[163] | 1 | /* -*- c++ -*- |
---|
| 2 | |
---|
| 3 | C++ xml parser template classes |
---|
| 4 | |
---|
| 5 | This file is part of the dpp library of C++ template classes |
---|
| 6 | |
---|
| 7 | doc: http://diaxen.ssji.net/dpp/index.html |
---|
| 8 | repo: https://www.ssji.net/svn/projets/trunk/libdpp |
---|
| 9 | |
---|
| 10 | This program is free software: you can redistribute it and/or |
---|
| 11 | modify it under the terms of the GNU Lesser General Public License |
---|
| 12 | as published by the Free Software Foundation, either version 3 of |
---|
| 13 | the License, or (at your option) any later version. |
---|
| 14 | |
---|
| 15 | This program is distributed in the hope that it will be useful, but |
---|
| 16 | WITHOUT ANY WARRANTY; without even the implied warranty of |
---|
| 17 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
---|
| 18 | Lesser General Public License for more details. |
---|
| 19 | |
---|
| 20 | You should have received a copy of the GNU Lesser General Public |
---|
| 21 | License along with this program. If not, see |
---|
| 22 | <http://www.gnu.org/licenses/>. |
---|
| 23 | |
---|
| 24 | (c) 2011 Alexandre Becoulet <alexandre.becoulet@free.fr> |
---|
| 25 | |
---|
| 26 | */ |
---|
| 27 | |
---|
| 28 | #ifndef DPP_XML_HH_ |
---|
| 29 | #define DPP_XML_HH_ |
---|
| 30 | |
---|
| 31 | /** @file @module{XML} */ |
---|
| 32 | |
---|
| 33 | #include <map> |
---|
| 34 | #include <vector> |
---|
| 35 | #include <string> |
---|
| 36 | #include <sstream> |
---|
| 37 | #include <exception> |
---|
| 38 | #include <cassert> |
---|
| 39 | #include <cctype> |
---|
| 40 | |
---|
| 41 | namespace dpp { |
---|
| 42 | |
---|
| 43 | template <class String> class xml_scanner_ns; |
---|
| 44 | template <class String> class xml_scanner; |
---|
| 45 | |
---|
| 46 | /** |
---|
| 47 | @short XML error class |
---|
| 48 | @module {XML} |
---|
| 49 | @header dpp/xml |
---|
| 50 | */ |
---|
| 51 | class xml_error : public std::exception |
---|
| 52 | { |
---|
| 53 | std::string _what; |
---|
| 54 | |
---|
| 55 | public: |
---|
| 56 | |
---|
| 57 | const char* what() const throw() |
---|
| 58 | { |
---|
| 59 | return _what.c_str(); |
---|
| 60 | } |
---|
| 61 | |
---|
| 62 | xml_error(const char *msg, unsigned int line) |
---|
| 63 | { |
---|
| 64 | std::ostringstream s; |
---|
| 65 | s << "XML:" << msg << " at line " << line; |
---|
| 66 | _what = s.str(); |
---|
| 67 | } |
---|
| 68 | |
---|
| 69 | ~xml_error() throw() |
---|
| 70 | { |
---|
| 71 | } |
---|
| 72 | }; |
---|
| 73 | |
---|
| 74 | ////////////////////////////////////////////////////////////////////// |
---|
| 75 | // XML scanner |
---|
| 76 | ////////////////////////////////////////////////////////////////////// |
---|
| 77 | |
---|
| 78 | /** @internal */ |
---|
| 79 | template <class String> |
---|
| 80 | class xml_scanner_base |
---|
| 81 | { |
---|
| 82 | template <class> friend class xml_scanner_ns; |
---|
| 83 | template <class> friend class xml_scanner; |
---|
| 84 | |
---|
| 85 | public: |
---|
| 86 | /** @This starts XML document scanning. */ |
---|
| 87 | void scan() throw(xml_error) |
---|
| 88 | { |
---|
| 89 | process_root(); |
---|
| 90 | } |
---|
| 91 | |
---|
| 92 | protected: |
---|
| 93 | /** @internal */ |
---|
| 94 | xml_scanner_base(std::istream &i) |
---|
| 95 | : _i(&i), |
---|
| 96 | _depth(0), |
---|
| 97 | _line(1) |
---|
| 98 | { |
---|
| 99 | } |
---|
| 100 | |
---|
| 101 | /** @This is called on proper document end */ |
---|
| 102 | virtual void scan_end_doc() |
---|
| 103 | { |
---|
| 104 | } |
---|
| 105 | |
---|
| 106 | /** @This is called on document start */ |
---|
| 107 | virtual void scan_start_doc() |
---|
| 108 | { |
---|
| 109 | } |
---|
| 110 | |
---|
| 111 | /** @This is called for each XML header attribute found */ |
---|
| 112 | virtual void scan_header_value(const String &name, const String &value) |
---|
| 113 | { |
---|
| 114 | } |
---|
| 115 | |
---|
| 116 | /** @This is called when a @tt CDATA section has been scand */ |
---|
| 117 | virtual void scan_cdata(const String &cdata) |
---|
| 118 | { |
---|
| 119 | } |
---|
| 120 | |
---|
| 121 | /** @This is called when a comment has been scand */ |
---|
| 122 | virtual void scan_comment(const String &comment) |
---|
| 123 | { |
---|
| 124 | } |
---|
| 125 | |
---|
| 126 | /** @This is called when a @tt{<! >} directive has been scand */ |
---|
| 127 | virtual void scan_directive(const String &dir) |
---|
| 128 | { |
---|
| 129 | } |
---|
| 130 | |
---|
| 131 | /** @This is called when a @tt{<? ?>} instruction has been scand */ |
---|
| 132 | virtual void scan_instruction(const String &ins) |
---|
| 133 | { |
---|
| 134 | } |
---|
| 135 | |
---|
| 136 | /** @This is called with content found before the next start/end tag */ |
---|
| 137 | virtual void scan_content(const String &content) |
---|
| 138 | { |
---|
| 139 | } |
---|
| 140 | |
---|
| 141 | /** @This is called to substitute entity reference strings. The default |
---|
| 142 | implementation handles predefined entities (@em amp, @em apos, |
---|
| 143 | @em lt, @em gt and @em quot). */ |
---|
| 144 | virtual void scan_get_reference(String &ref) |
---|
| 145 | { |
---|
| 146 | if (ref == "amp") |
---|
| 147 | ref = "&"; |
---|
| 148 | else if (ref == "apos") |
---|
| 149 | ref = "'"; |
---|
| 150 | else if (ref == "gt") |
---|
| 151 | ref = ">"; |
---|
| 152 | else if (ref == "lt") |
---|
| 153 | ref = "<"; |
---|
| 154 | else if (ref == "quot") |
---|
| 155 | ref = "\""; |
---|
| 156 | else |
---|
| 157 | ref = ""; |
---|
| 158 | } |
---|
| 159 | |
---|
| 160 | /** @This returns current tags nesting level. */ |
---|
| 161 | unsigned int depth() |
---|
| 162 | { |
---|
| 163 | return _depth; |
---|
| 164 | } |
---|
| 165 | |
---|
| 166 | /** @This returns current line number. */ |
---|
| 167 | unsigned int line() |
---|
| 168 | { |
---|
| 169 | return _line; |
---|
| 170 | } |
---|
| 171 | |
---|
| 172 | /** @internal throw */ |
---|
| 173 | void error(const char *msg) |
---|
| 174 | { |
---|
