[1] | 1 | |
---|
| 2 | #ifdef WIN32 |
---|
| 3 | #include <direct.h> |
---|
| 4 | #define popen _popen |
---|
| 5 | #define pclose _pclose |
---|
| 6 | #endif |
---|
| 7 | |
---|
| 8 | #include "cmt_awk.h" |
---|
| 9 | #include "cmt_system.h" |
---|
| 10 | |
---|
| 11 | class Parser |
---|
| 12 | { |
---|
| 13 | public: |
---|
| 14 | Parser (Awk* awk, const cmt_string pattern, const cmt_regexp* expression) : |
---|
| 15 | m_pattern (pattern), m_expression (expression), m_awk(awk) |
---|
| 16 | { |
---|
| 17 | } |
---|
| 18 | |
---|
| 19 | /** |
---|
| 20 | * this first level parsing function extracts individual lines |
---|
| 21 | * from the text, taking care of both Unix and Windows EOL styles. |
---|
| 22 | * |
---|
| 23 | * Then the second level parsing function parse_line is called. |
---|
| 24 | */ |
---|
| 25 | Awk::condition parse (const cmt_string& text) |
---|
| 26 | { |
---|
| 27 | Awk::condition result = Awk::ok; |
---|
| 28 | |
---|
| 29 | cmt_string line; |
---|
| 30 | int pos; |
---|
| 31 | int max_pos; |
---|
| 32 | |
---|
| 33 | pos = 0; |
---|
| 34 | max_pos = text.size (); |
---|
| 35 | |
---|
| 36 | m_accumulator.erase (0); |
---|
| 37 | |
---|
| 38 | for (pos = 0; pos < max_pos;) |
---|
| 39 | { |
---|
| 40 | int eol = text.find (pos, '\n'); |
---|
| 41 | |
---|
| 42 | if (eol == cmt_string::npos) |
---|
| 43 | { |
---|
| 44 | // Last line, since there is no eol at all |
---|
| 45 | text.substr (pos, line); |
---|
| 46 | pos = max_pos; |
---|
| 47 | } |
---|
| 48 | else |
---|
| 49 | { |
---|
| 50 | int length = 1; |
---|
| 51 | |
---|
| 52 | int cr = text.find (pos, "\r\n"); |
---|
| 53 | |
---|
| 54 | if (cr == (eol-1)) |
---|
| 55 | { |
---|
| 56 | eol = cr; |
---|
| 57 | length = 2; |
---|
| 58 | } |
---|
| 59 | |
---|
| 60 | if (eol == pos) |
---|
| 61 | { |
---|
| 62 | // this is an empty line |
---|
| 63 | line = ""; |
---|
| 64 | pos += length; |
---|
| 65 | } |
---|
| 66 | else |
---|
| 67 | { |
---|
| 68 | // The eol was found beyond the current position |
---|
| 69 | // (ie. this is a non empty line) |
---|
| 70 | text.substr (pos, eol - pos, line); |
---|
| 71 | pos = eol + length; |
---|
| 72 | } |
---|
| 73 | } |
---|
| 74 | |
---|
| 75 | if (m_awk != 0) m_awk->inc_line_number (); |
---|
| 76 | |
---|
| 77 | //cout << "parse> line=[" << line << "]" << endl; |
---|
| 78 | |
---|
| 79 | result = parse_line (line); |
---|
| 80 | if (result != Awk::ok) break; |
---|
| 81 | } |
---|
| 82 | |
---|
| 83 | return (result); |
---|
| 84 | } |
---|
| 85 | |
---|
| 86 | /** |
---|
| 87 | * This second level parsing function accumulates individual lines |
---|
| 88 | * with real trailing back slashes. |
---|
| 89 | * Eventually the possible text pattern or regular expression is |
---|
| 90 | * checked and the Awk::filter function is called in case of |
---|
| 91 | * succesful match onto the accumulated line. |
---|
| 92 | */ |
---|
| 93 | Awk::condition parse_line (const cmt_string& line) |
