1
mirror of https://invent.kde.org/network/falkon.git synced 2024-12-20 10:46:35 +01:00

[Code] Rewritten AdBlockRule to use enums for rule options.

Instead of using boolean variables for every option, use one QFlag
for options and second for exceptions.
This commit is contained in:
nowrep 2013-02-27 21:26:41 +01:00
parent 4d82e307ad
commit 9d6e34e3e5
2 changed files with 172 additions and 154 deletions

View File

@ -97,27 +97,12 @@ static QString toSecondLevelDomain(const QUrl &url)
AdBlockRule::AdBlockRule(const QString &filter, AdBlockSubscription* subscription) AdBlockRule::AdBlockRule(const QString &filter, AdBlockSubscription* subscription)
: m_subscription(subscription) : m_subscription(subscription)
, m_enabled(true) , m_type(StringContainsMatchRule)
, m_cssRule(false)
, m_exception(false)
, m_internalDisabled(false)
, m_domainRestricted(false)
, m_regExp(0)
, m_useDomainMatch(false)
, m_useEndsMatch(false)
, m_thirdParty(false)
, m_thirdPartyException(false)
, m_object(false)
, m_objectException(false)
, m_subdocument(false)
, m_subdocumentException(false)
, m_xmlhttprequest(false)
, m_xmlhttprequestException(false)
, m_image(false)
, m_imageException(false)
, m_document(false)
, m_elemhide(false)
, m_caseSensitivity(Qt::CaseInsensitive) , m_caseSensitivity(Qt::CaseInsensitive)
, m_isEnabled(true)
, m_isException(false)
, m_isInternalDisabled(false)
, m_regExp(0)
{ {
setFilter(filter); setFilter(filter);
} }
@ -145,7 +130,7 @@ void AdBlockRule::setFilter(const QString &filter)
bool AdBlockRule::isCssRule() const bool AdBlockRule::isCssRule() const
{ {
return m_cssRule; return m_type == CssRule;
} }
QString AdBlockRule::cssSelector() const QString AdBlockRule::cssSelector() const
@ -155,22 +140,22 @@ QString AdBlockRule::cssSelector() const
bool AdBlockRule::isDocument() const bool AdBlockRule::isDocument() const
{ {
return m_document; return hasOption(DocumentOption);
} }
bool AdBlockRule::isElemhide() const bool AdBlockRule::isElemhide() const
{ {
return m_elemhide; return hasOption(ElementHideOption);
} }
bool AdBlockRule::isDomainRestricted() const bool AdBlockRule::isDomainRestricted() const
{ {
return m_domainRestricted; return hasOption(DomainRestrictedOption);
} }
bool AdBlockRule::isException() const bool AdBlockRule::isException() const
{ {
return m_exception; return m_isException;
} }
bool AdBlockRule::isComment() const bool AdBlockRule::isComment() const
@ -180,12 +165,12 @@ bool AdBlockRule::isComment() const
bool AdBlockRule::isEnabled() const bool AdBlockRule::isEnabled() const
{ {
return m_enabled; return m_isEnabled;
} }
void AdBlockRule::setEnabled(bool enabled) void AdBlockRule::setEnabled(bool enabled)
{ {
m_enabled = enabled; m_isEnabled = enabled;
} }
bool AdBlockRule::isSlow() const bool AdBlockRule::isSlow() const
@ -195,72 +180,12 @@ bool AdBlockRule::isSlow() const
bool AdBlockRule::isInternalDisabled() const bool AdBlockRule::isInternalDisabled() const
{ {
return m_internalDisabled; return m_isInternalDisabled;
}
bool AdBlockRule::networkMatch(const QNetworkRequest &request, const QString &domain, const QString &encodedUrl) const
{
if (m_cssRule || !m_enabled || m_internalDisabled) {
return false;
}
bool matched = false;
if (m_useDomainMatch) {
matched = isMatchingDomain(domain, m_matchString);
}
else if (m_useEndsMatch) {
matched = encodedUrl.endsWith(m_matchString, m_caseSensitivity);
}
else if (m_regExp) {
if (!isMatchingRegExpStrings(encodedUrl)) {
return false;
}
matched = (m_regExp->indexIn(encodedUrl) != -1);
}
else {
