[1] | 1 | //----------------------------------------------------------- |
---|
| 2 | // Copyright Christian Arnault LAL-Orsay CNRS |
---|
| 3 | // arnault@lal.in2p3.fr |
---|
| 4 | // See the complete license in cmt_license.txt "http://www.cecill.info". |
---|
| 5 | //----------------------------------------------------------- |
---|
| 6 | |
---|
| 7 | #include "cmt_model.h" |
---|
| 8 | #include "cmt_fragment.h" |
---|
| 9 | #include "cmt_symbol.h" |
---|
| 10 | |
---|
| 11 | /** |
---|
| 12 | * Filters out all predefined XML constructs. |
---|
| 13 | */ |
---|
| 14 | void CmtModel::filter (cmt_string& text) |
---|
| 15 | { |
---|
| 16 | text.replace_all ("<cmt:tab/>", "\t"); |
---|
| 17 | text.replace_all ("<cmt:cr/>", "\r"); |
---|
| 18 | text.replace_all ("<cmt:lf/>", "\n"); |
---|
| 19 | } |
---|
| 20 | |
---|
| 21 | void CmtModel::display (cmt_string& text) |
---|
| 22 | { |
---|
| 23 | text.replace_all ("<", "<"); |
---|
| 24 | text.replace_all (">", ">"); |
---|
| 25 | text.replace_all ("\\\"", "\""); |
---|
| 26 | text.replace_all ("\\\'", "\'"); |
---|
| 27 | text.replace_all ("<cmt:null/>", ""); |
---|
| 28 | Symbol::expand (text); |
---|
| 29 | cout << text; |
---|
| 30 | } |
---|
| 31 | |
---|
| 32 | void CmtModel::expand (const cmt_string& input_text) |
---|
| 33 | { |
---|
| 34 | int openmarker; |
---|
| 35 | int closemarker; |
---|
| 36 | |
---|
| 37 | CmtSystem::cmt_string_vector subargs; |
---|
| 38 | |
---|
| 39 | cmt_string text = input_text; |
---|
| 40 | cmt_string remaining; |
---|
| 41 | |
---|
| 42 | filter (text); |
---|
| 43 | |
---|
| 44 | for (;;) |
---|
| 45 | { |
---|
| 46 | /** |
---|
| 47 | * Look for the next < ... /> pattern |
---|
| 48 | * If not found, then simply dump the text as it is. |
---|
| 49 | */ |
---|
| 50 | |
---|
| 51 | openmarker = text.find ("<"); |
---|
| 52 | if (openmarker == cmt_string::npos) break; |
---|
| 53 | |
---|
| 54 | closemarker = text.find (openmarker, "/>"); |
---|
| 55 | if (closemarker == cmt_string::npos) break; |
---|
| 56 | |
---|
| 57 | if (CmtSystem::testenv ("CMTTESTMODEL")) |
---|
| 58 | { |
---|
| 59 | cerr << "text=[" << text << "]" << endl; |
---|
| 60 | } |
---|
| 61 | |
---|
| 62 | |
---|
| 63 | /** |
---|
| 64 | * Extract the command from the pattern |
---|
| 65 | */ |
---|
| 66 | cmt_string command; |
---|
| 67 | text.substr (openmarker + 1, closemarker - openmarker - 1, command); |
---|
| 68 | |
---|
| 69 | if (CmtSystem::testenv ("CMTTESTMODEL")) |
---|
| 70 | { |
---|
| 71 | cerr << "command=[" << command << "]" << endl; |
---|
| 72 | } |
---|
| 73 | |
---|
| 74 | /** |
---|
| 75 | * Get what is left from the original text beyond the pattern |
---|
| 76 | */ |
---|
| 77 | text.substr (closemarker + 2, remaining); |
---|
| 78 | |
---|
| 79 | /** |
---|
| 80 | * Cut the original text up to the pattern start |
---|
| 81 | */ |
---|
| 82 | text.erase (openmarker); |
---|
| 83 | |
---|
| 84 | /** |
---|
| 85 | * Now display it |
---|
| 86 | */ |
---|
| 87 | display (text); |
---|
| 88 | |
---|
| 89 | /** |
---|