| 175 | throw xml_error(msg, _line); |
---|
| 176 | } |
---|
| 177 | |
---|
| 178 | private: |
---|
| 179 | int getc_eof() |
---|
| 180 | { |
---|
| 181 | char c; |
---|
| 182 | |
---|
| 183 | _i->get(c); |
---|
| 184 | |
---|
| 185 | if (_i->eof()) |
---|
| 186 | return -1; |
---|
| 187 | |
---|
| 188 | if (!_i->good()) |
---|
| 189 | error("io error reading xml stream"); |
---|
| 190 | |
---|
| 191 | if (c == '\n') |
---|
| 192 | _line++; |
---|
| 193 | |
---|
| 194 | return (unsigned char)c; |
---|
| 195 | } |
---|
| 196 | |
---|
| 197 | int getc() |
---|
| 198 | { |
---|
| 199 | int c = getc_eof(); |
---|
| 200 | |
---|
| 201 | if (c < 0) |
---|
| 202 | error("unexpected end of xml stream"); |
---|
| 203 | |
---|
| 204 | return c; |
---|
| 205 | } |
---|
| 206 | |
---|
| 207 | void putc(String &str, int c) |
---|
| 208 | { |
---|
| 209 | str += c; |
---|
| 210 | } |
---|
| 211 | |
---|
| 212 | void putc_data(String &str, int c) |
---|
| 213 | { |
---|
| 214 | switch (c) |
---|
| 215 | { |
---|
| 216 | case '&': |
---|
| 217 | switch (c = getc()) |
---|
| 218 | { |
---|
| 219 | // character reference |
---|
| 220 | case '#': { |
---|
| 221 | int base = 10; |
---|
| 222 | int r = 0; |
---|
| 223 | while ((c = getc()) != ';') |
---|
| 224 | { |
---|
| 225 | switch (c) |
---|
| 226 | { |
---|
| 227 | case 'x': |
---|
| 228 | base = 16; |
---|
| 229 | break; |
---|
| 230 | case '0': case '1': case '2': case '3': case '4': |
---|
| 231 | case '5': case '6': case '7': case '8': case '9': |
---|
| 232 | r = r * base + c - '0'; |
---|
| 233 | break; |
---|
| 234 | case 'A': case 'B': case 'C': |
---|
| 235 | case 'D': case 'E': case 'F': |
---|
| 236 | r = r * base + 10 + c - 'A'; |
---|
| 237 | break; |
---|
| 238 | case 'a': case 'b': case 'c': |
---|
| 239 | case 'd': case 'e': case 'f': |
---|
| 240 | r = r * base + 10 + c - 'a'; |
---|
| 241 | break; |
---|
| 242 | default: |
---|
| 243 | error("bad escaping"); |
---|
| 244 | } |
---|
| 245 | } |
---|
| 246 | return putc(str, r); |
---|
| 247 | } |
---|
| 248 | |
---|
| 249 | // entity reference |
---|
| 250 | default: { |
---|
| 251 | String ref; |
---|
| 252 | do { |
---|
| 253 | ref += c; |
---|
| 254 | c = getc(); |
---|
| 255 | } while (c != ';'); |
---|
| 256 | scan_get_reference(ref); |
---|
| 257 | str += ref; |
---|
| 258 | return; |
---|
| 259 | } |
---|
| 260 | } |
---|
| 261 | |
---|
| 262 | default: |
---|
| 263 | return putc(str, c); |
---|
| 264 | } |
---|
| 265 | } |
---|
| 266 | |
---|
| 267 | void process_attr_value(String &value, int c) |
---|
| 268 | { |
---|
| 269 | for (; c != '='; c = getc()) |
---|
| 270 | if (c > 32) |
---|
| 271 | error("unexpected char in attribute name"); |
---|
| 272 | for (c = getc(); c != '"' && c != '\''; c = getc()) |
---|
| 273 | if (c > 32) |
---|
| 274 | error("unexpected char in attribute value"); |
---|
| 275 | int q = c; |
---|
| 276 | for (c = getc(); c != q; c = getc()) |
---|
| 277 | putc_data(value, c); |
---|
| 278 | } |
---|
| 279 | |
---|
| 280 | void process_instruction() |
---|
| 281 | { |
---|
| 282 | _misc.clear(); |
---|
| 283 | while (1) |
---|
| 284 | { |
---|
| 285 | int c = getc(); |
---|
| 286 | |
---|
| 287 | switch (c) |
---|
| 288 | { |
---|
| 289 | case '?': |
---|
| 290 | if (getc() != '>') |
---|
| 291 | error("unexpected char in instruction"); |
---|
| 292 | return scan_instruction(_misc); |
---|
| 293 | |
---|
| 294 | case '"': |
---|
| 295 | case '\'': { |
---|
| 296 | int q = c; |
---|
| 297 | putc_data(_misc, c); |
---|
| 298 | for (c = getc(); c != q; c = getc()) |
---|
| 299 | putc_data(_misc, c); |
---|
| 300 | } |
---|
| 301 | |
---|
| 302 | default: |
---|
| 303 | putc(_misc, c); |
---|
| 304 | } |
---|
| 305 | } |
---|
| 306 | } |
---|
| 307 | |
---|
| 308 | void process_comment() |
---|
| 309 | { |
---|
| 310 | int c1 = 0, c2 = 0; |
---|
| 311 | while (1) |
---|
| 312 | { |
---|
| 313 | int c = getc(); |
---|
| 314 | |
---|
| 315 | switch (c) |
---|
| 316 | { |
---|
| 317 | case '>': |
---|
| 318 | if (c1 == '-' && c2 == '-') |
---|
| 319 | { |
---|
| 320 | _misc.resize(_misc.size()-2); |
---|
| 321 | return scan_comment(_misc); |
---|
| 322 | } |
---|
| 323 | |
---|
| 324 | default: |
---|
| 325 | c2 = c1; |
---|
| 326 | c1 = c; |
---|
| 327 | putc(_misc, c); |
---|
| 328 | } |
---|
| 329 | } |
---|
| 330 | } |
---|
| 331 | |
---|
| 332 | void process_cdata() |
---|
| 333 | { |
---|
| 334 | int c1 = 0, c2 = 0; |
---|
| 335 | while (1) |
---|
| 336 | { |
---|
| 337 | int c = getc(); |
---|
| 338 | |
---|
| 339 | switch (c) |
---|
| 340 | { |
---|
| 341 | case '>': |
---|
| 342 | if (c1 == ']' && c2 == ']') |
---|
| 343 | { |
---|
| 344 | _misc.resize(_misc.size()-2); |
---|
| 345 | return scan_cdata(_misc); |
---|
| 346 | } |
---|
| 347 | |
---|
| 348 | default: |
---|
| 349 | c2 = c1; |
---|
| 350 | c1 = c; |
---|
| 351 | putc(_misc, c); |
---|
| 352 | } |
---|
| 353 | } |
---|
| 354 | } |
---|
| 355 | |
---|
| 356 | void process_directive() |
---|
| 357 | { |
---|
| 358 | _misc.clear(); |
---|
| 359 | while (1) |
---|
| 360 | { |
---|
| 361 | int c = getc(); |
---|
| 362 | |
---|
| 363 | switch (c) |
---|
| 364 | { |
---|
| 365 | case '[': |
---|
| 366 | for (const char *s = "CDATA["; *s; s++) |
---|
| 367 | if (getc() != *s) |
---|
| 368 | error("bad directive"); |
---|
| 369 | return process_cdata(); |
---|
| 370 | |
---|
| 371 | case '-': |
---|
| 372 | if (getc() != '-') |
---|
| 373 | error("bad directive"); |
---|
| 374 | return process_comment(); |
---|
| 375 | |
---|
| 376 | default: { |
---|
| 377 | int r = 0; |
---|
| 378 | if (_depth) |
---|
| 379 | error("unexpected directive inside tag"); |
---|
| 380 | do { |
---|
| 381 | putc(_misc, c); |
---|
| 382 | switch (c) |
---|
| 383 | { |
---|
| 384 | case '"': |
---|
| 385 | case '\'': { |
---|
| 386 | int q = c; |
---|
| 387 | for (c = getc(); c != q; c = getc()) |
---|
| 388 | putc_data(_misc, c); |
---|
| 389 | putc_data(_misc, c); |
---|
| 390 | break; |
---|
| 391 | } |
---|
| 392 | case '[': |
---|
| 393 | r++; |
---|
| 394 | break; |