matched = encodedUrl.contains(m_matchString, m_caseSensitivity);
}
if (matched) {
// Check domain restrictions
if (m_domainRestricted && !matchDomain(domain)) {
return false;
}
// Check third-party restriction
if (m_thirdParty && !matchThirdParty(request)) {
return false;
}
// Check object restrictions
if (m_object && !matchObject(request)) {
return false;
}
// Check subdocument restriction
if (m_subdocument && !matchSubdocument(request)) {
return false;
}
// Check xmlhttprequest restriction
if (m_xmlhttprequest && !matchXmlHttpRequest(request)) {
return false;
}
// Check image restriction
if (m_image && !matchImage(encodedUrl)) {
return false;
}
}
return matched;
} }
bool AdBlockRule::urlMatch(const QUrl &url) const bool AdBlockRule::urlMatch(const QUrl &url) const
{ {
if (!m_document && !m_elemhide) { if (!hasOption(DocumentOption) && !hasOption(ElementHideOption)) {
return false; return false;
} }
@ -270,13 +195,73 @@ bool AdBlockRule::urlMatch(const QUrl &url) const
return networkMatch(QNetworkRequest(url), domain, encodedUrl); return networkMatch(QNetworkRequest(url), domain, encodedUrl);
} }
bool AdBlockRule::matchDomain(const QString &domain) const bool AdBlockRule::networkMatch(const QNetworkRequest &request, const QString &domain, const QString &encodedUrl) const
{ {
if (!m_enabled) { if (m_type == CssRule || !m_isEnabled || m_isInternalDisabled) {
return false; return false;
} }
if (!m_domainRestricted) { bool matched = false;
if (m_type == StringContainsMatchRule) {
matched = encodedUrl.contains(m_matchString, m_caseSensitivity);
}
else if (m_type == DomainMatchRule) {
matched = isMatchingDomain(domain, m_matchString);
}
else if (m_type == StringEndsMatchRule) {
matched = encodedUrl.endsWith(m_matchString, m_caseSensitivity);
}
else if (m_type == RegExpMatchRule) {
if (!isMatchingRegExpStrings(encodedUrl)) {
return false;
}
matched = (m_regExp->indexIn(encodedUrl) != -1);
}
if (matched) {
// Check domain restrictions
if (hasOption(DomainRestrictedOption) && !matchDomain(domain)) {
return false;
}
// Check third-party restriction
if (hasOption(ThirdPartyOption) && !matchThirdParty(request)) {
return false;
}
// Check object restrictions
if (hasOption(ObjectOption) && !matchObject(request)) {
return false;
}
// Check subdocument restriction
if (hasOption(SubdocumentOption) && !matchSubdocument(request)) {
return false;
}
// Check xmlhttprequest restriction
if (hasOption(XMLHttpRequestOption) && !matchXmlHttpRequest(request)) {
return false;
}
// Check image restriction
if (hasOption(ImageOption) && !matchImage(encodedUrl)) {
return false;
}
}
return matched;
}
bool AdBlockRule::matchDomain(const QString &domain) const
{
if (!m_isEnabled) {
return false;
}
if (!hasOption(DomainRestrictedOption)) {
return true; return true;
} }
@ -326,14 +311,14 @@ bool AdBlockRule::matchThirdParty(const QNetworkRequest &request) const
bool match = refererHost != host; bool match = refererHost != host;
return m_thirdPartyException ? !match : match; return hasException(ThirdPartyOption) ? !match : match;
} }
bool AdBlockRule::matchObject(const QNetworkRequest &request) const bool AdBlockRule::matchObject(const QNetworkRequest &request) const
{ {
bool match = request.attribute(QNetworkRequest::Attribute(QNetworkRequest::User + 150)).toString() == QLatin1String("object"); bool match = request.attribute(QNetworkRequest::Attribute(QNetworkRequest::User + 150)).toString() == QLatin1String("object");
return m_objectException ? !match : match; return hasException(ObjectOption) ? !match : match;
} }