| 90 | * Convert the extracted command into words |
---|
| 91 | */ |
---|
| 92 | CmtSystem::split (command, " ", subargs); |
---|
| 93 | |
---|
| 94 | /** |
---|
| 95 | * Recursively expand it |
---|
| 96 | */ |
---|
| 97 | expand (subargs); |
---|
| 98 | |
---|
| 99 | /** |
---|
| 100 | * The next iteration will operate on the remaining text |
---|
| 101 | */ |
---|
| 102 | text = remaining; |
---|
| 103 | } |
---|
| 104 | |
---|
| 105 | if (CmtSystem::testenv ("CMTTESTMODEL")) |
---|
| 106 | { |
---|
| 107 | cerr << "text=[" << text << "]" << endl; |
---|
| 108 | } |
---|
| 109 | |
---|
| 110 | /** |
---|
| 111 | * Display what is left after extracting the last pattern |
---|
| 112 | */ |
---|
| 113 | display (text); |
---|
| 114 | cout << endl; |
---|
| 115 | } |
---|
| 116 | |
---|
| 117 | void CmtModel::strict_expand (const cmt_string& input_text) |
---|
| 118 | { |
---|
| 119 | CmtSystem::cmt_string_vector subargs; |
---|
| 120 | |
---|
| 121 | cmt_string text = input_text; |
---|
| 122 | cmt_string remaining; |
---|
| 123 | |
---|
| 124 | // The patterns for syntax parsing |
---|
| 125 | |
---|
| 126 | cmt_string open_element = "<cmts:"; |
---|
| 127 | cmt_string open_elements = "<cmtv:"; |
---|
| 128 | cmt_string close_element = "/>"; |
---|
| 129 | |
---|
| 130 | cmt_string skip_spaces = "[ \t]+"; |
---|
| 131 | cmt_string word = "[^ \t/>]*"; |
---|
| 132 | |
---|
| 133 | cmt_string attribute = "[^=/>]+=((\"([^\"]|\\\\\")*\")|(\'([^\']|\\\\\')*\'))"; |
---|
| 134 | |
---|
| 135 | //cout << "attribute=[" << attribute << "]" << endl; |
---|
| 136 | |
---|
| 137 | cmt_regexp exp_open_element (open_element); |
---|
| 138 | cmt_regexp exp_open_elements (open_elements); |
---|
| 139 | cmt_regexp exp_close_element (close_element); |
---|
| 140 | |
---|
| 141 | cmt_regexp exp_skip_spaces (skip_spaces); |
---|
| 142 | cmt_regexp exp_word (word); |
---|
| 143 | |
---|
| 144 | cmt_regexp exp_attribute (attribute); |
---|
| 145 | |
---|
| 146 | cmt_regexp::iterator it; |
---|
| 147 | cmt_regexp::iterator its; |
---|
| 148 | |
---|
| 149 | static int expand_level = 0; |
---|
| 150 | |
---|
| 151 | //cout << "Text=[" << text << "]" << endl; |
---|
| 152 | |
---|
| 153 | filter (text); |
---|
| 154 | |
---|
| 155 | //cout << "Text=[" << text << "]" << endl; |
---|
| 156 | |
---|
| 157 | int pos = 0; |
---|
| 158 | |
---|
| 159 | for (;;) |
---|
| 160 | { |
---|
| 161 | bool multiple = false; |
---|
| 162 | |
---|
| 163 | it = exp_open_element.begin (text, pos); |
---|
| 164 | its = exp_open_elements.begin (text, pos); |
---|
| 165 | |
---|
| 166 | if ((it == exp_open_element.end ()) && ((its == exp_open_elements.end ()))) |
---|
| 167 | { |
---|
| 168 | // no more opening element |
---|
| 169 | |
---|
| 170 | // Display the remaining text. The parsing is now completed. |
---|
| 171 | |
---|
| 172 | remaining = text.substr (pos); |
---|
| 173 | display (remaining); |
---|
| 174 | |
---|
| 175 | break; |
---|
| 176 | } |
---|
| 177 | |
---|
| 178 | if (it == exp_open_element.end ()) |
---|
| 179 | { |
---|
| 180 | // no single element -> we have a multiple one |
