---|
| 1690 | n/a | if (ptr1 == end1) |
|---|
| 1691 | n/a | return 0; |
|---|
| 1692 | n/a | if (!CHAR_MATCHES(enc, ptr1, *ptr2)) |
|---|
| 1693 | n/a | return 0; |
|---|
| 1694 | n/a | } |
|---|
| 1695 | n/a | return ptr1 == end1; |
|---|
| 1696 | n/a | } |
|---|
| 1697 | n/a | |
|---|
| 1698 | n/a | static int PTRFASTCALL |
|---|
| 1699 | n/a | PREFIX(nameLength)(const ENCODING *enc, const char *ptr) |
|---|
| 1700 | n/a | { |
|---|
| 1701 | n/a | const char *start = ptr; |
|---|
| 1702 | n/a | for (;;) { |
|---|
| 1703 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 1704 | n/a | #define LEAD_CASE(n) \ |
|---|
| 1705 | n/a | case BT_LEAD ## n: ptr += n; break; |
|---|
| 1706 | n/a | LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4) |
|---|
| 1707 | n/a | #undef LEAD_CASE |
|---|
| 1708 | n/a | case BT_NONASCII: |
|---|
| 1709 | n/a | case BT_NMSTRT: |
|---|
| 1710 | n/a | #ifdef XML_NS |
|---|
| 1711 | n/a | case BT_COLON: |
|---|
| 1712 | n/a | #endif |
|---|
| 1713 | n/a | case BT_HEX: |
|---|
| 1714 | n/a | case BT_DIGIT: |
|---|
| 1715 | n/a | case BT_NAME: |
|---|
| 1716 | n/a | case BT_MINUS: |
|---|
| 1717 | n/a | ptr += MINBPC(enc); |
|---|
| 1718 | n/a | break; |
|---|
| 1719 | n/a | default: |
|---|
| 1720 | n/a | return (int)(ptr - start); |
|---|
| 1721 | n/a | } |
|---|
| 1722 | n/a | } |
|---|
| 1723 | n/a | } |
|---|
| 1724 | n/a | |
|---|
| 1725 | n/a | static const char * PTRFASTCALL |
|---|
| 1726 | n/a | PREFIX(skipS)(const ENCODING *enc, const char *ptr) |
|---|
| 1727 | n/a | { |
|---|
| 1728 | n/a | for (;;) { |
|---|
| 1729 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 1730 | n/a | case BT_LF: |
|---|
| 1731 | n/a | case BT_CR: |
|---|
| 1732 | n/a | case BT_S: |
|---|
| 1733 | n/a | ptr += MINBPC(enc); |
|---|
| 1734 | n/a | break; |
|---|
| 1735 | n/a | default: |
|---|
| 1736 | n/a | return ptr; |
|---|
| 1737 | n/a | } |
|---|
| 1738 | n/a | } |
|---|
| 1739 | n/a | } |
|---|
| 1740 | n/a | |
|---|
| 1741 | n/a | static void PTRCALL |
|---|
| 1742 | n/a | PREFIX(updatePosition)(const ENCODING *enc, |
|---|
| 1743 | n/a | const char *ptr, |
|---|
| 1744 | n/a | const char *end, |
|---|
| 1745 | n/a | POSITION *pos) |
|---|
| 1746 | n/a | { |
|---|
| 1747 | n/a | while (ptr < end) { |
|---|
| 1748 | n/a | switch (BYTE_TYPE(enc, ptr)) { |
|---|
| 1749 | n/a | #define LEAD_CASE(n) \ |
|---|
| 1750 | n/a | case BT_LEAD ## n: \ |
|---|
| 1751 | n/a | ptr += n; \ |
|---|
| 1752 | n/a | break; |
|---|
| 1753 | n/a | LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4) |
|---|
| 1754 | n/a | #undef LEAD_CASE |
|---|
| 1755 | n/a | case BT_LF: |
|---|
| 1756 | n/a | pos->columnNumber = (XML_Size)-1; |
|---|
| 1757 | n/a | pos->lineNumber++; |
|---|
| 1758 | n/a | ptr += MINBPC(enc); |
|---|
| 1759 | n/a | break; |
|---|
| 1760 | n/a | case BT_CR: |
|---|
| 1761 | n/a | pos->lineNumber++; |
|---|
| 1762 | n/a | ptr += MINBPC(enc); |
|---|
| 1763 | n/a | if (ptr != end && BYTE_TYPE(enc, ptr) == BT_LF) |
|---|
| 1764 | n/a | ptr += MINBPC(enc); |
|---|
| 1765 | n/a | pos->columnNumber = (XML_Size)-1; |
|---|
| 1766 | n/a | break; |
|---|
| 1767 | n/a | default: |
|---|
| 1768 | n/a | ptr += MINBPC(enc); |
|---|
| 1769 | n/a | break; |
|---|
| 1770 | n/a | } |
|---|
| 1771 | n/a | pos->columnNumber++; |
|---|
| 1772 | n/a | } |
|---|
| 1773 | n/a | } |
|---|
| 1774 | n/a | |
|---|
| 1775 | n/a | #undef DO_LEAD_CASE |
|---|
| 1776 | n/a | #undef MULTIBYTE_CASES |
|---|
| 1777 | n/a | #undef INVALID_CASES |
|---|
| 1778 | n/a | #undef CHECK_NAME_CASE |
|---|
| 1779 | n/a | #undef CHECK_NAME_CASES |
|---|
| 1780 | n/a | #undef CHECK_NMSTRT_CASE |
|---|
| 1781 | n/a | #undef CHECK_NMSTRT_CASES |
|---|
| 1782 | n/a | |
|---|
| 1783 | n/a | #endif /* XML_TOK_IMPL_C */ |
|---|