diff --git a/extensions/dict_legacy_xml/CMakeLists.txt b/extensions/dict_legacy_xml/CMakeLists.txt
new file mode 100644
index 0000000..561fbc5
--- /dev/null
+++ b/extensions/dict_legacy_xml/CMakeLists.txt
@@ -0,0 +1,35 @@
+# The dict_legacy_xml extension
+PROJECT("Legacy XML dictionary files support" C)
+
+# Parser files
+BISON_FILE(dict_lxml.y)
+FLEX_FILE(dict_lxml.l)
+SET_SOURCE_FILES_PROPERTIES(lex.dict_lxml.c dict_lxml.tab.c PROPERTIES COMPILE_FLAGS "-I ${CMAKE_CURRENT_SOURCE_DIR}")
+
+# We use LibXml2 (from GNOME) to parse XML files
+FIND_PACKAGE(LibXml2 REQUIRED)
+
+# List of source files
+SET( DICT_LXML_SRC
+	dict_lxml.c
+	dict_lxml_xml.c
+	dict_lxml.h
+	lex.dict_lxml.c
+	dict_lxml.tab.c
+	dict_lxml.tab.h
+)
+INCLUDE_DIRECTORIES(${CMAKE_CURRENT_BINARY_DIR})
+INCLUDE_DIRECTORIES(${LIBXML2_INCLUDE_DIR})
+
+# Compile these files as a freeDiameter extension
+FD_ADD_EXTENSION(dict_legacy_xml ${DICT_LXML_SRC})
+
+TARGET_LINK_LIBRARIES(dict_legacy_xml ${LIBXML2_LIBRARIES} )
+
+
+####
+## INSTALL section ##
+
+INSTALL(TARGETS dict_legacy_xml
+	LIBRARY DESTINATION ${INSTALL_EXTENSIONS_SUFFIX}
+	COMPONENT freeDiameter-dictionary-legacy)
diff --git a/extensions/dict_legacy_xml/dict_lxml.c b/extensions/dict_legacy_xml/dict_lxml.c
new file mode 100644
index 0000000..e4b8d8f
--- /dev/null
+++ b/extensions/dict_legacy_xml/dict_lxml.c
@@ -0,0 +1,54 @@
+/*********************************************************************************************************
+* Software License Agreement (BSD License)                                                               *
+* Author: Sebastien Decugis <sdecugis@freediameter.net>							 *
+*													 *
+* Copyright (c) 2011, WIDE Project and NICT								 *
+* All rights reserved.											 *
+* 													 *
+* Redistribution and use of this software in source and binary forms, with or without modification, are  *
+* permitted provided that the following conditions are met:						 *
+* 													 *
+* * Redistributions of source code must retain the above 						 *
+*   copyright notice, this list of conditions and the 							 *
+*   following disclaimer.										 *
+*    													 *
+* * Redistributions in binary form must reproduce the above 						 *
+*   copyright notice, this list of conditions and the 							 *
+*   following disclaimer in the documentation and/or other						 *
+*   materials provided with the distribution.								 *
+* 													 *
+* * Neither the name of the WIDE Project or NICT nor the 						 *
+*   names of its contributors may be used to endorse or 						 *
+*   promote products derived from this software without 						 *
+*   specific prior written permission of WIDE Project and 						 *
+*   NICT.												 *
+* 													 *
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED *
+* WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A *
+* PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR *
+* ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT 	 *
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 	 *
+* INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR *
+* TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF   *
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.								 *
+*********************************************************************************************************/
+
+/* 
+ * Legacy XML dictionary files support for freeDiameter.
+ */
+
+#include "dict_lxml.h"
+
+/* entry point */
+static int dict_lxml_entry(char * conffile)
+{
+	TRACE_ENTRY("%p", conffile);
+	
+	/* Parse the configuration file -- everything happens there */
+	CHECK_FCT( dict_lxml_handle(conffile) );
+	
+	/* We're done */
+	return 0;
+}
+
+EXTENSION_ENTRY("dict_legacy_xml", dict_lxml_entry);
diff --git a/extensions/dict_legacy_xml/dict_lxml.h b/extensions/dict_legacy_xml/dict_lxml.h
new file mode 100644
index 0000000..e862d62
--- /dev/null
+++ b/extensions/dict_legacy_xml/dict_lxml.h
@@ -0,0 +1,48 @@
+/*********************************************************************************************************
+* Software License Agreement (BSD License)                                                               *
+* Author: Sebastien Decugis <sdecugis@freediameter.net>							 *
+*													 *
+* Copyright (c) 2011, WIDE Project and NICT								 *
+* All rights reserved.											 *
+* 													 *
+* Redistribution and use of this software in source and binary forms, with or without modification, are  *
+* permitted provided that the following conditions are met:						 *
+* 													 *
+* * Redistributions of source code must retain the above 						 *
+*   copyright notice, this list of conditions and the 							 *
+*   following disclaimer.										 *
+*    													 *
+* * Redistributions in binary form must reproduce the above 						 *
+*   copyright notice, this list of conditions and the 							 *
+*   following disclaimer in the documentation and/or other						 *
+*   materials provided with the distribution.								 *
+* 													 *
+* * Neither the name of the WIDE Project or NICT nor the 						 *
+*   names of its contributors may be used to endorse or 						 *
+*   promote products derived from this software without 						 *
+*   specific prior written permission of WIDE Project and 						 *
+*   NICT.												 *
+* 													 *
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED *
+* WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A *
+* PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR *
+* ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT 	 *
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 	 *
+* INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR *
+* TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF   *
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.								 *
+*********************************************************************************************************/
+
+/* Header file for the dict_legacy_xml extension. 
+ *
+ *  See the dict_legacy_xml.conf.sample file for the format of the configuration file.
+ */
+ 
+/* FreeDiameter's common include file */
+#include <freeDiameter/extension.h>
+
+/* Parse the configuration file */
+int dict_lxml_handle(char * conffile);
+
+/* Parse an XML file and return the number of dictionary objects or -1 on error */
+int dict_lxml_parse(char * xmlfilename);
diff --git a/extensions/dict_legacy_xml/dict_lxml.l b/extensions/dict_legacy_xml/dict_lxml.l
new file mode 100644
index 0000000..fc612bb
--- /dev/null
+++ b/extensions/dict_legacy_xml/dict_lxml.l
@@ -0,0 +1,98 @@
+/*********************************************************************************************************
+* Software License Agreement (BSD License)                                                               *
+* Author: Sebastien Decugis <sdecugis@freediameter.net>							 *
+*													 *
+* Copyright (c) 2013, WIDE Project and NICT								 *
+* All rights reserved.											 *
+* 													 *
+* Redistribution and use of this software in source and binary forms, with or without modification, are  *
+* permitted provided that the following conditions are met:						 *
+* 													 *
+* * Redistributions of source code must retain the above 						 *
+*   copyright notice, this list of conditions and the 							 *
+*   following disclaimer.										 *
+*    													 *
+* * Redistributions in binary form must reproduce the above 						 *
+*   copyright notice, this list of conditions and the 							 *
+*   following disclaimer in the documentation and/or other						 *
+*   materials provided with the distribution.								 *
+* 													 *
+* * Neither the name of the WIDE Project or NICT nor the 						 *
+*   names of its contributors may be used to endorse or 						 *
+*   promote products derived from this software without 						 *
+*   specific prior written permission of WIDE Project and 						 *
+*   NICT.												 *
+* 													 *
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED *
+* WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A *
+* PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR *
+* ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT 	 *
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 	 *
+* INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR *
+* TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF   *
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.								 *
+*********************************************************************************************************/
+
+/* Tokenizer
+ *
+ */
+
+%{
+#include "dict_lxml.h"
+/* Include yacc tokens definitions */
+#include "dict_lxml.tab.h"
+
+/* Update the column information */
+#define YY_USER_ACTION { 						\
+	yylloc->first_column = yylloc->last_column + 1; 		\
+	yylloc->last_column = yylloc->first_column + yyleng - 1;	\
+}
+
+/* Avoid warning with newer flex */
+#define YY_NO_INPUT
+
+%}
+
+qstring		\"[^\"\n]*\"
+
+
+%option bison-bridge bison-locations
+%option noyywrap
+%option nounput
+
+%%
+
+	/* Update the line count */
+\n			{
+				yylloc->first_line++; 
+				yylloc->last_line++; 
+				yylloc->last_column=0; 
+			}
+	 
+	/* Eat all spaces but not new lines */
+([[:space:]]{-}[\n])+	;
+	/* Eat all comments */
+#.*$			;
+
+	/* Recognize quoted strings */
+{qstring}		{
+				/* Match a quoted string. */
+				CHECK_MALLOC_DO( yylval->string = strdup(yytext+1), 
+				{
+					TRACE_DEBUG(INFO, "Unable to copy the string '%s': %s", yytext, strerror(errno));
+					return LEX_ERROR; /* trig an error in yacc parser */
+				} );
+				yylval->string[strlen(yytext) - 2] = '\0';
+				return QSTRING;
+			}
+	
+	/* Valid single characters for yyparse */
+[;]			{ return yytext[0]; }
+
+	/* Unrecognized sequence, if it did not match any previous pattern */
+[^[:space:]\";\n]+	{ 
+				fd_log_debug("Unrecognized text on line %d col %d: '%s'.", yylloc->first_line, yylloc->first_column, yytext);
+			 	return LEX_ERROR; 
+			}
+
+%%
diff --git a/extensions/dict_legacy_xml/dict_lxml.y b/extensions/dict_legacy_xml/dict_lxml.y
new file mode 100644
index 0000000..4cafb4f
--- /dev/null
+++ b/extensions/dict_legacy_xml/dict_lxml.y
@@ -0,0 +1,141 @@
+/*********************************************************************************************************
+* Software License Agreement (BSD License)                                                               *
+* Author: Sebastien Decugis <sdecugis@freediameter.net>							 *
+*													 *
+* Copyright (c) 2013, WIDE Project and NICT								 *
+* All rights reserved.											 *
+* 													 *
+* Redistribution and use of this software in source and binary forms, with or without modification, are  *
+* permitted provided that the following conditions are met:						 *
+* 													 *
+* * Redistributions of source code must retain the above 						 *
+*   copyright notice, this list of conditions and the 							 *
+*   following disclaimer.										 *
+*    													 *
+* * Redistributions in binary form must reproduce the above 						 *
+*   copyright notice, this list of conditions and the 							 *
+*   following disclaimer in the documentation and/or other						 *
+*   materials provided with the distribution.								 *
+* 													 *
+* * Neither the name of the WIDE Project or NICT nor the 						 *
+*   names of its contributors may be used to endorse or 						 *
+*   promote products derived from this software without 						 *
+*   specific prior written permission of WIDE Project and 						 *
+*   NICT.												 *
+* 													 *
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED *
+* WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A *
+* PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR *
+* ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT 	 *
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 	 *
+* INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR *
+* TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF   *
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.								 *
+*********************************************************************************************************/
+
+/* Yacc extension's configuration parser.
