make RDF parser inherit from FeedParser, enhance ATOM and RDF detection reliability

src/librssguard/services/standard/atomparser.cpp

@@ -20,8 +20,6 @@ AtomParser::AtomParser(const QString& data) : FeedParser(data) {
   }
 }
 
-AtomParser::~AtomParser() = default;
-
 QString AtomParser::feedAuthor() const {
   QDomNodeList top_level_nodes = m_xml.documentElement().childNodes();
   QStringList author_str;
@@ -148,6 +146,10 @@ QString AtomParser::messageAuthor(const QDomElement& msg_element) const {
   return author_str.join(", ");
 }
 
+QString AtomParser::atomNamespace() const {
+  return m_atomNamespace;
+}
+
 QDomNodeList AtomParser::messageElements() {
   return m_xml.elementsByTagNameNS(m_atomNamespace, QSL("entry"));
 }

src/librssguard/services/standard/atomparser.h

@@ -13,7 +13,8 @@
 class AtomParser : public FeedParser {
   public:
     explicit AtomParser(const QString& data);
-    virtual ~AtomParser();
+
+    QString atomNamespace() const;
 
   private:
     QDomNodeList messageElements();

src/librssguard/services/standard/feedparser.cpp

@@ -19,8 +19,6 @@ FeedParser::FeedParser(QString data) : m_xmlData(std::move(data)), m_mrssNamespa
   }
 }
 
-FeedParser::~FeedParser() = default;
-
 QList<Message> FeedParser::messages() {
   QString feed_author = feedAuthor();
   QList<Message> messages;

src/librssguard/services/standard/feedparser.h

@@ -12,7 +12,6 @@
 class FeedParser {
   public:
     explicit FeedParser(QString data);
-    virtual ~FeedParser();
 
     virtual QList<Message> messages();
 

src/librssguard/services/standard/rdfparser.cpp

@@ -2,6 +2,7 @@
 
 #include "services/standard/rdfparser.h"
 
+#include "exceptions/applicationexception.h"
 #include "miscellaneous/application.h"
 #include "miscellaneous/textfactory.h"
 #include "network-web/webfactory.h"
@@ -9,33 +10,32 @@
 
 #include <QDomDocument>
 
-RdfParser::RdfParser() = default;
+RdfParser::RdfParser(const QString& data)
+  : FeedParser(data),
+  m_rdfNamespace(QSL("http://www.w3.org/1999/02/22-rdf-syntax-ns#")),
+  m_rssNamespace(QSL("http://purl.org/rss/1.0/")) {}
 
-RdfParser::~RdfParser() = default;
+QDomNodeList RdfParser::messageElements() {
-
-QList<Message> RdfParser::parseXmlData(const QString& data) {
-  QList<Message> messages;
   QDomDocument xml_file;
-  QDateTime current_time = QDateTime::currentDateTime();
 
-  xml_file.setContent(data, true);
+  xml_file.setContent(m_xmlData, true);
 
   // Pull out all messages.
-  QDomNodeList messages_in_xml = xml_file.elementsByTagName(QSL("item"));
+  return xml_file.elementsByTagName(QSL("item"));
+}
 
-  for (int i = 0; i < messages_in_xml.size(); i++) {
+Message RdfParser::extractMessage(const QDomElement& msg_element, QDateTime current_time) const {
-    QDomNode message_item = messages_in_xml.item(i);
   Message new_message;
 
   // Deal with title and description.
-    QString elem_title = message_item.namedItem(QSL("title")).toElement().text().simplified();
+  QString elem_title = msg_element.namedItem(QSL("title")).toElement().text().simplified();
-    QString elem_description = message_item.namedItem(QSL("description")).toElement().text();
+  QString elem_description = rawXmlChild(msg_element.namedItem(QSL("description")).toElement());
 
   // Now we obtained maximum of information for title & description.
   if (elem_title.isEmpty()) {
     if (elem_description.isEmpty()) {
       // BOTH title and description are empty, skip this message.
-        continue;
+      throw ApplicationException(QSL("Not enough data for the message."));
     }
     else {
       // Title is empty but description is not.
@@ -54,18 +54,18 @@ QList<Message> RdfParser::parseXmlData(const QString& data) {
 
   str.setCodec(DEFAULT_FEED_ENCODING);
 
-    message_item.save(str, 0, QDomNode::EncodingPolicy::EncodingFromTextStream);
+  msg_element.save(str, 0, QDomNode::EncodingPolicy::EncodingFromTextStream);
   new_message.m_rawContents = raw_contents;
 
   // Deal with link and author.
-    new_message.m_url = message_item.namedItem(QSL("link")).toElement().text();
+  new_message.m_url = msg_element.namedItem(QSL("link")).toElement().text();
-    new_message.m_author = message_item.namedItem(QSL("creator")).toElement().text();
+  new_message.m_author = msg_element.namedItem(QSL("creator")).toElement().text();
 
   // Deal with creation date.
-    QString elem_updated = message_item.namedItem(QSL("date")).toElement().text();
+  QString elem_updated = msg_element.namedItem(QSL("date")).toElement().text();
 
   if (elem_updated.isEmpty()) {
-      elem_updated = message_item.namedItem(QSL("dc:date")).toElement().text();
+    elem_updated = msg_element.namedItem(QSL("dc:date")).toElement().text();
   }
 