bool AdBlockRule::matchSubdocument(const QNetworkRequest &request) const bool AdBlockRule::matchSubdocument(const QNetworkRequest &request) const
@ -350,14 +335,14 @@ bool AdBlockRule::matchSubdocument(const QNetworkRequest &request) const
bool match = !(originatingFrame == page->mainFrame()); bool match = !(originatingFrame == page->mainFrame());
return m_subdocumentException ? !match : match; return hasException(SubdocumentOption) ? !match : match;
} }
bool AdBlockRule::matchXmlHttpRequest(const QNetworkRequest &request) const bool AdBlockRule::matchXmlHttpRequest(const QNetworkRequest &request) const
{ {
bool match = request.rawHeader("X-Requested-With") == QByteArray("XMLHttpRequest"); bool match = request.rawHeader("X-Requested-With") == QByteArray("XMLHttpRequest");
return m_xmlhttprequestException ? !match : match; return hasException(XMLHttpRequestOption) ? !match : match;
} }
bool AdBlockRule::matchImage(const QString &encodedUrl) const bool AdBlockRule::matchImage(const QString &encodedUrl) const
@ -367,7 +352,7 @@ bool AdBlockRule::matchImage(const QString &encodedUrl) const
encodedUrl.endsWith(QLatin1String(".gif")) || encodedUrl.endsWith(QLatin1String(".gif")) ||
encodedUrl.endsWith(QLatin1String(".jpeg")); encodedUrl.endsWith(QLatin1String(".jpeg"));
return m_imageException ? !match : match; return hasException(ImageOption) ? !match : match;
} }
void AdBlockRule::parseFilter() void AdBlockRule::parseFilter()
@ -376,13 +361,13 @@ void AdBlockRule::parseFilter()
// Empty rule or just comment // Empty rule or just comment
if (m_filter.trimmed().isEmpty() || m_filter.startsWith(QLatin1Char('!'))) { if (m_filter.trimmed().isEmpty() || m_filter.startsWith(QLatin1Char('!'))) {
m_enabled = false; m_isEnabled = false;
return; return;
} }
// CSS Element hiding rule // CSS Element hiding rule
if (parsedLine.contains(QLatin1String("##"))) { if (parsedLine.contains(QLatin1String("##"))) {
m_cssRule = true; m_type = CssRule;
int pos = parsedLine.indexOf(QLatin1String("##")); int pos = parsedLine.indexOf(QLatin1String("##"));
// Domain restricted rule // Domain restricted rule
@ -400,7 +385,7 @@ void AdBlockRule::parseFilter()
// Exception always starts with @@ // Exception always starts with @@
if (parsedLine.startsWith(QLatin1String("@@"))) { if (parsedLine.startsWith(QLatin1String("@@"))) {
m_exception = true; m_isException = true;
parsedLine = parsedLine.mid(2); parsedLine = parsedLine.mid(2);
} }
@ -420,47 +405,47 @@ void AdBlockRule::parseFilter()
++handledOptions; ++handledOptions;
} }
else if (option.endsWith(QLatin1String("third-party"))) { else if (option.endsWith(QLatin1String("third-party"))) {
m_thirdParty = true; setOption(ThirdPartyOption);
m_thirdPartyException = option.startsWith(QLatin1Char('~')); setException(ThirdPartyOption, option.startsWith(QLatin1Char('~')));
++handledOptions; ++handledOptions;
} }
else if (option.endsWith(QLatin1String("object"))) { else if (option.endsWith(QLatin1String("object"))) {
m_object = true; setOption(ObjectOption);
m_objectException = option.startsWith(QLatin1Char('~')); setException(ObjectOption, option.startsWith(QLatin1Char('~')));
++handledOptions; ++handledOptions;
} }
else if (option.endsWith(QLatin1String("subdocument"))) { else if (option.endsWith(QLatin1String("subdocument"))) {
m_subdocument = true; setOption(SubdocumentOption);
m_subdocumentException = option.startsWith(QLatin1Char('~')); setException(SubdocumentOption, option.startsWith(QLatin1Char('~')));
++handledOptions; ++handledOptions;
} }