+ * See doc/dict_legacy_xml.conf.sample for configuration file format
+ */
+
+/* For development only : */
+%debug 
+%error-verbose
+
+/* The parser receives the configuration file filename as parameter */
+%parse-param {char * conffile}
+
+/* Keep track of location */
+%locations 
+%pure-parser
+
+%{
+#include "dict_lxml.h"
+#include "dict_lxml.tab.h"	/* bison is not smart enough to define the YYLTYPE before including this code, so... */
+
+/* Forward declaration */
+int yyparse(char * conffile);
+
+static int nb_files = 0;
+static int nb_dict = 0;
+
+/* Parse the configuration file */
+int dict_lxml_handle(char * conffile)
+{
+	extern FILE * dict_lxmlin;
+	int ret;
+	
+	TRACE_ENTRY("%p", conffile);
+	
+	TRACE_DEBUG (FULL, "Parsing configuration file: %s...", conffile);
+	
+	dict_lxmlin = fopen(conffile, "r");
+	if (dict_lxmlin == NULL) {
+		ret = errno;
+		fd_log_debug("Unable to open extension configuration file %s for reading: %s", conffile, strerror(ret));
+		TRACE_DEBUG (INFO, "Error occurred, message logged -- configuration file.");
+		return ret;
+	}
+
+	ret = yyparse(conffile);
+
+	fclose(dict_lxmlin);
+
+	if (ret != 0) {
+		TRACE_DEBUG (INFO, "Unable to parse the configuration file.");
+		return EINVAL;
+	} else {
+		TRACE_DEBUG(FULL, "%d XML dictionary files parsed successfully, %d dictionary objects added.", nb_files, nb_dict);
+	}
+	
+	return 0;
+}
+
+/* The Lex parser prototype */
+int dict_lxmllex(YYSTYPE *lvalp, YYLTYPE *llocp);
+
+/* Function to report the errors */
+void yyerror (YYLTYPE *ploc, char * conffile, char const *s)
+{
+	TRACE_DEBUG(INFO, "Error in configuration parsing");
+	
+	if (ploc->first_line != ploc->last_line)
+		fd_log_debug("%s:%d.%d-%d.%d : %s", conffile, ploc->first_line, ploc->first_column, ploc->last_line, ploc->last_column, s);
+	else if (ploc->first_column != ploc->last_column)
+		fd_log_debug("%s:%d.%d-%d : %s", conffile, ploc->first_line, ploc->first_column, ploc->last_column, s);
+	else
+		fd_log_debug("%s:%d.%d : %s", conffile, ploc->first_line, ploc->first_column, s);
+}
+
+%}
+
+/* Values returned by lex for token */
+%union {
+	char 		*string;	/* The string is allocated by strdup in lex.*/
+}
+
+/* In case of error in the lexical analysis */
+%token 		LEX_ERROR
+
+/* A (de)quoted string (malloc'd in lex parser; it must be freed after use) */
+%token <string>	QSTRING
+
+/* -------------------------------------- */
+%%
+
+	/* The grammar definition */
+conffile:		/* empty grammar is OK */
+			| conffile xmlfile
+			;
+			
+	/* a RULE entry */
+xmlfile:		QSTRING ';'
+			{
+				int ret = dict_lxml_parse($1);
+				if (ret < 0) {
+					yyerror (&yylloc, conffile, "An error occurred while parsing a file, aborting...");
+					YYERROR;
+				}
+				nb_files++;
+				nb_dict += ret;
+			}
+			;
diff --git a/extensions/dict_legacy_xml/dict_lxml_xml.c b/extensions/dict_legacy_xml/dict_lxml_xml.c
new file mode 100644
index 0000000..748c43f
--- /dev/null
+++ b/extensions/dict_legacy_xml/dict_lxml_xml.c
@@ -0,0 +1,1816 @@
+/*********************************************************************************************************
+* Software License Agreement (BSD License)                                                               *
+* Author: Sebastien Decugis <sdecugis@freediameter.net>							 *
+*													 *
+* Copyright (c) 2013, WIDE Project and NICT								 *
+* All rights reserved.											 *
+* 													 *
+* Redistribution and use of this software in source and binary forms, with or without modification, are  *
+* permitted provided that the following conditions are met:						 *
+* 													 *
+* * Redistributions of source code must retain the above 						 *
+*   copyright notice, this list of conditions and the 							 *
+*   following disclaimer.										 *
+*    													 *
+* * Redistributions in binary form must reproduce the above 						 *
+*   copyright notice, this list of conditions and the 							 *
+*   following disclaimer in the documentation and/or other						 *
+*   materials provided with the distribution.								 *
+* 													 *
+* * Neither the name of the WIDE Project or NICT nor the 						 *
+*   names of its contributors may be used to endorse or 						 *
+*   promote products derived from this software without 						 *
+*   specific prior written permission of WIDE Project and 						 *
+*   NICT.												 *
+* 													 *
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED *
+* WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A *
+* PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR *
+* ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT 	 *
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 	 *
+* INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR *
+* TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF   *
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.								 *
+*********************************************************************************************************/
+
+#include "dict_lxml.h"
+
+/*
+The internal freeDiameter dictionary has strong dependency relations between
+the different objects, as follow:
+
+           vendor
+           /    \
+   application   \
+     /    \      |
+ command   \     |
+  |       type   |
+  |       /   \  |
+   \  enumval  \ | 
+    \          avp
+     \    _____/    
+      \  /
+      rule
+
+It means an AVP cannot be defined unless the parent TYPE has already been defined, 
+in turn depending on parent APPLICATION, etc. (top-to-bottom dependencies on the graph)
+
+On the other hand, the hierarchy of the XML format described in draft-frascone-xml-dictionary-00 
+does not enforce most of these dependencies, the structure is as follows:
+
+ vendor     application
+	   /     |     \
+      command    |     avp 
+	 /      type     \
+       rule	        enumval
+
+(in addition if DTD validation was performed, command and avp refer to vendor, avp refers to type, 
+but we do not do it for larger compatibility -- we just report when errors are found)
+
+As a consequence of this difference, it is impossible to parse the XML tree and create the dictionary objects in freeDiameter
+in only 1 pass. To avoid parsing the tree several times, we use a temporary structure in memory to contain all the data
+from the XML file, and when the parsing is complete we store all the objects in the dictionary. 
+*/
+
+/* We use the SAX interface of libxml2 (from GNOME) to parse the XML file. */ 
+#include <libxml/parser.h>
+
+/*******************************************/
+ /* Helper functions */
+static int xmltoint(xmlChar * xmlinteger, uint32_t * conv) {
+	TRACE_ENTRY("%p %p", xmlinteger, conv);
+	
+	/* Attempt at converting the string to an integer */
+	if (sscanf((char *)xmlinteger, "%u", conv) != 1) {
+		TRACE_DEBUG(INFO, "Unable to convert '%s' to integer.", (char *)xmlinteger)
+		return EINVAL;
+	}
+	
+	return 0;
+}
+
+
+/*******************************************
+ The temporary structure that is being built when the XML file is parsed 
+ *******************************************/
+
+/* VENDOR */
+struct t_vend {
+	struct fd_list 	chain; /* link in the t_dictionary->vendors */
+	uint32_t	id;
+	uint8_t *	name;
+};
+
+static int new_vendor(struct fd_list * parent, xmlChar * xmlid, xmlChar * xmlname) {
+	struct t_vend * new;
+	uint32_t id = 0;
+	
+	TRACE_ENTRY("%p %p %p", parent, xmlid, xmlname);
+	CHECK_PARAMS( parent && xmlid && xmlname );
+	
+	CHECK_FCT( xmltoint(xmlid, &id) );
+	
+	CHECK_MALLOC( new = malloc(sizeof(struct t_vend)) );
+	memset(new, 0, sizeof(struct t_vend));
+	fd_list_init(&new->chain, NULL);
+	new->id = id;
+	CHECK_MALLOC( new->name = (uint8_t *)strdup((char *)xmlname) );
+	
+	fd_list_insert_before(parent, &new->chain);
+	
+	return 0;
+}
+
+static void dump_vendor(struct t_vend * v) {
+	fd_log_debug(" Vendor %d:'%s'", v->id, (char *)v->name);
+}
+
+static void del_vendor_contents(struct t_vend * v) {
+	TRACE_ENTRY("%p", v);
+	free(v->name);
+}
+
+
+/* RULE */
+struct t_rule {
+	struct fd_list  chain;	/* link in either t_cmd or t_avp */
+	uint8_t *	avpname;
+	int		max;
+	int		min;
+};
+
+static int new_rule(struct fd_list * parent, xmlChar * xmlname, /* position is never used */ xmlChar * xmlmaximum, xmlChar * xmlminimum) {
+	struct t_rule * new;
+	uint32_t min, max;
+	
+	TRACE_ENTRY("%p %p %p %p", parent, xmlname, xmlmaximum, xmlminimum);
+	CHECK_PARAMS( parent && xmlname );
+	
+	CHECK_MALLOC( new = malloc(sizeof(struct t_rule)) );
+	memset(new, 0, sizeof(struct t_rule));
+	fd_list_init(&new->chain, NULL);
+	if (xmlminimum) {
+		CHECK_FCT( xmltoint(xmlminimum, &min) );
+		new->min = (int) min;
+	} else {
+		new->min = -1;
+	}
+	if (xmlmaximum) {
+		CHECK_FCT( xmltoint(xmlmaximum, &max) );
