Properly sanitize everything in the XML we send to the XMPP server

in this order:
- Make sure it is utf-8 encoded
- Remove all chars that are invalid in XML
- Escape all XML special chars (&'"<>)

2 files changed, 12 insertions, 4 deletions

diff --git a/src/xmpp/xmpp_stanza.cpp b/src/xmpp/xmpp_stanza.cpp
index 4290fc7..df19105 100644
--- a/src/xmpp/xmpp_stanza.cpp
+++ b/src/xmpp/xmpp_stanza.cpp
@@ -218,13 +218,12 @@ std::string XmlNode::to_string() const
   std::string res("<");
   res += this->name;
   for (const auto& it: this->attributes)
-    res += " " + utils::remove_invalid_xml_chars(it.first) + "='" +
-      utils::remove_invalid_xml_chars(it.second) + "'";
+    res += " " + it.first + "='" + sanitize(it.second) + "'";
   if (this->closed && !this->has_children() && this->inner.empty())
     res += "/>";
   else
     {
-      res += ">" + utils::remove_invalid_xml_chars(this->inner);
+      res += ">" + sanitize(this->inner);
       for (const auto& child: this->children)
         res += child->to_string();
       if (this->closed)
@@ -232,7 +231,7 @@ std::string XmlNode::to_string() const
           res += "</" + this->get_name() + ">";
         }
     }
-  res += utils::remove_invalid_xml_chars(this->tail);
+  res += sanitize(this->tail);
   return res;
 }
 
@@ -265,3 +264,11 @@ std::string& XmlNode::operator[](const std::string& name)
 {
   return this->attributes[name];
 }
+
+std::string sanitize(const std::string& data)
+{
+  if (utils::is_valid_utf8(data.data()))
+    return xml_escape(utils::remove_invalid_xml_chars(data));
+  else
+    return xml_escape(utils::remove_invalid_xml_chars(utils::convert_to_utf8(data, "ISO-8859-1")));
+}
diff --git a/src/xmpp/xmpp_stanza.hpp b/src/xmpp/xmpp_stanza.hpp
index 9229ae6..f1a6a0f 100644
--- a/src/xmpp/xmpp_stanza.hpp
+++ b/src/xmpp/xmpp_stanza.hpp
@@ -7,6 +7,7 @@
 
 std::string xml_escape(const std::string& data);
 std::string xml_unescape(const std::string& data);
+std::string sanitize(const std::string& data);
 
 /**
  * Represent an XML node. It has