---|
| 94 | { |
---|
| 95 | Awk::condition result = Awk::ok; |
---|
| 96 | int length; |
---|
| 97 | cmt_string temp_line = line; |
---|
| 98 | |
---|
| 99 | // |
---|
| 100 | // We scan the line for handling backslashes. |
---|
| 101 | // |
---|
| 102 | // Really terminating backslashes (ie those only followed by spaces/tabs |
---|
| 103 | // mean continued line |
---|
| 104 | // |
---|
| 105 | // |
---|
| 106 | |
---|
| 107 | bool finished = true; |
---|
| 108 | |
---|
| 109 | length = temp_line.size (); |
---|
| 110 | |
---|
| 111 | if (length == 0) |
---|
| 112 | { |
---|
| 113 | // An empty line following a backslash terminates the continuation. |
---|
| 114 | finished = true; |
---|
| 115 | } |
---|
| 116 | else |
---|
| 117 | { |
---|
| 118 | int back_slash = temp_line.find_last_of ('\\'); |
---|
| 119 | |
---|
| 120 | if (back_slash != cmt_string::npos) |
---|
| 121 | { |
---|
| 122 | // |
---|
| 123 | // This is the last backslash |
---|
| 124 | // check if there are only space chars after it |
---|
| 125 | // |
---|
| 126 | |
---|
| 127 | bool at_end = true; |
---|
| 128 | |
---|
| 129 | for (int i = (back_slash + 1); i < length; i++) |
---|
| 130 | { |
---|
| 131 | char c = temp_line[i]; |
---|
| 132 | if ((c != ' ') && (c != '\t')) |
---|
| 133 | { |
---|
| 134 | at_end = false; |
---|
| 135 | break; |
---|
| 136 | } |
---|
| 137 | } |
---|
| 138 | |
---|
| 139 | if (at_end) |
---|
| 140 | { |
---|
| 141 | temp_line.erase (back_slash); |
---|
| 142 | finished = false; |
---|
| 143 | } |
---|
| 144 | else |
---|
| 145 | { |
---|
| 146 | // This was not a trailing backslash. |
---|
| 147 | finished = true; |
---|
| 148 | } |
---|
| 149 | } |
---|
| 150 | |
---|
| 151 | m_accumulator += temp_line; |
---|
| 152 | } |
---|
| 153 | |
---|
| 154 | //cout << "parse_line1> accumulator=[" << m_accumulator << "]" << endl; |
---|
| 155 | //cout << "parse_line1> finished=[" << finished << "]" << endl; |
---|
| 156 | |
---|
| 157 | if (!finished) |
---|
| 158 | { |
---|
| 159 | // We still need to accumulate forthcoming lines |
---|
| 160 | // before parsing the resulting text. |
---|
| 161 | return (Awk::ok); |
---|
| 162 | } |
---|
| 163 | |
---|
| 164 | // now filter the complete accumulated line (if non empty) |
---|
| 165 | |
---|
| 166 | if (m_accumulator != "") |
---|
| 167 | { |
---|
| 168 | bool ok = false; |
---|
| 169 | |
---|
| 170 | if (m_expression != 0) |
---|
| 171 | { |
---|
| 172 | if (m_expression->match (m_accumulator)) |
---|
| 173 | { |
---|
| 174 | ok = true; |
---|
| 175 | } |
---|
| 176 | } |
---|
| 177 | else |
---|
| 178 | { |
---|
| 179 | if ((m_pattern == "") || |
---|
| 180 | (m_accumulator.find (m_pattern) != cmt_string::npos)) |
---|
| 181 | { |
---|
| 182 | ok = true; |
---|
| 183 | } |
---|
| 184 | } |
---|
| 185 | |
---|
| 186 | if (ok && (m_awk != 0)) |
---|
| 187 | { |
---|
| 188 | //cout << "parse_line> accumulator=[" << m_accumulator << "]" << endl; |
