Add an XmppParser, and Stanza classes

Generate events on stanza and stream open/close.
Create Stanza and serialize them.

Note: XML namespaces are not handled yet.

5 files changed, 444 insertions, 1 deletions

diff --git a/CMakeLists.txt b/CMakeLists.txt
index 9f1eef6..1c21b84 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -23,7 +23,7 @@ add_library(network STATIC ${source_network})
 file(GLOB source_irc
   src/irc/*.[hc]pp)
 add_library(irc STATIC ${source_irc})
-target_link_libraries(irc network ${CRYPTO++_LIBRARIES})
+target_link_libraries(irc network ${CRYPTO++_LIBRARIES} expatpp)
 
 #
 ## xmpplib
diff --git a/src/xmpp/xmpp_parser.cpp b/src/xmpp/xmpp_parser.cpp
new file mode 100644
index 0000000..00714b3
--- /dev/null
+++ b/src/xmpp/xmpp_parser.cpp
@@ -0,0 +1,103 @@
+#include <xmpp/xmpp_parser.hpp>
+#include <xmpp/xmpp_stanza.hpp>
+
+#include <iostream>
+
+XmppParser::XmppParser():
+  level(0),
+  current_node(nullptr)
+{
+}
+
+XmppParser::~XmppParser()
+{
+  if (this->current_node)
+    delete this->current_node;
+}
+
+void XmppParser::startElement(const XML_Char* name, const XML_Char** attribute)
+{
+  level++;
+
+  XmlNode* new_node = new XmlNode(name, this->current_node);
+  if (this->current_node)
+    this->current_node->add_child(new_node);
+  this->current_node = new_node;
+  for (size_t i = 0; attribute[i]; i += 2)
+    this->current_node->set_attribute(attribute[i], attribute[i+1]);
+  if (this->level == 1)
+    this->stream_open_event(*this->current_node);
+}
+
+void XmppParser::endElement(const XML_Char* name)
+{
+  assert(name == this->current_node->get_name());
+  level--;
+  this->current_node->close();
+  if (level == 1)
+    {
+      this->stanza_event(*this->current_node);
+    }
+  if (level == 0)
+    {
+      this->stream_close_event(*this->current_node);
+      delete this->current_node;
+      this->current_node = nullptr;
+    }
+  else
+    this->current_node = this->current_node->get_parent();
+  if (level == 1)
+    this->current_node->delete_all_children();
+}
+
+void XmppParser::charData(const XML_Char* data, int len)
+{
+  if (this->current_node->has_children())
+    this->current_node->get_last_child()->set_tail(std::string(data, len));
+  else
+    this->current_node->set_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
+{
+  for (const auto& callback: this->stanza_callbacks)
+    callback(stanza);
+}
+
+void XmppParser::stream_open_event(const XmlNode& node) const
+{
+  for (const auto& callback: this->stream_open_callbacks)
+    callback(node);
+}
+
+void XmppParser::stream_close_event(const XmlNode& node) const
+{
+  for (const auto& callback: this->stream_close_callbacks)
+    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));
+}
+
+void XmppParser::add_stream_open_callback(std::function<void(const XmlNode&)>&& callback)
+{
+  this->stream_open_callbacks.emplace_back(std::move(callback));
+}
+
+void XmppParser::add_stream_close_callback(std::function<void(const XmlNode&)>&& callback)
+{
+  this->stream_close_callbacks.emplace_back(std::move(callback));
+}
diff --git a/src/xmpp/xmpp_parser.hpp b/src/xmpp/xmpp_parser.hpp
new file mode 100644
index 0000000..2e83bc3
--- /dev/null
+++ b/src/xmpp/xmpp_parser.hpp
@@ -0,0 +1,107 @@
+#ifndef XMPP_PARSER_INCLUDED
+# define XMPP_PARSER_INCLUDED
+
+#include <functional>
+#include <stack>
+
+#include <xmpp/xmpp_stanza.hpp>
+
+#include <expatpp.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)
+ *
+ * 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
+ * stanza is reasonnably short.
+ *
