Implement IRC format to xhtml-im conversion

The generated XML is very verbose because each IRC formatting tag makes us
close a <span/> element and reopen it with the new style applied.  However,
this works quite well and is easy to implement.

src/bridge/bridge.cpp

@@ -21,14 +21,14 @@ Bridge::~Bridge()
 {
 }
 
-std::string Bridge::sanitize_for_xmpp(const std::string& str)
+Xmpp::body Bridge::make_xmpp_body(const std::string& str)
 {
   std::string res;
   if (utils::is_valid_utf8(str.c_str()))
     res = str;
   else
     res = utils::convert_to_utf8(str, "ISO-8859-1");
-  return res;
+  return irc_format_to_xhtmlim(res);
 }
 
 IrcClient* Bridge::get_irc_client(const std::string& hostname, const std::string& username)
@@ -102,7 +102,8 @@ void Bridge::send_channel_message(const Iid& iid, const std::string& body)
     irc->send_channel_message(iid.chan, *it);
   // We do not need to convert body to utf-8: it comes from our XMPP server,
   // so it's ok to send it back
-  this->xmpp->send_muc_message(iid.chan + "%" + iid.server, irc->get_own_nick(), body, this->user_jid);
+  this->xmpp->send_muc_message(iid.chan + "%" + iid.server, irc->get_own_nick(),
+                               this->make_xmpp_body(body), this->user_jid);
 }
 
 void Bridge::send_private_message(const Iid& iid, const std::string& body)
@@ -137,22 +138,17 @@ void Bridge::send_irc_kick(const Iid& iid, const std::string& target, const std:
 
 void Bridge::send_message(const Iid& iid, const std::string& nick, const std::string& body, const bool muc)
 {
-  std::string utf8_body = this->sanitize_for_xmpp(body);
-  if (utf8_body.substr(0, action_prefix_len) == action_prefix)
-    { // Special case for ACTION (/me) messages:
-      // "\01ACTION goes out\01" == "/me goes out"
-      utf8_body = std::string("/me ") +
-        utf8_body.substr(action_prefix_len, utf8_body.size() - action_prefix_len - 1);
-    }
   if (muc)
-    this->xmpp->send_muc_message(iid.chan + "%" + iid.server, nick, utf8_body, this->user_jid);
+    this->xmpp->send_muc_message(iid.chan + "%" + iid.server, nick,
+                                 this->make_xmpp_body(body), this->user_jid);
   else
-    this->xmpp->send_message(iid.chan + "%" + iid.server, utf8_body, this->user_jid);
+    this->xmpp->send_message(iid.chan + "%" + iid.server,
+                             this->make_xmpp_body(body), this->user_jid);
 }
 
 void Bridge::send_muc_leave(Iid&& iid, std::string&& nick, const std::string& message, const bool self)
 {
-  this->xmpp->send_muc_leave(std::move(iid.chan) + "%" + std::move(iid.server), std::move(nick), this->sanitize_for_xmpp(message), this->user_jid, self);
+  this->xmpp->send_muc_leave(std::move(iid.chan) + "%" + std::move(iid.server), std::move(nick), this->make_xmpp_body(message), this->user_jid, self);
 }
 
 void Bridge::send_nick_change(Iid&& iid, const std::string& old_nick, const std::string& new_nick, const bool self)
@@ -168,7 +164,7 @@ void Bridge::send_xmpp_message(const std::string& from, const std::string& autho
     body = std::string("[") + author + std::string("] ") + msg;
   else
     body = msg;
-  this->xmpp->send_message(from, this->sanitize_for_xmpp(body), this->user_jid);
+  this->xmpp->send_message(from, this->make_xmpp_body(body), this->user_jid);
 }
 
 void Bridge::send_user_join(const std::string& hostname, const std::string& chan_name, const std::string nick)
@@ -183,7 +179,7 @@ void Bridge::send_self_join(const std::string& hostname, const std::string& chan
 
 void Bridge::send_topic(const std::string& hostname, const std::string& chan_name, const std::string topic)
 {
-  this->xmpp->send_topic(chan_name + "%" + hostname, this->sanitize_for_xmpp(topic), this->user_jid);
+  this->xmpp->send_topic(chan_name + "%" + hostname, this->make_xmpp_body(topic), this->user_jid);
 }
 
 std::string Bridge::get_own_nick(const Iid& iid)

src/bridge/bridge.hpp

@@ -2,6 +2,7 @@
 # define BRIDGE_INCLUDED
 
 #include <irc/irc_client.hpp>
+#include <bridge/colors.hpp>
 #include <irc/iid.hpp>
 