+		new->max = (int) max;
+	} else {
+		new->max = -1;
+	}
+	CHECK_MALLOC( new->avpname = (uint8_t *)strdup((char *)xmlname) );
+	
+	fd_list_insert_before(parent, &new->chain);
+	
+	return 0;
+}
+
+static void dump_rule(struct t_rule * r, char * prefix) {
+	fd_log_debug("%s ", prefix);
+	if (r->min != -1)
+		fd_log_debug("m:%d ", r->min);
+	if (r->max != -1)
+		fd_log_debug("M:%d ", r->max);
+	fd_log_debug("%s", (char *)r->avpname);
+}
+
+static void del_rule_contents(struct t_rule * r) {
+	TRACE_ENTRY("%p",r);
+	free(r->avpname);
+}
+
+
+/* COMMAND */
+struct t_cmd {
+	struct fd_list  chain;    /* link in t_appl->commands */
+	uint32_t	code;
+	uint8_t *	name;
+	uint8_t		flags;
+	uint8_t		fmask;
+	struct fd_list  reqrules_fixed;     /* list of t_rule */
+	struct fd_list  reqrules_required;  /* list of t_rule */
+	struct fd_list  reqrules_optional;  /* list of t_rule */
+	struct fd_list  ansrules_fixed;     /* list of t_rule */
+	struct fd_list  ansrules_required;  /* list of t_rule */
+	struct fd_list  ansrules_optional;  /* list of t_rule */
+};
+
+static int new_cmd(struct fd_list * parent, xmlChar * xmlcode, xmlChar * xmlname /*, ignore the vendor id because we don't use it */, xmlChar * xmlpbit, struct t_cmd **ret) {
+	struct t_cmd * new;
+	uint32_t code;
+	uint32_t flag = 0;
+	uint32_t fmask = 0;
+	
+	TRACE_ENTRY("%p %p %p %p", parent, xmlcode, xmlname, xmlpbit);
+	CHECK_PARAMS( parent && xmlcode && xmlname );
+	
+	CHECK_FCT( xmltoint(xmlcode, &code) );
+	
+	if (xmlpbit) {
+		uint32_t val;
+		CHECK_FCT( xmltoint(xmlpbit, &val) );
+		fmask |= CMD_FLAG_PROXIABLE;
+		if (val)
+			flag |= CMD_FLAG_PROXIABLE;
+	}
+	
+	CHECK_MALLOC( new = malloc(sizeof(struct t_cmd)) );
+	memset(new, 0, sizeof(struct t_cmd));
+	fd_list_init(&new->chain, NULL);
+	new->code = code;
+	CHECK_MALLOC( new->name = (uint8_t *)strdup((char *)xmlname) );
+	new->flags = flag;
+	new->fmask = fmask;
+	fd_list_init(&new->reqrules_fixed, NULL);
+	fd_list_init(&new->reqrules_required, NULL);
+	fd_list_init(&new->reqrules_optional, NULL);
+	fd_list_init(&new->ansrules_fixed, NULL);
+	fd_list_init(&new->ansrules_required, NULL);
+	fd_list_init(&new->ansrules_optional, NULL);
+	
+	fd_list_insert_before(parent, &new->chain);
+	
+	*ret = new;
+	
+	return 0;
+}
+
+static void dump_cmd(struct t_cmd * c) {
+	struct fd_list * li;
+	fd_log_debug("  Command %d %s: %s", c->code, 
+		c->fmask ? ( c->flags ? "[P=1]" : "[P=0]") : "", c->name);
+	for (li = c->reqrules_fixed.next; li != &c->reqrules_fixed; li = li->next)
+		dump_rule((struct t_rule *)li, "    Request fixed    AVP:");
+	for (li = c->reqrules_required.next; li != &c->reqrules_required; li = li->next)
+		dump_rule((struct t_rule *)li, "    Request required AVP:");
+	for (li = c->reqrules_optional.next; li != &c->reqrules_optional; li = li->next)
+		dump_rule((struct t_rule *)li, "    Request optional AVP:");
+	for (li = c->ansrules_fixed.next; li != &c->ansrules_fixed; li = li->next)
+		dump_rule((struct t_rule *)li, "    Answer fixed    AVP:");
+	for (li = c->ansrules_required.next; li != &c->ansrules_required; li = li->next)
+		dump_rule((struct t_rule *)li, "    Answer required AVP:");
+	for (li = c->ansrules_optional.next; li != &c->ansrules_optional; li = li->next)
+		dump_rule((struct t_rule *)li, "    Answer optional AVP:");
+}
+
+static void del_cmd_contents(struct t_cmd * c) {
+	TRACE_ENTRY("%p", c);
+	free(c->name);
+	while (!FD_IS_LIST_EMPTY(&c->reqrules_fixed)) {
+		struct fd_list * li = c->reqrules_fixed.next;
+		fd_list_unlink(li);
+		del_rule_contents((struct t_rule *)li);
+		free(li);
+	}
+	while (!FD_IS_LIST_EMPTY(&c->reqrules_required)) {
+		struct fd_list * li = c->reqrules_required.next;
+		fd_list_unlink(li);
+		del_rule_contents((struct t_rule *)li);
+		free(li);
+	}
+	while (!FD_IS_LIST_EMPTY(&c->reqrules_optional)) {
+		struct fd_list * li = c->reqrules_optional.next;
+		fd_list_unlink(li);
+		del_rule_contents((struct t_rule *)li);
+		free(li);
+	}
+	while (!FD_IS_LIST_EMPTY(&c->ansrules_fixed)) {
+		struct fd_list * li = c->ansrules_fixed.next;
+		fd_list_unlink(li);
+		del_rule_contents((struct t_rule *)li);
+		free(li);
+	}
+	while (!FD_IS_LIST_EMPTY(&c->ansrules_required)) {
+		struct fd_list * li = c->ansrules_required.next;
+		fd_list_unlink(li);
+		del_rule_contents((struct t_rule *)li);
+		free(li);
+	}
+	while (!FD_IS_LIST_EMPTY(&c->ansrules_optional)) {
+		struct fd_list * li = c->ansrules_optional.next;
+		fd_list_unlink(li);
+		del_rule_contents((struct t_rule *)li);
+		free(li);
+	}
+}
+
+/* TYPE */
+struct t_typedefn {
+	struct fd_list  chain;	/* link in t_appl->types */
+	uint8_t *	name;
+	uint8_t *	parent_name;
+};
+
+static int new_type(struct fd_list * parent, xmlChar * xmlname, xmlChar * xmlparent /*, xmlChar * xmldescription -- ignore */) {
+	struct t_typedefn * new;
+	
+	TRACE_ENTRY("%p %p %p", parent, xmlname, xmlparent);
+	CHECK_PARAMS( parent && xmlname );
+	
+	CHECK_MALLOC( new = malloc(sizeof(struct t_typedefn)) );
+	memset(new, 0, sizeof(struct t_typedefn));
+	fd_list_init(&new->chain, NULL);
+	CHECK_MALLOC( new->name = (uint8_t *)strdup((char *)xmlname) );
+	if (xmlparent) {
+		CHECK_MALLOC( new->parent_name = (uint8_t *)strdup((char *)xmlparent) );
+	}
+	
+	fd_list_insert_before(parent, &new->chain);
+	
+	return 0;
+}
+
+static void dump_type(struct t_typedefn * t) {
+	fd_log_debug("  Type %s%s%s%s", (char *)t->name,
+		     t->parent_name ? "(parent: " : "", 
+		     t->parent_name ? (char *)t->parent_name : "", 
+		     t->parent_name ? ")" : "");
+}
+
+static void del_type_contents(struct t_typedefn * t) {
+	TRACE_ENTRY("%p", t);
+	free(t->name);
+	free(t->parent_name);
+}
+	
+
+/* TYPE INSIDE AVP */
+struct t_avptype {
+	struct fd_list  chain;  /* link in t_avp->type */
+	uint8_t *	type_name;
+};
+
+static int new_avptype(struct fd_list * parent, xmlChar * xmlname) {
+	struct t_avptype * new;
+	
+	TRACE_ENTRY("%p %p", parent, xmlname);
+	CHECK_PARAMS( parent && xmlname );
+	
+	CHECK_MALLOC( new = malloc(sizeof(struct t_avptype)) );
+	memset(new, 0, sizeof(struct t_avptype));
+	fd_list_init(&new->chain, NULL);
+	CHECK_MALLOC( new->type_name = (uint8_t *)strdup((char *)xmlname) );
+	
+	fd_list_insert_before(parent, &new->chain);
+	
+	return 0;
+}
+
+static void dump_avptype(struct t_avptype * t) {
+	fd_log_debug("    data type: %s", t->type_name);
+}
+
+static void del_avptype_contents(struct t_avptype * t) {
+	TRACE_ENTRY("%p", t);
+	free(t->type_name);
+}
+	
+
+/* ENUM */
+struct t_enum {
+	struct fd_list  chain;  /* link in t_avp->enums */
+	uint32_t	code;
+	uint8_t *	name;
+};
+
+static int new_enum(struct fd_list * parent, xmlChar * xmlcode, xmlChar * xmlname) {
+	struct t_enum * new;
+	uint32_t code = 0;
+	
+	TRACE_ENTRY("%p %p %p", parent, xmlcode, xmlname);
+	CHECK_PARAMS( parent && xmlcode && xmlname );
+	
+	CHECK_FCT( xmltoint(xmlcode, &code) );
+	
+	CHECK_MALLOC( new = malloc(sizeof(struct t_enum)) );
+	memset(new, 0, sizeof(struct t_enum));
+	fd_list_init(&new->chain, NULL);
+	new->code = code;
+	CHECK_MALLOC( new->name = (uint8_t *)strdup((char *)xmlname) );
+	
+	fd_list_insert_before(parent, &new->chain);
+	
+	return 0;
+}
+
+static void dump_enum(struct t_enum * e) {
+	fd_log_debug("    Value: %d == %s", e->code, e->name);
+}	
+
+static void del_enum_contents(struct t_enum * e) {
+	TRACE_ENTRY("%p", e);
+	free(e->name);
+}
+
+/* AVP */
+struct t_avp {
+	struct fd_list  chain;  /* link in t_appl->avps */
+	uint32_t	code;
+	uint8_t *	name;
+	uint8_t		flags;
+	uint8_t		fmask;
+	uint32_t	vendor;
+	struct fd_list  type;             /* list of t_avptype -- there must be at max 1 item in the list */
+	struct fd_list  enums;            /* list of t_enum */
+	struct fd_list  grouped_fixed;    /* list of t_rule */
+	struct fd_list  grouped_required; /* list of t_rule */
+	struct fd_list  grouped_optional; /* list of t_rule */
+};
+
+static int new_avp(struct fd_list * parent, xmlChar * xmlcode, xmlChar * xmlname, xmlChar * xmlmandatory, xmlChar * xmlvendor, struct t_avp **ret) {
+	/* we ignore description, may-encrypt, protected, ... */
+	struct t_avp * new;
+	uint32_t code;
+	uint32_t vendor = 0;
+	uint32_t flag = 0;
+	uint32_t fmask = 0;
+	
+	TRACE_ENTRY("%p %p %p %p %p", parent, xmlcode, xmlname, xmlmandatory, xmlvendor);
+	CHECK_PARAMS( parent && xmlcode && xmlname );
+	
+	CHECK_FCT( xmltoint(xmlcode, &code) );
+	
+	if (xmlmandatory && !strcasecmp((char *)xmlmandatory, "must")) {
+		flag |= AVP_FLAG_MANDATORY;
+		fmask |= AVP_FLAG_MANDATORY;
+	}
+	
+	if (xmlvendor) {
+		CHECK_FCT( xmltoint(xmlvendor, &vendor) );
+		if (vendor)
+			flag |= AVP_FLAG_VENDOR;
+		fmask |= AVP_FLAG_VENDOR;
+	}
+	
+	CHECK_MALLOC( new = malloc(sizeof(struct t_avp)) );
+	memset(new, 0, sizeof(struct t_avp));
+	fd_list_init(&new->chain, NULL);
