Add black/whitelisting of URLs

3 files changed, 172 insertions, 50 deletions

diff --git a/src/main.cpp b/src/main.cpp
index b828d60..208a635 100644
--- a/src/main.cpp
+++ b/src/main.cpp
@@ -3,6 +3,7 @@
 #include <QApplication>
 #include <QKeyEvent>
 #include <QCommandLineParser>
+#include <QTextStream>
 
 class KeyHandler : public QObject
 {
@@ -18,6 +19,8 @@ public:
 	}
 };
 
+QStringList loadUrlList(const QString &file);
+
 /**
  * MAIN
  */
@@ -26,23 +29,61 @@ int main(int argc, char** argv)
 	QApplication app(argc, argv);
 	QCommandLineParser parser;
 	parser.addHelpOption();
-	parser.addPositionalArgument("url", "URL to load");
-	QCommandLineOption ignoreSsl("insecure", "Ignore SSL errors");
-	QCommandLineOption fullscreen("fullscreen", "Run browser in full screen");
-	QCommandLineOption reloadInterval("reload-interval", "Reload displayed page every X seconds", "seconds");
+	parser.addPositionalArgument("url", "URL to load.");
+	QCommandLineOption ignoreSsl("insecure", "Ignore SSL errors.");
+	QCommandLineOption fullscreen("fullscreen", "Run browser in full screen.");
+	QCommandLineOption reloadInterval("reload-interval", "Reload displayed page every X seconds.", "seconds");
+	QCommandLineOption whitelist("whitelist", "Path to a file of allowed URLs. Mutually exclusive with blacklist.", "file");
+	QCommandLineOption blacklist("blacklist", "Path to a file of disallowed URLs. Mutually exclusive with whitelist.", "file");
 	parser.addOption(ignoreSsl);
 	parser.addOption(fullscreen);
 	parser.addOption(reloadInterval);
+	parser.addOption(whitelist);
+	parser.addOption(blacklist);
 	parser.process(app);
 	QStringList list(parser.positionalArguments());
 	if (list.empty()) {
-		QMessageBox::critical(NULL, "Error", "Need one argument: file name");
+		QMessageBox::critical(nullptr, "Error", "Need one argument: file name");
 		return 1;
 	}
-	QString url(list[0]);
-	SLXbrowser main(url, parser.isSet(fullscreen), parser.isSet(ignoreSsl), parser.value(reloadInterval).toInt());
+	if (parser.isSet(whitelist) && parser.isSet(blacklist)) {
+		QMessageBox::critical(nullptr, "Error", "Need either blacklist or whitelist, not both");
+		return 2;
+	}
+	BrowserSettings settings;
+	settings.url = list[0];
+	settings.fullscreen = parser.isSet(fullscreen);
+	settings.ignoreSslErrors = parser.isSet(ignoreSsl);
+	settings.reloadInterval = parser.value(reloadInterval).toInt();
+	if (parser.isSet(whitelist)) {
+		settings.urlList = loadUrlList(parser.value(whitelist));
+	} else if (parser.isSet(blacklist)) {
+		settings.urlList = loadUrlList(parser.value(blacklist));
+	}
+	settings.isWhitelist = parser.isSet(whitelist);
+	SLXbrowser main(settings);
 	main.show();
 	app.installEventFilter(new KeyHandler());
 	app.exec();
 	return 0;
 }
+
+QStringList loadUrlList(const QString &file)
+{
+	QStringList stringList;
+	QFile textFile(file);
+	if (!textFile.open(QFile::ReadOnly)) {
+		QTextStream(stdout) << "Cannot open URL list\n";
+		return QStringList();
+	}
+	QTextStream textStream(&textFile);
+	while (true)
+	{
+		QString line = textStream.readLine();
+		if (line.isNull())
+			break;
+		else
+			stringList.append(line);
+	}
+	return stringList;
+}
diff --git a/src/slxbrowser.cpp b/src/slxbrowser.cpp
index 0690102..5ffe281 100644
--- a/src/slxbrowser.cpp
+++ b/src/slxbrowser.cpp
@@ -1,19 +1,24 @@
 #include "slxbrowser.h"
 #include "webview.h"
+#include "nam.h"
 #include <QtWebKitWidgets>
 #include <QWebPage>
 #include <QNetworkReply>
 #include <QSslConfiguration>
 #include <QProgressBar>
+#include <QDateTime>
 
