mirrors/qBittorrent
1
0
Fork 0
mirror of https://github.com/qbittorrent/qBittorrent.git synced 2024-11-28 05:27:15 +03:00
Code Issues Projects Releases Packages Wiki Activity

Cache rule regular expressions for performance

Browse source 
--HG--
branch : magao-dev
This commit is contained in:
Tim Delaney 2017-02-11 18:43:50 +11:00
parent f9abd254f4
commit a844ccb06a
2 changed files with 38 additions and 7 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

41
src/base/rss/rssdownloadrule.cpp
View file

@ -30,6 +30,7 @@
#include <QDebug> #include <QDebug>
#include <QDir> #include <QDir>
#include <QHash>
#include <QRegExp> #include <QRegExp>
#include <QRegularExpression> #include <QRegularExpression>
#include <QString> #include <QString>
@ -49,9 +50,29 @@ DownloadRule::DownloadRule()
, m_useRegex(false) , m_useRegex(false)
, m_apstate(USE_GLOBAL) , m_apstate(USE_GLOBAL)
, m_ignoreDays(0) , m_ignoreDays(0)
, m_cachedRegexes(new QHash<QString, QRegularExpression>)
{ {
} }
DownloadRule::~DownloadRule()
{
delete m_cachedRegexes;
}
QRegularExpression DownloadRule::getRegex(const QString &expression, bool isRegex) const
{
// Use a cache of regexes so we don't have to continually recompile - big performance increase.
// The cache is cleared whenever the regex/wildcard, must or must not contain fields or
// episode filter are modified.
Q_ASSERT(!expression.isEmpty());
QRegularExpression regex((*m_cachedRegexes)[expression]);
if (!regex.pattern().isEmpty())
return regex;
return (*m_cachedRegexes)[expression] = QRegularExpression(isRegex ? expression : Utils::String::wildcardToRegex(expression), QRegularExpression::CaseInsensitiveOption);
}
bool DownloadRule::matches(const QString &articleTitle, const QString &expression) const bool DownloadRule::matches(const QString &articleTitle, const QString &expression) const
{ {
static QRegularExpression whitespace("\\s+"); static QRegularExpression whitespace("\\s+");
@ -61,14 +82,14 @@ bool DownloadRule::matches(const QString &articleTitle, const QString &expressio
return true; return true;
} }
else if (m_useRegex) { else if (m_useRegex) {
QRegularExpression reg(expression, QRegularExpression::CaseInsensitiveOption); QRegularExpression reg(getRegex(expression));
return reg.match(articleTitle).hasMatch(); return reg.match(articleTitle).hasMatch();
} }
else { else {
// Only match if every wildcard token (separated by spaces) is present in the article name. // Only match if every wildcard token (separated by spaces) is present in the article name.
// Order of wildcard tokens is unimportant (if order is important, they should have used *). // Order of wildcard tokens is unimportant (if order is important, they should have used *).
foreach (const QString &wildcard, expression.split(whitespace, QString::SplitBehavior::SkipEmptyParts)) { foreach (const QString &wildcard, expression.split(whitespace, QString::SplitBehavior::SkipEmptyParts)) {
QRegularExpression reg(Utils::String::wildcardToRegex(wildcard), QRegularExpression::CaseInsensitiveOption); QRegularExpression reg(getRegex(wildcard, false));
if (!reg.match(articleTitle).hasMatch()) if (!reg.match(articleTitle).hasMatch())
return false; return false;
@ -126,7 +147,7 @@ bool DownloadRule::matches(const QString &articleTitle) const
if (!m_episodeFilter.isEmpty()) { if (!m_episodeFilter.isEmpty()) {
qDebug() << "Checking episode filter:" << m_episodeFilter; qDebug() << "Checking episode filter:" << m_episodeFilter;
QRegularExpression f("(^\\d{1,4})x(.*;$)"); QRegularExpression f(getRegex("(^\\d{1,4})x(.*;$)"));
QRegularExpressionMatch matcher = f.match(m_episodeFilter); QRegularExpressionMatch matcher = f.match(m_episodeFilter);
bool matched = matcher.hasMatch(); bool matched = matcher.hasMatch();
@ -148,7 +169,7 @@ bool DownloadRule::matches(const QString &articleTitle) const
if (ep.indexOf('-') != -1) { // Range detected if (ep.indexOf('-') != -1) { // Range detected
QString partialPattern1 = "\\bs0?(\\d{1,4})[ -_\\.]?e(0?\\d{1,4})(?:\\D|\\b)"; QString partialPattern1 = "\\bs0?(\\d{1,4})[ -_\\.]?e(0?\\d{1,4})(?:\\D|\\b)";
QString partialPattern2 = "\\b(\\d{1,4})x(0?\\d{1,4})(?:\\D|\\b)"; QString partialPattern2 = "\\b(\\d{1,4})x(0?\\d{1,4})(?:\\D|\\b)";
QRegularExpression reg(partialPattern1, QRegularExpression::CaseInsensitiveOption); QRegularExpression reg(getRegex(partialPattern1));
if (ep.endsWith('-')) { // Infinite range if (ep.endsWith('-')) { // Infinite range
int epOurs = ep.left(ep.size() - 1).toInt(); int epOurs = ep.left(ep.size() - 1).toInt();
@ -158,7 +179,7 @@ bool DownloadRule::matches(const QString &articleTitle) const
matched = matcher.hasMatch(); matched = matcher.hasMatch();
if (!matched) { if (!matched) {
reg = QRegularExpression(partialPattern2, QRegularExpression::CaseInsensitiveOption); reg = QRegularExpression(getRegex(partialPattern2));
matcher = reg.match(articleTitle); matcher = reg.match(articleTitle);
matched = matcher.hasMatch(); matched = matcher.hasMatch();
} }
@ -187,7 +208,7 @@ bool DownloadRule::matches(const QString &articleTitle) const
matched = matcher.hasMatch(); matched = matcher.hasMatch();
if (!matched) { if (!matched) {
reg = QRegularExpression(partialPattern2, QRegularExpression::CaseInsensitiveOption); reg = QRegularExpression(getRegex(partialPattern2));
matcher = reg.match(articleTitle); matcher = reg.match(articleTitle);
matched = matcher.hasMatch(); matched = matcher.hasMatch();
} }
@ -205,7 +226,7 @@ bool DownloadRule::matches(const QString &articleTitle) const
} }
else { // Single number else { // Single number
QString expStr("\\b(?:s0?" + s + "[ -_\\.]?" + "e0?" + ep + "|" + s + "x" + "0?" + ep + ")(?:\\D|\\b)"); QString expStr("\\b(?:s0?" + s + "[ -_\\.]?" + "e0?" + ep + "|" + s + "x" + "0?" + ep + ")(?:\\D|\\b)");
QRegularExpression reg(expStr, QRegularExpression::CaseInsensitiveOption); QRegularExpression reg(getRegex(expStr));
if (reg.match(articleTitle).hasMatch()) { if (reg.match(articleTitle).hasMatch()) {
qDebug() << "Matched episode:" << ep; qDebug() << "Matched episode:" << ep;
qDebug() << "Matched article:" << articleTitle; qDebug() << "Matched article:" << articleTitle;
@ -223,6 +244,8 @@ bool DownloadRule::matches(const QString &articleTitle) const
void DownloadRule::setMustContain(const QString &tokens) void DownloadRule::setMustContain(const QString &tokens)
{ {
m_cachedRegexes->clear();
if (m_useRegex) if (m_useRegex)
m_mustContain = QStringList() << tokens; m_mustContain = QStringList() << tokens;
else else
@ -235,6 +258,8 @@ void DownloadRule::setMustContain(const QString &tokens)
void DownloadRule::setMustNotContain(const QString &tokens) void DownloadRule::setMustNotContain(const QString &tokens)
{ {
m_cachedRegexes->clear();
if (m_useRegex) if (m_useRegex)
m_mustNotContain = QStringList() << tokens; m_mustNotContain = QStringList() << tokens;
else else
@ -384,6 +409,7 @@ bool DownloadRule::useRegex() const
void DownloadRule::setUseRegex(bool enabled) void DownloadRule::setUseRegex(bool enabled)
{ {
m_useRegex = enabled; m_useRegex = enabled;
m_cachedRegexes->clear();
} }
QString DownloadRule::episodeFilter() const QString DownloadRule::episodeFilter() const
@ -394,6 +420,7 @@ QString DownloadRule::episodeFilter() const
void DownloadRule::setEpisodeFilter(const QString &e) void DownloadRule::setEpisodeFilter(const QString &e)
{ {
m_episodeFilter = e; m_episodeFilter = e;
m_cachedRegexes->clear();
} }
QStringList DownloadRule::findMatchingArticles(const FeedPtr &feed) const QStringList DownloadRule::findMatchingArticles(const FeedPtr &feed) const