---|
| 395 | case ']': |
---|
| 396 | r--; |
---|
| 397 | break; |
---|
| 398 | } |
---|
| 399 | c = getc(); |
---|
| 400 | } while (r || c != '>'); |
---|
| 401 | |
---|
| 402 | return scan_directive(_misc); |
---|
| 403 | } |
---|
| 404 | } |
---|
| 405 | } |
---|
| 406 | } |
---|
| 407 | |
---|
| 408 | virtual void process_assignment(int c) = 0; |
---|
| 409 | |
---|
| 410 | bool process_attributes() |
---|
| 411 | { |
---|
| 412 | while (1) |
---|
| 413 | { |
---|
| 414 | int c = getc(); |
---|
| 415 | |
---|
| 416 | switch (c) |
---|
| 417 | { |
---|
| 418 | case '>': |
---|
| 419 | return false; |
---|
| 420 | |
---|
| 421 | case '/': |
---|
| 422 | c = getc(); |
---|
| 423 | if (c != '>') |
---|
| 424 | error("unexpected char in empty tag"); |
---|
| 425 | return true; |
---|
| 426 | |
---|
| 427 | case ' ': |
---|
| 428 | case '\t': |
---|
| 429 | case '\r': |
---|
| 430 | case '\n': |
---|
| 431 | continue; |
---|
| 432 | |
---|
| 433 | default: |
---|
| 434 | process_assignment(c); |
---|
| 435 | } |
---|
| 436 | } |
---|
| 437 | } |
---|
| 438 | |
---|
| 439 | virtual void process_start_tag_name(int c) = 0; |
---|
| 440 | virtual void process_end_tag_name() = 0; |
---|
| 441 | |
---|
| 442 | void process_end_tag_tail(int c) |
---|
| 443 | { |
---|
| 444 | while (c != '>') |
---|
| 445 | { |
---|
| 446 | if (c > 32) |
---|
| 447 | error("unexpected char in end tag"); |
---|
| 448 | c = getc(); |
---|
| 449 | } |
---|
| 450 | } |
---|
| 451 | |
---|
| 452 | void process_tag() |
---|
| 453 | { |
---|
| 454 | while (1) |
---|
| 455 | { |
---|
| 456 | int c = getc(); |
---|
| 457 | |
---|
| 458 | switch (c) |
---|
| 459 | { |
---|
| 460 | case '?': |
---|
| 461 | return process_instruction(); |
---|
| 462 | case '!': |
---|
| 463 | return process_directive(); |
---|
| 464 | |
---|
| 465 | case '/': { |
---|
| 466 | |
---|
| 467 | if (!_depth--) |
---|
| 468 | error("unexpected end tag"); |
---|
| 469 | |
---|
| 470 | process_end_tag_name(); |
---|
| 471 | return; |
---|
| 472 | } |
---|
| 473 | |
---|
| 474 | case ' ': |
---|
| 475 | case '\t': |
---|
| 476 | case '\r': |
---|
| 477 | case '\n': |
---|
| 478 | continue; |
---|
| 479 | |
---|
| 480 | default: |
---|
| 481 | return process_start_tag_name(c); |
---|
| 482 | } |
---|
| 483 | } |
---|
| 484 | } |
---|
| 485 | |
---|
| 486 | void process_root() |
---|
| 487 | { |
---|
| 488 | int c; |
---|
| 489 | |
---|
| 490 | scan_start_doc(); |
---|
| 491 | |
---|
| 492 | do { |
---|
| 493 | c = getc(); |
---|
| 494 | } while (c < 32); |
---|
| 495 | for (const char *s = "<?xml"; *s; c = getc()) |
---|
| 496 | if (c > 32 && std::tolower(c) != *s++) |
---|
| 497 | error("bad header"); |
---|
| 498 | while (1) |
---|
| 499 | { |
---|
| 500 | c = getc(); |
---|
| 501 | switch (c) |
---|
| 502 | { |
---|
| 503 | case '?': |
---|
| 504 | goto hdr_done; |
---|
| 505 | |
---|
| 506 | case ' ': |
---|
| 507 | case '\t': |
---|
| 508 | case '\r': |
---|
| 509 | case '\n': |
---|
| 510 | break; |
---|
| 511 | |
---|
| 512 | default: { |
---|
| 513 | String attr, value; |
---|
| 514 | |
---|
| 515 | do { |
---|
| 516 | putc(attr, c); |
---|
| 517 | c = getc(); |
---|
| 518 | } while (c > 32 && c != '='); |
---|
| 519 | |
---|
| 520 | process_attr_value(value, c); |
---|
| 521 | scan_header_value(attr, value); |
---|
| 522 | } |
---|
| 523 | } |
---|
| 524 | } |
---|
| 525 | hdr_done: |
---|
| 526 | if (getc() != '>') |
---|
| 527 | error("bad header end"); |
---|
| 528 | |
---|
| 529 | c = getc_eof(); |
---|
| 530 | while (1) |
---|
| 531 | { |
---|
| 532 | switch (c) |
---|
| 533 | { |
---|
| 534 | case '<': |
---|
| 535 | process_tag(); |
---|
| 536 | break; |
---|
| 537 | |
---|
| 538 | case -1: |
---|
| 539 | if (_depth) |
---|
| 540 | error("unexpected end of xml stream"); |
---|
| 541 | scan_end_doc(); |
---|
| 542 | return; |
---|
| 543 | |
---|
| 544 | default: { |
---|
| 545 | String content; |
---|
| 546 | do { |
---|
| 547 | putc_data(content, c); |
---|
| 548 | c = getc_eof(); |
---|
| 549 | } while (c > 0 && c != '<'); |
---|
| 550 | scan_content(content); |
---|
| 551 | continue; |
---|
| 552 | } |
---|
| 553 | } |
---|
| 554 | c = getc_eof(); |
---|
| 555 | } |
---|
| 556 | } |
---|
| 557 | |
---|
| 558 | private: |
---|
| 559 | std::istream *_i; |
---|
| 560 | unsigned int _depth; |
---|
| 561 | unsigned int _line; |
---|
| 562 | String _misc; |
---|
| 563 | }; |
---|
| 564 | |
---|
| 565 | /** |
---|
| 566 | @short XML scanner |
---|
| 567 | @module {XML} |
---|
| 568 | @header dpp/xml |
---|
| 569 | @see xml_event_parser |
---|
| 570 | |
---|
| 571 | @This implements an XML scanner. Derived classes must override |
---|
| 572 | some functions to handle XML scanning events. |
---|
| 573 | |
---|
| 574 | @ref xml_error exception is thrown on syntax error. Start and |
---|
| 575 | end tags count must be balanced but tags name missmatch is not checked |
---|
| 576 | here. |
---|
| 577 | |
---|
| 578 | The @tt String template parameter may be used to specify a |
---|
| 579 | string class with optional wide characters support. |
---|
| 580 | |
---|
| 581 | @section {Example} |
---|
| 582 | @example test/test_xml_scanner.cc:scanner P |
---|
| 583 | @example test/test_xml_scanner.cc:use |
---|
| 584 | @end section |
---|
| 585 | */ |
---|
| 586 | template <class String> |
---|
| 587 | class xml_scanner : public xml_scanner_base<String> |
---|
| 588 | { |
---|
| 589 | typedef xml_scanner_base<String> base; |
---|
| 590 | std::istringstream _ss; |
---|
| 591 | |
---|
| 592 | public: |
---|
| 593 | /** @This create an xml scanner for processing from input stream. */ |
---|
| 594 | xml_scanner(std::istream &i) |
---|
| 595 | : xml_scanner_base<String>(i) |
---|
| 596 | { |
---|
| 597 | } |
---|
| 598 | |
---|
| 599 | /** @This create an xml scanner for processing from string content. */ |
---|
| 600 | xml_scanner(const std::string &s) |
---|
| 601 | : _ss(s), |
---|
| 602 | xml_scanner_base<String>(_ss) |
---|
| 603 | { |
---|
| 604 | } |
---|
| 605 | |
---|
| 606 | private: |
---|
