Make internal linking more adhere to HTML spec.

see #4406

* Support lookup through decoded url fragement
* Escape special characters when building query selector

src/lib/pdfconverter.cc

@@ -39,6 +39,7 @@
 #include <io.h>
 #endif
 
+#include "utilities.hh"
 #include "dllbegin.inc"
 using namespace wkhtmltopdf;
 using namespace wkhtmltopdf::settings;
@@ -560,18 +561,25 @@ void PdfConverterPrivate::findLinks(QWebFrame * frame, QVector<QPair<QWebElement
 			if (href.isEmpty()) continue;
 			href=frame->baseUrl().resolved(href);
 			QString key = QUrl::fromPercentEncoding(href.toString(QUrl::RemoveFragment).toLocal8Bit());
+			QString decodedFragement = QUrl::fromPercentEncoding(href.fragment().toLocal8Bit());
+
+			QString escapedFragement = escapeCSS(href.fragment());
+			QString escapedDecodedFragement = escapeCSS(decodedFragement);
+
 			if (urlToPageObj.contains(key)) {
 				if (ulocal) {
 					PageObject * p = urlToPageObj[key];
 					QWebElement e;
 					if (!href.hasFragment())
 						e = p->page->mainFrame()->findFirstElement("body");
 					else {
-						e = p->page->mainFrame()->findFirstElement("a[name=\""+href.fragment()+"\"]");
+						e = p->page->mainFrame()->findFirstElement("#"+escapedFragement);
+						if (e.isNull())
+							e = p->page->mainFrame()->findFirstElement("a[name=\""+escapedFragement+"\"]");
 						if (e.isNull())
-							e = p->page->mainFrame()->findFirstElement("*[id=\""+href.fragment()+"\"]");
+							e = p->page->mainFrame()->findFirstElement("#"+escapedDecodedFragement);
 						if (e.isNull())
-							e = p->page->mainFrame()->findFirstElement("*[name=\""+href.fragment()+"\"]");
+							e = p->page->mainFrame()->findFirstElement("a[name=\""+escapedDecodedFragement+"\"]");
 					}
 					if (!e.isNull()) {
 						p->anchors[href.toString()] = e;

src/lib/utilities.cc

@@ -25,6 +25,9 @@
 #include <QMetaEnum>
 #include <QNetworkReply>
 
+#define CSS_ESCAPE_CHARACTER(character) (QString("\\") + character)
+#define CSS_ESCAPE_UNICODE_AS_CODEPOINT(unicode) QString("\\%1").arg(unicode, 0, 16)
+
 void loadSvg(QSvgRenderer * & ptr, const QString & path, const char * def, int w, int h) {
 	 delete ptr;
 	 ptr = 0;
@@ -179,6 +182,48 @@ int handleError(bool success, int errorCode) {
 	return success?EXIT_SUCCESS:EXIT_FAILURE;
 }
 
+// Escape a CSS selector as per CSS.escape() in CSS Object Model spec
+// see https://www.w3.org/TR/cssom-1/#the-css.escape()-method
+QString escapeCSS(QString input) {
+	QString output;
+	QTextStream outputStream(&output);
+
+	if (input == "-")
+		return QString("\\-");
+
+	for (int i=0; i<input.length(); i++) {
+		QChar character = input.at(i);
+		ushort unicode = character.unicode();
+
+		if (unicode == 0)
+			outputStream << QChar(0xFFFD);
+
+		else if (
+			unicode < 0x1F || unicode == 0x7F ||
+			(
+				(i == 0 || (i == 1 && input[0].unicode() == 0x2d)) &&
+				unicode >= 0x30 && unicode <= 0x39
+			)
+		)
+			outputStream << CSS_ESCAPE_UNICODE_AS_CODEPOINT(character);
+
+		else if (
+			unicode >= 0x80 || unicode == 0x2D || unicode == 0x5F ||
+			(unicode >= 0x30 && unicode <= 0x39) ||
+			(unicode >= 0x41 && unicode <= 0x5A) ||
+			(unicode >= 0x61 && unicode <= 0x7A)
+		)
+			outputStream << character;
+
+		else
+			outputStream << CSS_ESCAPE_CHARACTER(character);
+
+	}
+
+	return output;
+}
+
+
 QSvgRenderer * MyLooksStyle::checkbox = 0;
 QSvgRenderer * MyLooksStyle::checkbox_checked = 0;
 QSvgRenderer * MyLooksStyle::radiobutton = 0;

src/lib/utilities.hh

@@ -56,5 +56,7 @@ public slots:
 
 DLL_PUBLIC int handleError(bool success, int errorCode);
 
+QString escapeCSS(QString input);
+
 #include <dllend.inc>
 #endif //__UTILITIES_HH__