+ * 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
+{
+public:
+  explicit XmppParser();
+  ~XmppParser();
+
+public:
+  /**
+   * Add one callback for the various events that this parser can spawn.
+   */
+  void add_stanza_callback(std::function<void(const Stanza&)>&& callback);
+  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
+   * the previous "current" node. We have all the element's attributes in
+   * this event.
+   *
+   * 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);
+  /**
+   * 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
+   * a level-2 element we spawn a stanza_event with this node.
+   *
+   * And we then delete the stanza (and everything under it, its children,
+   * attribute, etc).
+   */
+  void endElement(const XML_Char* name);
+  /**
+   * Some inner or tail data has been parsed
+   */
+  void charData(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.
+   */
+  void stanza_event(const Stanza& stanza) const;
+  /**
+   * Calls all the stream_open_callbacks one by one. Note: the passed node is not
+   * closed yet.
+   */
+  void stream_open_event(const XmlNode& node) const;
+  /**
+   * Calls all the stream_close_callbacks one by one.
+   */
+  void stream_close_event(const XmlNode& node) const;
+
+  /**
+   * The current depth in the XML document
+   */
+  size_t level;
+  /**
+   * The deepest XML node opened but not yet closed (to which we are adding
+   * new children, inner or tail)
+   */
+  XmlNode* current_node;
+  /**
+   * A list of callbacks to be called on an *_event, receiving the
+   * concerned Stanza/XmlNode.
+   */
+  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
diff --git a/src/xmpp/xmpp_stanza.cpp b/src/xmpp/xmpp_stanza.cpp
new file mode 100644
index 0000000..2c98acc
--- /dev/null
+++ b/src/xmpp/xmpp_stanza.cpp
@@ -0,0 +1,127 @@
+#include <xmpp/xmpp_stanza.hpp>
+
+#include <iostream>
+
+XmlNode::XmlNode(const std::string& name, XmlNode* parent):
+  name(name),
+  parent(parent),
+  closed(false)
+{
+}
+
+XmlNode::XmlNode(const std::string& name):
+  XmlNode(name, nullptr)
+{
+}
+
+XmlNode::~XmlNode()
+{
+  this->delete_all_children();
+}
+
+void XmlNode::delete_all_children()
+{
+  for (auto& child: this->children)
+    {
+      delete child;
+    }
+  this->children.clear();
+}
+
+void XmlNode::set_attribute(const std::string& name, const std::string& value)
+{
+  this->attributes[name] = value;
+}
+
+void XmlNode::set_tail(const std::string& data)
+{
+  this->tail = data;
+}
+
+void XmlNode::set_inner(const std::string& data)
+{
+  this->inner = data;
+}
+
+void XmlNode::add_child(XmlNode* child)
+{
+  this->children.push_back(child);
+}
+
+void XmlNode::add_child(XmlNode&& child)
+{
+  XmlNode* new_node = new XmlNode(std::move(child));
+  this->add_child(new_node);
+}
+
+XmlNode* XmlNode::get_last_child() const
+{
+  return this->children.back();
+}
+
+void XmlNode::close()
+{
+  if (this->closed)
+    throw std::runtime_error("Closing an already closed XmlNode");
+  this->closed = true;
+}
+
+XmlNode* XmlNode::get_parent() const
+{
+  return this->parent;
+}
+
+const std::string& XmlNode::get_name() const
+{
+  return this->name;
+}
+
+std::string XmlNode::to_string() const
+{
+  std::string res("<");
+  res += this->name;
+  for (const auto& it: this->attributes)
+    res += " " + it.first + "='" + it.second + "'";
+  if (this->closed && !this->has_children() && this->inner.empty())
+    res += "/>";
+  else
+    {
+      res += ">" + this->inner;
+      for (const auto& child: this->children)
+        res += child->to_string();