| 607 | |
---|
| 608 | void process_assignment(int c) |
---|
| 609 | { |
---|
| 610 | String attr, value; |
---|
| 611 | |
---|
| 612 | do { |
---|
| 613 | base::putc(attr, c); |
---|
| 614 | c = base::getc(); |
---|
| 615 | } while (c > 32 && c != '='); |
---|
| 616 | |
---|
| 617 | base::process_attr_value(value, c); |
---|
| 618 | scan_attribute(attr, value); |
---|
| 619 | } |
---|
| 620 | |
---|
| 621 | void process_start_tag_name(int c) |
---|
| 622 | { |
---|
| 623 | String stag; |
---|
| 624 | do { |
---|
| 625 | base::putc(stag, c); |
---|
| 626 | c = base::getc(); |
---|
| 627 | } while (c > 32 && c != '>' && c != '/'); |
---|
| 628 | |
---|
| 629 | if (c == '\n') |
---|
| 630 | base::_line--; |
---|
| 631 | base::_i->unget(); |
---|
| 632 | |
---|
| 633 | scan_start_tag(stag); |
---|
| 634 | if (base::process_attributes()) |
---|
| 635 | scan_end_tag(stag); |
---|
| 636 | else |
---|
| 637 | base::_depth++; |
---|
| 638 | } |
---|
| 639 | |
---|
| 640 | void process_end_tag_name() |
---|
| 641 | { |
---|
| 642 | String etag; |
---|
| 643 | int c; |
---|
| 644 | |
---|
| 645 | while ((c = base::getc()) != '>' && c > 32) |
---|
| 646 | base::putc(etag, c); |
---|
| 647 | |
---|
| 648 | base::process_end_tag_tail(c); |
---|
| 649 | scan_end_tag(etag); |
---|
| 650 | } |
---|
| 651 | |
---|
| 652 | protected: |
---|
| 653 | |
---|
| 654 | /** @This is called for each start tag */ |
---|
| 655 | virtual void scan_start_tag(const String &start_tag) |
---|
| 656 | { |
---|
| 657 | } |
---|
| 658 | |
---|
| 659 | /** @This is called for each end tag */ |
---|
| 660 | virtual void scan_end_tag(const String &end_tag) |
---|
| 661 | { |
---|
| 662 | } |
---|
| 663 | |
---|
| 664 | /** @This is called for each attribute found in the preceding start tag */ |
---|
| 665 | virtual void scan_attribute(const String &name, const String &value) |
---|
| 666 | { |
---|
| 667 | } |
---|
| 668 | }; |
---|
| 669 | |
---|
| 670 | /** |
---|
| 671 | @short XML scanner with namespace support |
---|
| 672 | @module {XML} |
---|
| 673 | @header dpp/xml |
---|
| 674 | @see xml_scanner |
---|
| 675 | |
---|
| 676 | This class implements an XML scanner. This class is similar to |
---|
| 677 | @xref{xml_scanner} but tag and attribute namespace name is |
---|
| 678 | extracted. |
---|
| 679 | */ |
---|
| 680 | template <class String> |
---|
| 681 | class xml_scanner_ns : public xml_scanner_base<String> |
---|
| 682 | { |
---|
| 683 | typedef xml_scanner_base<String> base; |
---|
| 684 | |
---|
| 685 | std::istringstream _ss; |
---|
| 686 | |
---|
| 687 | public: |
---|
| 688 | |
---|
| 689 | /** @This create an xml scanner for processing from input stream. */ |
---|
| 690 | xml_scanner_ns(std::istream &i) |
---|
| 691 | : xml_scanner_base<String>(i) |
---|
| 692 | { |
---|
| 693 | } |
---|
| 694 | |
---|
| 695 | /** @This create an xml scanner for processing from string content. */ |
---|
| 696 | xml_scanner_ns(const std::string &s) |
---|
| 697 | : _ss(s), |
---|
| 698 | xml_scanner_base<String>(_ss) |
---|
| 699 | { |
---|
| 700 | } |
---|
| 701 | |
---|
| 702 | private: |
---|
| 703 | void process_assignment(int c) |
---|
| 704 | { |
---|
| 705 | String xmlns, attr, value; |
---|
| 706 | |
---|
| 707 | do { |
---|
| 708 | base::putc(xmlns, c); |
---|
| 709 | c = base::getc(); |
---|
| 710 | } while (c > 32 && c != '=' && c != ':'); |
---|
| 711 | |
---|
| 712 | if (c == ':') { |
---|
| 713 | c = base::getc(); |
---|
| 714 | do { |
---|
| 715 | base::putc(attr, c); |
---|
| 716 | c = base::getc(); |
---|
| 717 | } while (c > 32 && c != '='); |
---|
| 718 | } |
---|
| 719 | else |
---|
| 720 | attr.swap(xmlns); |
---|
| 721 | |
---|
| 722 | base::process_attr_value(value, c); |
---|
| 723 | scan_attribute(xmlns, attr, value); |
---|
| 724 | } |
---|
| 725 | |
---|
| 726 | void process_start_tag_name(int c) |
---|
| 727 | { |
---|
| 728 | String xmlns; |
---|
| 729 | String stag; |
---|
| 730 | do { |
---|
| 731 | base::putc(xmlns, c); |
---|
| 732 | c = base::getc(); |
---|
| 733 | } while (c > 32 && c != '>' && c != '/' && c != ':'); |
---|
| 734 | |
---|
| 735 | if (c == ':') { |
---|
| 736 | c = base::getc(); |
---|
| 737 | do { |
---|
| 738 | base::putc(stag, c); |
---|
| 739 | c = base::getc(); |
---|
| 740 | } while (c > 32 && c != '>' && c != '/'); |
---|
| 741 | } |
---|
| 742 | else |
---|
| 743 | stag.swap(xmlns); |
---|
| 744 | |
---|
| 745 | if (c == '\n') |
---|
| 746 | base::_line--; |
---|
| 747 | base::_i->unget(); |
---|
| 748 | |
---|
| 749 | scan_start_tag(xmlns, stag); |
---|
| 750 | if (base::process_attributes()) |
---|
| 751 | scan_end_tag(xmlns, stag); |
---|
| 752 | else |
---|
| 753 | base::_depth++; |
---|
| 754 | } |
---|
| 755 | |
---|
| 756 | void process_end_tag_name() |
---|
| 757 | { |
---|
| 758 | String xmlns; |
---|
| 759 | String etag; |
---|
| 760 | int c; |
---|
| 761 | |
---|
| 762 | while (1) |
---|
| 763 | { |
---|
| 764 | c = base::getc(); |
---|
| 765 | if (!((c != '>' && c != ':' && c > 32))) |
---|
| 766 | break; |
---|
| 767 | base::putc(xmlns, c); |
---|
| 768 | } |
---|
| 769 | |
---|
| 770 | if (c == ':') { |
---|
| 771 | while (1) |
---|
| 772 | { |
---|
| 773 | c = base::getc(); |
---|
| 774 | if (!((c != '>' && c > 32))) |
---|
| 775 | break; |
---|
| 776 | base::putc(etag, c); |
---|
| 777 | } |
---|
| 778 | } |
---|
| 779 | else |
---|
| 780 | etag.swap(xmlns); |
---|
| 781 | |
---|
| 782 | base::process_end_tag_tail(c); |
---|
| 783 | scan_end_tag(xmlns, etag); |
---|
| 784 | } |
---|
| 785 | |
---|
| 786 | protected: |
---|
| 787 | |
---|
| 788 | /** @This is called for each start tag */ |
---|
| 789 | virtual void scan_start_tag(const String &xmlns, const String &start_tag) |
---|
| 790 | { |
---|
| 791 | } |
---|
| 792 | |
---|
| 793 | /** @This is called for each end tag */ |
---|
| 794 | virtual void scan_end_tag(const String &xmlns, const String &end_tag) |
---|
| 795 | { |
---|
| 796 | } |
---|
| 797 | |
---|
| 798 | /** @This is called for each attribute found in the preceding start tag */ |