---|
| 189 | |
---|
| 190 | m_awk->filter (m_accumulator); |
---|
| 191 | result = m_awk->get_last_condition (); |
---|
| 192 | } |
---|
| 193 | |
---|
| 194 | m_accumulator.erase (0); |
---|
| 195 | } |
---|
| 196 | |
---|
| 197 | return (result); |
---|
| 198 | } |
---|
| 199 | |
---|
| 200 | private: |
---|
| 201 | |
---|
| 202 | cmt_string m_accumulator; |
---|
| 203 | cmt_string m_pattern; |
---|
| 204 | const cmt_regexp* m_expression; |
---|
| 205 | Awk* m_awk; |
---|
| 206 | }; |
---|
| 207 | |
---|
| 208 | //------------------------------------------------ |
---|
| 209 | Awk::Awk () |
---|
| 210 | { |
---|
| 211 | m_condition = ok; |
---|
| 212 | } |
---|
| 213 | |
---|
| 214 | //------------------------------------------------ |
---|
| 215 | Awk::~Awk () |
---|
| 216 | { |
---|
| 217 | } |
---|
| 218 | |
---|
| 219 | //------------------------------------------------ |
---|
| 220 | Awk::condition Awk::run (const cmt_string& text, |
---|
| 221 | const cmt_string& pattern) |
---|
| 222 | { |
---|
| 223 | m_line_number = 0; |
---|
| 224 | m_condition = ok; |
---|
| 225 | |
---|
| 226 | begin (); |
---|
| 227 | if (m_condition != ok) return (m_condition); |
---|
| 228 | |
---|
| 229 | if (CmtSystem::testenv ("CMTTESTAWK")) |
---|
| 230 | { |
---|
| 231 | Parser p (this, pattern, 0); |
---|
| 232 | |
---|
| 233 | m_condition = p.parse (text); |
---|
| 234 | if (m_condition != ok) return (m_condition); |
---|
| 235 | } |
---|
| 236 | else |
---|
| 237 | { |
---|
| 238 | cmt_string line; |
---|
| 239 | int pos = 0; |
---|
| 240 | int max_pos; |
---|
| 241 | |
---|
| 242 | max_pos = text.size (); |
---|
| 243 | |
---|
| 244 | for (pos = 0; pos < max_pos;) |
---|
| 245 | { |
---|
| 246 | int cr = text.find (pos, "\r\n"); |
---|
| 247 | int nl = text.find (pos, '\n'); |
---|
| 248 | |
---|
| 249 | // Get the first end-of-line (either lf or cr-lf) |
---|
| 250 | |
---|
| 251 | //-------------------- |
---|
| 252 | // |
---|
| 253 | // cr 1 0 |
---|
| 254 | // nl |
---|
| 255 | // |
---|
| 256 | // 1 a b |
---|
| 257 | // |
---|
| 258 | // 0 c d |
---|
| 259 | // |
---|
| 260 | //-------------------- |
---|
| 261 | |
---|
| 262 | int first = nl; |
---|
| 263 | |
---|
| 264 | if (cr != cmt_string::npos) |
---|
| 265 | { |
---|
| 266 | // cases a or c |
---|
| 267 | |
---|
| 268 | if (nl == cmt_string::npos) |
---|
| 269 | { |
---|
| 270 | // case a |
---|
| 271 | first = cr; |
---|
| 272 | } |
---|
| 273 | else |
---|
| 274 | { |
---|
| 275 | // case c |
---|
| 276 | first = (nl < cr) ? nl : cr; |
---|
| 277 | } |
---|
| 278 | } |
---|
| 279 | |
---|
| 280 | if (first == cmt_string::npos) |
---|
| 281 | { |
---|
| 282 | // This is likely the last line since there is no end-of-line |
---|
| 283 | text.substr (pos, line); |
---|
| 284 | pos = max_pos; |
---|
| 285 | } |
---|
| 286 | else if (first > pos) |
---|
| 287 | { |
---|
| 288 | // The eol was found beyond the current position |
---|
