11 files changed, 103 insertions, 15 deletions
diff --git a/Swiften/Parser/ExpatParser.cpp b/Swiften/Parser/ExpatParser.cpp
index a50949b..640d561 100644
--- a/Swiften/Parser/ExpatParser.cpp
+++ b/Swiften/Parser/ExpatParser.cpp
@@ -66,27 +66,43 @@ static void handleXMLDeclaration(void*, const XML_Char*, const XML_Char*, int) {
 
 static void handleNamespaceDeclaration(void* parser, const XML_Char* prefix, const XML_Char* uri) {
     static_cast<XMLParser*>(parser)->getClient()->handleNamespaceDeclaration(std::string(prefix ? prefix : ""), std::string(uri ? uri : ""));
 }
 
 static void handleEntityDeclaration(void* parser, const XML_Char*, int, const XML_Char*, int, const XML_Char*, const XML_Char*, const XML_Char*, const XML_Char*) {
     static_cast<ExpatParser*>(parser)->stopParser();
 }
 
+static void handleComment(void* parser, const XML_Char* /*data*/) {
+    if (!static_cast<ExpatParser*>(parser)->allowsComments()) {
+        static_cast<ExpatParser*>(parser)->stopParser();
+    }
+}
+
+static void handleProcessingInstruction(void* parser, const XML_Char* /*target*/, const XML_Char* /*data*/) {
+    static_cast<ExpatParser*>(parser)->stopParser();
+}
+
+static void handleDoctypeDeclaration(void* parser, const XML_Char* /*doctypeName*/, const XML_Char* /*sysid*/, const XML_Char* /*pubid*/, int /*has_internal_subset*/) {
+    static_cast<ExpatParser*>(parser)->stopParser();
+}
 
-ExpatParser::ExpatParser(XMLParserClient* client) : XMLParser(client), p(new Private()) {
+ExpatParser::ExpatParser(XMLParserClient* client, bool allowComments) : XMLParser(client, allowComments), p(new Private()) {
     p->parser_ = XML_ParserCreateNS("UTF-8", NAMESPACE_SEPARATOR);
     XML_SetUserData(p->parser_, this);
     XML_SetElementHandler(p->parser_, handleStartElement, handleEndElement);
     XML_SetCharacterDataHandler(p->parser_, handleCharacterData);
     XML_SetXmlDeclHandler(p->parser_, handleXMLDeclaration);
     XML_SetEntityDeclHandler(p->parser_, handleEntityDeclaration);
     XML_SetNamespaceDeclHandler(p->parser_, handleNamespaceDeclaration, nullptr);
+    XML_SetCommentHandler(p->parser_, handleComment);
+    XML_SetProcessingInstructionHandler(p->parser_, handleProcessingInstruction);
+    XML_SetDoctypeDeclHandler(p->parser_, handleDoctypeDeclaration, nullptr);
 }
 
 ExpatParser::~ExpatParser() {
     XML_ParserFree(p->parser_);
 }
 
 bool ExpatParser::parse(const std::string& data, bool finalData) {
     if (data.size() > std::numeric_limits<int>::max()) {
         return false;
diff --git a/Swiften/Parser/ExpatParser.h b/Swiften/Parser/ExpatParser.h
index 7583339..34d790d 100644
--- a/Swiften/Parser/ExpatParser.h
+++ b/Swiften/Parser/ExpatParser.h
@@ -10,19 +10,19 @@
 
 #include <boost/noncopyable.hpp>
 
 #include <Swiften/Base/API.h>
 #include <Swiften/Parser/XMLParser.h>
 
 namespace Swift {
     class SWIFTEN_API ExpatParser : public XMLParser, public boost::noncopyable {
         public:
-            ExpatParser(XMLParserClient* client);
+            ExpatParser(XMLParserClient* client, bool allowComments = false);
             ~ExpatParser();
 
             bool parse(const std::string& data, bool finalData = false);
 
             void stopParser();
 
         private:
             struct Private;
             const std::unique_ptr<Private> p;
diff --git a/Swiften/Parser/LibXMLParser.cpp b/Swiften/Parser/LibXMLParser.cpp
index 192f44b..4e02059 100644
--- a/Swiften/Parser/LibXMLParser.cpp
+++ b/Swiften/Parser/LibXMLParser.cpp
@@ -53,46 +53,68 @@ static void handleStartElement(void* parser, const xmlChar* name, const xmlChar*
 static void handleEndElement(void *parser, const xmlChar* name, const xmlChar*, const xmlChar* xmlns) {
     static_cast<XMLParser*>(parser)->getClient()->handleEndElement(reinterpret_cast<const char*>(name), (xmlns ? reinterpret_cast<const char*>(xmlns) : std::string()));
 }
 
 static void handleCharacterData(void* parser, const xmlChar* data, int len) {
     assert(len >= 0);
     static_cast<XMLParser*>(parser)->getClient()->handleCharacterData(std::string(reinterpret_cast<const char*>(data), static_cast<size_t>(len)));
 }
 
+static void handleComment(void* parser, const xmlChar* /*data*/) {
+    if (!static_cast<LibXMLParser*>(parser)->allowsComments()) {
+        static_cast<LibXMLParser*>(parser)->stopParser();
+    }
+}
+
+static void handleEntityDeclaration(void * parser, const xmlChar* /*name*/, int /*type*/, const xmlChar* /*publicId*/, const xmlChar* /*systemId*/, xmlChar* /*content*/) {
+    static_cast<LibXMLParser*>(parser)->stopParser();
+}
+
+static void handleProcessingInstruction(void* parser, const xmlChar* /*target*/, const xmlChar* /*data*/) {
+    static_cast<LibXMLParser*>(parser)->stopParser();
+}
+
+static void handleExternalSubset(void* parser, const xmlChar * /*name*/, const xmlChar * /*ExternalID*/, const xmlChar * /*SystemID*/) {
+    static_cast<LibXMLParser*>(parser)->stopParser();
+}
+
 static void handleError(void*, const char* /*m*/, ... ) {
     /*
     va_list args;
     va_start(args, m);
     vfprintf(stdout, m, args);
     va_end(args);
     */
 }
 
 static void handleWarning(void*, const char*, ... ) {
 }
 
 bool LibXMLParser::initialized = false;
 
-LibXMLParser::LibXMLParser(XMLParserClient* client) : XMLParser(client), p(new Private()) {
+LibXMLParser::LibXMLParser(XMLParserClient* client, bool allowComments) : XMLParser(client, allowComments), p(new Private()) {
     // Initialize libXML for multithreaded applications
     if (!initialized) {
         xmlInitParser();
         initialized = true;
     }
 
     memset(&p->handler_, 0, sizeof(p->handler_) );
     p->handler_.initialized = XML_SAX2_MAGIC;
     p->handler_.startElementNs = &handleStartElement;
     p->handler_.endElementNs = &handleEndElement;
     p->handler_.characters = &handleCharacterData;
     p->handler_.warning = &handleWarning;
     p->handler_.error = &handleError;
+    p->handler_.comment = &handleComment;
+    p->handler_.entityDecl = &handleEntityDeclaration;
+    p->handler_.processingInstruction = &handleProcessingInstruction;
+    p->handler_.externalSubset = &handleExternalSubset;
 
     p->context_ = xmlCreatePushParserCtxt(&p->handler_, this, nullptr, 0, nullptr);
     xmlCtxtUseOptions(p->context_, XML_PARSE_NOENT);
     assert(p->context_);
 }
 
 LibXMLParser::~LibXMLParser() {
     if (p->context_) {
         xmlFreeParserCtxt(p->context_);
@@ -100,19 +122,25 @@ LibXMLParser::~LibXMLParser() {
 }
 
 bool LibXMLParser::parse(const std::string& data, bool finalData) {
     if (data.size() > std::numeric_limits<int>::max()) {
         return false;
     }
     if (xmlParseChunk(p->context_, data.c_str(), static_cast<int>(data.size()), finalData) == XML_ERR_OK) {
         return true;
     }
+    if (stopped_) return false;
     xmlError* error = xmlCtxtGetLastError(p->context_);
     if (error->code == XML_WAR_NS_URI || error->code == XML_WAR_NS_URI_RELATIVE) {
         xmlCtxtResetLastError(p->context_);
         p->context_->errNo = XML_ERR_OK;
         return true;
     }
     return false;
 }
 
+void LibXMLParser::stopParser() {
+    stopped_ = true;
+    xmlStopParser(p->context_);
+}
+
 }
diff --git a/Swiften/Parser/LibXMLParser.h b/Swiften/Parser/LibXMLParser.h
index a863867..e21770d 100644
--- a/Swiften/Parser/LibXMLParser.h
+++ b/Swiften/Parser/LibXMLParser.h
@@ -13,21 +13,24 @@
 #include <Swiften/Parser/XMLParser.h>
 
 namespace Swift {
     /**
      * Warning: This constructor is not thread-safe, because it depends on global state to
      * check whether it is initialized.
      */
     class LibXMLParser : public XMLParser, public boost::noncopyable {
         public:
-            LibXMLParser(XMLParserClient* client);
+            LibXMLParser(XMLParserClient* client, bool allowComments = false);
             virtual ~LibXMLParser();
 
             bool parse(const std::string& data, bool finalData = false);
 
+            void stopParser();
+
         private:
             static bool initialized;
+            bool stopped_ = false;
 
             struct Private;
             const std::unique_ptr<Private> p;
     };
 }
diff --git a/Swiften/Parser/PayloadParsers/UnitTest/PayloadsParserTester.h b/Swiften/Parser/PayloadParsers/UnitTest/PayloadsParserTester.h
index dcdbffa..8f9e0e1 100644
--- a/Swiften/Parser/PayloadParsers/UnitTest/PayloadsParserTester.h
+++ b/Swiften/Parser/PayloadParsers/UnitTest/PayloadsParserTester.h
@@ -13,19 +13,19 @@
 #include <Swiften/Parser/PayloadParsers/FullPayloadParserFactoryCollection.h>
 #include <Swiften/Parser/PlatformXMLParserFactory.h>
 #include <Swiften/Parser/XMLParser.h>
 #include <Swiften/Parser/XMLParserClient.h>
 
 namespace Swift {
     class PayloadsParserTester : public XMLParserClient {
         public:
             PayloadsParserTester() : level(0) {
-                xmlParser = PlatformXMLParserFactory().createXMLParser(this);
+                xmlParser = PlatformXMLParserFactory().createXMLParser(this, false);
             }
 
             bool parse(const std::string& data) {
                 return xmlParser->parse(data);
             }
 
             virtual void handleStartElement(const std::string& element, const std::string& ns, const AttributeMap& attributes) {
                 if (level == 0) {
                     assert(!payloadParser.get());
diff --git a/Swiften/Parser/PlatformXMLParserFactory.cpp b/Swiften/Parser/PlatformXMLParserFactory.cpp
index bf66734..a424aca 100644
--- a/Swiften/Parser/PlatformXMLParserFactory.cpp
+++ b/Swiften/Parser/PlatformXMLParserFactory.cpp
@@ -14,18 +14,18 @@
 #include <Swiften/Parser/ExpatParser.h>
 #endif
 
 
 namespace Swift {
 
 PlatformXMLParserFactory::PlatformXMLParserFactory() {
 }
 
-std::unique_ptr<XMLParser> PlatformXMLParserFactory::createXMLParser(XMLParserClient* client) {
+std::unique_ptr<XMLParser> PlatformXMLParserFactory::createXMLParser(XMLParserClient* client, bool allowComments) {
 #ifdef HAVE_LIBXML
-    return std::make_unique<LibXMLParser>(client);
+    return std::make_unique<LibXMLParser>(client, allowComments);
 #else
-    return std::make_unique<ExpatParser>(client);
+    return std::make_unique<ExpatParser>(client, allowComments);
 #endif
 }
 
 }
diff --git a/Swiften/Parser/PlatformXMLParserFactory.h b/Swiften/Parser/PlatformXMLParserFactory.h
index fa3ca19..d72a513 100644
--- a/Swiften/Parser/PlatformXMLParserFactory.h
+++ b/Swiften/Parser/PlatformXMLParserFactory.h
@@ -8,12 +8,12 @@
 
 #include <Swiften/Base/API.h>
 #include <Swiften/Parser/XMLParserFactory.h>
 
 namespace Swift {
     class SWIFTEN_API PlatformXMLParserFactory : public XMLParserFactory {
         public:
             PlatformXMLParserFactory();
 
-            virtual std::unique_ptr<XMLParser> createXMLParser(XMLParserClient*);
+            virtual std::unique_ptr<XMLParser> createXMLParser(XMLParserClient*, bool allowComments = false);
     };
 }
diff --git a/Swiften/Parser/UnitTest/XMLParserTest.cpp b/Swiften/Parser/UnitTest/XMLParserTest.cpp
index c026b4b..63d30ea 100644
--- a/Swiften/Parser/UnitTest/XMLParserTest.cpp
+++ b/Swiften/Parser/UnitTest/XMLParserTest.cpp
@@ -33,18 +33,22 @@ class XMLParserTest : public CppUnit::TestFixture {
         CPPUNIT_TEST(testParse_InErrorState);
         CPPUNIT_TEST(testParse_Incremental);
         CPPUNIT_TEST(testParse_WhitespaceInAttribute);
         CPPUNIT_TEST(testParse_AttributeWithoutNamespace);
         CPPUNIT_TEST(testParse_AttributeWithNamespace);
         CPPUNIT_TEST(testParse_BillionLaughs);
         CPPUNIT_TEST(testParse_InternalEntity);
         //CPPUNIT_TEST(testParse_UndefinedPrefix);
         //CPPUNIT_TEST(testParse_UndefinedAttributePrefix);
+        CPPUNIT_TEST(testParse_AllowCommentsInXML);
+        CPPUNIT_TEST(testParse_DisallowCommentsInXML);
+        CPPUNIT_TEST(testParse_Doctype);
+        CPPUNIT_TEST(testParse_ProcessingInstructions);
         CPPUNIT_TEST_SUITE_END();
 
     public:
         void testParse_NestedElements() {
             ParserType testling(&client_);
 
             CPPUNIT_ASSERT(testling.parse(
                 "<iq type=\"get\">"
                 "<query xmlns='jabber:iq:version'/>"
@@ -179,19 +183,19 @@ class XMLParserTest : public CppUnit::TestFixture {
             CPPUNIT_ASSERT_EQUAL(std::string("y"), client_.events[2].data);
             CPPUNIT_ASSERT_EQUAL(std::string("bla"), client_.events[2].ns);
 
             CPPUNIT_ASSERT_EQUAL(Client::EndElement, client_.events[3].type);
             CPPUNIT_ASSERT_EQUAL(std::string("x"), client_.events[3].data);
             CPPUNIT_ASSERT_EQUAL(std::string("bla"), client_.events[3].ns);
         }
 
         void testParse_UnhandledXML() {
-            ParserType testling(&client_);
+            ParserType testling(&client_, true);
 
             CPPUNIT_ASSERT(testling.parse("<iq><!-- Testing --></iq>"));
 
             CPPUNIT_ASSERT_EQUAL(static_cast<size_t>(2), client_.events.size());
 
             CPPUNIT_ASSERT_EQUAL(Client::StartElement, client_.events[0].type);
             CPPUNIT_ASSERT_EQUAL(std::string("iq"), client_.events[0].data);
 
             CPPUNIT_ASSERT_EQUAL(Client::EndElement, client_.events[1].type);
@@ -325,25 +329,56 @@ class XMLParserTest : public CppUnit::TestFixture {
 
             CPPUNIT_ASSERT_EQUAL(Client::EndElement, client_.events[3].type);
             CPPUNIT_ASSERT_EQUAL(std::string("foo:bar"), client_.events[3].data);
             CPPUNIT_ASSERT_EQUAL(std::string(""), client_.events[3].ns);
         }
 
         void testParse_UndefinedAttributePrefix() {
             ParserType testling(&client_);
 
-            CPPUNIT_ASSERT(testling.parse(
-                "<foo bar:baz='bla'/>"));
+            CPPUNIT_ASSERT(testling.parse("<foo bar:baz='bla'/>"));
 
             CPPUNIT_ASSERT_EQUAL(static_cast<size_t>(1), client_.events[0].attributes.getEntries().size());
             CPPUNIT_ASSERT_EQUAL(std::string("bar:baz"), client_.events[0].attributes.getEntries()[0].getAttribute().getName());
         }
 