---|
| 799 | virtual void scan_attribute(const String &xmlns, const String &name, const String &value) |
---|
| 800 | { |
---|
| 801 | } |
---|
| 802 | }; |
---|
| 803 | |
---|
| 804 | /** @internal */ |
---|
| 805 | struct empty |
---|
| 806 | { |
---|
| 807 | }; |
---|
| 808 | |
---|
| 809 | ////////////////////////////////////////////////////////////////////// |
---|
| 810 | // XML event |
---|
| 811 | ////////////////////////////////////////////////////////////////////// |
---|
| 812 | |
---|
| 813 | template <class String> |
---|
| 814 | class xml_event_elements |
---|
| 815 | { |
---|
| 816 | protected: |
---|
| 817 | typedef std::map<String, String> attributes_map_t; |
---|
| 818 | |
---|
| 819 | /** @This is called for each parsed start tag. |
---|
| 820 | |
---|
| 821 | Attributes list is passed as a non-const reference so that it |
---|
| 822 | can be swapped with an other empty @ref std::map to avoid data copy. |
---|
| 823 | |
---|
| 824 | @This is called once the new tag @tt Node object has |
---|
| 825 | been pushed onto the processing stack, it can be accessed from |
---|
| 826 | @this as well as other parsing functions by using the @ref |
---|
| 827 | stack_top function. |
---|
| 828 | |
---|
| 829 | @see parse_start_doc @see parse_end_tag |
---|
| 830 | */ |
---|
| 831 | virtual void parse_start_tag(const String &name, attributes_map_t &attributes) |
---|
| 832 | { |
---|
| 833 | } |
---|
| 834 | |
---|
| 835 | /** @This is called for each parsed end tag. |
---|
| 836 | |
---|
| 837 | @This is called before the ended tag @tt Node object |
---|
| 838 | is poped from the processing stack. |
---|
| 839 | |
---|
| 840 | @see parse_start_tag |
---|
| 841 | */ |
---|
| 842 | virtual void parse_end_tag(const String &name) |
---|
| 843 | { |
---|
| 844 | } |
---|
| 845 | |
---|
| 846 | /** @This is called with content found before the next start/end tag */ |
---|
| 847 | virtual void parse_content(const String &content) |
---|
| 848 | { |
---|
| 849 | } |
---|
| 850 | |
---|
| 851 | /** @This is called when a @tt CDATA section has been parsed */ |
---|
| 852 | virtual void parse_cdata(const String &cdata) |
---|
| 853 | { |
---|
| 854 | } |
---|
| 855 | }; |
---|
| 856 | |
---|
| 857 | /** |
---|
| 858 | @short XML event parser |
---|
| 859 | @module {XML} |
---|
| 860 | @header dpp/xml |
---|
| 861 | @main |
---|
| 862 | @see xml_scanner |
---|
| 863 | |
---|
| 864 | @This implements an XML event based parser. A user @tt Node |
---|
| 865 | type may be embedded in the internal processing stack to ease |
---|
| 866 | building of in memory tree representation of XML documents (see @xref{example 2}). |
---|
| 867 | |
---|
| 868 | The @tt String template parameter may be used to specify a |
---|
| 869 | string class with optional wide characters support. |
---|
| 870 | |
---|
| 871 | @ref xml_error exception is thrown on input error. |
---|
| 872 | |
---|
| 873 | @section {Example 1} |
---|
| 874 | |
---|
| 875 | In this example we simply reprint the XML document during parsing. |
---|
| 876 | |
---|
| 877 | @example test/test_xml_event_parser.cc:parser P |
---|
| 878 | @example test/test_xml_event_parser.cc:use |
---|
| 879 | @end section |
---|
| 880 | |
---|
| 881 | @section {Example 2} |
---|
| 882 | |
---|
| 883 | In this example we use a @ref ref smart pointer to our @tt MyNode |
---|
| 884 | structure to build a tree from XML input. The smart pointer |
---|
| 885 | ensures that our tree is automatically freed even if an exception |
---|
| 886 | is thrown during parsing. |
---|
| 887 | |
---|
| 888 | @example test/test_xml_event_parser_tree.cc:node P |
---|
| 889 | @example test/test_xml_event_parser_tree.cc:parser |
---|
| 890 | @example test/test_xml_event_parser_tree.cc:use |
---|
| 891 | @end section |
---|
| 892 | */ |
---|
| 893 | template <class String, class Node = empty> |
---|
| 894 | class xml_event_parser |
---|
| 895 | : protected xml_scanner<String> |
---|
| 896 | , public xml_event_elements<String> |
---|
| 897 | { |
---|
| 898 | typedef xml_scanner<String> base; |
---|
| 899 | public: |
---|
| 900 | typedef std::map<String, String> attributes_map_t; |
---|
| 901 | |
---|
| 902 | /** @This forwards argument to @ref xml_scanner constructor */ |
---|
| 903 | template <typename T> |
---|
| 904 | xml_event_parser(T &t) |
---|
| 905 | : base(t) |
---|
| 906 | { |
---|
| 907 | } |
---|
| 908 | |
---|
| 909 | /** @This starts XML parsing and returns document root @tt Node. */ |
---|
| 910 | Node parse() throw(xml_error) |
---|
| 911 | { |
---|
| 912 | try { |
---|
| 913 | xml_scanner<String>::scan(); |
---|
| 914 | } catch (...) { |
---|
| 915 | parse_error(); |
---|
| 916 | cleanup(); |
---|
| 917 | throw; |
---|
| 918 | } |
---|
| 919 | |
---|
| 920 | Node r = _stack.back()._tc; |
---|
| 921 | cleanup(); |
---|
| 922 | return r; |
---|
| 923 | } |
---|
| 924 | |
---|
| 925 | protected: |
---|
| 926 | |
---|
| 927 | using base::line; |
---|
| 928 | using base::depth; |
---|
| 929 | |
---|
| 930 | /** @This is called on XML document start. |
---|
| 931 | |
---|
| 932 | XML header attributes list is passed as a non-const reference |
---|
| 933 | so that it can be swaped with an other empty @ref std::map to |
---|
| 934 | avoid data copy. |
---|
| 935 | |
---|
| 936 | @This is called once the new tag @tt Node object has |
---|
| 937 | been pushed onto the processing stack, it can be accessed by using |
---|
| 938 | the @ref stack_top function. |
---|
| 939 | |
---|
| 940 | @see parse_start_tag @see parse_end_doc |
---|
| 941 | */ |
---|
| 942 | virtual void parse_start_doc(attributes_map_t &header_values) |
---|
| 943 | { |
---|
| 944 | } |
---|
| 945 | |
---|
| 946 | /** @This is called on XML document parse end */ |
---|
| 947 | virtual void parse_end_doc() |
---|
| 948 | { |
---|
| 949 | } |
---|
| 950 | |
---|
| 951 | /** @This is called when a "@tt{<! >}" directive has been parsed */ |
---|
| 952 | virtual void parse_directive(const String &dir) |
---|
| 953 | { |
---|
| 954 | } |
---|
| 955 | |
---|
