Mercurial Repositories > eric / file comparison
comparison: WebBrowser/AdBlock/AdBlockRule.py
WebBrowser/AdBlock/AdBlockRule.py
- branch
- QtWebEngine
- changeset 4858
- 19dff9c9cf26
- parent 4631
- 5c1a96925da4
- child 4860
- 0a44aff88bfa
equal
deleted
inserted
replaced
| 4857:8dba5fb92f12 | 4858:19dff9c9cf26 |
|---|---|
| 1 # -*- coding: utf-8 -*- | |
| 2 | |
| 3 # Copyright (c) 2009 - 2016 Detlev Offenbach <detlev@die-offenbachs.de> | |
| 4 # | |
| 5 | |
| 6 """ | |
| 7 Module implementing the AdBlock rule class. | |
| 8 """ | |
| 9 | |
| 10 from __future__ import unicode_literals | |
| 11 | |
| 12 import re | |
| 13 | |
| 14 from PyQt5.QtCore import Qt, QRegExp, QUrl | |
| 15 from PyQt5.QtNetwork import QNetworkRequest | |
| 16 from PyQt5.QtWebEngineCore import QWebEngineUrlRequestInfo | |
| 17 | |
| 18 | |
| 19 def toSecondLevelDomain(url): | |
| 20 """ | |
| 21 Module function to get a second level domain from the given URL. | |
| 22 | |
| 23 @param url URL to extract domain from (QUrl) | |
| 24 @return name of second level domain (string) | |
| 25 """ | |
| 26 topLevelDomain = url.topLevelDomain() | |
| 27 urlHost = url.host() | |
| 28 | |
| 29 if not topLevelDomain or not urlHost: | |
| 30 return "" | |
| 31 | |
| 32 domain = urlHost[:len(urlHost) - len(topLevelDomain)] | |
| 33 if domain.count(".") == 0: | |
| 34 return urlHost | |
| 35 | |
| 36 while domain.count(".") != 0: | |
| 37 domain = domain[domain.find(".") + 1:] | |
| 38 | |
| 39 return domain + topLevelDomain | |
| 40 | |
| 41 | |
| 42 class AdBlockRule(object): | |
| 43 """ | |
| 44 Class implementing the AdBlock rule. | |
| 45 """ | |
| 46 def __init__(self, filter="", subscription=None): | |
| 47 """ | |
| 48 Constructor | |
| 49 | |
| 50 @param filter filter string of the rule (string) | |
| 51 @param subscription reference to the subscription object | |
| 52 (AdBlockSubscription) | |
| 53 """ | |
| 54 self.__subscription = subscription | |
| 55 | |
| 56 self.__regExp = QRegExp() | |
| 57 self.__options = [] | |
| 58 self.__blockedDomains = [] | |
| 59 self.__allowedDomains = [] | |
| 60 | |
| 61 self.__enabled = True | |
| 62 self.__cssRule = False | |
| 63 self.__exception = False | |
| 64 self.__internalDisabled = False | |
| 65 self.__domainRestricted = False | |
| 66 self.__useRegExp = False | |
| 67 self.__useDomainMatch = False | |
| 68 self.__useEndsMatch = False | |
| 69 self.__thirdParty = False | |
| 70 self.__thirdPartyException = False | |
| 71 self.__object = False | |
| 72 self.__objectException = False | |
| 73 self.__subdocument = False | |
| 74 self.__subdocumentException = False | |
| 75 self.__xmlhttprequest = False | |
| 76 self.__xmlhttprequestException = False | |
| 77 self.__document = False | |
| 78 self.__elemhide = False | |
| 79 self.__caseSensitivity = Qt.CaseInsensitive | |
| 80 self.__image = False | |
| 81 self.__imageException = False | |
| 82 self.__script = False | |
| 83 self.__scriptException = False | |
| 84 self.__stylesheet = False | |
| 85 self.__stylesheetException = False | |
| 86 self.__objectSubrequest = False | |
| 87 self.__objectSubrequestException = False | |
| 88 | |
| 89 self.setFilter(filter) | |
| 90 | |
| 91 def subscription(self): | |
| 92 """ | |