+	new->code = code;
+	CHECK_MALLOC( new->name = (uint8_t *)strdup((char *)xmlname) );
+	new->flags = flag;
+	new->fmask = fmask;
+	new->vendor= vendor; 
+	fd_list_init(&new->type, NULL);
+	fd_list_init(&new->enums, NULL);
+	fd_list_init(&new->grouped_fixed, NULL);
+	fd_list_init(&new->grouped_required, NULL);
+	fd_list_init(&new->grouped_optional, NULL);
+	
+	fd_list_insert_before(parent, &new->chain);
+	
+	*ret = new;
+	
+	return 0;
+}
+
+static void dump_avp(struct t_avp * a) {
+	struct fd_list * li;
+	fd_log_debug("  AVP %d %s%s: %s", a->code, 
+		a->fmask & AVP_FLAG_MANDATORY ? ( a->flags & AVP_FLAG_MANDATORY ? "[M=1]" : "[M=0]") : "", 
+		a->fmask & AVP_FLAG_VENDOR ? ( a->flags & AVP_FLAG_VENDOR ? "[V=1]" : "[V=0]") : "", 
+		a->name);
+	if (a->fmask & AVP_FLAG_VENDOR)
+		fd_log_debug("    vendor: %d", a->vendor);
+	for (li = a->type.next; li != &a->type; li = li->next)
+		dump_avptype((struct t_avptype *)li);
+	for (li = a->enums.next; li != &a->enums; li = li->next)
+		dump_enum((struct t_enum *)li);
+	for (li = a->grouped_fixed.next; li != &a->grouped_fixed; li = li->next)
+		dump_rule((struct t_rule *)li, "    Grouped, fixed    AVP:");
+	for (li = a->grouped_required.next; li != &a->grouped_required; li = li->next)
+		dump_rule((struct t_rule *)li, "    Grouped, required AVP:");
+	for (li = a->grouped_optional.next; li != &a->grouped_optional; li = li->next)
+		dump_rule((struct t_rule *)li, "    Grouped, optional AVP:");
+}
+
+static void del_avp_contents(struct t_avp * a) {
+	TRACE_ENTRY("%p", a);
+	free(a->name);
+	while (!FD_IS_LIST_EMPTY(&a->type)) {
+		struct fd_list * li = a->type.next;
+		fd_list_unlink(li);
+		del_avptype_contents((struct t_avptype *)li);
+		free(li);
+	}
+	while (!FD_IS_LIST_EMPTY(&a->enums)) {
+		struct fd_list * li = a->enums.next;
+		fd_list_unlink(li);
+		del_enum_contents((struct t_enum *)li);
+		free(li);
+	}
+	while (!FD_IS_LIST_EMPTY(&a->grouped_fixed)) {
+		struct fd_list * li = a->grouped_fixed.next;
+		fd_list_unlink(li);
+		del_rule_contents((struct t_rule *)li);
+		free(li);
+	}
+	while (!FD_IS_LIST_EMPTY(&a->grouped_required)) {
+		struct fd_list * li = a->grouped_required.next;
+		fd_list_unlink(li);
+		del_rule_contents((struct t_rule *)li);
+		free(li);
+	}
+	while (!FD_IS_LIST_EMPTY(&a->grouped_optional)) {
+		struct fd_list * li = a->grouped_optional.next;
+		fd_list_unlink(li);
+		del_rule_contents((struct t_rule *)li);
+		free(li);
+	}
+}
+
+
+/* APPLICATION */
+struct t_appl {
+	struct fd_list 	chain; /* link in the t_dictionary->base_and_applications, the sentinel corresponds to "base" */
+	uint32_t	id;
+	uint8_t *	name;
+	struct fd_list  commands; /* list of t_cmd */
+	struct fd_list	types;    /* list of t_typedefn */
+	struct fd_list  avps;     /* list of t_avp */
+};
+
+static int new_appl(struct fd_list * parent, xmlChar * xmlid, xmlChar * xmlname /* We ignore the URI */, struct t_appl **ret) {
+	struct t_appl * new;
+	uint32_t id = 0;
+	
+	TRACE_ENTRY("%p %p %p", parent, xmlid, xmlname);
+	CHECK_PARAMS( parent && xmlid && xmlname );
+	
+	CHECK_FCT( xmltoint(xmlid, &id) );
+	
+	CHECK_MALLOC( new = malloc(sizeof(struct t_appl)) );
+	memset(new, 0, sizeof(struct t_appl));
+	fd_list_init(&new->chain, NULL);
+	new->id = id;
+	CHECK_MALLOC( new->name = (uint8_t *)strdup((char *)xmlname) );
+	
+	fd_list_init(&new->commands, NULL);
+	fd_list_init(&new->types, NULL);
+	fd_list_init(&new->avps, NULL);
+	
+	fd_list_insert_before(parent, &new->chain);
+	
+	*ret = new;
+	
+	return 0;
+}
+
+static void dump_appl(struct t_appl * a) {
+	struct fd_list * li;
+	fd_log_debug(" Application %d: %s", a->id, a->name);
+	for (li = a->commands.next; li != &a->commands; li = li->next)
+		dump_cmd((struct t_cmd *)li);
+	for (li = a->types.next; li != &a->types; li = li->next)
+		dump_type((struct t_typedefn *)li);
+	for (li = a->avps.next; li != &a->avps; li = li->next)
+		dump_avp((struct t_avp *)li);
+}
+
+static void del_appl_contents(struct t_appl * a) {
+	TRACE_ENTRY("%p", a);
+	free(a->name);
+	while (!FD_IS_LIST_EMPTY(&a->commands)) {
+		struct fd_list * li = a->commands.next;
+		fd_list_unlink(li);
+		del_cmd_contents((struct t_cmd *)li);
+		free(li);
+	}
+	while (!FD_IS_LIST_EMPTY(&a->types)) {
+		struct fd_list * li = a->types.next;
+		fd_list_unlink(li);
+		del_type_contents((struct t_typedefn *)li);
+		free(li);
+	}
+	while (!FD_IS_LIST_EMPTY(&a->avps)) {
+		struct fd_list * li = a->avps.next;
+		fd_list_unlink(li);
+		del_avp_contents((struct t_avp *)li);
+		free(li);
+	}
+}
+
+/* DICTIONARY */	
+struct t_dictionary {
+	struct fd_list  vendors;
+	struct t_appl   base_and_applications;
+};
+
+static void dump_dict(struct t_dictionary * d) {
+	struct fd_list * li;
+	for (li = d->vendors.next; li != &d->vendors; li = li->next)
+		dump_vendor((struct t_vend *)li);
+	dump_appl(&d->base_and_applications);
+	for (li = d->base_and_applications.chain.next; li != &d->base_and_applications.chain; li = li->next)
+		dump_appl((struct t_appl *)li);
+}
+
+static void del_dict_contents(struct t_dictionary * d) {
+	TRACE_ENTRY("%p", d);
+	while (!FD_IS_LIST_EMPTY(&d->vendors)) {
+		struct fd_list * li = d->vendors.next;
+		fd_list_unlink(li);
+		del_vendor_contents((struct t_vend *)li);
+		free(li);
+	}
+	while (!FD_IS_LIST_EMPTY(&d->base_and_applications.chain)) {
+		struct fd_list * li = d->base_and_applications.chain.next;
+		fd_list_unlink(li);
+		del_appl_contents((struct t_appl *)li);
+		free(li);
+	}
+	d->base_and_applications.name = NULL;
+	del_appl_contents(&d->base_and_applications);
+}
+
+/*********************************************/
+	
+/* The states for the SAX parser, corresponding roughly to the expected structure of the XML file. 
+We use the states mostly to validate the XML file. */
+enum state {
+	INIT = 0,
+	START, /* In "dictionary" */
+	 IN_VENDOR,
+	 IN_APPLICATION,        /* note that "base" is equivalent to "application" for our state machine */
+	  IN_COMMAND,
+	   IN_REQRULES,
+	    IN_REQRULES_FIXED,
+	    IN_REQRULES_REQUIRED,
+	    IN_REQRULES_OPTIONAL,
+	   IN_ANSRULES,
+	    IN_ANSRULES_FIXED,
+	    IN_ANSRULES_REQUIRED,
+	    IN_ANSRULES_OPTIONAL,
+	  IN_TYPEDEFN,
+	  IN_AVP,
+	   IN_AVP_TYPE,
+	   IN_AVP_ENUM,
+	   IN_AVP_GROUPED,
+	    IN_AVP_GROUPED_FIXED,
+	    IN_AVP_GROUPED_REQUIRED,
+	    IN_AVP_GROUPED_OPTIONAL
+};
+
+
+/* The context passed to the SAX parser */
+struct parser_ctx {
+	enum state 		state; 	     /* the current state */
+	int        		error_depth; /* if non 0, we are in an unexpected element, wait until the count goes back to 0 to resume normal parsing. */
+	struct t_dictionary 	dict;        /* The dictionary being built */
+	struct t_appl *		cur_app;
+	struct t_cmd  *		cur_cmd;
+	struct t_avp  *		cur_avp;
+	char * 			xmlfilename; /* Name of the file, for error messages */
+};
+
+/* Find an attribute with given name in the list */
+static void get_attr(const xmlChar ** atts_array, const char * attr_name, xmlChar ** attr_val) {
+	int i;
+	*attr_val = NULL;
+	if (atts_array == NULL)
+		return;
+	for (i=0; atts_array[i] != NULL; i+=2) {
+		if (!strcasecmp((char *)atts_array[i], attr_name)) {
+			/* found */
+			*attr_val = (xmlChar *)atts_array[i+1];
+			return;
+		}
+	}
+	/* not found */
+	return;
+}
+
+/* The following macro avoids duplicating a lot of code in the state machine */
+#define ADD_RULE( _parent_list ) { 				\
+	xmlChar *xname, *xmin, *xmax;				\
+	/* We are expecting an <avprule> tag at this point */	\
+	if (strcasecmp((char *)name, "avprule"))		\
+		goto xml_tree_error;				\
+	/* Search the expected attributes */			\
+	get_attr(atts, "name", &xname);				\
+	get_attr(atts, "maximum", &xmax);			\
+	get_attr(atts, "minimum", &xmin);			\
+	/* Check the mandatory name is here */			\
+	CHECK_PARAMS_DO(xname, 					\
+		{ TRACE_DEBUG(INFO, "Invalid 'avprule' tag found without 'name' attribute."); goto xml_tree_error; } );	\
+	/* Create the rule and add into the parent list */	\
+	CHECK_FCT_DO( new_rule((_parent_list), xname, xmax, xmin),\
+		{ TRACE_DEBUG(INFO, "An error occurred while parsing an avprule tag. Entry ignored."); goto xml_tree_error; } ); \
+	/* Done. we don't change the state */			\
+}
+	
+	
+/* The function called on each XML element start tag (startElementSAXFunc) */
+static void SAXstartelem (void * ctx, const xmlChar * name, const xmlChar ** atts)
+{
+	struct parser_ctx * data = ctx;
+	TRACE_ENTRY("%p %p %p", ctx, name, atts);
+	CHECK_PARAMS_DO( ctx && name, { return; } );
+	
+	TRACE_DEBUG(CALL, "Tag: <%s>", (char *)name);
+	
+	if (data->error_depth) /* we are in an unknown element, just skip until it is closed */
+		goto xml_tree_error;
+	
+	switch (data->state) {
+		case INIT: /* we are just starting. We only expect a <dictionary> tag, reject anything else. */