| 289 | // (ie. this is a non empty line) |
---|
| 290 | text.substr (pos, first - pos, line); |
---|
| 291 | pos = first + 1; |
---|
| 292 | } |
---|
| 293 | else |
---|
| 294 | { |
---|
| 295 | // an empty line |
---|
| 296 | line = ""; |
---|
| 297 | pos++; |
---|
| 298 | } |
---|
| 299 | |
---|
| 300 | m_line_number++; |
---|
| 301 | |
---|
| 302 | if (line != "") |
---|
| 303 | { |
---|
| 304 | if ((pattern == "") || |
---|
| 305 | (line.find (pattern) != cmt_string::npos)) |
---|
| 306 | { |
---|
| 307 | filter (line); |
---|
| 308 | if (m_condition != ok) return (m_condition); |
---|
| 309 | } |
---|
| 310 | } |
---|
| 311 | } |
---|
| 312 | } |
---|
| 313 | |
---|
| 314 | end (); |
---|
| 315 | |
---|
| 316 | return (m_condition); |
---|
| 317 | } |
---|
| 318 | |
---|
| 319 | //------------------------------------------------ |
---|
| 320 | Awk::condition Awk::run (const cmt_string& text, |
---|
| 321 | const cmt_regexp& expression) |
---|
| 322 | { |
---|
| 323 | m_line_number = 0; |
---|
| 324 | m_condition = ok; |
---|
| 325 | |
---|
| 326 | begin (); |
---|
| 327 | if (m_condition != ok) return (m_condition); |
---|
| 328 | |
---|
| 329 | Parser p (this, "", &expression); |
---|
| 330 | |
---|
| 331 | m_condition = p.parse (text); |
---|
| 332 | if (m_condition != ok) return (m_condition); |
---|
| 333 | |
---|
| 334 | /* |
---|
| 335 | if (CmtSystem::testenv ("CMTTESTAWK")) |
---|
| 336 | { |
---|
| 337 | } |
---|
| 338 | else |
---|
| 339 | { |
---|
| 340 | cmt_string line; |
---|
| 341 | int pos = 0; |
---|
| 342 | int max_pos; |
---|
| 343 | |
---|
| 344 | max_pos = text.size (); |
---|
| 345 | |
---|
| 346 | for (pos = 0; pos < max_pos;) |
---|
| 347 | { |
---|
| 348 | int cr = text.find (pos, "\r\n"); |
---|
| 349 | int nl = text.find (pos, '\n'); |
---|
| 350 | |
---|
| 351 | // Get the first end-of-line (either lf or cr-lf) |
---|
| 352 | |
---|
| 353 | int first = nl; |
---|
| 354 | |
---|
| 355 | if (cr != cmt_string::npos) |
---|
| 356 | { |
---|
| 357 | if (nl == cmt_string::npos) |
---|
| 358 | { |
---|
| 359 | first = cr; |
---|
| 360 | } |
---|
| 361 | else |
---|
| 362 | { |
---|
| 363 | first = (nl < cr) ? nl : cr; |
---|
| 364 | } |
---|
| 365 | } |
---|
| 366 | |
---|
| 367 | if (first == cmt_string::npos) |
---|
| 368 | { |
---|
| 369 | // This is likely the last line since there is no end-of-line |
---|
| 370 | text.substr (pos, line); |
---|
| 371 | pos = max_pos; |
---|
| 372 | } |
---|
| 373 | else if (first > pos) |
---|
| 374 | { |
---|
| 375 | // The eol was found beyond the current position |
---|
| 376 | // (ie. this is a non empty line) |
---|
| 377 | text.substr (pos, first - pos, line); |
---|
| 378 | pos = first + 1; |
---|
| 379 | } |
---|
| 380 | else |
---|
| 381 | { |
---|
| 382 | // an empty line |
---|
| 383 | line = ""; |
---|
| 384 | pos++; |
---|
| 385 | } |
---|
| 386 | |
---|
| 387 | m_line_number++; |
---|
| 388 | |
---|
| 389 | if (line != "") |
---|
| 390 | { |
---|
| 391 | if (expression.match (line)) |
---|
| 392 | { |
---|
| 393 | filter (line); |
---|
| 394 | if (m_condition != ok) return (m_condition); |
---|
| 395 | } |
---|
| 396 | } |
---|
| 397 | } |
---|
| 398 | } |
---|
| 399 | */ |
---|
| 400 | |
---|
| 401 | end (); |
---|
| 402 | |
---|
| 403 | return (m_condition); |
---|
| 404 | } |
---|
| 405 | |
---|
| 406 | //------------------------------------------------ |
---|
| 407 | void Awk::stop () |
---|
| 408 | { |
---|
| 409 | m_condition = stopped; |
---|
| 410 | } |
---|
| 411 | |
---|
| 412 | //------------------------------------------------ |
---|
| 413 | void Awk::abort () |
---|
| 414 | { |
---|
| 415 | m_condition = failed; |
---|
| 416 | } |
---|
| 417 | |
---|
| 418 | //------------------------------------------------ |
---|
| 419 | void Awk::allow_continuation () |
---|
| 420 | { |
---|
| 421 | m_continuation_allowed = true; |
---|
| 422 | } |
---|
| 423 | |
---|
| 424 | //------------------------------------------------ |
---|
| 425 | Awk::condition Awk::get_last_condition () const |
---|
| 426 | { |
---|
| 427 | return (m_condition); |
---|
| 428 | } |
---|
| 429 | |
---|
| 430 | //------------------------------------------------ |
---|
| 431 | void Awk::begin () |
---|
| 432 | { |
---|
| 433 | } |
---|
| 434 | |
---|
| 435 | //------------------------------------------------ |
---|
| 436 | void Awk::filter (const cmt_string& /*line*/) |
---|
| 437 | { |
---|
| 438 | //cout << "awk> " << line << endl; |
---|
| 439 | } |
---|
| 440 | |
---|
| 441 | //------------------------------------------------ |
---|
| 442 | void Awk::end () |
---|
| 443 | { |
---|
| 444 | } |
---|
| 445 | |
---|
| 446 | //------------------------------------------------ |
---|
| 447 | void Awk::inc_line_number () |
---|
| 448 | { |
---|
| 449 | m_line_number++; |
---|
| 450 | } |
---|
| 451 | |
---|
| 452 | //------------------------------------------------ |
---|
| 453 | Awk::condition FAwk::run (const cmt_string& file_name, |
---|
| 454 | const cmt_string& pattern) |
---|
| 455 | { |
---|
| 456 | if (!CmtSystem::test_file (file_name)) return (failed); |
---|
| 457 | |
---|
| 458 | CmtSystem::basename (file_name, m_file_name); |
---|
| 459 | CmtSystem::dirname (file_name, m_dir_name); |
---|
| 460 | |
---|
| 461 | cmt_string text; |
---|
| 462 | |
---|
| 463 | text.read (file_name); |
---|
| 464 | |
---|
| 465 | return (Awk::run (text, pattern)); |
---|
| 466 | } |
---|
| 467 | |
---|
| 468 | //------------------------------------------------ |
---|
| 469 | Awk::condition FAwk::run (const cmt_string& file_name, |
---|
| 470 | const cmt_regexp& expression) |
---|
| 471 | { |
---|
| 472 | if (!CmtSystem::test_file (file_name)) return (failed); |
---|
| 473 | |
---|
| 474 | CmtSystem::basename (file_name, m_file_name); |
---|
| 475 | CmtSystem::dirname (file_name, m_dir_name); |
---|
| 476 | |
---|
| 477 | cmt_string text; |
---|
| 478 | |
---|