| 93 Public method to get the subscription this rule belongs to. | |
| 94 | |
| 95 @return subscription of the rule (AdBlockSubscription) | |
| 96 """ | |
| 97 return self.__subscription | |
| 98 | |
| 99 def filter(self): | |
| 100 """ | |
| 101 Public method to get the rule filter string. | |
| 102 | |
| 103 @return rule filter string (string) | |
| 104 """ | |
| 105 return self.__filter | |
| 106 | |
| 107 def setFilter(self, filter): | |
| 108 """ | |
| 109 Public method to set the rule filter string. | |
| 110 | |
| 111 @param filter rule filter string (string) | |
| 112 """ | |
| 113 self.__filter = filter | |
| 114 self.__parseFilter() | |
| 115 | |
| 116 def __parseFilter(self): | |
| 117 """ | |
| 118 Private method to parse the filter pattern. | |
| 119 """ | |
| 120 parsedLine = self.__filter | |
| 121 | |
| 122 # empty rule or just a comment | |
| 123 if not parsedLine.strip() or parsedLine.startswith(("!", "[Adblock")): | |
| 124 self.__enabled = False | |
| 125 return | |
| 126 | |
| 127 # CSS element hiding rule | |
| 128 if "##" in parsedLine or "#@#" in parsedLine: | |
| 129 self.__cssRule = True | |
| 130 pos = parsedLine.find("#") | |
| 131 | |
| 132 # domain restricted rule | |
| 133 if not parsedLine.startswith("##"): | |
| 134 domains = parsedLine[:pos] | |
| 135 self.__parseDomains(domains, ",") | |
| 136 | |
| 137 self.__exception = parsedLine[pos + 1] == "@" | |
| 138 | |
| 139 if self.__exception: | |
| 140 self.__cssSelector = parsedLine[pos + 3:] | |
| 141 else: | |
| 142 self.__cssSelector = parsedLine[pos + 2:] | |
| 143 # CSS rule cannot have more options -> stop parsing | |
| 144 return | |
| 145 | |
| 146 # Exception always starts with @@ | |
| 147 if parsedLine.startswith("@@"): | |
| 148 self.__exception = True | |
| 149 parsedLine = parsedLine[2:] | |
| 150 | |
| 151 # Parse all options following '$' character | |
| 152 optionsIndex = parsedLine.find("$") | |
| 153 if optionsIndex >= 0: | |
| 154 options = parsedLine[optionsIndex + 1:].split(",") | |
| 155 | |
| 156 handledOptions = 0 | |
| 157 for option in options: | |
| 158 if option.startswith("domain="): | |
| 159 self.__parseDomains(option[7:], "|") | |
| 160 handledOptions += 1 | |
| 161 elif option == "match-case": | |
| 162 self.__caseSensitivity = Qt.CaseSensitive | |
| 163 handledOptions += 1 | |
| 164 elif option.endswith("third-party"): | |
| 165 self.__thirdParty = True | |
| 166 self.__thirdPartyException = option.startswith("~") | |
| 167 handledOptions += 1 | |
| 168 elif option.endswith("object"): | |
| 169 self.__object = True | |
| 170 self.__objectException = option.startswith("~") | |
| 171 handledOptions += 1 | |
| 172 elif option.endswith("subdocument"): | |
| 173 self.__subdocument = True | |
| 174 self.__subdocumentException = option.startswith("~") | |
| 175 handledOptions += 1 | |
| 176 elif option.endswith("xmlhttprequest"): | |
| 177 self.__xmlhttprequest = True | |
| 178 self.__xmlhttprequestException = option.startswith("~") | |
| 179 handledOptions += 1 | |
| 180 elif option.endswith("image"): | |
| 181 self.__image = True | |
| 182 self.__imageException = option.startswith("~") | |
| 183 elif option.endswith("script"): | |
| 184 self.__script = True | |
| 185 self.__scriptException = option.startswith("~") | |