+			if (strcasecmp((char *)name, "dictionary"))
+				goto xml_tree_error;
+			
+			data->state = START;
+			break;
+			
+		case START: 
+			/* We are in <dictionary> 
+				Valid tags are: <vendor>, <base>, <application> */
+			if (!strcasecmp((char *)name, "vendor")) {
+				xmlChar *xid, *xname;
+				
+				get_attr(atts, "id", &xid);
+				get_attr(atts, "name", &xname);
+				
+				/* id and name are required */
+				CHECK_PARAMS_DO(xid && xname, 
+					{ TRACE_DEBUG(INFO, "Invalid 'vendor' tag found without 'id' or 'name' attribute."); goto xml_tree_error; } );
+				
+				
+				CHECK_FCT_DO( new_vendor(&data->dict.vendors, xid, xname),
+					{ TRACE_DEBUG(INFO, "An error occurred while parsing a vendor tag. Entry ignored."); goto xml_tree_error; } )
+				
+				data->state = IN_VENDOR;
+				break;
+			}
+			
+			if (!strcasecmp((char *)name, "base")) {
+				/* we don't care for the 'uri' attribute */
+				data->cur_app = &data->dict.base_and_applications;
+				data->state = IN_APPLICATION;
+				break;
+			}
+			
+			if (!strcasecmp((char *)name, "application")) {
+				/* we don't care for the 'uri' attribute */
+				xmlChar *xid, *xname;
+				char buf[50];
+				
+				get_attr(atts, "id", &xid);
+				get_attr(atts, "name", &xname);
+				
+				CHECK_PARAMS_DO(xid, 
+					{ TRACE_DEBUG(INFO, "Invalid 'application' tag found without 'id' attribute."); goto xml_tree_error; } );
+				
+				/* Name is optional, if not provided we create a name */
+				if (!xname) {
+					snprintf(buf, sizeof(buf), "Application %s", xid);
+					xname = (xmlChar *)buf;
+				}
+				
+				CHECK_FCT_DO( new_appl(&data->dict.base_and_applications.chain, xid, xname, &data->cur_app),
+					{ TRACE_DEBUG(INFO, "An error occurred while parsing an application tag. Entry ignored."); goto xml_tree_error; } )
+				
+				data->state = IN_APPLICATION;
+				break;
+			}
+			
+			/* Other tags are errors */
+			goto xml_tree_error;
+			
+			
+		case IN_VENDOR: /* nothing is allowed inside <vendor> */
+			goto xml_tree_error;
+				
+		case IN_APPLICATION: 
+			/* We are in <base> or <application>
+				Valid tags are: <command>, <typedefn>, <avp> */
+			if (!strcasecmp((char *)name, "command")) {
+				/* we don't care for the 'vendor-id' attribute. */
+				xmlChar *xcode, *xname, *xpbit;
+				
+				get_attr(atts, "code", &xcode);
+				get_attr(atts, "name", &xname);
+				get_attr(atts, "pbit", &xpbit);
+				
+				/* code and name are required */
+				CHECK_PARAMS_DO(xcode && xname, 
+					{ TRACE_DEBUG(INFO, "Invalid 'command' tag found without 'code' or 'name' attribute."); goto xml_tree_error; } );
+				
+				CHECK_FCT_DO( new_cmd( &data->cur_app->commands, xcode, xname, xpbit, &data->cur_cmd),
+					{ TRACE_DEBUG(INFO, "An error occurred while parsing a command tag. Entry ignored."); goto xml_tree_error; } )
+				
+				data->state = IN_COMMAND;
+				break;
+			}
+			
+			if (!strcasecmp((char *)name, "typedefn")) {
+				/* we don't care for the 'description' attribute. */
+				xmlChar *xname, *xparent;
+				
+				get_attr(atts, "type-name", &xname);
+				get_attr(atts, "type-parent", &xparent);
+				
+				/* name is required */
+				CHECK_PARAMS_DO(xname, 
+					{ TRACE_DEBUG(INFO, "Invalid 'typedefn' tag found without 'name' attribute."); goto xml_tree_error; } );
+				
+				CHECK_FCT_DO( new_type( &data->cur_app->types, xname, xparent),
+					{ TRACE_DEBUG(INFO, "An error occurred while parsing a typedefn tag. Entry ignored."); goto xml_tree_error; } )
+				
+				data->state = IN_TYPEDEFN;
+				break;
+			}
+			
+			if (!strcasecmp((char *)name, "avp")) {
+				/* we don't care for the description, may-encrypt, and protected attributes */
+				xmlChar *xname, *xcode, *xmandatory, *xvendor;
+				
+				get_attr(atts, "name", &xname);
+				get_attr(atts, "code", &xcode);
+				get_attr(atts, "mandatory", &xmandatory);
+				get_attr(atts, "vendor-id", &xvendor);
+				
+				/* code and name are required */
+				CHECK_PARAMS_DO(xcode && xname, 
+					{ TRACE_DEBUG(INFO, "Invalid 'avp' tag found without 'code' or 'name' attribute."); goto xml_tree_error; } );
+				
+				CHECK_FCT_DO( new_avp(&data->cur_app->avps, xcode, xname, xmandatory, xvendor, &data->cur_avp),
+					{ TRACE_DEBUG(INFO, "An error occurred while parsing an avp tag. Entry ignored."); goto xml_tree_error; } )
+				
+				data->state = IN_AVP;
+				break;
+			}
+			/* Other tags are errors */
+			goto xml_tree_error;
+		
+			
+		case IN_COMMAND: 
+			/* We are in <command>
+				Valid tags are: <requestrules>, <answerrules> */
+			if (!strcasecmp((char *)name, "requestrules")) {
+				data->state = IN_REQRULES;
+				break;
+			}
+			if (!strcasecmp((char *)name, "answerrules")) {
+				data->state = IN_ANSRULES;
+				break;
+			}
+			/* Other tags are errors */
+			goto xml_tree_error;
+		
+		case IN_REQRULES: 
+			/* We are in <requestrules>
+				Valid tags are: <fixed>, <required>, <optional> */
+			if (!strcasecmp((char *)name, "fixed")) {
+				data->state = IN_REQRULES_FIXED;
+				break;
+			}
+			if (!strcasecmp((char *)name, "required")) {
+				data->state = IN_REQRULES_REQUIRED;
+				break;
+			}
+			if (!strcasecmp((char *)name, "optional")) {
+				data->state = IN_REQRULES_OPTIONAL;
+				break;
+			}
+			/* Other tags are errors */
+			goto xml_tree_error;
+		
+		case IN_ANSRULES: 
+			/* We are in <answerrules>
+				Valid tags are: <fixed>, <required>, <optional> */
+			if (!strcasecmp((char *)name, "fixed")) {
+				data->state = IN_ANSRULES_FIXED;
+				break;
+			}
+			if (!strcasecmp((char *)name, "required")) {
+				data->state = IN_ANSRULES_REQUIRED;
+				break;
+			}
+			if (!strcasecmp((char *)name, "optional")) {
+				data->state = IN_ANSRULES_OPTIONAL;
+				break;
+			}
+			/* Other tags are errors */
+			goto xml_tree_error;
+			
+		case IN_REQRULES_FIXED:
+			/* We are in <command><answerrules><fixed>
+				Valid tags are: <avprule> */
+			ADD_RULE( &data->cur_cmd->reqrules_fixed );
+			break;
+		case IN_REQRULES_REQUIRED:
+			ADD_RULE( &data->cur_cmd->reqrules_required );
+			break;
+		case IN_REQRULES_OPTIONAL:
+			ADD_RULE( &data->cur_cmd->reqrules_optional );
+			break;
+		case IN_ANSRULES_FIXED:
+			ADD_RULE( &data->cur_cmd->ansrules_fixed );
+			break;
+		case IN_ANSRULES_REQUIRED:
+			ADD_RULE( &data->cur_cmd->ansrules_required );
+			break;
+		case IN_ANSRULES_OPTIONAL:
+			ADD_RULE( &data->cur_cmd->ansrules_optional );
+			break;
+			
+		
+		case IN_TYPEDEFN: /* nothing is allowed inside <typedefn> */
+			goto xml_tree_error;
+				
+		
+		case IN_AVP: 
+			/* We are in <avp>
+				Valid tags are: <type>, <enum>, <grouped> */
+			if (!strcasecmp((char *)name, "type")) {
+				xmlChar *xname;
+				
+				get_attr(atts, "type-name", &xname);
+				
+				/* name is required */
+				CHECK_PARAMS_DO(xname, 
+					{ TRACE_DEBUG(INFO, "Invalid 'type' tag found without 'name' attribute."); goto xml_tree_error; } );
+				
+				/* Check there is only 1 type */
+				if (!FD_IS_LIST_EMPTY(&data->cur_avp->type)) {
+					TRACE_DEBUG(INFO, "Multiple 'type' tags found for AVP.");
+					goto xml_tree_error;
+				}
+				
+				/* Add the new type */
+				CHECK_FCT_DO( new_avptype(&data->cur_avp->type, xname),
+					{ TRACE_DEBUG(INFO, "An error occurred while parsing a type tag. Entry ignored."); goto xml_tree_error; } )
+				
+				data->state = IN_AVP_TYPE;
+				break;
+			}
+			if (!strcasecmp((char *)name, "enum")) {
+				xmlChar *xcode, *xname;
+				
+				get_attr(atts, "code", &xcode);
+				get_attr(atts, "name", &xname);
+				
+				/* code and name are required */
+				CHECK_PARAMS_DO(xcode && xname, 
+					{ TRACE_DEBUG(INFO, "Invalid 'enum' tag found without 'code' or 'name' attribute."); goto xml_tree_error; } );
+				
+				CHECK_FCT_DO( new_enum(&data->cur_avp->enums, xcode, xname),
+					{ TRACE_DEBUG(INFO, "An error occurred while parsing a command tag. Entry ignored."); goto xml_tree_error; } )
+			
+				data->state = IN_AVP_ENUM;
+				break;
+			}
+			if (!strcasecmp((char *)name, "grouped")) {
+				/* no attribute for this one */
+				data->state = IN_AVP_GROUPED;
+				break;
+			}
+			/* Other tags are errors */
+			goto xml_tree_error;
+		
+		case IN_AVP_TYPE: /* nothing is allowed inside <type> */
+			goto xml_tree_error;
+			
+		case IN_AVP_ENUM: /* nothing is allowed inside <enum> */
+			goto xml_tree_error;
+			
+		case IN_AVP_GROUPED: 
+			/* We are in <avp><grouped>
+				Valid tags are: <fixed>, <required>, <optional> */
+			if (!strcasecmp((char *)name, "fixed")) {
+				data->state = IN_AVP_GROUPED_FIXED;
+				break;
+			}
+			if (!strcasecmp((char *)name, "required")) {
+				data->state = IN_AVP_GROUPED_REQUIRED;
+				break;
+			}
+			if (!strcasecmp((char *)name, "optional")) {
+				data->state = IN_AVP_GROUPED_OPTIONAL;
+				break;
+			}
