Mercurial Repositories > eric / file comparison
comparison: Helpviewer/AdBlock/AdBlockRule.py
Helpviewer/AdBlock/AdBlockRule.py
- branch
- Py2 comp.
- changeset 3057
- 10516539f238
- parent 2525
- 8b507a9a2d40
- parent 3000
- 971d84f7a6d6
- child 3058
- 0a02c433f52d
equal
deleted
inserted
replaced
| 3056:9986ec0e559a | 3057:10516539f238 |
|---|---|
| 56 Class implementing the AdBlock rule. | 56 Class implementing the AdBlock rule. |
| 57 """ | 57 """ |
| 58 def __init__(self, filter="", subscription=None): | 58 def __init__(self, filter="", subscription=None): |
| 59 """ | 59 """ |
| 60 Constructor | 60 Constructor |
| 61 | |
| 62 @param filter filter string of the rule (string) | |
| 63 @param subscription reference to the subscription object | |
| 64 (AdBlockSubscription) | |
| 61 """ | 65 """ |
| 62 self.__subscription = subscription | 66 self.__subscription = subscription |
| 63 | 67 |
| 64 self.__regExp = QRegExp() | 68 self.__regExp = QRegExp() |
| 65 self.__options = [] | 69 self.__options = [] |
| 180 handledOptions += 1 | 184 handledOptions += 1 |
| 181 elif option == "collapse": | 185 elif option == "collapse": |
| 182 # Hiding placeholders of blocked elements | 186 # Hiding placeholders of blocked elements |
| 183 handledOptions += 1 | 187 handledOptions += 1 |
| 184 | 188 |
| 185 # If we don't handle all options, it's safer to just disable this rule | 189 # If we don't handle all options, it's safer to just disable |
| 190 # this rule | |
| 186 if handledOptions != len(options): | 191 if handledOptions != len(options): |
| 187 self.__internalDisabled = True | 192 self.__internalDisabled = True |
| 188 return | 193 return |
| 189 | 194 |
| 190 parsedLine = parsedLine[:optionsIndex] | 195 parsedLine = parsedLine[:optionsIndex] |
| 224 # If there is still a wildcard (*) or separator (^) or (|), | 229 # If there is still a wildcard (*) or separator (^) or (|), |
| 225 # the rule must be modified to comply with QRegExp. | 230 # the rule must be modified to comply with QRegExp. |
| 226 if "*" in parsedLine or "^" in parsedLine or "|" in parsedLine: | 231 if "*" in parsedLine or "^" in parsedLine or "|" in parsedLine: |
| 227 pattern = self.__convertPatternToRegExp(parsedLine) | 232 pattern = self.__convertPatternToRegExp(parsedLine) |
| 228 self.__useRegExp = True | 233 self.__useRegExp = True |
| 229 self.__regExp = QRegExp(pattern, self.__caseSensitivity, QRegExp.RegExp) | 234 self.__regExp = QRegExp(pattern, self.__caseSensitivity, |
| 235 QRegExp.RegExp) | |
| 230 return | 236 return |
| 231 | 237 |
| 232 # no regexp required | 238 # no regexp required |
| 233 self.__useRegExp = False | 239 self.__useRegExp = False |
| 234 self.__matchString = parsedLine | 240 self.__matchString = parsedLine |
| 271 matched = self.__regExp.indexIn(encodedUrl) != -1 | 277 matched = self.__regExp.indexIn(encodedUrl) != -1 |
| 272 elif self.__useDomainMatch: | 278 elif self.__useDomainMatch: |
| 273 matched = domain.endswith(self.__matchString) | 279 matched = domain.endswith(self.__matchString) |
| 274 elif self.__useEndsMatch: | 280 elif self.__useEndsMatch: |
| 275 if self.__caseSensitivity == Qt.CaseInsensitive: | 281 if self.__caseSensitivity == Qt.CaseInsensitive: |
| 276 matched = encodedUrl.lower().endswith(self.__matchString.lower()) | 282 matched = encodedUrl.lower().endswith( |
| 283 self.__matchString.lower()) | |
| 277 else: | 284 else: |
| 278 matched = encodedUrl.endswith(self.__matchString) | 285 matched = encodedUrl.endswith(self.__matchString) |
| 279 else: | 286 else: |
| 280 if self.__caseSensitivity == Qt.CaseInsensitive: | 287 if self.__caseSensitivity == Qt.CaseInsensitive: |
| 281 matched = self.__matchString.lower() in encodedUrl.lower() | 288 matched = self.__matchString.lower() in encodedUrl.lower() |
| 538 @param wildcardPattern string containing the wildcard pattern (string) | 545 @param wildcardPattern string containing the wildcard pattern (string) |
| 539 @return string containing a regular expression (string) | 546 @return string containing a regular expression (string) |
| 540 """ | 547 """ |
| 541 pattern = wildcardPattern | 548 pattern = wildcardPattern |
| 542 | 549 |
| 543 pattern = re.sub(r"\*+", "*", pattern) # remove multiple wildcards | 550 # remove multiple wildcards |
| 544 pattern = re.sub(r"\^\|$", "^", pattern) # remove anchors following separator | 551 pattern = re.sub(r"\*+", "*", pattern) |
| 545 # placeholder | 552 # remove anchors following separator placeholder |
| 546 pattern = re.sub(r"^(\*)", "", pattern) # remove leading wildcards | 553 pattern = re.sub(r"\^\|$", "^", pattern) |
| 547 pattern = re.sub(r"(\*)$", "", pattern) # remove trailing wildcards | 554 # remove leading wildcards |
| 548 pattern = re.sub(r"(\W)", r"\\\1", pattern) # escape special symbols | 555 pattern = re.sub(r"^(\*)", "", pattern) |
| 556 # remove trailing wildcards | |
| 557 pattern = re.sub(r"(\*)$", "", pattern) | |
| 558 # escape special symbols | |
| 559 pattern = re.sub(r"(\W)", r"\\\1", pattern) | |
| 560 # process extended anchor at expression start | |
| 549 pattern = re.sub(r"^\\\|\\\|", | 561 pattern = re.sub(r"^\\\|\\\|", |
| 550 r"^[\w\-]+:\/+(?!\/)(?:[^\/]+\.)?", pattern) # process extended anchor at | 562 r"^[\w\-]+:\/+(?!\/)(?:[^\/]+\.)?", pattern) |
| 551 # expression start | 563 # process separator placeholders |
| 552 pattern = re.sub(r"\\\^", | 564 pattern = re.sub(r"\\\^", r"(?:[^\w\d\-.%]|$)", pattern) |
| 553 r"(?:[^\w\d\-.%]|$)", pattern) # process separator placeholders | 565 # process anchor at expression start |
| 554 pattern = re.sub(r"^\\\|", "^", pattern) # process anchor at expression start | 566 pattern = re.sub(r"^\\\|", "^", pattern) |
| 555 pattern = re.sub(r"\\\|$", "$", pattern) # process anchor at expression end | 567 # process anchor at expression end |
| 556 pattern = re.sub(r"\\\*", ".*", pattern) # replace wildcards by .* | 568 pattern = re.sub(r"\\\|$", "$", pattern) |
| 569 # replace wildcards by .* | |
| 570 pattern = re.sub(r"\\\*", ".*", pattern) | |
| 557 | 571 |
| 558 return pattern | 572 return pattern |