| 186 elif option.endswith("stylesheet"): | |
| 187 self.__stylesheet = True | |
| 188 self.__stylesheetException = option.startswith("~") | |
| 189 elif option.endswith("object-subrequest"): | |
| 190 self.__objectSubrequest = True | |
| 191 self.__objectSubrequestException = option.startswith("~") | |
| 192 elif option == "document" and self.__exception: | |
| 193 self.__document = True | |
| 194 handledOptions += 1 | |
| 195 elif option == "elemhide" and self.__exception: | |
| 196 self.__elemhide = True | |
| 197 handledOptions += 1 | |
| 198 elif option == "collapse": | |
| 199 # Hiding placeholders of blocked elements | |
| 200 handledOptions += 1 | |
| 201 | |
| 202 # If we don't handle all options, it's safer to just disable | |
| 203 # this rule | |
| 204 if handledOptions != len(options): | |
| 205 self.__internalDisabled = True | |
| 206 return | |
| 207 | |
| 208 parsedLine = parsedLine[:optionsIndex] | |
| 209 | |
| 210 # Rule is classic regexp | |
| 211 if parsedLine.startswith("/") and parsedLine.endswith("/"): | |
| 212 parsedLine = parsedLine[1:-1] | |
| 213 self.__useRegExp = True | |
| 214 self.__regExp = QRegExp(parsedLine, self.__caseSensitivity, | |
| 215 QRegExp.RegExp) | |
| 216 return | |
| 217 | |
| 218 # Remove starting / ending wildcards | |
| 219 if parsedLine.startswith("*"): | |
| 220 parsedLine = parsedLine[1:] | |
| 221 if parsedLine.endswith("*"): | |
| 222 parsedLine = parsedLine[:-1] | |
| 223 | |
| 224 # Fast string matching for domain can be used | |
| 225 if parsedLine.startswith("||") and \ | |
| 226 parsedLine.endswith("^") and \ | |
| 227 QRegExp("[/:?=&\\*]").indexIn(parsedLine) == -1: | |
| 228 parsedLine = parsedLine[2:-1] | |
| 229 self.__useDomainMatch = True | |
| 230 self.__matchString = parsedLine | |
| 231 return | |
| 232 | |
| 233 # If rule contains '|' only at the end, string matching can be used | |
| 234 if parsedLine.endswith("|") and \ | |
| 235 QRegExp("[\\^\\*]").indexIn(parsedLine) == -1 and \ | |
| 236 parsedLine.count("|") == 1: | |
| 237 parsedLine = parsedLine[:-1] | |
| 238 self.__useEndsMatch = True | |
| 239 self.__matchString = parsedLine | |
| 240 return | |
| 241 | |
| 242 # If there is still a wildcard (*) or separator (^) or (|), | |
| 243 # the rule must be modified to comply with QRegExp. | |
| 244 if "*" in parsedLine or "^" in parsedLine or "|" in parsedLine: | |
| 245 pattern = self.__convertPatternToRegExp(parsedLine) | |
| 246 self.__useRegExp = True | |
| 247 self.__regExp = QRegExp(pattern, self.__caseSensitivity, | |
| 248 QRegExp.RegExp) | |
| 249 return | |
| 250 | |
| 251 # no regexp required | |
| 252 self.__useRegExp = False | |
| 253 self.__matchString = parsedLine | |
| 254 | |
| 255 def __parseDomains(self, domains, separator): | |
| 256 """ | |
| 257 Private method to parse a string with a domain list. | |
| 258 | |
| 259 @param domains list of domains (string) | |
| 260 @param separator separator character used by the list (string) | |
| 261 """ | |
| 262 domainsList = domains.split(separator) | |
| 263 | |
| 264 for domain in domainsList: | |
| 265 if not domain: | |
| 266 continue | |
| 267 if domain.startswith("~"): | |
| 268 self.__blockedDomains.append(domain[1:]) | |
| 269 else: | |