else if (option.endsWith(QLatin1String("xmlhttprequest"))) { else if (option.endsWith(QLatin1String("xmlhttprequest"))) {
m_xmlhttprequest = true; setOption(XMLHttpRequestOption);
m_xmlhttprequestException = option.startsWith(QLatin1Char('~')); setException(XMLHttpRequestOption, option.startsWith(QLatin1Char('~')));
++handledOptions; ++handledOptions;
} }
else if (option.endsWith(QLatin1String("image"))) { else if (option.endsWith(QLatin1String("image"))) {
m_image = true; setOption(ImageOption);
m_imageException = option.startsWith(QLatin1Char('~')); setException(ImageOption, option.startsWith(QLatin1Char('~')));
++handledOptions; ++handledOptions;
} }
else if (option == QLatin1String("document") && m_exception) { else if (option == QLatin1String("document") && m_isException) {
m_document = true; setOption(DocumentOption);
++handledOptions; ++handledOptions;
} }
else if (option == QLatin1String("elemhide") && m_exception) { else if (option == QLatin1String("elemhide") && m_isException) {
m_elemhide = true; setOption(ElementHideOption);
++handledOptions; ++handledOptions;
} }
else if (option == QLatin1String("collapse")) { else if (option == QLatin1String("collapse")) {
// Hiding placeholders of blocked elements // Hiding placeholders of blocked elements is enabled by default
++handledOptions; ++handledOptions;
} }
} }
// If we don't handle all options, it's safer to just disable this rule // If we don't handle all options, it's safer to just disable this rule
if (handledOptions != options.count()) { if (handledOptions != options.count()) {
m_internalDisabled = true; m_isInternalDisabled = true;
return; return;
} }
@ -472,6 +457,7 @@ void AdBlockRule::parseFilter()
parsedLine = parsedLine.mid(1); parsedLine = parsedLine.mid(1);
parsedLine = parsedLine.left(parsedLine.size() - 1); parsedLine = parsedLine.left(parsedLine.size() - 1);
m_type = RegExpMatchRule;
m_regExp = new QzRegExp(parsedLine, m_caseSensitivity); m_regExp = new QzRegExp(parsedLine, m_caseSensitivity);
return; return;
} }
@ -491,7 +477,7 @@ void AdBlockRule::parseFilter()
parsedLine = parsedLine.mid(2); parsedLine = parsedLine.mid(2);
parsedLine = parsedLine.left(parsedLine.size() - 1); parsedLine = parsedLine.left(parsedLine.size() - 1);
m_useDomainMatch = true; m_type = DomainMatchRule;
m_matchString = parsedLine; m_matchString = parsedLine;
return; return;
} }
@ -501,7 +487,7 @@ void AdBlockRule::parseFilter()
&& parsedLine.count(QLatin1Char('|')) == 1) { && parsedLine.count(QLatin1Char('|')) == 1) {
parsedLine = parsedLine.left(parsedLine.size() - 1); parsedLine = parsedLine.left(parsedLine.size() - 1);
m_useEndsMatch = true; m_type = StringEndsMatchRule;
m_matchString = parsedLine; m_matchString = parsedLine;
return; return;
} }
@ -525,12 +511,14 @@ void AdBlockRule::parseFilter()
.replace(QzRegExp(QLatin1String("\\\\\\|$")), QLatin1String("$")) // process anchor at expression end .replace(QzRegExp(QLatin1String("\\\\\\|$")), QLatin1String("$")) // process anchor at expression end
.replace(QzRegExp(QLatin1String("\\\\\\*")), QLatin1String(".*")); // replace wildcards by .* .replace(QzRegExp(QLatin1String("\\\\\\*")), QLatin1String(".*")); // replace wildcards by .*
m_type = RegExpMatchRule;
m_regExpStrings = parseRegExpFilter(parsedLine); m_regExpStrings = parseRegExpFilter(parsedLine);
m_regExp = new QzRegExp(parsedRegExp, m_caseSensitivity); m_regExp = new QzRegExp(parsedRegExp, m_caseSensitivity);
return; return;
} }
// We haven't found anything that needs use of regexp, yay! // We haven't found anything that needs use of regexp, yay!