4
src/base/rss/rssdownloadrule.h
View file

@ -36,6 +36,7 @@
#include <QSharedPointer> #include <QSharedPointer>
#include <QStringList> #include <QStringList>
template <class T, class U> class QHash;
class QRegularExpression; class QRegularExpression;
namespace Rss namespace Rss
@ -57,6 +58,7 @@ namespace Rss
}; };
DownloadRule(); DownloadRule();
~DownloadRule();
static DownloadRulePtr fromVariantHash(const QVariantHash &ruleHash); static DownloadRulePtr fromVariantHash(const QVariantHash &ruleHash);
QVariantHash toVariantHash() const; QVariantHash toVariantHash() const;
@ -91,6 +93,7 @@ namespace Rss
private: private:
bool matches(const QString &articleTitle, const QString &expression) const; bool matches(const QString &articleTitle, const QString &expression) const;
QRegularExpression getRegex(const QString &expression, bool isRegex = true) const;
QString m_name; QString m_name;
QStringList m_mustContain; QStringList m_mustContain;
@ -104,6 +107,7 @@ namespace Rss
AddPausedState m_apstate; AddPausedState m_apstate;
QDateTime m_lastMatch; QDateTime m_lastMatch;
int m_ignoreDays; int m_ignoreDays;
mutable QHash<QString, QRegularExpression> *m_cachedRegexes;
}; };
} }