| 270 self.__allowedDomains.append(domain) | |
| 271 | |
| 272 self.__domainRestricted = \ | |
| 273 bool(self.__blockedDomains) or bool(self.__allowedDomains) | |
| 274 | |
| 275 def networkMatch(self, request, domain, encodedUrl): | |
| 276 """ | |
| 277 Public method to check the rule for a match. | |
| 278 | |
| 279 @param request reference to the network request | |
| 280 @type QWebEngineUrlRequestInfo | |
| 281 @param domain domain name | |
| 282 @type str | |
| 283 @param encodedUrl string encoded URL to be checked | |
| 284 @type str | |
| 285 @return flag indicating a match | |
| 286 @rtype bool | |
| 287 """ | |
| 288 if self.__cssRule or not self.__enabled or self.__internalDisabled: | |
| 289 return False | |
| 290 | |
| 291 matched = False | |
| 292 | |
| 293 if self.__useRegExp: | |
| 294 matched = self.__regExp.indexIn(encodedUrl) != -1 | |
| 295 elif self.__useDomainMatch: | |
| 296 matched = domain.endswith(self.__matchString) | |
| 297 elif self.__useEndsMatch: | |
| 298 if self.__caseSensitivity == Qt.CaseInsensitive: | |
| 299 matched = encodedUrl.lower().endswith( | |
| 300 self.__matchString.lower()) | |
| 301 else: | |
| 302 matched = encodedUrl.endswith(self.__matchString) | |
| 303 else: | |
| 304 if self.__caseSensitivity == Qt.CaseInsensitive: | |
| 305 matched = self.__matchString.lower() in encodedUrl.lower() | |
| 306 else: | |
| 307 matched = self.__matchString in encodedUrl | |
| 308 | |
| 309 if matched: | |
| 310 # check domain restrictions | |
| 311 if self.__domainRestricted and \ | |
| 312 not self.matchDomain(request.firstPartyUrl().host()): | |
| 313 return False | |
| 314 | |
| 315 # check third-party restrictions | |
| 316 if self.__thirdParty and not self.matchThirdParty(request): | |
| 317 return False | |
| 318 | |
| 319 # check object restrictions | |
| 320 if self.__object and not self.matchObject(request): | |
| 321 return False | |
| 322 | |
| 323 # check subdocument restrictions | |
| 324 if self.__subdocument and not self.matchSubdocument(request): | |
| 325 return False | |
| 326 | |
| 327 # check xmlhttprequest restriction | |
| 328 if self.__xmlhttprequest and not self.matchXmlHttpRequest(request): | |
| 329 return False | |
| 330 | |
| 331 # check image restriction | |
| 332 if self.__image and not self.matchImage(request): | |
| 333 return False | |
| 334 | |
| 335 # check script restriction | |
| 336 if self.__script and not self.matchScript(request): | |
| 337 return False | |
| 338 | |
| 339 # check stylesheet restriction | |
| 340 if self.__stylesheet and not self.matchStyleSheet(request): | |
| 341 return False | |
| 342 | |
| 343 # check object-subrequest restriction | |
| 344 if self.__objectSubrequest and \ | |
| 345 not self.matchObjectSubrequest(request): | |
| 346 return False | |
| 347 | |
| 348 return matched | |
| 349 | |
| 350 def urlMatch(self, url): | |
| 351 """ | |
| 352 Public method to check an URL against the rule. | |
| 353 | |
| 354 @param url URL to check (QUrl) | |
| 355 @return flag indicating a match (boolean) | |
| 356 """ | |
| 357 if not self.__document and not self.__elemhide: | |
| 358 return False | |
| 359 | |
| 360 encodedUrl = bytes(url.toEncoded()).decode() | |
| 361 domain = url.host() | |