| 479 | text.read (file_name); |
---|
| 480 | |
---|
| 481 | return (Awk::run (text, expression)); |
---|
| 482 | } |
---|
| 483 | |
---|
| 484 | //------------------------------------------------ |
---|
| 485 | Awk::condition PAwk::run (const cmt_string& command, |
---|
| 486 | const cmt_string& pattern) |
---|
| 487 | { |
---|
| 488 | cmt_string line; |
---|
| 489 | |
---|
| 490 | m_line_number = 0; |
---|
| 491 | m_condition = ok; |
---|
| 492 | |
---|
| 493 | begin (); |
---|
| 494 | if (m_condition != ok) return (m_condition); |
---|
| 495 | |
---|
| 496 | FILE* f = popen (command.c_str (), "r"); |
---|
| 497 | |
---|
| 498 | if (f == 0) return (failed); |
---|
| 499 | |
---|
| 500 | char buffer[8192]; |
---|
| 501 | char* ptr; |
---|
| 502 | |
---|
| 503 | while ((ptr = fgets (buffer, sizeof (buffer), f)) != NULL) |
---|
| 504 | { |
---|
| 505 | line = ptr; |
---|
| 506 | |
---|
| 507 | if (line.find ("\n") == cmt_string::npos) |
---|
| 508 | { |
---|
| 509 | cerr << "#CMT> Warning : Line too long and truncated in PAwk::run for command " << command << endl; |
---|
| 510 | } |
---|
| 511 | |
---|
| 512 | line.replace ("\n", ""); |
---|
| 513 | |
---|
| 514 | m_line_number++; |
---|
| 515 | |
---|
| 516 | if (line != "") |
---|
| 517 | { |
---|
| 518 | if ((pattern == "") || |
---|
| 519 | (line.find (pattern) != cmt_string::npos)) |
---|
| 520 | { |
---|
| 521 | filter (line); |
---|
| 522 | if (m_condition != ok) return (m_condition); |
---|
| 523 | } |
---|
| 524 | } |
---|
| 525 | } |
---|
| 526 | |
---|
| 527 | pclose (f); |
---|
| 528 | |
---|
| 529 | end (); |
---|
| 530 | |
---|
| 531 | return (m_condition); |
---|
| 532 | } |
---|
| 533 | |
---|
| 534 | //------------------------------------------------ |
---|
| 535 | Awk::condition PAwk::run (const cmt_string& command, |
---|
| 536 | const cmt_regexp& expression) |
---|
| 537 | { |
---|
| 538 | cmt_string line; |
---|
| 539 | |
---|
| 540 | m_line_number = 0; |
---|
| 541 | m_condition = ok; |
---|
| 542 | |
---|
| 543 | begin (); |
---|
| 544 | if (m_condition != ok) return (m_condition); |
---|
| 545 | |
---|
| 546 | FILE* f = popen (command.c_str (), "r"); |
---|
| 547 | |
---|
| 548 | if (f == 0) return (failed); |
---|
| 549 | |
---|
| 550 | char buffer[256]; |
---|
| 551 | char* ptr; |
---|
| 552 | |
---|
| 553 | while ((ptr = fgets (buffer, sizeof (buffer), f)) != NULL) |
---|
| 554 | { |
---|
| 555 | line = ptr; |
---|
| 556 | |
---|
| 557 | line.replace ("\n", ""); |
---|
| 558 | |
---|
| 559 | m_line_number++; |
---|
| 560 | |
---|
| 561 | if (line != "") |
---|
| 562 | { |
---|
| 563 | if (expression.match (line)) |
---|
| 564 | { |
---|
| 565 | filter (line); |
---|
| 566 | if (m_condition != ok) return (m_condition); |
---|
| 567 | } |
---|
| 568 | } |
---|
| 569 | } |
---|
| 570 | |
---|
| 571 | pclose (f); |
---|
| 572 | |
---|
| 573 | end (); |
---|
| 574 | |
---|
| 575 | return (m_condition); |
---|
| 576 | } |
---|
| 577 | |
---|