---|
| 181 | it = its; |
---|
| 182 | multiple = true; |
---|
| 183 | } |
---|
| 184 | else if (its == exp_open_elements.end ()) |
---|
| 185 | { |
---|
| 186 | // no multiple element -> we have a single one |
---|
| 187 | } |
---|
| 188 | else if (it < its) |
---|
| 189 | { |
---|
| 190 | // the single is before the multiple |
---|
| 191 | } |
---|
| 192 | else |
---|
| 193 | { |
---|
| 194 | // the multiple is before the single |
---|
| 195 | it = its; |
---|
| 196 | multiple = true; |
---|
| 197 | } |
---|
| 198 | |
---|
| 199 | remaining = text.substr (pos, it._pos - pos); |
---|
| 200 | display (remaining); |
---|
| 201 | |
---|
| 202 | //cout << "pos=" << pos << " open_element -> p=" << it._pos << " l=" << it._length << endl; |
---|
| 203 | pos = it._pos + it._length; |
---|
| 204 | |
---|
| 205 | it = exp_word.begin (text, pos); |
---|
| 206 | if (it._pos != pos) |
---|
| 207 | { |
---|
| 208 | cerr << "Syntax error: no element name" << endl; |
---|
| 209 | break; |
---|
| 210 | } |
---|
| 211 | |
---|
| 212 | //cout << "pos=" << pos << " word -> p=" << it._pos << " l=" << it._length << endl; |
---|
| 213 | pos += it._length; |
---|
| 214 | |
---|
| 215 | cmt_string element = it (text); |
---|
| 216 | //cout << "pos=" << pos << " element=" << element << endl; |
---|
| 217 | |
---|
| 218 | subargs.clear (); |
---|
| 219 | |
---|
| 220 | { |
---|
| 221 | cmt_string& s = subargs.add (); |
---|
| 222 | s = element; |
---|
| 223 | } |
---|
| 224 | |
---|
| 225 | for (;;) |
---|
| 226 | { |
---|
| 227 | it = exp_skip_spaces.begin (text, pos); |
---|
| 228 | if (it._pos == pos) |
---|
| 229 | { |
---|
| 230 | //cout << "pos=" << pos << " skip_spaces -> p=" << it._pos << " l=" << it._length << endl; |
---|
| 231 | pos += it._length; |
---|
| 232 | } |
---|
| 233 | |
---|
| 234 | //cout << "pos=" << pos << endl; |
---|
| 235 | |
---|
| 236 | it = exp_close_element.begin (text, pos); |
---|
| 237 | if (it._pos == pos) |
---|
| 238 | { |
---|
| 239 | //cout << "Recursing to [" << element << "]" << endl; |
---|
| 240 | |
---|
| 241 | int i; |
---|
| 242 | |
---|
| 243 | Variable::VariableVector variables; |
---|
| 244 | |
---|
| 245 | /** |
---|
| 246 | * We start by decoding all [variable=value] pairs from the arguments |
---|
| 247 | * A vector of Variables is filled from them. |
---|
| 248 | */ |
---|
| 249 | |
---|
| 250 | int multiple_count = 0; |
---|
| 251 | CmtSystem::cmt_string_vector values; |
---|
| 252 | cmt_string name; |
---|
| 253 | cmt_string value; |
---|
| 254 | |
---|
| 255 | for (i = 1; i < subargs.size (); i++) |
---|
| 256 | { |
---|
| 257 | const cmt_string& arg = subargs[i]; |
---|
| 258 | |
---|
| 259 | int p = arg.find ("="); |
---|
| 260 | |
---|
| 261 | arg.substr (0, p, name); |
---|
| 262 | arg.substr (p + 1, value); |
---|
| 263 | if (value[0] == '"') |
---|
| 264 | { |
---|
| 265 | value.erase (0, 1); |
---|
| 266 | if (value[value.size()-1] == '"') value.erase (value.size ()-1); |
---|
| 267 | } |
---|
| 268 | else if (value[0] == '\'') |
---|
| 269 | { |
---|
| 270 | value.erase (0, 1); |
---|
| 271 | if (value[value.size()-1] == '\'') value.erase (value.size ()-1); |
---|
| 272 | } |
---|
| 273 | |
---|
| 274 | //cout << "+ name = [" << value << "]" << endl; |
---|
| 275 | |
---|
| 276 | if (multiple) |
---|
| 277 | { |
---|
| 278 | values.clear (); |
---|
| 279 | CmtSystem::split (value, " \t", values); |
---|
| 280 | int n = values.size (); |
---|
| 281 | if (n > multiple_count) multiple_count = n; |
---|
| 282 | } |
---|
| 283 | |
---|
| 284 | Variable* v = Variable::find (variables, name); |
---|
| 285 | if (v == 0) |
---|
| 286 | { |
---|
| 287 | v = &(variables.add ()); |
---|
| 288 | v->set (name); |
---|
| 289 | } |
---|
| 290 | |
---|
| 291 | (*v) = value; |
---|
| 292 | } |
---|
| 293 | |
---|
| 294 | cmt_string sub_text; |
---|
| 295 | |
---|
| 296 | FragmentHandle fragment (element); |
---|
| 297 | |
---|
| 298 | if (multiple) |
---|
| 299 | { |
---|
| 300 | //cout << "+ count = " << multiple_count << endl; |
---|
| 301 | for (int n = 0; n < multiple_count; n++) |
---|
| 302 | { |
---|
| 303 | for (i = 1; i < subargs.size (); i++) |
---|
| 304 | { |
---|
| 305 | const cmt_string& arg = subargs[i]; |
---|
| 306 | |
---|
| 307 | int p = arg.find ("="); |
---|
| 308 | |
---|
| 309 | arg.substr (p + 1, value); |
---|
| 310 | if (value[0] == '"') |
---|
| 311 | { |
---|
| 312 | value.erase (0, 1); |
---|
| 313 | if (value[value.size()-1] == '"') value.erase (value.size ()-1); |
---|
| 314 | } |
---|
| 315 | else if (value[0] == '\'') |
---|
| 316 | { |
---|
| 317 | value.erase (0, 1); |
---|
| 318 | if (value[value.size()-1] == '\'') value.erase (value.size ()-1); |
---|
| 319 | } |
---|
| 320 | |
---|
| 321 | //cout << "+ values = [" << value << "]" << endl; |
---|
| 322 | |
---|
| 323 | values.clear (); |
---|
| 324 | CmtSystem::split (value, " \t", values); |
---|
| 325 | if (n < values.size ()) |
---|
| 326 | { |
---|
| 327 | value = values[n]; |
---|
| 328 | } |
---|
| 329 | else |
---|
| 330 | { |
---|
| 331 | value = ""; |
---|
| 332 | } |
---|
| 333 | |
---|
| 334 | Variable& v = variables[i-1]; |
---|
| 335 | |
---|
| 336 | //cout << "+ value = [" << value << "]" << endl; |
---|
| 337 | |
---|
| 338 | v = value; |
---|
| 339 | } |
---|
| 340 | |
---|
| 341 | fragment.copy (sub_text, variables, 0); |
---|
| 342 | |
---|
| 343 | expand_level++; |
---|
| 344 | strict_expand (sub_text); |
---|
| 345 | expand_level--; |
---|
| 346 | } |
---|
| 347 | } |
---|
| 348 | else |
---|
| 349 | { |
---|
| 350 | fragment.copy (sub_text, variables, 0); |
---|
| 351 | |
---|
| 352 | expand_level++; |
---|
| 353 | strict_expand (sub_text); |
---|
| 354 | expand_level--; |
---|
| 355 | } |
---|
| 356 | |
---|
| 357 | pos += it._length; |
---|
| 358 | break; |
---|
| 359 | } |
---|
| 360 | |
---|
| 361 | //cout << "Try attribute at pos=" << pos << " [" << text.substr (pos) << "]" << endl; |
---|
| 362 | |
---|
| 363 | it = exp_attribute.begin (text, pos); |
---|
| 364 | if (it._pos != pos) |
---|
| 365 | { |
---|