+			/* Other tags are errors */
+			goto xml_tree_error;
+			
+		case IN_AVP_GROUPED_FIXED:
+			/* We are in <avp><grouped><fixed>
+				Valid tags are: <avprule> */
+			ADD_RULE( &data->cur_avp->grouped_fixed );
+			break;
+		case IN_AVP_GROUPED_REQUIRED:
+			ADD_RULE( &data->cur_avp->grouped_required );
+			break;
+		case IN_AVP_GROUPED_OPTIONAL:
+			ADD_RULE( &data->cur_avp->grouped_optional );
+			break;
+		
+			
+		default:
+			TRACE_DEBUG(INFO, "Internal parsing error, unexpected state %d.", data->state);
+	}
+	
+	return;
+	
+xml_tree_error:
+	if (!data->error_depth) {
+		TRACE_DEBUG(INFO, "Unexpected XML element found: '%s'. Ignoring...", name);
+	}
+	data->error_depth += 1;
+	if (data->cur_app || data->cur_cmd || data->cur_avp) {
+		TRACE_DEBUG(INFO, "Error encountered while parsing tag of:");
+		if (data->cur_app)
+			fd_log_debug("  Application: '%s'", data->cur_app->name);
+		if (data->cur_cmd)
+			fd_log_debug("  Command    : '%s'", data->cur_cmd->name);
+		if (data->cur_avp)
+			fd_log_debug("  AVP        : '%s'", data->cur_avp->name);
+	}
+	return;
+}
+
+/* The function called on each XML element end tag (endElementSAXFunc) */
+static void SAXendelem (void * ctx, const xmlChar * name)
+{
+	struct parser_ctx * data = ctx;
+	TRACE_ENTRY("%p %p", ctx, name);
+	CHECK_PARAMS_DO( ctx && name, { return; } );
+	
+	TRACE_DEBUG(CALL, "Tag: </%s>", (char *)name);
+	
+	if (data->error_depth) {
+		/* we are recovering from an erroneous element */
+		data->error_depth -= 1;
+		return;
+	}
+	
+	switch (data->state) {
+		case INIT: 
+			goto state_machine_error;
+			
+		case START:
+			if (strcasecmp((char *)name, "dictionary"))
+				goto state_machine_error;
+			
+			data->state = 0;
+			break;
+		
+		case IN_VENDOR:
+			if (strcasecmp((char *)name, "vendor"))
+				goto state_machine_error;
+			
+			data->state = START;
+			break;
+		
+		case IN_APPLICATION:
+			if (strcasecmp((char *)name, "base") && strcasecmp((char *)name, "application"))
+				goto state_machine_error;
+			
+			data->cur_app = NULL;
+			data->state = START;
+			break;
+		
+		case IN_COMMAND:
+			if (strcasecmp((char *)name, "command"))
+				goto state_machine_error;
+			
+			data->cur_cmd = NULL;
+			data->state = IN_APPLICATION;
+			break;
+			
+		case IN_REQRULES:
+			if (strcasecmp((char *)name, "requestrules"))
+				goto state_machine_error;
+			
+			data->state = IN_COMMAND;
+			break;
+		
+		case IN_REQRULES_FIXED:
+			if (!strcasecmp((char *)name, "avprule"))
+				/* we don't have a special state for these, just ignore */
+				return;
+			if (strcasecmp((char *)name, "fixed"))
+				goto state_machine_error;
+			data->state = IN_REQRULES;
+			break;
+		case IN_REQRULES_REQUIRED:
+			if (!strcasecmp((char *)name, "avprule"))
+				/* we don't have a special state for these, just ignore */
+				return;
+			if (strcasecmp((char *)name, "required"))
+				goto state_machine_error;
+			data->state = IN_REQRULES;
+			break;
+		case IN_REQRULES_OPTIONAL:
+			if (!strcasecmp((char *)name, "avprule"))
+				/* we don't have a special state for these, just ignore */
+				return;
+			if (strcasecmp((char *)name, "optional"))
+				goto state_machine_error;
+			data->state = IN_REQRULES;
+			break;
+			
+		case IN_ANSRULES: 
+			if (strcasecmp((char *)name, "answerrules"))
+				goto state_machine_error;
+			
+			data->state = IN_COMMAND;
+			break;			
+		case IN_ANSRULES_FIXED:
+			if (!strcasecmp((char *)name, "avprule"))
+				/* we don't have a special state for these, just ignore */
+				return;
+			if (strcasecmp((char *)name, "fixed"))
+				goto state_machine_error;
+			data->state = IN_ANSRULES;
+			break;
+		case IN_ANSRULES_REQUIRED:
+			if (!strcasecmp((char *)name, "avprule"))
+				/* we don't have a special state for these, just ignore */
+				return;
+			if (strcasecmp((char *)name, "required"))
+				goto state_machine_error;
+			data->state = IN_ANSRULES;
+			break;
+		case IN_ANSRULES_OPTIONAL:
+			if (!strcasecmp((char *)name, "avprule"))
+				/* we don't have a special state for these, just ignore */
+				return;
+			if (strcasecmp((char *)name, "optional"))
+				goto state_machine_error;
+			data->state = IN_ANSRULES;
+			break;
+		
+		
+		case IN_TYPEDEFN:
+			if (strcasecmp((char *)name, "typedefn"))
+				goto state_machine_error;
+			
+			data->state = IN_APPLICATION;
+			break;
+		
+		case IN_AVP: 
+			if (strcasecmp((char *)name, "avp"))
+				goto state_machine_error;
+			
+			data->cur_avp = NULL;
+			data->state = IN_APPLICATION;
+			break;
+		
+		case IN_AVP_TYPE:
+			if (strcasecmp((char *)name, "type"))
+				goto state_machine_error;
+			
+			data->state = IN_AVP;
+			break;
+		
+		case IN_AVP_ENUM:
+			if (strcasecmp((char *)name, "enum"))
+				goto state_machine_error;
+			
+			data->state = IN_AVP;
+			break;
+		
+		case IN_AVP_GROUPED:
+			if (strcasecmp((char *)name, "grouped"))
+				goto state_machine_error;
+			
+			data->state = IN_AVP;
+			break;
+		
+		case IN_AVP_GROUPED_FIXED:
+			if (!strcasecmp((char *)name, "avprule"))
+				/* we don't have a special state for these, just ignore */
+				return;
+			if (strcasecmp((char *)name, "fixed"))
+				goto state_machine_error;
+			data->state = IN_AVP_GROUPED;
+			break;
+		case IN_AVP_GROUPED_REQUIRED:
+			if (!strcasecmp((char *)name, "avprule"))
+				return;
+			if (strcasecmp((char *)name, "required"))
+				goto state_machine_error;
+			data->state = IN_AVP_GROUPED;
+			break;
+		case IN_AVP_GROUPED_OPTIONAL:
+			if (!strcasecmp((char *)name, "avprule"))
+				return;
+			if (strcasecmp((char *)name, "optional"))
+				goto state_machine_error;
+			data->state = IN_AVP_GROUPED;
+			break;
+			
+		default:
+			TRACE_DEBUG(INFO, "Internal parsing error, unexpected state %d.", data->state);
+	}
+	
+	return;
+	
+state_machine_error:
+	TRACE_DEBUG(INFO, "Internal parsing error, ignored [state %d, closing tag '%s'].", data->state, name);
+	return;
+}
+
+/* The SAX parser sends a warning, error, fatalerror -- do we need these ?
+static void SAXwarning (void * ctx, const char * msg, ...)
+{
+
+}
+static void SAXerror (void * ctx, const char * msg, ...)
+{
+
+}
+static void SAXfatal (void * ctx, const char * msg, ...)
+{
+
+}
+*/
+
+
+
+
+/*********************************************/
+ /* 2nd pass: from memory to fD dictionary */
+/*********************************************/
+
+/* Find or create a vendor */
+static int vend_to_fD(struct t_vend * v, struct dictionary * fD_dict, struct dict_object ** fd_v, int * nb_added)
+{
+	int ret;
+	struct dict_object * prev = NULL;
+	struct dict_vendor_data vd;
+
+	TRACE_ENTRY("%p %p %p %p", v, fD_dict, fd_v, nb_added);
+	
+	CHECK_PARAMS(v && fD_dict);
+	
+	/* Prepare the data in fD's format */
+	memset(&vd, 0, sizeof(vd));
+	vd.vendor_id   = v->id;
+	vd.vendor_name = (char *)v->name;
+	
+	/* Create or search in the dictionary */
+	ret = fd_dict_new ( fD_dict, DICT_VENDOR, &vd, NULL, &prev );
+	if (fd_v)
+		*fd_v = prev;
+	if (ret == EEXIST) {
+		/* Conflict with existing entry */
+		CHECK_FCT( fd_dict_getval(prev, &vd) );
+		TRACE_DEBUG(INFO, "[dict_legacy_xml] Warning: Conflicting entry.");
+		TRACE_DEBUG(INFO, "[dict_legacy_xml]   New entry (ignored): %u - '%s'", v->id, (char *)v->name);
+		TRACE_DEBUG(INFO, "[dict_legacy_xml]   Old entry          : %u - '%s'", vd.vendor_id, vd.vendor_name);
+		return 0;
+	} else {
+		/* other errors are stoppers */
+		CHECK_FCT(ret);
+	}
+	
+	/* Update count */
+	if (nb_added)
+		*nb_added += 1;
+	
+	/* Done */
+	return 0;
+}
+
+/* Find the base fD type from a type name */
+static int resolve_base_type(struct dictionary * fD_dict, uint8_t * type_name, enum dict_avp_basetype * basetype, struct dict_object **type)
+{
+	int ret;
+	struct dict_type_data td;
+	struct dict_object *t;
+	
+	TRACE_ENTRY("%p, %p %p", fD_dict, type_name, basetype);
+	CHECK_PARAMS( fD_dict && type_name && basetype );
+	
+	/* First, check if the type is already in the dictionary */
+	ret = fd_dict_search ( fD_dict, DICT_TYPE, TYPE_BY_NAME, type_name, &t, ENOENT);
+	switch (ret) {
+		case 0: /* the type is already in the dictionary */
+			CHECK_FCT( fd_dict_getval(t, &td) );
+			*basetype = td.type_base;
+			if (type)
+				*type = t;
+			return 0;
+		
+		case ENOENT: /* We did not find it, it is maybe normal */
+			break;
+			
+		default:
+			/* An unexpected error occurred */
+			CHECK_FCT(ret);
+	}
+	
+	/* at this point we did not find the type in the dictionary */
+#define PREDEF_TYPES( _typename_, _basetype_ )			\
+	if (!strcasecmp((char *)type_name, (_typename_))) {	\
+		*basetype = (_basetype_);			\
+		return 0;					\
+	}
+	
+	PREDEF_TYPES( "OctetString", AVP_TYPE_OCTETSTRING );
+	PREDEF_TYPES( "Integer32",   AVP_TYPE_INTEGER32   );
+	PREDEF_TYPES( "Integer64",   AVP_TYPE_INTEGER64   );