   // Deal with creation date.
@@ -85,8 +85,13 @@ QList<Message> RdfParser::parseXmlData(const QString& data) {
     new_message.m_url = "";
   }
 
-    messages.append(new_message);
+  return new_message;
-  }
+}
-
+
-  return messages;
+QString RdfParser::rssNamespace() const {
+  return m_rssNamespace;
+}
+
+QString RdfParser::rdfNamespace() const {
+  return m_rdfNamespace;
 }

src/librssguard/services/standard/rdfparser.h

@@ -3,16 +3,25 @@
 #ifndef RDFPARSER_H
 #define RDFPARSER_H
 
+#include "services/standard/feedparser.h"
+
 #include "core/message.h"
 
 #include <QList>
 
-class RdfParser {
+class RdfParser : public FeedParser {
   public:
-    explicit RdfParser();
+    explicit RdfParser(const QString& data);
-    virtual ~RdfParser();
 
-    QList<Message> parseXmlData(const QString& data);
+    QString rdfNamespace() const;
+    QString rssNamespace() const;
+
+  private:
+    QDomNodeList messageElements();
+    Message extractMessage(const QDomElement& msg_element, QDateTime current_time) const;
+
+    QString m_rdfNamespace;
+    QString m_rssNamespace;
 };
 
 #endif // RDFPARSER_H

src/librssguard/services/standard/rssparser.cpp

@@ -14,8 +14,6 @@
 
 RssParser::RssParser(const QString& data) : FeedParser(data) {}
 
-RssParser::~RssParser() = default;
-
 QDomNodeList RssParser::messageElements() {
   QDomNode channel_elem = m_xml.namedItem(QSL("rss")).namedItem(QSL("channel"));
 

src/librssguard/services/standard/rssparser.h

@@ -12,7 +12,6 @@
 class RssParser : public FeedParser {
   public:
     explicit RssParser(const QString& data);
-    virtual ~RssParser();
 
   private:
     QDomNodeList messageElements();

src/librssguard/services/standard/standardfeed.cpp

@@ -352,6 +352,7 @@ StandardFeed* StandardFeed::guessFeed(StandardFeed::SourceType source_type,
     int error_line, error_column;
 
     if (!xml_document.setContent(xml_contents_encoded,
+                                 true,
                                  &error_msg,
                                  &error_line,
                                  &error_column)) {
@@ -362,23 +363,24 @@ StandardFeed* StandardFeed::guessFeed(StandardFeed::SourceType source_type,
     feed->setEncoding(encod);
 
     QDomElement root_element = xml_document.documentElement();
-    QString root_tag_name = root_element.tagName();
+    RdfParser rdf(QSL("<a/>"));
+    AtomParser atom(QSL("<a/>"));
 
-    if (root_tag_name == QL1S("rdf:RDF")) {
+    if (root_element.namespaceURI() == rdf.rdfNamespace()) {
       // We found RDF feed.
-      QDomElement channel_element = root_element.namedItem(QSL("channel")).toElement();
+      QDomElement channel_element = root_element.elementsByTagNameNS(rdf.rssNamespace(), QSL("channel")).at(0).toElement();
 
       feed->setType(Type::Rdf);
-      feed->setTitle(channel_element.namedItem(QSL("title")).toElement().text());
+      feed->setTitle(channel_element.elementsByTagNameNS(rdf.rssNamespace(), QSL("title")).at(0).toElement().text());
-      feed->setDescription(channel_element.namedItem(QSL("description")).toElement().text());
+      feed->setDescription(channel_element.elementsByTagNameNS(rdf.rssNamespace(), QSL("description")).at(0).toElement().text());
 
-      QString home_page = channel_element.namedItem(QSL("link")).toElement().text();
+      QString home_page = channel_element.elementsByTagNameNS(rdf.rssNamespace(), QSL("link")).at(0).toElement().text();
 
       if (!home_page.isEmpty()) {
         icon_possible_locations.prepend({ home_page, false });
       }
     }
-    else if (root_tag_name == QL1S("rss")) {
+    else if (root_element.tagName() == QL1S("rss")) {
       // We found RSS 0.91/0.92/0.93/2.0/2.0.1 feed.
       QString rss_type = root_element.attribute("version", "2.0");
 
@@ -410,7 +412,7 @@ StandardFeed* StandardFeed::guessFeed(StandardFeed::SourceType source_type,
         icon_possible_locations.prepend({ home_page, false });
       }
     }
-    else if (root_tag_name == QL1S("feed")) {
+    else if (root_element.namespaceURI() == atom.atomNamespace()) {
       // We found ATOM feed.
       feed->setType(Type::Atom10);
       feed->setTitle(root_element.namedItem(QSL("title")).toElement().text());

src/librssguard/services/standard/standardserviceroot.cpp

@@ -244,7 +244,7 @@ QList<Message> StandardServiceRoot::obtainNewMessages(const QList<Feed*>& feeds)
         break;
 
       case StandardFeed::Type::Rdf:
-        messages = RdfParser().parseXmlData(formatted_feed_contents);
+        messages = RdfParser(formatted_feed_contents).messages();
         break;
 
       case StandardFeed::Type::Atom10: