Use the Expat library directly instead of relying on expatpp

And now we handle namespaces, yay.
And a nice little test.

CMakeLists.txt

@@ -13,11 +13,14 @@ set(CMAKE_CXX_FLAGS_DEBUG "${CMAKE_CXX_FLAGS_DEBUG} -Og -fsanitize=address")
 #
 set(CMAKE_MODULE_PATH ${CMAKE_MODULE_PATH} "${CMAKE_SOURCE_DIR}/cmake/Modules/")
 find_package(Cryptopp REQUIRED)
+include(FindEXPAT)
+find_package(EXPAT REQUIRED)
 
 include_directories("src/")
 # the SYSTEM flag tells the compiler that we don't care about warnings
 # coming from these headers.
 include_directories(SYSTEM ${CRYPTO++_INCLUDE_DIR})
+include_directories(SYSTEM ${EXPAT_INCLUDE_DIRS})
 
 #
 ## utils
@@ -47,7 +50,8 @@ target_link_libraries(irc network utils)
 file(GLOB source_xmpp
   src/xmpp/*.[hc]pp)
 add_library(xmpp STATIC ${source_xmpp})
-target_link_libraries(xmpp bridge network utils ${CRYPTO++_LIBRARIES} expatpp)
+target_link_libraries(xmpp bridge network utils
+  ${CRYPTO++_LIBRARIES} ${EXPAT_LIBRARIES})
 
 #
 ## bridge

src/test.cpp

@@ -10,7 +10,7 @@
 #include <utils/encoding.hpp>
 #include <string.h>
 
-#include <fstream>
+#include <xmpp/xmpp_parser.hpp>
 
 int main()
 {
@@ -44,5 +44,23 @@ int main()
   std::string coucou("\u0002\u0002COUCOU\u0003");
   remove_irc_colors(coucou);
   assert(coucou == "COUCOU");
+
+  /**
+   * XML parsing
+   */
+  XmppParser xml;
+  const std::string doc = "<stream xmlns='stream_ns'><stanza b='c'>inner<child1/><child2 xmlns='child2_ns'/>tail</stanza></stream>";
+  xml.add_stanza_callback([](const Stanza& stanza)
+      {
+        assert(stanza.get_name() == "stream_ns:stanza");
+        assert(stanza["b"] == "c");
+        assert(stanza.get_inner() == "inner");
+        assert(stanza.get_tail() == "");
+        assert(stanza.get_child("stream_ns:child1") != nullptr);
+        assert(stanza.get_child("stream_ns:child2") == nullptr);
+        assert(stanza.get_child("child2_ns:child2") != nullptr);
+        assert(stanza.get_child("child2_ns:child2")->get_tail() == "tail");
+      });
+  xml.feed(doc.data(), doc.size(), true);
   return 0;
 }

src/xmpp/xmpp_component.cpp

@@ -10,6 +10,15 @@
 #include <hex.h>
 #include <sha.h>
 
+#define STREAM_NS        "http://etherx.jabber.org/streams"
+#define COMPONENT_NS     "jabber:component:accept"
+#define MUC_NS           "http://jabber.org/protocol/muc"
+#define MUC_USER_NS      MUC_NS"#user"
+#define DISCO_NS         "http://jabber.org/protocol/disco"
+#define DISCO_ITEMS_NS   DISCO_NS"#items"
+#define DISCO_INFO_NS    DISCO_NS"#info"
+
+
 XmppComponent::XmppComponent(const std::string& hostname, const std::string& secret):
   served_hostname(hostname),
   secret(secret),
@@ -21,11 +30,11 @@ XmppComponent::XmppComponent(const std::string& hostname, const std::string& sec
                                                   std::placeholders::_1));
   this->parser.add_stream_close_callback(std::bind(&XmppComponent::on_remote_stream_close, this,
                                                   std::placeholders::_1));
-  this->stanza_handlers.emplace("handshake",
+  this->stanza_handlers.emplace(COMPONENT_NS":handshake",
                                 std::bind(&XmppComponent::handle_handshake, this,std::placeholders::_1));
-  this->stanza_handlers.emplace("presence",
+  this->stanza_handlers.emplace(COMPONENT_NS":presence",
                                 std::bind(&XmppComponent::handle_presence, this,std::placeholders::_1));
-  this->stanza_handlers.emplace("message",
+  this->stanza_handlers.emplace(COMPONENT_NS":message",
                                 std::bind(&XmppComponent::handle_message, this,std::placeholders::_1));
 }
 
@@ -49,8 +58,8 @@ void XmppComponent::on_connected()
 {
   std::cout << "connected to XMPP server" << std::endl;
   XmlNode node("stream:stream", nullptr);
-  node["xmlns"] = "jabber:component:accept";
+  node["xmlns"] = COMPONENT_NS;
-  node["xmlns:stream"] = "http://etherx.jabber.org/streams";
+  node["xmlns:stream"] = STREAM_NS;
   node["to"] = "irc.abricot";
   this->send_stanza(node);
 }
@@ -62,7 +71,7 @@ void XmppComponent::on_connection_close()
 
 void XmppComponent::parse_in_buffer()
 {
-  this->parser.XML_Parse(this->in_buf.data(), this->in_buf.size(), false);
+  this->parser.feed(this->in_buf.data(), this->in_buf.size(), false);
   this->in_buf.clear();
 }
 
@@ -122,7 +131,7 @@ void XmppComponent::send_stream_error(const std::string& name, const std::string
 {
   XmlNode node("stream:error", nullptr);
   XmlNode error(name, nullptr);
-  error["xmlns"] = "urn:ietf:params:xml:ns:xmpp-streams";
+  error["xmlns"] = STREAM_NS;
   if (!explanation.empty())
     error.set_inner(explanation);
   error.close();
@@ -161,7 +170,7 @@ void XmppComponent::handle_presence(const Stanza& stanza)
         bridge->join_irc_channel(iid, to.resource);
       else if (type == "unavailable")
         {
-          XmlNode* status = stanza.get_child("status");
+          XmlNode* status = stanza.get_child(MUC_USER_NS":status");
           bridge->leave_irc_channel(std::move(iid), status ? std::move(status->get_inner()) : "");
         }
     }
@@ -172,7 +181,7 @@ void XmppComponent::handle_message(const Stanza& stanza)
   Bridge* bridge = this->get_user_bridge(stanza["from"]);
   Jid to(stanza["to"]);
   Iid iid(to.local);
-  XmlNode* body = stanza.get_child("body");
+  XmlNode* body = stanza.get_child(COMPONENT_NS":body");
   if (stanza["type"] == "groupchat")
     {
       if (to.resource.empty())
@@ -214,7 +223,7 @@ void XmppComponent::send_user_join(const std::string& from, const std::string& n
   node["from"] = from + "@" + this->served_hostname + "/" + nick;
 
   XmlNode x("x");
-  x["xmlns"] = "http://jabber.org/protocol/muc#user";
+  x["xmlns"] = MUC_USER_NS;
 
   // TODO: put real values here
   XmlNode item("item");
@@ -235,7 +244,7 @@ void XmppComponent::send_self_join(const std::string& from, const std::string& n
   node["from"] = from + "@" + this->served_hostname + "/" + nick;
 
   XmlNode x("x");
-  x["xmlns"] = "http://jabber.org/protocol/muc#user";
+  x["xmlns"] = MUC_USER_NS;
 
   // TODO: put real values here
   XmlNode item("item");

src/xmpp/xmpp_parser.cpp

@@ -1,21 +1,56 @@
 #include <xmpp/xmpp_parser.hpp>
 #include <xmpp/xmpp_stanza.hpp>
 
-#include <iostream>
+/**
+ * Expat handlers. Called by the Expat library, never by ourself.
+ * They just forward the call to the XmppParser corresponding methods.
+ */
+
+static void start_element_handler(void* user_data, const XML_Char* name, const XML_Char** atts)
+{
+  static_cast<XmppParser*>(user_data)->start_element(name, atts);
+}
+
+static void end_element_handler(void* user_data, const XML_Char* name)
+{
+  static_cast<XmppParser*>(user_data)->end_element(name);
+}
+
+static void character_data_handler(void *user_data, const XML_Char *s, int len)
+{
+  static_cast<XmppParser*>(user_data)->char_data(s, len);
+}
+
+/**
+ * XmppParser class
+ */
 
 XmppParser::XmppParser():
   level(0),
   current_node(nullptr)
 {
+  // Create the expat parser
+  this->parser = XML_ParserCreateNS("UTF-8", ':');
+  XML_SetUserData(this->parser, static_cast<void*>(this));
+
+  // Install Expat handlers
+  XML_SetElementHandler(this->parser, &start_element_handler, &end_element_handler);
+  XML_SetCharacterDataHandler(this->parser, &character_data_handler);
 }
 
 XmppParser::~XmppParser()
 {
   if (this->current_node)
     delete this->current_node;
+  XML_ParserFree(this->parser);
 }
 
-void XmppParser::startElement(const XML_Char* name, const XML_Char** attribute)
+void XmppParser::feed(const char* data, const int len, const bool is_final)
+{
+  XML_Parse(this->parser, data, len, is_final);
+}
+
+void XmppParser::start_element(const XML_Char* name, const XML_Char** attribute)
 {
   level++;
 
@@ -29,9 +64,9 @@ void XmppParser::startElement(const XML_Char* name, const XML_Char** attribute)
     this->stream_open_event(*this->current_node);
 }
 
-void XmppParser::endElement(const XML_Char* name)
+void XmppParser::end_element(const XML_Char* name)
 {
-  assert(name == this->current_node->get_name());
+  (void)name;
   level--;
   this->current_node->close();
   if (level == 1)
@@ -50,18 +85,12 @@ void XmppParser::endElement(const XML_Char* name)
     this->current_node->delete_all_children();
 }
 
-void XmppParser::charData(const XML_Char* data, int len)
+void XmppParser::char_data(const XML_Char* data, int len)
 {
   if (this->current_node->has_children())
-    this->current_node->get_last_child()->set_tail(std::string(data, len));
+    this->current_node->get_last_child()->add_to_tail(std::string(data, len));
   else
-    this->current_node->set_inner(std::string(data, len));
+    this->current_node->add_to_inner(std::string(data, len));
-}
-
-void XmppParser::startNamespace(const XML_Char* prefix, const XML_Char* uri)
-{
-  std::cout << "startNamespace: " << prefix << ":" << uri << std::endl;
-  this->namespaces.emplace(std::make_pair(prefix, uri));
 }
 
 void XmppParser::stanza_event(const Stanza& stanza) const
@@ -82,11 +111,6 @@ void XmppParser::stream_close_event(const XmlNode& node) const
     callback(node);
 }
 
-void XmppParser::endNamespace(const XML_Char* coucou)
-{
-  std::cout << "endNamespace: " << coucou << std::endl;
-}
-
 void XmppParser::add_stanza_callback(std::function<void(const Stanza&)>&& callback)
 {
   this->stanza_callbacks.emplace_back(std::move(callback));

src/xmpp/xmpp_parser.hpp

@@ -1,35 +1,43 @@
 #ifndef XMPP_PARSER_INCLUDED
 # define XMPP_PARSER_INCLUDED
 
-#include <functional>
-#include <stack>
-
 #include <xmpp/xmpp_stanza.hpp>
 
-#include <expatpp.h>
+#include <functional>
+
+#include <expat.h>
 
 /**
  * A SAX XML parser that builds XML nodes and spawns events when a complete
  * stanza is received (an element of level 2), or when the document is
- * opened (an element of level 1)
+ * opened/closed (an element of level 1)
  *
  * After a stanza_event has been spawned, we delete the whole stanza. This
  * means that even with a very long document (in XMPP the document is
- * potentially infinite), the memory then is never exhausted as long as each
+ * potentially infinite), the memory is never exhausted as long as each
  * stanza is reasonnably short.
  *
+ * The element names generated by expat contain the namespace of the
+ * element, a colon (':') and then the actual name of the element.  To get
+ * an element "x" with a namespace of "http://jabber.org/protocol/muc", you
+ * just look for an XmlNode named "http://jabber.org/protocol/muc:x"
+ *
  * TODO: enforce the size-limit for the stanza (limit the number of childs
  * it can contain). For example forbid the parser going further than level
  * 20 (arbitrary number here), and each XML node to have more than 15 childs
  * (arbitrary number again).
  */
-class XmppParser: public expatpp
+class XmppParser
 {
 public:
   explicit XmppParser();
   ~XmppParser();
 
 public:
+  /**
+   * Feed the parser with some XML data
+   */
+  void feed(const char* data, const int len, const bool is_final);
   /**
    * Add one callback for the various events that this parser can spawn.
    */
@@ -37,7 +45,6 @@ public:
   void add_stream_open_callback(std::function<void(const XmlNode&)>&& callback);
   void add_stream_close_callback(std::function<void(const XmlNode&)>&& callback);
 
-private:
   /**
    * Called when a new XML element has been opened. We instanciate a new
    * XmlNode and set it as our current node. The parent of this new node is
@@ -46,7 +53,7 @@ private:
    *
    * We spawn a stream_event with this node if this is a level-1 element.
    */
-  void startElement(const XML_Char* name, const XML_Char** attribute);
+  void start_element(const XML_Char* name, const XML_Char** attribute);
   /**
    * Called when an XML element has been closed. We close the current_node,
    * set our current_node as the parent of the current_node, and if that was
@@ -55,19 +62,11 @@ private:
    * And we then delete the stanza (and everything under it, its children,
    * attribute, etc).
    */
-  void endElement(const XML_Char* name);
+  void end_element(const XML_Char* name);
   /**
    * Some inner or tail data has been parsed
    */
-  void charData(const XML_Char* data, int len);
+  void char_data(const XML_Char* data, int len);
-  /**
-   * TODO use that.
-   */
-  void startNamespace(const XML_Char* prefix, const XML_Char* uri);
-  /**
-   * TODO and that.
-   */
-  void endNamespace(const XML_Char* prefix);
   /**
    * Calls all the stanza_callbacks one by one.
    */
@@ -82,6 +81,11 @@ private:
    */
   void stream_close_event(const XmlNode& node) const;
 
+private:
+  /**
+   * Expat structure.
+   */
+  XML_Parser parser;
   /**
    * The current depth in the XML document
    */
@@ -98,10 +102,6 @@ private:
   std::vector<std::function<void(const Stanza&)>> stanza_callbacks;
   std::vector<std::function<void(const XmlNode&)>> stream_open_callbacks;
   std::vector<std::function<void(const XmlNode&)>> stream_close_callbacks;
-  /**
-   * TODO: also use that.
-   */
-  std::stack<std::pair<std::string, std::string>> namespaces;
 };
 
 #endif // XMPP_PARSER_INCLUDED

src/xmpp/xmpp_stanza.cpp

@@ -72,16 +72,31 @@ void XmlNode::set_tail(const std::string& data)
   this->tail = data;
 }
 
+void XmlNode::add_to_tail(const std::string& data)
+{
+  this->tail += data;
+}
+
 void XmlNode::set_inner(const std::string& data)
 {
   this->inner = xml_escape(data);
 }
 
+void XmlNode::add_to_inner(const std::string& data)
+{
+  this->inner += xml_escape(data);
+}
+
 std::string XmlNode::get_inner() const
 {
   return this->inner;
 }
 
+std::string XmlNode::get_tail() const
+{
+  return this->tail;
+}
+
 XmlNode* XmlNode::get_child(const std::string& name) const
 {
   for (auto& child: this->children)

src/xmpp/xmpp_stanza.hpp

@@ -5,8 +5,6 @@
 #include <string>
 #include <vector>
 
-#include <expatpp.h>
-
 std::string xml_escape(const std::string& data);
 
 /**
@@ -52,15 +50,29 @@ public:
    */
   void set_tail(const std::string& data);
   /**
-   * Set the content of the inner, that is the text inside this node
+   * Append the given data to the content of the tail. This exists because
-   * TODO: escape it here.
+   * the expat library may provide the complete text of an element in more
+   * than one call
+   */
+  void add_to_tail(const std::string& data);
+  /**
+   * Set the content of the inner, that is the text inside this node.
    */
   void set_inner(const std::string& data);
+  /**
+   * Append the given data to the content of the inner. For the reason
+   * described in add_to_tail comment.
+   */
+  void add_to_inner(const std::string& data);
   /**
    * Get the content of inner
    * TODO: unescape it here.
    */
   std::string get_inner() const;
+  /**
+   * Get the content of the tail
+   */
+  std::string get_tail() const;
   /**
    * Get a pointer to the first child element with that name
    */