m_type = StringContainsMatchRule;
m_matchString = parsedLine; m_matchString = parsedLine;
} }
@ -550,7 +538,9 @@ void AdBlockRule::parseDomains(const QString &domains, const QChar &separator)
} }
} }
m_domainRestricted = (!m_blockedDomains.isEmpty() || !m_allowedDomains.isEmpty()); if (!m_blockedDomains.isEmpty() || !m_allowedDomains.isEmpty()) {
setOption(DomainRestrictedOption);
}
} }
bool AdBlockRule::isMatchingDomain(const QString &domain, const QString &filter) const bool AdBlockRule::isMatchingDomain(const QString &domain, const QString &filter) const
@ -590,6 +580,28 @@ QStringList AdBlockRule::parseRegExpFilter(const QString &parsedFilter) const
return list; return list;
} }
bool AdBlockRule::hasOption(const AdBlockRule::RuleOption &opt) const
{
return (m_options & opt);
}
bool AdBlockRule::hasException(const AdBlockRule::RuleOption &opt) const
{
return (m_exceptions & opt);
}
void AdBlockRule::setOption(const AdBlockRule::RuleOption &opt)
{
m_options |= opt;
}
void AdBlockRule::setException(const AdBlockRule::RuleOption &opt, bool on)
{
if (on) {
m_exceptions |= opt;
}
}
AdBlockRule::~AdBlockRule() AdBlockRule::~AdBlockRule()
{ {
delete m_regExp; delete m_regExp;

View File

@ -85,8 +85,8 @@ public:
bool isSlow() const; bool isSlow() const;
bool isInternalDisabled() const; bool isInternalDisabled() const;
bool networkMatch(const QNetworkRequest &request, const QString &domain, const QString &encodedUrl) const;
bool urlMatch(const QUrl &url) const; bool urlMatch(const QUrl &url) const;
bool networkMatch(const QNetworkRequest &request, const QString &domain, const QString &encodedUrl) const;
bool matchDomain(const QString &domain) const; bool matchDomain(const QString &domain) const;
bool matchThirdParty(const QNetworkRequest &request) const; bool matchThirdParty(const QNetworkRequest &request) const;
@ -101,52 +101,58 @@ protected:
QStringList parseRegExpFilter(const QString &parsedFilter) const; QStringList parseRegExpFilter(const QString &parsedFilter) const;
private: private:
enum RuleType {
CssRule = 0,
DomainMatchRule = 1,
RegExpMatchRule = 2,
StringEndsMatchRule = 3,
StringContainsMatchRule = 4
};
enum RuleOption {
DomainRestrictedOption = 1,
ThirdPartyOption = 2,
ObjectOption = 4,
SubdocumentOption = 8,
XMLHttpRequestOption = 16,
ImageOption = 32,
// Exception only options
DocumentOption = 64,
ElementHideOption = 128
};
Q_DECLARE_FLAGS(RuleOptions, RuleOption)
inline bool hasOption(const RuleOption &opt) const;
inline bool hasException(const RuleOption &opt) const;
inline void setOption(const RuleOption &opt);
inline void setException(const RuleOption &opt, bool on);
void parseFilter(); void parseFilter();
void parseDomains(const QString &domains, const QChar &separator); void parseDomains(const QString &domains, const QChar &separator);
AdBlockSubscription* m_subscription; AdBlockSubscription* m_subscription;
RuleType m_type;
RuleOptions m_options;
RuleOptions m_exceptions;
QString m_filter; QString m_filter;
QString m_cssSelector;
QString m_matchString;
Qt::CaseSensitivity m_caseSensitivity;
bool m_enabled; bool m_isEnabled;
bool m_cssRule; bool m_isException;
bool m_exception; bool m_isInternalDisabled;
bool m_internalDisabled;
bool m_domainRestricted;
QzRegExp* m_regExp; QzRegExp* m_regExp;
QStringList m_regExpStrings; QStringList m_regExpStrings;
bool m_useDomainMatch;
bool m_useEndsMatch;
QString m_cssSelector;
QString m_matchString;
// Rule $options
QStringList m_allowedDomains; QStringList m_allowedDomains;
QStringList m_blockedDomains; QStringList m_blockedDomains;
bool m_thirdParty;
bool m_thirdPartyException;
bool m_object;
bool m_objectException;
bool m_subdocument;
bool m_subdocumentException;
bool m_xmlhttprequest;
bool m_xmlhttprequestException;
bool m_image;
bool m_imageException;
// Exception only options
bool m_document;
bool m_elemhide;
Qt::CaseSensitivity m_caseSensitivity;
}; };
#endif // ADBLOCKRULE_H #endif // ADBLOCKRULE_H