 #include <unordered_map>
@@ -23,7 +24,7 @@ public:
   explicit Bridge(const std::string& user_jid, XmppComponent* xmpp, Poller* poller);
   ~Bridge();
 
-  static std::string sanitize_for_xmpp(const std::string& str);
+  static Xmpp::body make_xmpp_body(const std::string& str);
   /***
    **
    ** From XMPP to IRC.

src/bridge/colors.cpp

@@ -1,7 +1,169 @@
 #include <bridge/colors.hpp>
+#include <xmpp/xmpp_stanza.hpp>
+#include <utils/make_unique.hpp>
+
 #include <algorithm>
 
 #include <iostream>
 
+#include <string.h>
+
+static const char IRC_NUM_COLORS = 16;
+
+static const char* irc_colors_to_css[IRC_NUM_COLORS] = {
+  "white",
+  "black",
+  "blue",
+  "green",
+  "indianred",
+  "red",
+  "magenta",
+  "brown",
+  "yellow",
+  "lightgreen",
+  "cyan",
+  "lightcyan",
+  "lightblue",
+  "lightmagenta",
+  "gray",
+  "white",
+};
+
+#define XHTML_NS "http://www.w3.org/1999/xhtml"
+
+struct styles_t
 {
+  bool strong;
+  bool underline;
+  bool italic;
+  int fg;
+  int bg;
+};
+
+/** We keep the currently-applied CSS styles in a structure. Each time a tag
+ * is found, update this style list, then close the current span XML element
+ * (if it is open), then reopen it with all the new styles in it.  This is
+ * done this way because IRC formatting does not map well with XML
+ * (hierarchical tags), it’s a lot easier and cleaner to remove all styles
+ * and reapply them for each tag, instead of trying to keep a consistent
+ * hierarchy of span, strong, em etc tags.  The generated XML is one-level
+ * deep only.
+*/
+Xmpp::body irc_format_to_xhtmlim(const std::string& s)
+{
+  if (s.find_first_of(irc_format_char) == std::string::npos)
+    // there is no special formatting at all
+    return std::make_tuple(s, nullptr);
+
+  std::string cleaned;
+
+  styles_t styles = {false, false, false, -1, -1};
+
+  std::unique_ptr<XmlNode> result = std::make_unique<XmlNode>("body");
+  (*result)["xmlns"] = XHTML_NS;
+
+  XmlNode* current_node = result.get();
+  std::string::size_type pos_start = 0;
+  std::string::size_type pos_end;
+
+  while ((pos_end = s.find_first_of(irc_format_char, pos_start)) != std::string::npos)
+    {
+      const std::string txt = s.substr(pos_start, pos_end-pos_start);
+      cleaned += txt;
+      if (current_node->has_children())
+        current_node->get_last_child()->set_tail(txt);
+      else
+        current_node->set_inner(txt);
+
+      if (s[pos_end] == IRC_FORMAT_BOLD_CHAR)
+        styles.strong = !styles.strong;
+      else if (s[pos_end] == IRC_FORMAT_UNDERLINE_CHAR)
+        styles.underline = !styles.underline;
+      else if (s[pos_end] == IRC_FORMAT_ITALIC_CHAR)
+        styles.italic = !styles.italic;
+      else if (s[pos_end] == IRC_FORMAT_RESET_CHAR)
+        styles = {false, false, false, -1, -1};
+      else if (s[pos_end] == IRC_FORMAT_REVERSE_CHAR)
+        { }                      // TODO
+      else if (s[pos_end] == IRC_FORMAT_REVERSE2_CHAR)
+        { }                      // TODO
+      else if (s[pos_end] == IRC_FORMAT_FIXED_CHAR)
+        { }                      // TODO
+      else if (s[pos_end] == IRC_FORMAT_COLOR_CHAR)
+        {
+          size_t pos = pos_end + 1;
+          styles.fg = -1;
+          styles.bg = -1;
+          // get the first number following the format char
+          if (pos < s.size() && s[pos] >= '0' && s[pos] <= '9')
+            {                   // first digit
+              styles.fg = s[pos++] - '0';
+              if (pos < s.size() && s[pos] >= '0' && s[pos] <= '9')
+                // second digit
+                styles.fg = styles.fg * 10 + s[pos++] - '0';
+            }
+          if (pos < s.size() && s[pos] == ',')
+            {                   // get bg color after the comma
+              pos++;
+              if (pos < s.size() && s[pos] >= '0' && s[pos] <= '9')
+                {               // first digit
+                  styles.bg = s[pos++] - '0';
+                  if (pos < s.size() && s[pos] >= '0' && s[pos] <= '9')
+                    // second digit
+                    styles.bg = styles.bg * 10 + s[pos++] - '0';
+                }
+            }
+          pos_end = pos - 1;
+        }
+
+      // close opened span, if any
+      if (current_node != result.get())
+        {
+          current_node->close();
+          result->add_child(current_node);
+          current_node = result.get();
+        }
+      // Take all currently-applied style and create a new span with it
+      std::string styles_str;
+      if (styles.strong)
+        styles_str += "font-weight:bold;";
+      if (styles.underline)
+        styles_str += "text-decoration:underline;";
+      if (styles.italic)
+        styles_str += "font-style:italic;";
+      if (styles.fg != -1)
+        styles_str += std::string("color:") +
+          irc_colors_to_css[styles.fg % IRC_NUM_COLORS] + ";";
+      if (styles.bg != -1)
+        styles_str += std::string("background-color:") +
+          irc_colors_to_css[styles.bg % IRC_NUM_COLORS] + ";";
+      if (!styles_str.empty())
+        {
+          current_node = new XmlNode("span");
+          (*current_node)["style"] = styles_str;
+        }
+
+      pos_start = pos_end + 1;
+    }
+
+  // If some text remains, without any format char, just append that text at
+  // the end of the current node
+  const std::string txt = s.substr(pos_start, pos_end-pos_start);
+  cleaned += txt;
+  if (current_node->has_children())
+    current_node->get_last_child()->set_tail(txt);
+  else
+    current_node->set_inner(txt);
+
+  if (current_node != result.get())
+    {
+      current_node->close();
+      result->add_child(current_node);
+      current_node = result.get();
+    }
+
+
+  result->close();
+  Xmpp::body body_res = std::make_tuple(cleaned, std::move(result));
+  return body_res;
 }