| 956 | /** @This is called when a comment has been parsed */ |
---|
| 957 | virtual void parse_comment(const String &comment) |
---|
| 958 | { |
---|
| 959 | } |
---|
| 960 | |
---|
| 961 | /** @This is called when a "@tt{<? ?>}" instruction has been parsed */ |
---|
| 962 | virtual void parse_instruction(const String &ins) |
---|
| 963 | { |
---|
| 964 | } |
---|
| 965 | |
---|
| 966 | /** @This is called when a parse error occurs, before the internal |
---|
| 967 | stack is destroyed */ |
---|
| 968 | virtual void parse_error() |
---|
| 969 | { |
---|
| 970 | } |
---|
| 971 | |
---|
| 972 | /** @This returns node on stack relative to stack top, -1 is stack top. */ |
---|
| 973 | Node & stack_top(int i) |
---|
| 974 | { |
---|
| 975 | return _stack[_stack.size() + i]._tc; |
---|
| 976 | } |
---|
| 977 | |
---|
| 978 | /** @This returns node on stack relative to stack bottom, 0 is stack bottom. */ |
---|
| 979 | Node & stack_bottom(int i) |
---|
| 980 | { |
---|
| 981 | return _stack[i]._tc; |
---|
| 982 | } |
---|
| 983 | |
---|
| 984 | /** @This returns current stack size */ |
---|
| 985 | size_t stack_size() |
---|
| 986 | { |
---|
| 987 | return _stack.size(); |
---|
| 988 | } |
---|
| 989 | |
---|
| 990 | private: |
---|
| 991 | |
---|
| 992 | /** @override */ |
---|
| 993 | void scan_cdata(const String &cdata) |
---|
| 994 | { |
---|
| 995 | (this->*_do_start)(); |
---|
| 996 | parse_cdata(cdata); |
---|
| 997 | } |
---|
| 998 | |
---|
| 999 | /** @override */ |
---|
| 1000 | void scan_comment(const String &comment) |
---|
| 1001 | { |
---|
| 1002 | (this->*_do_start)(); |
---|
| 1003 | parse_comment(comment); |
---|
| 1004 | } |
---|
| 1005 | |
---|
| 1006 | /** @override */ |
---|
| 1007 | void scan_directive(const String &dir) |
---|
| 1008 | { |
---|
| 1009 | (this->*_do_start)(); |
---|
| 1010 | parse_directive(dir); |
---|
| 1011 | } |
---|
| 1012 | |
---|
| 1013 | /** @override */ |
---|
| 1014 | void scan_instruction(const String &ins) |
---|
| 1015 | { |
---|
| 1016 | (this->*_do_start)(); |
---|
| 1017 | parse_instruction(ins); |
---|
| 1018 | } |
---|
| 1019 | |
---|
| 1020 | /** @override */ |
---|
| 1021 | void scan_content(const String &content) |
---|
| 1022 | { |
---|
| 1023 | (this->*_do_start)(); |
---|
| 1024 | parse_content(content); |
---|
| 1025 | } |
---|
| 1026 | |
---|
| 1027 | /** @override */ |
---|
| 1028 | void scan_attribute(const String &name, const String &value) |
---|
| 1029 | { |
---|
| 1030 | _attr[name] = value; |
---|
| 1031 | } |
---|
| 1032 | |
---|
| 1033 | /** @override */ |
---|
| 1034 | void scan_start_tag(const String &start_tag) |
---|
| 1035 | { |
---|
| 1036 | (this->*_do_start)(); |
---|
| 1037 | _attr.clear(); |
---|
| 1038 | _started_tag = start_tag; |
---|
| 1039 | _do_start = &xml_event_parser::do_start_tag; |
---|
| 1040 | } |
---|
| 1041 | |
---|
| 1042 | /** @override */ |
---|
| 1043 | void scan_end_tag(const String &end_tag) |
---|
| 1044 | { |
---|
| 1045 | (this->*_do_start)(); |
---|
| 1046 | assert(_stack.size() > 1); |
---|
| 1047 | if (_stack.back()._name != end_tag) |
---|
| 1048 | base::error("end tag name doesn't match"); |
---|
| 1049 | parse_end_tag(_stack.back()._name); |
---|
| 1050 | _stack.pop_back(); |
---|
| 1051 | } |
---|
| 1052 | |
---|
| 1053 | /** @override */ |
---|
| 1054 | void scan_start_doc() |
---|
| 1055 | { |
---|
| 1056 | _do_start = &xml_event_parser::do_start_doc; |
---|
| 1057 | } |
---|
| 1058 | |
---|
| 1059 | /** @override */ |
---|
| 1060 | void scan_header_value(const String &name, const String &value) |
---|
| 1061 | { |
---|
| 1062 | _attr[name] = value; |
---|
| 1063 | } |
---|
| 1064 | |
---|
| 1065 | /** @override */ |
---|
| 1066 | void scan_end_doc() |
---|
| 1067 | { |
---|
| 1068 | assert(_stack.size() == 1); |
---|
| 1069 | parse_end_doc(); |
---|
| 1070 | } |
---|
| 1071 | |
---|
| 1072 | void do_start_none() |
---|
| 1073 | { |
---|
| 1074 | } |
---|
| 1075 | |
---|
| 1076 | void do_start_doc() |
---|
| 1077 | { |
---|
| 1078 | _do_start = &xml_event_parser::do_start_none; |
---|
| 1079 | _stack.push_back(stack_item("")); |
---|
| 1080 | parse_start_doc(_attr); |
---|
| 1081 | } |
---|
| 1082 | |
---|
| 1083 | void do_start_tag() |
---|
| 1084 | { |
---|
| 1085 | _do_start = &xml_event_parser::do_start_none; |
---|
| 1086 | _stack.push_back(stack_item(_started_tag)); |
---|
| 1087 | parse_start_tag(_started_tag, _attr); |
---|
| 1088 | } |
---|
| 1089 | |
---|
| 1090 | void cleanup() |
---|
| 1091 | { |
---|
| 1092 | _stack.clear(); |
---|
| 1093 | _attr.clear(); |
---|
| 1094 | _started_tag.clear(); |
---|
| 1095 | } |
---|
| 1096 | |
---|
| 1097 | struct stack_item |
---|
| 1098 | { |
---|
| 1099 | stack_item(const String &name) |
---|
| 1100 | : _name(name), _tc() |
---|
| 1101 | { |
---|
| 1102 | } |
---|
| 1103 | |
---|
| 1104 | String _name; |
---|
| 1105 | Node _tc; |
---|
| 1106 | }; |
---|
| 1107 | |
---|
| 1108 | std::vector<stack_item> _stack; |
---|
| 1109 | std::map<String, String> _attr; |
---|
| 1110 | void (xml_event_parser::*_do_start)(); |
---|
| 1111 | String _started_tag; |
---|
| 1112 | }; |
---|
| 1113 | |
---|
| 1114 | /** |
---|
| 1115 | @short XML event parser with namespace support |
---|
| 1116 | @module {XML} |
---|
| 1117 | @header dpp/xml |
---|
| 1118 | @main |
---|
| 1119 | @see xml_event_parser |
---|
| 1120 | |
---|
| 1121 | This class implements an XML event based parser. |
---|
| 1122 | */ |
---|
| 1123 | |
---|
| 1124 | ////////////////////////////////////////////////////////////////////// |
---|
| 1125 | // XML dom |
---|
| 1126 | ////////////////////////////////////////////////////////////////////// |
---|
| 1127 | |
---|
| 1128 | class xml_dom_node; |
---|
| 1129 | inline std::ostream & operator<<(std::ostream &o, const xml_dom_node &node); |
---|
| 1130 | |
---|
| 1131 | /** |
---|
| 1132 | @short XML dom node base class |
---|
| 1133 | @module {XML} |
---|
| 1134 | @header dpp/xml |
---|
| 1135 | */ |
---|
| 1136 | class xml_dom_node |
---|
| 1137 | { |
---|