| 362 return self.networkMatch(QNetworkRequest(url), domain, encodedUrl) | |
| 363 | |
| 364 def matchDomain(self, domain): | |
| 365 """ | |
| 366 Public method to match a domain. | |
| 367 | |
| 368 @param domain domain name to check (string) | |
| 369 @return flag indicating a match (boolean) | |
| 370 """ | |
| 371 if not self.__enabled: | |
| 372 return False | |
| 373 | |
| 374 if not self.__domainRestricted: | |
| 375 return True | |
| 376 | |
| 377 if len(self.__blockedDomains) == 0: | |
| 378 for dom in self.__allowedDomains: | |
| 379 if domain.endswith(dom): | |
| 380 return True | |
| 381 elif len(self.__allowedDomains) == 0: | |
| 382 for dom in self.__blockedDomains: | |
| 383 if domain.endswith(dom): | |
| 384 return False | |
| 385 return True | |
| 386 else: | |
| 387 for dom in self.__blockedDomains: | |
| 388 if domain.endswith(dom): | |
| 389 return False | |
| 390 for dom in self.__allowedDomains: | |
| 391 if domain.endswith(dom): | |
| 392 return True | |
| 393 | |
| 394 return False | |
| 395 | |
| 396 def matchThirdParty(self, req): | |
| 397 """ | |
| 398 Public slot to match a third-party rule. | |
| 399 | |
| 400 @param req request object to check (QWebEngineUrlRequestInfo) | |
| 401 @return flag indicating a match (boolean) | |
| 402 """ | |
| 403 # Third-party matching should be performed on second-level domains | |
| 404 firstPartyHost = toSecondLevelDomain(req.firstPartyUrl()) | |
| 405 host = toSecondLevelDomain(req.requestUrl()) | |
| 406 | |
| 407 match = firstPartyHost != host | |
| 408 | |
| 409 if self.__thirdPartyException: | |
| 410 return not match | |
| 411 else: | |
| 412 return match | |
| 413 | |
| 414 def matchObject(self, req): | |
| 415 """ | |
| 416 Public slot to match an object rule. | |
| 417 | |
| 418 @param req request object to check (QWebEngineUrlRequestInfo) | |
| 419 @return flag indicating a match (boolean) | |
| 420 """ | |
| 421 match = ( | |
| 422 req.resourceType() == QWebEngineUrlRequestInfo.ResourceTypeObject) | |
| 423 | |
| 424 if self.__objectException: | |
| 425 return not match | |
| 426 else: | |
| 427 return match | |
| 428 | |
| 429 def matchSubdocument(self, req): | |
| 430 """ | |
| 431 Public slot to match a sub-document rule. | |
| 432 | |
| 433 @param req request object to check (QWebEngineUrlRequestInfo) | |
| 434 @return flag indicating a match (boolean) | |
| 435 """ | |
| 436 match = ( | |
| 437 req.resourceType() == | |
| 438 QWebEngineUrlRequestInfo.ResourceTypeSubFrame) | |
| 439 | |
| 440 if self.__subdocumentException: | |
| 441 return not match | |
| 442 else: | |
| 443 return match | |
| 444 | |
| 445 def matchXmlHttpRequest(self, req): | |
| 446 """ | |
| 447 Public slot to match a XmlHttpRequest rule. | |
| 448 | |
| 449 @param req request object to check (QWebEngineUrlRequestInfo) | |
| 450 @return flag indicating a match (boolean) | |
| 451 """ | |
| 452 match = ( | |
| 453 req.resourceType() == QWebEngineUrlRequestInfo.ResourceTypeXhr) | |
| 454 | |
| 455 if self.__xmlhttprequestException: | |
| 456 return not match | |
| 457 else: | |
| 458 return match | |
| 459 | |
| 460 def matchImage(self, req): | |
| 461 """ | |
| 462 Public slot to match an Image rule. | |
| 463 | |