+        void testParse_AllowCommentsInXML() {
+            ParserType testling(&client_, true);
+
+            CPPUNIT_ASSERT(testling.parse("<message><!-- Some More Comments Testing --></message>"));
+
+            CPPUNIT_ASSERT_EQUAL(static_cast<size_t>(2), client_.events.size());
+
+            CPPUNIT_ASSERT_EQUAL(Client::StartElement, client_.events[0].type);
+            CPPUNIT_ASSERT_EQUAL(std::string("message"), client_.events[0].data);
+
+            CPPUNIT_ASSERT_EQUAL(Client::EndElement, client_.events[1].type);
+            CPPUNIT_ASSERT_EQUAL(std::string("message"), client_.events[1].data);
+        }
+
+        void testParse_DisallowCommentsInXML() {
+            ParserType testling(&client_);
+
+            CPPUNIT_ASSERT(!testling.parse("<message><!-- Some More Comments Testing --></message>"));
+        }
+
+        void testParse_Doctype() {
+            ParserType testling(&client_);
+
+            CPPUNIT_ASSERT(!testling.parse("<!DOCTYPE greeting SYSTEM \"hello.dtd\">"));
+        }
+
+        void testParse_ProcessingInstructions() {
+            ParserType testling(&client_);
+
+            CPPUNIT_ASSERT(!testling.parse("<?xml-stylesheet type=\"text/xsl\" href=\"Sample.xsl\"?>"));
+        }
+
     private:
         class Client : public XMLParserClient {
             public:
                 using NamespaceMap = std::unordered_map<std::string /* prefix */, std::string /* uri */>;
                 enum Type { StartElement, EndElement, CharacterData, NamespaceDefined };
                 struct Event {
                     Event(
                             Type type,
                             const std::string& data,
@@ -372,18 +407,19 @@ class XMLParserTest : public CppUnit::TestFixture {
                 }
 
                 void handleCharacterData(const std::string& data) override {
                     events.push_back(Event(CharacterData, data));
                 }
 
                 void handleNamespaceDeclaration(const std::string& prefix, const std::string& uri) override {
                     namespaces_[prefix] = uri;
                 }
+
                 std::vector<Event> events;
             private:
                 NamespaceMap namespaces_;
         } client_;
 };
 
 #ifdef HAVE_EXPAT
 CPPUNIT_TEST_SUITE_REGISTRATION(XMLParserTest<ExpatParser>);
 #endif
diff --git a/Swiften/Parser/XMLParser.cpp b/Swiften/Parser/XMLParser.cpp
index 8e92fe4..8a0799f 100644
--- a/Swiften/Parser/XMLParser.cpp
+++ b/Swiften/Parser/XMLParser.cpp
@@ -2,16 +2,16 @@
  * Copyright (c) 2010 Isode Limited.
  * All rights reserved.
  * See the COPYING file for more information.
  */
 
 #include <Swiften/Parser/XMLParser.h>
 
 namespace Swift {
 
-XMLParser::XMLParser(XMLParserClient* client) : client_(client) {
+XMLParser::XMLParser(XMLParserClient* client, bool allowComments) : client_(client), allowComments_(allowComments){
 }
 
 XMLParser::~XMLParser() {
 }
 
 }
diff --git a/Swiften/Parser/XMLParser.h b/Swiften/Parser/XMLParser.h
index ad79b2d..3b09d22 100644
--- a/Swiften/Parser/XMLParser.h
+++ b/Swiften/Parser/XMLParser.h
@@ -9,22 +9,27 @@
 #include <string>
 
 #include <Swiften/Base/API.h>
 
 namespace Swift {
     class XMLParserClient;
 
     class SWIFTEN_API XMLParser {
         public:
-            XMLParser(XMLParserClient* client);
+            XMLParser(XMLParserClient* client, bool allowComments = false);
             virtual ~XMLParser();
 
             virtual bool parse(const std::string& data, bool finalData = false) = 0;
 
             XMLParserClient* getClient() const {
                 return client_;
             }
 
+            bool allowsComments() const {
+                return allowComments_;
+            }
+
         private:
             XMLParserClient* client_;
+            const bool allowComments_ = false;
     };
 }
diff --git a/Swiften/Parser/XMLParserFactory.h b/Swiften/Parser/XMLParserFactory.h
index 595512b..ae3c90e 100644
--- a/Swiften/Parser/XMLParserFactory.h
+++ b/Swiften/Parser/XMLParserFactory.h
@@ -12,12 +12,12 @@
 
 namespace Swift {
     class XMLParser;
     class XMLParserClient;
 
     class SWIFTEN_API XMLParserFactory {
         public:
             virtual ~XMLParserFactory();
 
-            virtual std::unique_ptr<XMLParser> createXMLParser(XMLParserClient*) = 0;
+            virtual std::unique_ptr<XMLParser> createXMLParser(XMLParserClient*, bool allowComments = false) = 0;
     };
 }