| 366 | cerr << "Syntax error : no attribute before closing the element [" << text.substr (pos) << "]" << endl; |
---|
| 367 | break; |
---|
| 368 | } |
---|
| 369 | |
---|
| 370 | //cout << "pos=" << pos << " attribute -> p=" << it._pos << " l=" << it._length << endl; |
---|
| 371 | //cout << "pos=" << pos << " attr=[" << it (text) << "]" << endl; |
---|
| 372 | |
---|
| 373 | { |
---|
| 374 | cmt_string& s = subargs.add (); |
---|
| 375 | s = it (text); |
---|
| 376 | } |
---|
| 377 | |
---|
| 378 | pos += it._length; |
---|
| 379 | } |
---|
| 380 | } |
---|
| 381 | |
---|
| 382 | if (expand_level == 0) cout << endl; |
---|
| 383 | } |
---|
| 384 | |
---|
| 385 | void CmtModel::test_regexp (const cmt_string& pattern, const cmt_string& input_text) |
---|
| 386 | { |
---|
| 387 | cout << "Testing pattern [" << pattern << "] against [" << input_text << "]" << endl; |
---|
| 388 | |
---|
| 389 | cmt_regexp exp (pattern); |
---|
| 390 | |
---|
| 391 | cmt_regexp::iterator it; |
---|
| 392 | |
---|
| 393 | for (int pos = 0; pos < 10; pos++) |
---|
| 394 | { |
---|
| 395 | it = exp.begin (input_text, pos); |
---|
| 396 | |
---|
| 397 | if (it == exp.end ()) |
---|
| 398 | { |
---|
| 399 | cout << "No match" << endl; |
---|
| 400 | } |
---|
| 401 | else |
---|
| 402 | { |
---|
| 403 | cout << "p=" << it._pos << " l=" << it._length << " -> [" << it (input_text) << "]" << endl; |
---|
| 404 | } |
---|
| 405 | } |
---|
| 406 | } |
---|
| 407 | |
---|
| 408 | |
---|
| 409 | /** |
---|
| 410 | * Expands a model file |
---|
| 411 | * Arguments are : |
---|
| 412 | * model-name : name of a model file |
---|
| 413 | * var=value : variable value to be expanded |
---|
| 414 | * (will take the form ${var} ) |
---|
| 415 | */ |
---|
| 416 | void CmtModel::expand (const CmtSystem::cmt_string_vector& arguments) |
---|
| 417 | { |
---|
| 418 | int i; |
---|
| 419 | |
---|
| 420 | Variable::VariableVector variables; |
---|
| 421 | |
---|
| 422 | /** |
---|
| 423 | * We start by decoding all [variable=value] pairs from the arguments |
---|
| 424 | * A vector of Variables is filled from them. |
---|
| 425 | */ |
---|
| 426 | for (i = 0; i < arguments.size (); i++) |
---|
| 427 | { |
---|
| 428 | const cmt_string& arg = arguments[i]; |
---|
| 429 | |
---|
| 430 | if (arg.find ("=") != cmt_string::npos) |
---|
| 431 | { |
---|
| 432 | cmt_string name; |
---|
| 433 | cmt_string value; |
---|
| 434 | int pos = arg.find ("="); |
---|
| 435 | |
---|
| 436 | arg.substr (0, pos, name); |
---|
| 437 | arg.substr (pos + 1, value); |
---|
| 438 | |
---|
| 439 | Variable* v = Variable::find (variables, name); |
---|
| 440 | if (v == 0) |
---|
| 441 | { |
---|
| 442 | v = &(variables.add ()); |
---|
| 443 | v->set (name); |
---|
| 444 | } |
---|
| 445 | |
---|
| 446 | (*v) = value; |
---|
| 447 | } |
---|
| 448 | } |
---|
| 449 | |
---|
| 450 | /** |
---|
| 451 | * Then model names are extracted. |
---|
| 452 | * Each model may contain a set of <...> patterns. |
---|
| 453 | * The expected syntax for each of them is : |
---|
| 454 | * <model-name variable-name=value ...> |
---|
| 455 | * |
---|