+	PREDEF_TYPES( "Unsigned32",  AVP_TYPE_UNSIGNED32  );
+	PREDEF_TYPES( "Enumerated",  AVP_TYPE_INTEGER32  );
+	PREDEF_TYPES( "Unsigned64",  AVP_TYPE_UNSIGNED64  );
+	PREDEF_TYPES( "Float32",     AVP_TYPE_FLOAT32     );
+	PREDEF_TYPES( "Float64",     AVP_TYPE_FLOAT64     );
+	
+	/* When we reach this point, we have not yet found this type anywhere. */
+	TODO("Type not found. Maybe search in whole xmldictionary if it is defined later?");
+	TRACE_DEBUG(INFO, "The type '%s' could not be resolved. Please check it is defined before use.", type_name);
+	return ENOENT;
+}
+
+/* Find or create a type. */
+static int typdefn_to_fD(struct t_typedefn * t, struct dictionary * fD_dict, struct dict_object * fd_appl, struct dict_object ** fd_t, int * nb_added)
+{
+	int ret;
+	struct dict_object * prev = NULL;
+	struct dict_type_data td;
+
+	TRACE_ENTRY("%p %p %p %p %p", t, fD_dict, fd_appl, fd_t, nb_added);
+	
+	CHECK_PARAMS(t && fD_dict);
+	
+	/* Prepare the data in fD's format */
+	memset(&td, 0, sizeof(td));
+	td.type_name = (char *)t->name;
+	
+	/* infer td.type_base from t->parent_name */
+	CHECK_FCT( resolve_base_type(fD_dict, t->parent_name, &td.type_base, NULL) );
+	
+	/* Create or search in the dictionary */
+	ret = fd_dict_new ( fD_dict, DICT_TYPE, &td, fd_appl, &prev );
+	if (fd_t)
+		*fd_t = prev;
+	if (ret == EEXIST) {
+		/* Conflict with existing entry */
+		enum dict_avp_basetype xmlbt = td.type_base;
+		extern const char * type_base_name[]; /* in libfreeDiameter/dictionary.c */
+		CHECK_FCT( fd_dict_getval(prev, &td) );
+		TRACE_DEBUG(INFO, "[dict_legacy_xml] Warning: Conflicting entry.");
+		TRACE_DEBUG(INFO, "[dict_legacy_xml]   New entry (ignored): '%s' (%d - %s)", t->name, xmlbt, type_base_name[xmlbt] );
+		TRACE_DEBUG(INFO, "[dict_legacy_xml]   Old entry          : '%s' (%d - %s)", td.type_name, td.type_base, type_base_name[td.type_base]);
+		return 0;
+	} else {
+		/* other errors are stoppers */
+		CHECK_FCT(ret);
+	}
+	
+	/* Update count */
+	if (nb_added)
+		*nb_added += 1;
+	
+	/* Done */
+	return 0;
+}
+
+/* Process one list of rules */
+static int rules_to_fD_onelist(struct dictionary * fD_dict, struct dict_object * parent, enum rule_position position, struct fd_list * list, int * nb_added)
+{
+	struct dict_rule_data rd;
+	struct fd_list * li;
+	int order = 0;
+	int ret;
+
+	TRACE_ENTRY("%p %p %d %p %p", fD_dict, parent, position, list, nb_added);
+	
+	CHECK_PARAMS(fD_dict && parent && position && list);
+	
+	for (li = list->next; li != list; li = li->next) {
+		struct t_rule * r = (struct t_rule *)li;
+		
+		/* The [AVP] rule in all ABNF definitions is implicit in freeDiameter, skip it */
+		if (!strcmp((char *)r->avpname, "AVP"))
+			continue;
+		
+		/* Prepare rule data */
+		memset(&rd, 0, sizeof(rd));
+		rd.rule_position = position;
+		rd.rule_order = ++order; /* actually only used for fixed rules, but no harm for others */
+		rd.rule_min = r->min;
+		rd.rule_max = r->max;
+		
+		/* Resolve the AVP */
+		ret = fd_dict_search(fD_dict, DICT_AVP, AVP_BY_NAME_ALL_VENDORS, r->avpname, &rd.rule_avp, ENOENT);
+		if (ret == ENOENT) {
+			TRACE_DEBUG(INFO, "[dict_legacy_xml] Error: AVP '%s' used in a rule before being defined.", r->avpname);
+		}
+		CHECK_FCT(ret);
+		
+		/* Now create the new rule */
+		CHECK_FCT_DO( ret = fd_dict_new ( fD_dict, DICT_RULE, &rd, parent, NULL ),
+			{ TRACE_DEBUG(INFO, "Error creating rule for sub-AVP '%s'", r->avpname); return ret; } );
+		if (nb_added)
+			*nb_added += 1;
+	}
+	
+	return 0;
+}
+
+/* Process lists of rules */
+static int rules_to_fD(struct dictionary * fD_dict, struct dict_object * parent, struct fd_list * fixed, struct fd_list * required, struct fd_list * optional, int * nb_added)
+{
+	int ret;
+	
+	TRACE_ENTRY("%p %p %p %p %p %p", fD_dict, parent, fixed, required, optional, nb_added);
+	
+	/* Process the rules */
+	CHECK_FCT_DO( ret = rules_to_fD_onelist(fD_dict, parent, RULE_FIXED_HEAD, fixed, nb_added),
+		{ TRACE_DEBUG(INFO, "Error processing FIXED rules"); return ret; } );
+	CHECK_FCT_DO( ret = rules_to_fD_onelist(fD_dict, parent, RULE_REQUIRED, required, nb_added),
+		{ TRACE_DEBUG(INFO, "Error processing REQUIRED rules"); return ret; } );
+	CHECK_FCT_DO( ret = rules_to_fD_onelist(fD_dict, parent, RULE_OPTIONAL, optional, nb_added),
+		{ TRACE_DEBUG(INFO, "Error processing OPTIONAL rules"); return ret; } );
+	
+	return 0;
+}
+
+/* Find or create an AVP (and dependent objects) */
+static int avp_to_fD(struct t_avp * a, struct dictionary * fD_dict, struct dict_object * fd_appl, struct dict_object ** fd_a, int * nb_added)
+{
+	int ret;
+	struct dict_object * prev = NULL, *type = NULL;
+	struct dict_avp_data ad;
+	struct fd_list * li;
+
+	TRACE_ENTRY("%p %p %p %p %p", a, fD_dict, fd_appl, fd_a, nb_added);
+	
+	CHECK_PARAMS(a && fD_dict);
+	
+	/* Prepare the data in fD's format */
+	memset(&ad, 0, sizeof(ad));
+	ad.avp_code   = a->code;
+	ad.avp_vendor = a->vendor;
+	ad.avp_name   = (char *)a->name;
+	ad.avp_flag_mask = a->fmask | AVP_FLAG_VENDOR;
+	ad.avp_flag_val  = a->flags;
+	
+	if (!FD_IS_LIST_EMPTY(&a->type)) {
+		/* special exception: we use per-AVP enumerated types in fD */
+		if (!strcasecmp("Enumerated", (char *)((struct t_avptype *)a->type.next)->type_name))
+			goto enumerated;
+		/* Let's allow "Integer32" instead of "Enumerated" also... */
+		if ((!FD_IS_LIST_EMPTY(&a->enums)) && (!strcasecmp("Integer32", (char *)((struct t_avptype *)a->type.next)->type_name)))
+			goto enumerated;
+		
+		/* The type was explicitly specified, resolve it */
+		CHECK_FCT( resolve_base_type(fD_dict, ((struct t_avptype *)a->type.next)->type_name, &ad.avp_basetype, &type) );
+	} else {
+		/* The type was not specified, try to infer it from provided data */
+		if (       !FD_IS_LIST_EMPTY(&a->grouped_optional)
+			|| !FD_IS_LIST_EMPTY(&a->grouped_required)
+			|| !FD_IS_LIST_EMPTY(&a->grouped_fixed) ) {
+			/* The AVP has rules, it is a grouped AVP */
+			CHECK_PARAMS_DO( FD_IS_LIST_EMPTY(&a->enums), 
+				{ TRACE_DEBUG(INFO, "Conflict: The AVP '%s' has both enum values and rules.", ad.avp_name); return EINVAL; } );
+			ad.avp_basetype = AVP_TYPE_GROUPED;
+		} else {
+			/* It should be an enumerated AVP... */
+			if (FD_IS_LIST_EMPTY(&a->enums)) {
+				TRACE_DEBUG(INFO, "Error: Missing type information for AVP '%s'", ad.avp_name);
+				return EINVAL;
+			} else {
+				/* We create a new type to hold the enumerated values -- fD specifics */
+				char typename[256];
+				struct dict_type_data 	tdata;
+				
+enumerated:
+				snprintf(typename, sizeof(typename), "Enumerated(%s)", ad.avp_name);
+				memset(&tdata, 0, sizeof(tdata));
+				tdata.type_base = AVP_TYPE_INTEGER32;
+				tdata.type_name = &typename[0];
+				CHECK_FCT( fd_dict_new ( fD_dict, DICT_TYPE, &tdata, fd_appl, &type ) );
+				if (nb_added)
+					*nb_added += 1;
+				
+				ad.avp_basetype = AVP_TYPE_INTEGER32;
+			}
+		}
+	}
+	
+	/* At this point, ad.avp_basetype is defined and type might also be */
+	
+	/* Create or search in the dictionary */
+	ret = fd_dict_new ( fD_dict, DICT_AVP, &ad, type, &prev );
+	if (fd_a)
+		*fd_a = prev;
+	if (ret == EEXIST) {
+		/* Conflict with existing entry */
+		CHECK_FCT( fd_dict_getval(prev, &ad) );
+		TRACE_DEBUG(INFO, "[dict_legacy_xml] Warning: Conflicting entry.");
+		TRACE_DEBUG(INFO, "[dict_legacy_xml]   New entry (ignored): %u - '%s'", a->code, (char *)a->name);
+		TRACE_DEBUG(INFO, "[dict_legacy_xml]   Old entry          : %u - '%s'", ad.avp_code, ad.avp_name);
+		goto inside;
+	} else {
+		/* other errors are stoppers */
+		CHECK_FCT(ret);
+	}
+	
+	/* Update count */
+	if (nb_added)
+		*nb_added += 1;
+	
+inside:
+	/* Now, the inner elements, if any */
+	
+	if ( (!FD_IS_LIST_EMPTY(&a->enums)) && (ad.avp_basetype != AVP_TYPE_UNSIGNED32)) {
+		TRACE_DEBUG(INFO, "AVP '%s' type is not an Unsigned32 but it has enum values (invalid in this extension).", ad.avp_name);
+		return EINVAL;
+	}
+
+	/* In case of enumeration, define the enum values */
+	for (li = a->enums.next; li != &a->enums; li = li->next) {
+		struct t_enum * e = (struct t_enum *)li;
+		struct dict_enumval_data ed;
+		
+		memset(&ed, 0, sizeof(ed));
+		ed.enum_name = (char *)e->name;
+		ed.enum_value.u32 = e->code;
+		
+		CHECK_FCT_DO( ret = fd_dict_new ( fD_dict, DICT_ENUMVAL, &ed, type, NULL ),
+			{
+				TRACE_DEBUG(INFO, "Error defining constant value '%s' for AVP '%s': %s", ed.enum_name, ad.avp_name, strerror(ret));
+				return ret;
+			} );
+		if (nb_added)
+			*nb_added += 1;
+	}
+	
+	/* In case of grouped AVP, check the type is really grouped */
+	if ( !FD_IS_LIST_EMPTY(&a->grouped_optional)
+	  || !FD_IS_LIST_EMPTY(&a->grouped_required)
+	  || !FD_IS_LIST_EMPTY(&a->grouped_fixed) ) {
+		CHECK_PARAMS_DO( ad.avp_basetype == AVP_TYPE_GROUPED, 