| 464 @param req request object to check (QWebEngineUrlRequestInfo) | |
| 465 @return flag indicating a match (boolean) | |
| 466 """ | |
| 467 match = ( | |
| 468 req.resourceType() == QWebEngineUrlRequestInfo.ResourceTypeImage) | |
| 469 | |
| 470 if self.__imageException: | |
| 471 return not match | |
| 472 else: | |
| 473 return match | |
| 474 | |
| 475 def matchScript(self, req): | |
| 476 """ | |
| 477 Public slot to match a Script rule. | |
| 478 | |
| 479 @param req request object to check (QWebEngineUrlRequestInfo) | |
| 480 @return flag indicating a match (boolean) | |
| 481 """ | |
| 482 match = ( | |
| 483 req.resourceType() == QWebEngineUrlRequestInfo.ResourceTypeScript) | |
| 484 | |
| 485 if self.__scriptException: | |
| 486 return not match | |
| 487 else: | |
| 488 return match | |
| 489 | |
| 490 def matchStyleSheet(self, req): | |
| 491 """ | |
| 492 Public slot to match a StyleSheet rule. | |
| 493 | |
| 494 @param req request object to check (QWebEngineUrlRequestInfo) | |
| 495 @return flag indicating a match (boolean) | |
| 496 """ | |
| 497 match = ( | |
| 498 req.resourceType() == | |
| 499 QWebEngineUrlRequestInfo.ResourceTypeStylesheet) | |
| 500 | |
| 501 if self.__stylesheetException: | |
| 502 return not match | |
| 503 else: | |
| 504 return match | |
| 505 | |
| 506 def matchObjectSubrequest(self, req): | |
| 507 """ | |
| 508 Public slot to match an Object Subrequest rule. | |
| 509 | |
| 510 @param req request object to check (QWebEngineUrlRequestInfo) | |
| 511 @return flag indicating a match (boolean) | |
| 512 """ | |
| 513 match = ( | |
| 514 req.resourceType() == | |
| 515 QWebEngineUrlRequestInfo.ResourceTypeSubResource) | |
| 516 | |
| 517 if self.__objectSubrequestException: | |
| 518 return not match | |
| 519 else: | |
| 520 return match | |
| 521 | |
| 522 def isException(self): | |
| 523 """ | |
| 524 Public method to check, if the rule defines an exception. | |
| 525 | |
| 526 @return flag indicating an exception (boolean) | |
| 527 """ | |
| 528 return self.__exception | |
| 529 | |
| 530 def setException(self, exception): | |
| 531 """ | |
| 532 Public method to set the rule's exception flag. | |
| 533 | |
| 534 @param exception flag indicating an exception rule (boolean) | |
| 535 """ | |
| 536 self.__exception = exception | |
| 537 | |
| 538 def isEnabled(self): | |
| 539 """ | |
| 540 Public method to check, if the rule is enabled. | |
| 541 | |
| 542 @return flag indicating enabled state (boolean) | |
| 543 """ | |
| 544 return self.__enabled | |
| 545 | |
| 546 def setEnabled(self, enabled): | |
| 547 """ | |
| 548 Public method to set the rule's enabled state. | |
| 549 | |
| 550 @param enabled flag indicating the new enabled state (boolean) | |
| 551 """ | |
| 552 self.__enabled = enabled | |
| 553 if not enabled: | |
| 554 self.__filter = "!" + self.__filter | |
| 555 else: | |
| 556 self.__filter = self.__filter[1:] | |
| 557 | |
| 558 def isCSSRule(self): | |
| 559 """ | |
| 560 Public method to check, if the rule is a CSS rule. | |
| 561 | |
| 562 @return flag indicating a CSS rule (boolean) | |
| 563 """ | |
| 564 return self.__cssRule | |
| 565 | |
| 566 def cssSelector(self): | |
| 567 """ | |
| 568 Public method to get the CSS selector of the rule. | |
| 569 | |