src/bridge/colors.hpp

@@ -1,20 +1,54 @@
 #ifndef COLORS_INCLUDED
 # define COLORS_INCLUDED
 
-#include <string>
-
 /**
- * A module handling the conversion between IRC colors and XHTML-IM, and vice versa.
+ * A module handling the conversion between IRC colors and XHTML-IM, and
+ * vice versa.
  */
 
-#define IRC_COLOR_BOLD_CHAR       '\x02'
-#define IRC_COLOR_COLOR_CHAR      '\x03'
-#define IRC_COLOR_RESET_CHAR      '\x0F'
-#define IRC_COLOR_FIXED_CHAR      '\x11'
-#define IRC_COLOR_REVERSE_CHAR    '\x12'
-#define IRC_COLOR_REVERSE2_CHAR   '\x16'
-#define IRC_COLOR_ITALIC_CHAR     '\x1D'
-#define IRC_COLOR_UNDERLINE_CHAR  '\x1F'
+#include <string>
+#include <memory>
+#include <tuple>
 
+class XmlNode;
+
+namespace Xmpp
+{
+// Contains:
+// - an XMPP-valid UTF-8 body
+// - an XML node representing the XHTML-IM body, or null
+  typedef std::tuple<const std::string, std::unique_ptr<XmlNode>> body;
+}
+
+#define IRC_FORMAT_BOLD_CHAR       '\x02' // done
+#define IRC_FORMAT_COLOR_CHAR      '\x03' // done
+#define IRC_FORMAT_RESET_CHAR      '\x0F' // done
+#define IRC_FORMAT_FIXED_CHAR      '\x11' // ??
+#define IRC_FORMAT_REVERSE_CHAR    '\x12' // maybe one day
+#define IRC_FORMAT_REVERSE2_CHAR   '\x16' // wat
+#define IRC_FORMAT_ITALIC_CHAR     '\x1D' // done
+#define IRC_FORMAT_UNDERLINE_CHAR  '\x1F' // done
+
+static const char irc_format_char[] = {
+  IRC_FORMAT_BOLD_CHAR,
+  IRC_FORMAT_COLOR_CHAR,
+  IRC_FORMAT_RESET_CHAR,
+  IRC_FORMAT_FIXED_CHAR,
+  IRC_FORMAT_REVERSE_CHAR,
+  IRC_FORMAT_REVERSE2_CHAR,
+  IRC_FORMAT_ITALIC_CHAR,
+  IRC_FORMAT_UNDERLINE_CHAR,
+  '\x00'
+};
+
+/**
+ * Convert the passed string into an XML tree representing the XHTML version
+ * of the message, converting the IRC colors symbols into xhtml-im
+ * formatting.
+ *
+ * Returns the body cleaned from any IRC formatting (but without any xhtml),
+ * and the body as XHTML-IM
+ */
+Xmpp::body irc_format_to_xhtmlim(const std::string& str);
 
 #endif // COLORS_INCLUDED