+			{ TRACE_DEBUG(INFO, "Got rules for non-grouped AVP '%s'", ad.avp_name); return EINVAL;} );
+		CHECK_FCT_DO( ret = rules_to_fD(fD_dict, prev, &a->grouped_fixed, &a->grouped_required, &a->grouped_optional, nb_added),
+			{ TRACE_DEBUG(INFO, "Error processing rules for AVP '%s': %s", ad.avp_name, strerror(ret)); return ret; } );
+	}
+	
+	/* done! */
+	return 0;
+}
+
+/* Find or create a command. */
+static int cmd_to_fD(struct t_cmd * c, struct dictionary * fD_dict, struct dict_object * fd_appl, struct dict_object ** fd_req, int * nb_added)
+{
+	int ret;
+	struct dict_object * req = NULL, *ans = NULL;
+	struct dict_cmd_data cd;
+	char cmdname[512];
+
+	TRACE_ENTRY("%p %p %p %p %p", c, fD_dict, fd_appl, fd_req, nb_added);
+	
+	CHECK_PARAMS(c && fD_dict);
+	
+	/* Prepare the request data in fD's format */
+	memset(&cd, 0, sizeof(cd));
+	cd.cmd_code = c->code;
+	snprintf(cmdname, sizeof(cmdname), "%s-Request", (char *)c->name);
+	cd.cmd_name = &cmdname[0];
+	cd.cmd_flag_mask = c->fmask | CMD_FLAG_REQUEST | CMD_FLAG_ERROR;
+	cd.cmd_flag_val  = c->flags | CMD_FLAG_REQUEST;
+	
+	/* Create or search in the dictionary */
+	ret = fd_dict_new ( fD_dict, DICT_COMMAND, &cd, fd_appl, &req );
+	if (fd_req)
+		*fd_req = req;
+	if (ret == EEXIST) {
+		struct dict_cmd_data prevcd;
+		/* Conflict with existing entry */
+		CHECK_FCT( fd_dict_getval(req, &prevcd) );
+		TRACE_DEBUG(INFO, "[dict_legacy_xml] Warning: Conflicting entry.");
+		TRACE_DEBUG(INFO, "[dict_legacy_xml]   New entry (ignored): %u - '%s'", cd.cmd_code, cd.cmd_name);
+		TRACE_DEBUG(INFO, "[dict_legacy_xml]   Old entry          : %u - '%s'", prevcd.cmd_code, prevcd.cmd_name);
+		goto answer;
+	} else {
+		/* other errors are stoppers */
+		CHECK_FCT(ret);
+	}
+	
+	/* Update count */
+	if (nb_added)
+		*nb_added += 1;
+	
+answer:
+	/* update data for the answer */
+	snprintf(cmdname, sizeof(cmdname), "%s-Answer", (char *)c->name);
+	cd.cmd_flag_val &= ~CMD_FLAG_REQUEST;
+	cd.cmd_flag_mask &= ~CMD_FLAG_ERROR;
+	
+	ret = fd_dict_new ( fD_dict, DICT_COMMAND, &cd, fd_appl, &ans );
+	if (ret == EEXIST) {
+		struct dict_cmd_data prevcd;
+		/* Conflict with existing entry */
+		CHECK_FCT( fd_dict_getval(ans, &prevcd) );
+		TRACE_DEBUG(INFO, "[dict_legacy_xml] Warning: Conflicting entry.");
+		TRACE_DEBUG(INFO, "[dict_legacy_xml]   New entry (ignored): %u - '%s'", cd.cmd_code, cd.cmd_name);
+		TRACE_DEBUG(INFO, "[dict_legacy_xml]   Old entry          : %u - '%s'", prevcd.cmd_code, prevcd.cmd_name);
+		goto rules;
+	} else {
+		/* other errors are stoppers */
+		CHECK_FCT(ret);
+	}
+	
+	/* Update count */
+	if (nb_added)
+		*nb_added += 1;
+	
+rules:
+	/* Now process the rules inside the command */
+	CHECK_FCT_DO( ret = rules_to_fD(fD_dict, req, &c->reqrules_fixed, &c->reqrules_required, &c->reqrules_optional, nb_added),
+			{
+				TRACE_DEBUG(INFO, "Error converting data from request rules: %s", strerror(ret));
+				return ret;
+			}   );
+	CHECK_FCT_DO( ret = rules_to_fD(fD_dict, ans, &c->ansrules_fixed, &c->ansrules_required, &c->ansrules_optional, nb_added),
+			{
+				TRACE_DEBUG(INFO, "Error converting data from answer rules: %s", strerror(ret));
+				return ret;
+			}   );
+	
+	/* Done */
+	return 0;
+}
+
+/* Find or create an application (and dependent objects) */
+static int appl_to_fD(struct t_appl * a, struct dictionary * fD_dict, struct dict_object ** fd_a, int * nb_added)
+{
+	int ret;
+	struct dict_object * prev = NULL;
+	struct dict_application_data ad;
+	struct fd_list * li;
+
+	TRACE_ENTRY("%p %p %p %p", a, fD_dict, fd_a, nb_added);
+	
+	CHECK_PARAMS(a && fD_dict);
+	
+	if (a->id) { /* skip app 0 */
+	
+		/* Prepare the data in fD's format */
+		memset(&ad, 0, sizeof(ad));
+		ad.application_id   = a->id;
+		ad.application_name = (char *)a->name;
+
+		/* Create or search in the dictionary */
+		ret = fd_dict_new ( fD_dict, 
+					DICT_APPLICATION, 
+					&ad, 
+					NULL /* we don't have a parent vendor in XML files, so currently everything links to no vendor */, 
+					&prev );
+		if (fd_a)
+			*fd_a = prev;
+		if (ret == EEXIST) {
+			/* Conflict with existing entry */
+			CHECK_FCT( fd_dict_getval(prev, &ad) );
+			TRACE_DEBUG(INFO, "[dict_legacy_xml] Warning: Conflicting entry.");
+			TRACE_DEBUG(INFO, "[dict_legacy_xml]   New entry (ignored): %u - '%s'", a->id, (char *)a->name);
+			TRACE_DEBUG(INFO, "[dict_legacy_xml]   Old entry          : %u - '%s'", ad.application_id, ad.application_name);
+			goto inside;
+		} else {
+			/* other errors are stoppers */
+			CHECK_FCT(ret);
+		}
+
+		/* Update count */
+		if (nb_added)
+			*nb_added += 1;
+	}
+	
+inside:
+	/* Now, the inner elements */
+	
+	/* First, define all the types */
+	for (li = a->types.next; li != &a->types; li = li->next) {
+		CHECK_FCT_DO( ret = typdefn_to_fD((struct t_typedefn *)li, fD_dict, prev, NULL, nb_added),
+			{
+				TRACE_DEBUG(INFO, "Error converting data from typedefn '%s': %s", ((struct t_typedefn *)li)->name, strerror(ret));
+				return ret;
+			}  );
+	}
+	
+	/* Then, AVPs, enums, and grouped AVP rules */
+	for (li = a->avps.next; li != &a->avps; li = li->next) {
+		CHECK_FCT_DO( ret = avp_to_fD((struct t_avp *)li, fD_dict, prev, NULL, nb_added),
+			{
+				TRACE_DEBUG(INFO, "Error converting data from AVP '%s': %s", ((struct t_avp *)li)->name, strerror(ret));
+				return ret;
+			}  );
+	}
+	
+	/* Finally, the commands and rules */
+	for (li = a->commands.next; li != &a->commands; li = li->next) {
+		CHECK_FCT_DO( ret = cmd_to_fD((struct t_cmd *)li, fD_dict, prev, NULL, nb_added),
+			{
+				TRACE_DEBUG(INFO, "Error converting data from command '%s': %s", ((struct t_cmd *)li)->name, strerror(ret));
+				return ret;
+			}  );
+	}
+	
+	/* done! */
+	return 0;
+}
+
+
+static int dict_to_fD(struct dictionary * fD_dict, struct t_dictionary * xmldict, int * nb_added)
+{
+	struct fd_list * li;
+	int ret;
+	
+	TRACE_ENTRY("%p %p %p", fD_dict, xmldict, nb_added);
+	
+	CHECK_PARAMS(fD_dict && xmldict && nb_added);
+	
+	*nb_added = 0;
+	
+	/* Create all the vendors */
+	for (li = xmldict->vendors.next; li != &xmldict->vendors; li = li->next) {
+		CHECK_FCT_DO( ret = vend_to_fD((struct t_vend *)li, fD_dict, NULL, nb_added),
+			{
+				TRACE_DEBUG(INFO, "Error converting data from vendor '%s': %s", ((struct t_vend *)li)->name, strerror(ret));
+				return ret;
+			} );
+	}
+	
+	/* Now, process each application */
+	CHECK_FCT_DO( ret = appl_to_fD(&xmldict->base_and_applications, fD_dict, NULL, nb_added),
+			{
+				TRACE_DEBUG(INFO, "Error converting data from Base application: %s", strerror(ret));
+				return ret;
+			} );
+	for (li = xmldict->base_and_applications.chain.next; li != &xmldict->base_and_applications.chain; li = li->next) {
+		CHECK_FCT_DO( ret = appl_to_fD((struct t_appl *) li, fD_dict, NULL, nb_added),
+			{
+				TRACE_DEBUG(INFO, "Error converting data from application '%s': %s", ((struct t_appl *)li)->name, strerror(ret));
+				return ret;
+			}  );
+	}
+	
+	/* Complete! */
+	return 0;
+}
+
+
+
+
+
+/*********************************************/
+
+int dict_lxml_parse(char * xmlfilename)
+{
+	xmlSAXHandler handler;
+	struct parser_ctx data;
+	int ret;
+
+	TRACE_ENTRY("%p", xmlfilename);
+	
+	CHECK_PARAMS_DO(xmlfilename, { return -1; } );
+	
+	TRACE_DEBUG(FULL, "Parsing next XML file: %s...", xmlfilename);
+	
+	/* Initialize the parser */
+	memset(&handler, 0, sizeof(handler));
+	handler.startElement = SAXstartelem;
+	handler.endElement   = SAXendelem;
+	
+	/* Initialize the data */
+	memset(&data, 0, sizeof(data));
+	fd_list_init( &data.dict.vendors, NULL );
+	fd_list_init( &data.dict.base_and_applications.chain, NULL );
+	data.dict.base_and_applications.name = (uint8_t *)"[Diameter Base Protocol]";
+	fd_list_init( &data.dict.base_and_applications.commands, NULL );
+	fd_list_init( &data.dict.base_and_applications.types, NULL );
+	fd_list_init( &data.dict.base_and_applications.avps, NULL );
+	data.xmlfilename = xmlfilename;
+	
+	/* Parse the file */
+	ret = xmlSAXUserParseFile(&handler, &data, xmlfilename);
+	if (ret < 0) {
+		TRACE_DEBUG(INFO, "An error occurred while parsing %s, aborting.", xmlfilename);
+		del_dict_contents(&data.dict);
+		return -1;
+	}
+	
+	TRACE_DEBUG(FULL, "XML file parsing, 1st pass completed.");
+	if (TRACE_BOOL(ANNOYING)) {
+		dump_dict(&data.dict);
+	}
+	
+	/* Now, convert all the objects from the temporary tree into the freeDiameter dictionary */
+	CHECK_FCT_DO( dict_to_fD(fd_g_config->cnf_dict, &data.dict, &ret), 
+		{ 
+			TRACE_DEBUG(INFO, "Error while converting data read from file '%s'", xmlfilename);
+			del_dict_contents(&data.dict); 
+			return -1; 
+		} );
+	
+	TRACE_DEBUG(FULL, "Conversion from '%s' to freeDiameter internal format complete.", xmlfilename);
+	
+	/* Done */
+	del_dict_contents(&data.dict);
+	
+	return ret;
+}