| 1138 | friend std::ostream & operator<<(std::ostream &o, const xml_dom_node &node); |
---|
| 1139 | template <class> friend class xml_dom_parser; |
---|
| 1140 | template <class> friend class xml_dom_parent_node; |
---|
| 1141 | template <class> friend class xml_dom_element; |
---|
| 1142 | |
---|
| 1143 | protected: |
---|
| 1144 | virtual ~xml_dom_node() |
---|
| 1145 | { |
---|
| 1146 | } |
---|
| 1147 | |
---|
| 1148 | void push_back(xml_dom_node *b) |
---|
| 1149 | { |
---|
| 1150 | xml_dom_node *a = this; |
---|
| 1151 | b->_prev = a; |
---|
| 1152 | a->_next->_prev = b; |
---|
| 1153 | b->_next = a->_next; |
---|
| 1154 | a->_next = b; |
---|
| 1155 | } |
---|
| 1156 | |
---|
| 1157 | void push_front(xml_dom_node *b) |
---|
| 1158 | { |
---|
| 1159 | xml_dom_node *a = this; |
---|
| 1160 | b->_next = a; |
---|
| 1161 | a->_prev->_next = b; |
---|
| 1162 | b->_prev = a->_prev; |
---|
| 1163 | a->_prev = b; |
---|
| 1164 | } |
---|
| 1165 | |
---|
| 1166 | bool empty() const |
---|
| 1167 | { |
---|
| 1168 | return _next == _prev; |
---|
| 1169 | } |
---|
| 1170 | |
---|
| 1171 | private: |
---|
| 1172 | virtual void print(std::ostream &o) const |
---|
| 1173 | { |
---|
| 1174 | } |
---|
| 1175 | |
---|
| 1176 | xml_dom_node *_next; |
---|
| 1177 | xml_dom_node *_prev; |
---|
| 1178 | }; |
---|
| 1179 | |
---|
| 1180 | inline std::ostream & operator<<(std::ostream &o, const xml_dom_node &node) |
---|
| 1181 | { |
---|
| 1182 | node.print(o); |
---|
| 1183 | return o; |
---|
| 1184 | } |
---|
| 1185 | |
---|
| 1186 | /** @internal */ |
---|
| 1187 | template <class String> |
---|
| 1188 | class xml_dom_parent_node : public xml_dom_node |
---|
| 1189 | { |
---|
| 1190 | template <class> friend class xml_dom_parser; |
---|
| 1191 | |
---|
| 1192 | protected: |
---|
| 1193 | typedef xml_dom_node chld_base; |
---|
| 1194 | typedef xml_dom_node root_base; |
---|
| 1195 | typedef std::map<String, xml_dom_node *> child_hash_t; |
---|
| 1196 | |
---|
| 1197 | xml_dom_parent_node() |
---|
| 1198 | { |
---|
| 1199 | _root._next = &_root; |
---|
| 1200 | _root._prev = &_root; |
---|
| 1201 | } |
---|
| 1202 | |
---|
| 1203 | ~xml_dom_parent_node() |
---|
| 1204 | { |
---|
| 1205 | cleanup(); |
---|
| 1206 | } |
---|
| 1207 | |
---|
| 1208 | void print(std::ostream &o) const |
---|
| 1209 | { |
---|
| 1210 | for (xml_dom_node *i = _root._next; i != &_root; i = i->_next) |
---|
| 1211 | i->print(o); |
---|
| 1212 | } |
---|
| 1213 | |
---|
| 1214 | void cleanup() |
---|
| 1215 | { |
---|
| 1216 | xml_dom_node *j; |
---|
| 1217 | for (xml_dom_node *i = _root._next; i != &_root; i = j) |
---|
| 1218 | { |
---|
| 1219 | j = i->_next; |
---|
| 1220 | delete i; |
---|
| 1221 | } |
---|
| 1222 | } |
---|
| 1223 | |
---|
| 1224 | xml_dom_node _root; |
---|
| 1225 | child_hash_t _children; |
---|
| 1226 | }; |
---|
| 1227 | |
---|
| 1228 | /** @internal */ |
---|
| 1229 | template <class String> |
---|
| 1230 | class xml_dom_attr_node |
---|
| 1231 | { |
---|
| 1232 | typedef std::map<String, String> attr_hash_t; |
---|
| 1233 | |
---|
| 1234 | public: |
---|
| 1235 | /** @This returns a reference to attributes map */ |
---|
| 1236 | const attr_hash_t &get_attributes() const |
---|
| 1237 | { |
---|
| 1238 | return _attrs; |
---|
| 1239 | } |
---|
| 1240 | |
---|
| 1241 | /** @This returns a modifiable reference to attributes map */ |
---|
| 1242 | attr_hash_t &get_attributes() |
---|
| 1243 | { |
---|
| 1244 | return _attrs; |
---|
| 1245 | } |
---|
| 1246 | |
---|
| 1247 | /** @This returns value of the specified attribute or default a value */ |
---|
| 1248 | const String &get_attribute(const String &name, const String &default_ = String()) |
---|
| 1249 | { |
---|
| 1250 | typename attr_hash_t::const_iterator i = _attrs.find(name); |
---|
| 1251 | |
---|
| 1252 | if (i == _attrs.end()) |
---|
| 1253 | return default_; |
---|
| 1254 | |
---|
| 1255 | return i->second; |
---|
| 1256 | } |
---|
| 1257 | |
---|
| 1258 | protected: |
---|
| 1259 | |
---|
| 1260 | void print_attr(std::ostream &o) const |
---|
| 1261 | { |
---|
| 1262 | for (typename attr_hash_t::const_iterator i = _attrs.begin(); |
---|
| 1263 | i != _attrs.end(); i++) |
---|
| 1264 | o << " " << i->first << "=\"" << i->second << "\""; |
---|
| 1265 | } |
---|
| 1266 | |
---|
| 1267 | attr_hash_t _attrs; |
---|
| 1268 | }; |
---|
| 1269 | |
---|
| 1270 | /** |
---|
| 1271 | @short XML dom document root node |
---|
| 1272 | @module {XML} |
---|
| 1273 | @header dpp/xml |
---|
| 1274 | */ |
---|
| 1275 | template <class String> |
---|
| 1276 | class xml_dom_doc |
---|
| 1277 | : public xml_dom_parent_node<String> |
---|
| 1278 | , public xml_dom_attr_node<String> |
---|
| 1279 | { |
---|
| 1280 | template <class> friend class xml_dom_parser; |
---|
| 1281 | typedef xml_dom_parent_node<String> base; |
---|
| 1282 | |
---|
| 1283 | void print(std::ostream &o) const |
---|
| 1284 | { |
---|
| 1285 | o << "<?xml"; |
---|
| 1286 | xml_dom_attr_node<String>::print_attr(o); |
---|
| 1287 | o << " ?>"; |
---|
| 1288 | base::print(o); |
---|
| 1289 | } |
---|
| 1290 | }; |
---|
| 1291 | |
---|
| 1292 | /** |
---|
| 1293 | @short XML dom tag element |
---|
| 1294 | @module {XML} |
---|
| 1295 | @header dpp/xml |
---|
| 1296 | */ |
---|
| 1297 | template <class String> |
---|
| 1298 | class xml_dom_element |
---|
| 1299 | : public xml_dom_parent_node<String> |
---|
| 1300 | , public xml_dom_attr_node<String> |
---|
| 1301 | { |
---|
| 1302 | template <class> friend class xml_dom_parser; |
---|
| 1303 | typedef xml_dom_parent_node<String> base; |
---|
| 1304 | |
---|
| 1305 | public: |
---|
| 1306 | typedef std::map<String, String> attr_hash_t; |
---|
| 1307 | |
---|
| 1308 | /** @This returns element name */ |
---|
| 1309 | const String &get_name() const |
---|
| 1310 | { |
---|
| 1311 | } |
---|
| 1312 | |
---|
| 1313 | /** @This changes element name */ |
---|
| 1314 | void set_name(const String &name) const |
---|
| 1315 | { |
---|
| 1316 | _name = name; |
---|
| 1317 | } |
---|
| 1318 | |
---|