| 570 @return CSS selector (string) | |
| 571 """ | |
| 572 return self.__cssSelector | |
| 573 | |
| 574 def isDocument(self): | |
| 575 """ | |
| 576 Public method to check, if this is a document rule. | |
| 577 | |
| 578 @return flag indicating a document rule (boolean) | |
| 579 """ | |
| 580 return self.__document | |
| 581 | |
| 582 def isElementHiding(self): | |
| 583 """ | |
| 584 Public method to check, if this is an element hiding rule. | |
| 585 | |
| 586 @return flag indicating an element hiding rule (boolean) | |
| 587 """ | |
| 588 return self.__elemhide | |
| 589 | |
| 590 def isDomainRestricted(self): | |
| 591 """ | |
| 592 Public method to check, if this rule is restricted by domain. | |
| 593 | |
| 594 @return flag indicating a domain restriction (boolean) | |
| 595 """ | |
| 596 return self.__domainRestricted | |
| 597 | |
| 598 def isComment(self): | |
| 599 """ | |
| 600 Public method to check, if this is a comment. | |
| 601 | |
| 602 @return flag indicating a comment (boolean) | |
| 603 """ | |
| 604 return self.__filter.startswith("!") | |
| 605 | |
| 606 def isHeader(self): | |
| 607 """ | |
| 608 Public method to check, if this is a header. | |
| 609 | |
| 610 @return flag indicating a header (boolean) | |
| 611 """ | |
| 612 return self.__filter.startswith("[Adblock") | |
| 613 | |
| 614 def isSlow(self): | |
| 615 """ | |
| 616 Public method to check, if this is a slow rule. | |
| 617 | |
| 618 @return flag indicating a slow rule (boolean) | |
| 619 """ | |
| 620 return self.__useRegExp | |
| 621 | |
| 622 def isInternalDisabled(self): | |
| 623 """ | |
| 624 Public method to check, if this rule was disabled internally. | |
| 625 | |
| 626 @return flag indicating an internally disabled rule (boolean) | |
| 627 """ | |
| 628 return self.__internalDisabled | |
| 629 | |
| 630 def __convertPatternToRegExp(self, wildcardPattern): | |
| 631 """ | |
| 632 Private method to convert a wildcard pattern to a regular expression. | |
| 633 | |
| 634 @param wildcardPattern string containing the wildcard pattern (string) | |
| 635 @return string containing a regular expression (string) | |
| 636 """ | |
| 637 pattern = wildcardPattern | |
| 638 | |
| 639 # remove multiple wildcards | |
| 640 pattern = re.sub(r"\*+", "*", pattern) | |
| 641 # remove anchors following separator placeholder | |
| 642 pattern = re.sub(r"\^\|$", "^", pattern) | |
| 643 # remove leading wildcards | |
| 644 pattern = re.sub(r"^(\*)", "", pattern) | |
| 645 # remove trailing wildcards | |
| 646 pattern = re.sub(r"(\*)$", "", pattern) | |
| 647 # escape special symbols | |
| 648 pattern = re.sub(r"(\W)", r"\\\1", pattern) | |
| 649 # process extended anchor at expression start | |
| 650 pattern = re.sub( | |
| 651 r"^\\\|\\\|", | |
| 652 r"^[\w\-]+:\/+(?!\/)(?:[^\/]+\.)?", pattern) | |
| 653 # process separator placeholders | |
| 654 pattern = re.sub(r"\\\^", r"(?:[^\w\d\-.%]|$)", pattern) | |
| 655 # process anchor at expression start | |
| 656 pattern = re.sub(r"^\\\|", "^", pattern) | |
| 657 # process anchor at expression end | |
| 658 pattern = re.sub(r"\\\|$", "$", pattern) | |
| 659 # replace wildcards by .* | |
| 660 pattern = re.sub(r"\\\*", ".*", pattern) | |
| 661 | |
| 662 return pattern |