+      if (this->closed)
+        {
+          res += "</" + this->name + ">";
+        }
+    }
+  res += this->tail;
+  return res;
+}
+
+void XmlNode::display() const
+{
+  std::cout << this->to_string() << std::endl;
+}
+
+bool XmlNode::has_children() const
+{
+  return !this->children.empty();
+}
+
+const std::string& XmlNode::operator[](const std::string& name) const
+{
+  try
+    {
+      const auto& value = this->attributes.at(name);
+      return value;
+    }
+  catch (const std::out_of_range& e)
+    {
+      throw AttributeNotFound();
+    }
+}
+
+std::string& XmlNode::operator[](const std::string& name)
+{
+  return this->attributes[name];
+}
diff --git a/src/xmpp/xmpp_stanza.hpp b/src/xmpp/xmpp_stanza.hpp
new file mode 100644
index 0000000..277b0db
--- /dev/null
+++ b/src/xmpp/xmpp_stanza.hpp
@@ -0,0 +1,106 @@
+#ifndef XMPP_STANZA_INCLUDED
+# define XMPP_STANZA_INCLUDED
+
+#include <unordered_map>
+#include <string>
+#include <vector>
+
+#include <expatpp.h>
+
+/**
+ * Raised on operator[] when the attribute does not exist
+ */
+class AttributeNotFound: public std::exception
+{
+};
+
+/**
+ * Represent an XML node. It has
+ * - A parent XML node (in the case of the first-level nodes, the parent is
+     nullptr)
+ * - zero, one or more children XML nodes
+ * - A name
+ * - attributes
+ * - inner data (inside the node)
+ * - tail data (just after the node)
+ */
+class XmlNode
+{
+public:
+  explicit XmlNode(const std::string& name, XmlNode* parent);
+  explicit XmlNode(const std::string& name);
+  XmlNode(XmlNode&& node):
+    name(std::move(node.name)),
+    parent(std::move(node.parent)),
+    closed(std::move(node.closed)),
+    attributes(std::move(node.attributes)),
+    children(std::move(node.children)),
+    inner(std::move(node.inner)),
+    tail(std::move(node.tail))
+  {
+    node.parent = nullptr;
+  }
+
+  ~XmlNode();
+
+  void delete_all_children();
+  void set_attribute(const std::string& name, const std::string& value);
+  /**
+   * Set the content of the tail, that is the text just after this node
+   */
+  void set_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);
+  void add_child(XmlNode* child);
+  void add_child(XmlNode&& child);
+  XmlNode* get_last_child() const;
+  /**
+   * Mark this node as closed, nothing else
+   */
+  void close();
+  XmlNode* get_parent() const;
+  const std::string& get_name() const;
+  /**
+   * Serialize the stanza into a string
+   */
+  std::string to_string() const;
+  void display() const;
+  /**
+   * Whether or not this node has at least one child (if not, this is a leaf
+   * node)
+   */
+  bool has_children() const;
+  /**
+   * Gets the value for the given attribute, raises AttributeNotFound if the
+   * node as no such attribute.
+   */
+  const std::string& operator[](const std::string& name) const;
+  /**
+   * Use this to set an attribute's value, like node["id"] = "12";
+   */
+  std::string& operator[](const std::string& name);
+
+private:
+  std::string name;
+  XmlNode* parent;
+  bool closed;
+  std::unordered_map<std::string, std::string> attributes;
+  std::vector<XmlNode*> children;
+  std::string inner;
+  std::string tail;
+
+  XmlNode(const XmlNode&) = delete;
+  XmlNode& operator=(const XmlNode&) = delete;
+  XmlNode& operator=(XmlNode&&) = delete;
+};
+
+/**
+ * An XMPP stanza is just an XML node of level 2 in the XMPP document (the
+ * level 1 ones are the <stream::stream/>, and the ones about 2 are just the
+ * content of the stanzas)
+ */
+typedef XmlNode Stanza;
+
+#endif // XMPP_STANZA_INCLUDED