| 456 | * Therefore the current expand function is recursively restarted. |
---|
| 457 | * |
---|
| 458 | * Text around patterns is displayed after replacements of all |
---|
| 459 | * variable values detected by ${variable-name} patterns |
---|
| 460 | */ |
---|
| 461 | cmt_string text; |
---|
| 462 | |
---|
| 463 | for (i = 0; i < arguments.size (); i++) |
---|
| 464 | { |
---|
| 465 | const cmt_string& arg = arguments[i]; |
---|
| 466 | |
---|
| 467 | if (arg.find ("=") == cmt_string::npos) |
---|
| 468 | { |
---|
| 469 | FragmentHandle fragment (arg); |
---|
| 470 | fragment.copy (text, variables, 0); |
---|
| 471 | |
---|
| 472 | filter (text); |
---|
| 473 | |
---|
| 474 | int openmarker; |
---|
| 475 | int closemarker; |
---|
| 476 | |
---|
| 477 | CmtSystem::cmt_string_vector subargs; |
---|
| 478 | |
---|
| 479 | cmt_string remaining; |
---|
| 480 | |
---|
| 481 | for (;;) |
---|
| 482 | { |
---|
| 483 | /** |
---|
| 484 | * Look for the next <...> pattern |
---|
| 485 | * If not found, then simply dump the text as it is. |
---|
| 486 | */ |
---|
| 487 | |
---|
| 488 | openmarker = text.find ("<"); |
---|
| 489 | if (openmarker == cmt_string::npos) break; |
---|
| 490 | |
---|
| 491 | closemarker = text.find (openmarker, "/>"); |
---|
| 492 | if (closemarker == cmt_string::npos) |
---|
| 493 | { |
---|
| 494 | /** |
---|
| 495 | * The opening < was in the text |
---|
| 496 | */ |
---|
| 497 | |
---|
| 498 | /** |
---|
| 499 | * Get what is left from the original text beyond the pattern |
---|
| 500 | */ |
---|
| 501 | text.substr (openmarker + 1, remaining); |
---|
| 502 | text.erase (openmarker + 1); |
---|
| 503 | |
---|
| 504 | /** |
---|
| 505 | * Now display it |
---|
| 506 | */ |
---|
| 507 | display (text); |
---|
| 508 | } |
---|
| 509 | else |
---|
| 510 | { |
---|
| 511 | /** |
---|
| 512 | * Extract the command from the pattern |
---|
| 513 | */ |
---|
| 514 | cmt_string command; |
---|
| 515 | text.substr (openmarker + 1, closemarker - openmarker - 1, command); |
---|
| 516 | |
---|
| 517 | /** |
---|
| 518 | * Get what is left from the original text beyond the pattern |
---|
| 519 | */ |
---|
| 520 | text.substr (closemarker + 2, remaining); |
---|
| 521 | text.erase (openmarker); |
---|
| 522 | |
---|
| 523 | /** |
---|
| 524 | * Now display it |
---|
| 525 | */ |
---|
| 526 | display (text); |
---|
| 527 | |
---|
| 528 | /** |
---|
| 529 | * Convert the extracted command into words |
---|
| 530 | */ |
---|
| 531 | CmtSystem::split (command, " ", subargs); |
---|
| 532 | |
---|
| 533 | /** |
---|
| 534 | * Recursively expand it |
---|
| 535 | */ |
---|
| 536 | expand (subargs); |
---|
| 537 | } |
---|
| 538 | |
---|
| 539 | /** |
---|
| 540 | * The next iteration will operate on the remaining text |
---|
| 541 | */ |
---|
| 542 | text = remaining; |
---|
| 543 | } |
---|
| 544 | |
---|
| 545 | /** |
---|
| 546 | * Display what is left after extracting the last pattern |
---|
| 547 | */ |
---|
| 548 | display (text); |
---|
| 549 | } |
---|
| 550 | } |
---|
| 551 | } |
---|
| 552 | |
---|