-SLXbrowser::SLXbrowser(QString url, bool fullscreen, bool ignoreSslErrors, int reloadIntervalSeconds)
+static QRegularExpression urlListToRegExp(const QStringList &list);
+
+SLXbrowser::SLXbrowser(BrowserSettings settings)
     : QMainWindow(nullptr),
-	  _url(url),
-	  _ignoreSslErrors(ignoreSslErrors),
+	  _settings(settings),
       _unsupportedUri(false),
-	  _reloadIntervalMs(reloadIntervalSeconds * 1000)
+	  _blockedSite(false),
+	  _lastPageLoad(0)
 {
-	if (fullscreen) {
+	_settings.reloadInterval *= 1000;
+	if (_settings.fullscreen) {
 		this->showFullScreen();
 		this->setWindowFlags(windowFlags() | Qt::WindowStaysOnTopHint);
 	}
@@ -40,12 +45,21 @@ SLXbrowser::SLXbrowser(QString url, bool fullscreen, bool ignoreSslErrors, int r
 	connect(_browser, SIGNAL(loadProgress(int)), this, SLOT(loadProgress(int)));
 	//
 	QWebPage *page = _browser->page();
-	QNetworkAccessManager *nam = new QNetworkAccessManager(this);
+	QNetworkAccessManager *nam;
+	if (_settings.urlList.isEmpty()) {
+		nam = new QNetworkAccessManager(this);
+	} else {
+		if (_settings.isWhitelist) {
+			// Just to be safe
+			_settings.urlList << _settings.url;
+		}
+		nam = new SlxNetworkAccessManager(urlListToRegExp(_settings.urlList), _settings.isWhitelist);
+	}
 	connect(nam, SIGNAL(sslErrors(QNetworkReply*, const QList<QSslError>&)),
 			this, SLOT(sslErrors(QNetworkReply*, const QList<QSslError>&)));
 	connect(nam, SIGNAL(finished(QNetworkReply*)), this, SLOT(requestFinished(QNetworkReply*)));
 	page->setNetworkAccessManager(nam);
-	page->mainFrame()->load(url);
+	page->mainFrame()->load(_settings.url);
 	//
 	_browser->show();
 }
@@ -57,8 +71,8 @@ SLXbrowser::~SLXbrowser()
 void SLXbrowser::loadStarted()
 {
 	_reset.stop();
-	if (_reloadIntervalMs > 0) {
-		_reset.start(_reloadIntervalMs + 10000);
+	if (_settings.reloadInterval > 0) {
+		_reset.start(_settings.reloadInterval + 10000);
 	}
 	_normalError.clear();
 	_sslErrors.clear();
@@ -69,37 +83,63 @@ void SLXbrowser::loadStarted()
 void SLXbrowser::loadFinished(bool ok)
 {
 	_progress->hide();
-    if (!_browser->wasAbortedDownload() && !_unsupportedUri && !ok) {
-		_browser->page()->mainFrame()->setHtml("<html><body style='background:blue;color:white'><br><br><center><h1>Page Load Error</h1><pre id='content'></pre></body></html>");
-		QWebElement el = _browser->page()->mainFrame()->documentElement().findFirst("#content");
-		QString str;
-		if (!_sslErrors.empty()) {
-			str.append("SSL Errors:\n");
-			for (QSslError err : _sslErrors) {
-				str.append(err.errorString());
-				str.append('\n');
+	qint64 now = QDateTime::currentMSecsSinceEpoch();
+	if (!_browser->wasAbortedDownload() && !ok) {
+
+		if (_unsupportedUri) {
+			QMessageBox::warning(this, QString::fromUtf8("Denied"),
+					QString::fromUtf8("This URL type is not supported.\n\n"
+							"Diese Art Link wird nicht unterstützt.\n\n"
+							"(z.B. Mail)"));
+		} else if (_blockedSite) {
+			int numElems = _browser->page()->mainFrame()->documentElement().findAll("*").count();
+			if (now - _lastPageLoad < 500 && numElems < 15) {
+				// Was probably a JS redirect, go back before displaying the error
+				_browser->page()->triggerAction(QWebPage::Back);
 			}
-		} else if (_normalError.length() != 0) {
-			str.append("Load Error:\n");
-			str.append(_normalError);
+			QTimer::singleShot(5, [=]() {
+				QMessageBox::warning(this, QString::fromUtf8("Denied"),
+						QString::fromUtf8("Target URL not allowed.\n\n"
+								"Dieser Link führt auf eine nicht erlaubte Seite."));
+			});
+			/*
+			_browser->page()->mainFrame()->setHtml("<html><body style='background:white;color:black'><br><br>"
+												   "<center><h1>Zugriff verweigert</h1>"
+												   "<p>Diese URL wurde gefiltert.<br>"
+												   "<a href='javascript:window.history.back()'>Zurück</a></p></center>"
+												   "</body></html>");
+			*/
 		} else {
-			str.append("Unknown Error");
+			_browser->page()->mainFrame()->setHtml("<html><body style='background:blue;color:white'><br><br>"
+												   "<center><h1>Page Load Error</h1><pre id='content'></pre></center>"
+												   "</body></html>");
+			QWebElement el = _browser->page()->mainFrame()->documentElement().findFirst("#content");
+			QString str;
+			if (!_sslErrors.empty()) {
+				str.append("SSL Errors:\n");
+				for (QSslError err : _sslErrors) {
+					str.append(err.errorString());
+					str.append('\n');
+				}
+			} else if (_normalError.length() != 0) {
+				str.append("Load Error:\n");
+				str.append(_normalError);
+			} else {
+				str.append("Unknown Error");
+			}
+			str.append("\n\n\n\n" + QDateTime::currentDateTime().toString());
+			el.setPlainText(str);
 		}
-		str.append("\n\n\n\n" + QDateTime::currentDateTime().toString());
-		el.setPlainText(str);
 		_sslErrors.clear();
 		_normalError.clear();
-		_reset.start(30000);
-	} else if (_reloadIntervalMs > 0) {
-        _reset.start(_reloadIntervalMs);
+		_reset.start(qMin(30000, _settings.reloadInterval > 0 ? _settings.reloadInterval : 30000));
+	} else if (_settings.reloadInterval > 0) {
+		_reset.start(_settings.reloadInterval);
 	}
-    if (_unsupportedUri && !ok) {
-        QMessageBox::warning(this, QString::fromUtf8("Denied"),
-                QString::fromUtf8("This URL type is not supported.\n\n"
-                        "Diese Art Link wird nicht unterstützt.\n\n"
-                        "(z.B. Mail)"));
-    }
+
     _unsupportedUri = false;
+	_blockedSite = false;
+	_lastPageLoad = now;
 }
 