| 1319 | private: |
---|
| 1320 | xml_dom_element(const String &name) |
---|
| 1321 | : _name(name) |
---|
| 1322 | { |
---|
| 1323 | } |
---|
| 1324 | |
---|
| 1325 | void print(std::ostream &o) const |
---|
| 1326 | { |
---|
| 1327 | o << "<" << _name; |
---|
| 1328 | xml_dom_attr_node<String>::print_attr(o); |
---|
| 1329 | if (base::_root.empty()) |
---|
| 1330 | o << " />"; |
---|
| 1331 | else |
---|
| 1332 | { |
---|
| 1333 | o << ">"; |
---|
| 1334 | base::print(o); |
---|
| 1335 | o << "</" << _name << ">"; |
---|
| 1336 | } |
---|
| 1337 | } |
---|
| 1338 | |
---|
| 1339 | String _name; |
---|
| 1340 | }; |
---|
| 1341 | |
---|
| 1342 | /** |
---|
| 1343 | @short XML dom text and CDATA content node |
---|
| 1344 | @module {XML} |
---|
| 1345 | @header dpp/xml |
---|
| 1346 | */ |
---|
| 1347 | template <class String> |
---|
| 1348 | class xml_dom_content : public xml_dom_node |
---|
| 1349 | { |
---|
| 1350 | template <class> friend class xml_dom_parser; |
---|
| 1351 | |
---|
| 1352 | public: |
---|
| 1353 | /** @this returns content */ |
---|
| 1354 | const String & get_content() const |
---|
| 1355 | { |
---|
| 1356 | return _content; |
---|
| 1357 | } |
---|
| 1358 | |
---|
| 1359 | /** @this changes content */ |
---|
| 1360 | void set_content(const String &content) const |
---|
| 1361 | { |
---|
| 1362 | _content = content; |
---|
| 1363 | } |
---|
| 1364 | |
---|
| 1365 | /** @this returns true if content is in a CDATA section in xml file */ |
---|
| 1366 | const bool is_cdata() const |
---|
| 1367 | { |
---|
| 1368 | return _is_cdata; |
---|
| 1369 | } |
---|
| 1370 | |
---|
| 1371 | private: |
---|
| 1372 | xml_dom_content(const String &content, bool is_cdata = false) |
---|
| 1373 | : _content(content), |
---|
| 1374 | _is_cdata(is_cdata) |
---|
| 1375 | { |
---|
| 1376 | } |
---|
| 1377 | |
---|
| 1378 | void print(std::ostream &o) const |
---|
| 1379 | { |
---|
| 1380 | if (_is_cdata) |
---|
| 1381 | o << "<![CDATA[" << _content << "]]>"; |
---|
| 1382 | else |
---|
| 1383 | o << _content; |
---|
| 1384 | } |
---|
| 1385 | |
---|
| 1386 | String _content; |
---|
| 1387 | bool _is_cdata; |
---|
| 1388 | }; |
---|
| 1389 | |
---|
| 1390 | /** |
---|
| 1391 | @short XML dom parser |
---|
| 1392 | @module {XML} |
---|
| 1393 | @header dpp/xml |
---|
| 1394 | @see xml_event_parser |
---|
| 1395 | @main |
---|
| 1396 | |
---|
| 1397 | @This implements an XML dom parser. It builds an in memory |
---|
| 1398 | tree representation of the XML input. |
---|
| 1399 | |
---|
| 1400 | The @ref parse function must be used to start XML input parsing. |
---|
| 1401 | Nodes can be used with @ref std::ostream objects to write XML |
---|
| 1402 | output. |
---|
| 1403 | |
---|
| 1404 | The @tt String template parameter may be used to specify a |
---|
| 1405 | string class with optional wide characters support. |
---|
| 1406 | |
---|
| 1407 | @ref xml_error exception is thrown on input error. |
---|
| 1408 | |
---|
| 1409 | @section {Example} |
---|
| 1410 | @example test/test_xml_dom_parser.cc:read P |
---|
| 1411 | @end section |
---|
| 1412 | */ |
---|
| 1413 | template <class String> |
---|
| 1414 | class xml_dom_parser : protected xml_event_parser<String, xml_dom_parent_node<String>*> |
---|
| 1415 | { |
---|
| 1416 | typedef xml_event_parser<String, xml_dom_parent_node<String>*> base; |
---|
| 1417 | typedef xml_dom_element<String> element_t; |
---|
| 1418 | typedef xml_dom_node node_t; |
---|
| 1419 | |
---|
| 1420 | public: |
---|
| 1421 | typedef std::map<String, String> attributes_map_t; |
---|
| 1422 | |
---|
| 1423 | /** @This forwards argument to @ref xml_event_parser constructor */ |
---|
| 1424 | template <typename T> |
---|
| 1425 | xml_dom_parser(T &t) |
---|
| 1426 | : base(t) |
---|
| 1427 | { |
---|
| 1428 | } |
---|
| 1429 | |
---|
| 1430 | /** @This parses input and returns the root node */ |
---|
| 1431 | xml_dom_doc<String> * parse() throw(xml_error) |
---|
| 1432 | { |
---|
| 1433 | xml_dom_doc<String> *r; |
---|
| 1434 | |
---|
| 1435 | try { |
---|
| 1436 | r = static_cast<xml_dom_doc<String>*>(base::parse()); |
---|
| 1437 | } catch (...) { |
---|
| 1438 | throw; |
---|
| 1439 | } |
---|
| 1440 | |
---|
| 1441 | return r; |
---|
| 1442 | } |
---|
| 1443 | |
---|
| 1444 | private: |
---|
| 1445 | |
---|
| 1446 | /** @override */ |
---|
| 1447 | void parse_start_doc(attributes_map_t &attrs) |
---|
| 1448 | { |
---|
| 1449 | xml_dom_doc<String> *r = new xml_dom_doc<String>(); |
---|
| 1450 | base::stack_top(-1) = r; |
---|
| 1451 | r->_attrs.swap(attrs); |
---|
| 1452 | } |
---|
| 1453 | |
---|
| 1454 | /** @override */ |
---|
| 1455 | void parse_start_tag(const String &name, attributes_map_t &attrs) |
---|
| 1456 | { |
---|
| 1457 | element_t *e = new xml_dom_element<String>(name); |
---|
| 1458 | base::stack_top(-1) = e; |
---|
| 1459 | e->_attrs.swap(attrs); |
---|
| 1460 | } |
---|
| 1461 | |
---|
| 1462 | /** @override */ |
---|
| 1463 | void parse_end_tag(const String &name) |
---|
| 1464 | { |
---|
| 1465 | base::stack_top(-2)->_root.push_front(base::stack_top(-1)); |
---|
| 1466 | } |
---|
| 1467 | |
---|
| 1468 | /** @override */ |
---|
| 1469 | void parse_content(const String &content) |
---|
| 1470 | { |
---|
| 1471 | xml_dom_content<String> *c = new xml_dom_content<String>(content, false); |
---|
| 1472 | base::stack_top(-1)->_root.push_front(c); |
---|
| 1473 | } |
---|
| 1474 | |
---|
| 1475 | /** @override */ |
---|
| 1476 | void parse_cdata(const String &content) |
---|
| 1477 | { |
---|
| 1478 | xml_dom_content<String> *c = new xml_dom_content<String>(content, true); |
---|
| 1479 | base::stack_top(-1)->_root.push_front(c); |
---|
| 1480 | } |
---|
| 1481 | |
---|
| 1482 | /** @override */ |
---|
| 1483 | void parse_error() |
---|
| 1484 | { |
---|
| 1485 | for (size_t i = 0; i < base::stack_size(); i++) |
---|
| 1486 | delete base::stack_bottom(i); |
---|
| 1487 | } |
---|
| 1488 | |
---|
| 1489 | }; |
---|
| 1490 | |
---|
| 1491 | } |
---|
| 1492 | |
---|
| 1493 | #endif |
---|
| 1494 | |
---|