 void SLXbrowser::loadProgress(int progress)
@@ -109,7 +149,7 @@ void SLXbrowser::loadProgress(int progress)
 
 void SLXbrowser::sslErrors(QNetworkReply* reply, const QList<QSslError>& errors)
 {
-	if (_ignoreSslErrors) {
+	if (_settings.ignoreSslErrors) {
 		reply->ignoreSslErrors();
 		return;
 	}
@@ -118,17 +158,48 @@ void SLXbrowser::sslErrors(QNetworkReply* reply, const QList<QSslError>& errors)
 
 void SLXbrowser::requestFinished(QNetworkReply *reply)
 {
-    if (reply->error() == QNetworkReply::NoError) {
-        _unsupportedUri = false;
-    } else if (reply->error() == QNetworkReply::ProtocolUnknownError) {
+	if (reply->error() == QNetworkReply::ProtocolUnknownError) {
         _unsupportedUri = true;
+	} else if (reply->error() == QNetworkReply::UnknownNetworkError) {
+		_blockedSite = true;
     } else {
-        qDebug() << reply->error();
+		//qDebug() << reply->error();
 		_normalError = reply->errorString();
 	}
 }
 
 void SLXbrowser::reloadInitial()
 {
-	_browser->page()->mainFrame()->load(_url);
+	_browser->page()->mainFrame()->load(_settings.url);
+}
+
+static QRegularExpression urlListToRegExp(const QStringList &list)
+{
+	// We search in the escaped string, so actually look for \*\* and \*
+	// Capture char before that because it must not be another backslash, as that
+	// means the star was already escaped in the list.
+	// Since these are C strings there are some additional backslashes here.
+	static const QRegularExpression STARSTAR("(^|[^\\\\])\\\\\\*\\\\\\*");
+	static const QRegularExpression STAR("(^|[^\\\\])\\\\\\*");
+	static const QRegularExpression QUEST("(^|[^\\\\])\\\\\\?");
+	static const QString STARSTAR_REP("\\1.*");
+	static const QString STAR_REP("\\1[^/]*");
+	static const QString QUEST_REP("\\1.?");
+	QStringList regexes; // All my regex's live in Regtexas
+	for (const QString &str : list) {
+		QString mangled;
+		if (str.contains(QLatin1String("//"))) {
+				mangled = str;
+		} else if (str.contains(QLatin1Char('/')) || str.contains(QLatin1String("**"))) {
+			mangled = "*//" + str;
+		} else {
+			mangled = "*//" + str + "/**";
+		}
+		mangled = QRegularExpression::escape(mangled);
+		mangled = mangled.replace(STARSTAR, STARSTAR_REP).replace(STAR, STAR_REP)
+				.replace(QUEST, QUEST_REP);
+		regexes << mangled;
+	}
+	qDebug() << regexes;
+	return QRegularExpression("^(" + regexes.join('|') + ")$");
 }
diff --git a/src/slxbrowser.h b/src/slxbrowser.h
index c036821..f68e186 100644
--- a/src/slxbrowser.h
+++ b/src/slxbrowser.h
@@ -11,11 +11,21 @@ class WebView;
 class QProgressBar;
 class QNetworkReply;
 
+struct BrowserSettings
+{
+	QString url;
+	bool fullscreen;
+	bool ignoreSslErrors;
+	int reloadInterval;
+	bool isWhitelist;
+	QStringList urlList;
+};
+
 class SLXbrowser : public QMainWindow
 {
 	Q_OBJECT
 public:
-	SLXbrowser(QString url, bool fullscreen, bool ignoreSslErrors, int reloadInterval);
+	SLXbrowser(BrowserSettings settings);
 	virtual ~SLXbrowser();
 
 private slots:
@@ -27,15 +37,15 @@ private slots:
 	void reloadInitial();
 
 private:
-	QString _url;
-	bool _ignoreSslErrors;
+	BrowserSettings _settings;
     bool _unsupportedUri;
+	bool _blockedSite;
 	WebView *_browser;
 	QProgressBar *_progress;
 	QTimer _reset;
 	QList<QSslError> _sslErrors;
 	QString _normalError;
-	int _reloadIntervalMs;
+	qint64 _lastPageLoad;
 };
 
 #endif /* SLXBROWSER_H_ */