[r2003-06-03 20:45:50 by cvscast] expat

Original author: cvscast
Date: 2003-06-03 20:45:50+00:00

CHANGES

+0.0.91
+  * Second alpha release
+  * Bug fixes and syntax changes (e.g. making "fun" optional in "let fun")
+  * Experimenting with expat support (see expat/README)
+ 
+
 0.0.9  
   * First public release (alpha release)

Makefile

@@ -38,7 +38,7 @@ website: web/files
 
 # Packaging
 
-DISTRIB = $(DIRS) tools web depend INSTALL INSTALL.WIN32 CHANGES LICENSE README doc
+DISTRIB = $(DIRS) expat tools web depend INSTALL INSTALL.WIN32 CHANGES LICENSE README doc
 
 PACKAGE = cduce-$(VERSION)
 .PHONY: package

README

@@ -124,11 +124,6 @@ There are several sources of information concerning CDuce:
 
 - and of course the interpreter source code !
 
-------------------------------------------------------------------------------
-Abstract
-------------------------------------------------------------------------------
-
-
 
 ------------------------------------------------------------------------------
 Feedback

expat/README

+This directory contains code written by Maas-Maarten Zeeman
+and included here for convenience.
+Home page: http://home.wanadoo.nl/maas/ocaml/
+
+To build support for expat, do (in CDuce root directory):
+
+  make clean; make expat
+
+Now, when the CDuce interpreter is run with the option --expat,
+it will use expat instead of PXP for load_xml.
+
+Note: the current wrapper for expat does not support inclusion
+of external entities. Moreover I encounter random segfaults...

expat/expat.ml

@@ -5,7 +5,7 @@
 (* LICENCE for details.                                                *)
 (***********************************************************************)
 
-(* $Id: expat.ml,v 1.1 2003/05/31 10:32:42 cvscast Exp $ *)
+(* $Id: expat.ml,v 1.2 2003/06/03 20:45:50 cvscast Exp $ *)
 
 type expat_parser
 
@@ -56,7 +56,8 @@ external parser_create_ns : encoding:string -> separator:char -> expat_parser =
 
 (* calls needed to parse *)
 external parse : expat_parser -> string -> unit =  "expat_XML_Parse"
-external parse_sub : expat_parser -> string -> int -> int -> unit =  "expat_XML_Parse_sub"
+external parse_sub : expat_parser -> string -> int -> int -> unit = 
+  "expat_XML_ParseSub"
 external final : expat_parser -> unit = "expat_XML_Final"
 
 (* start element handler calls *)
@@ -91,6 +92,24 @@ external set_comment_handler : expat_parser -> (string -> unit) -> unit =
 external reset_comment_handler : expat_parser -> unit =
   "expat_XML_ResetCommentHandler"
 
+(* start cdata handler *)
+external set_start_cdata_handler : expat_parser -> (unit -> unit) -> unit =
+  "expat_XML_SetStartCDataHandler"
+external reset_start_cdata_handler : expat_parser -> unit =
+  "expat_XML_ResetStartCDataHandler"
+
+(* end cdata handler *)
+external set_end_cdata_handler : expat_parser -> (unit -> unit) -> unit =
+  "expat_XML_SetEndCDataHandler"
+external reset_end_cdata_handler : expat_parser -> unit =
+  "expat_XML_ResetEndCDataHandler"
+
+(* default handler *)
+external set_default_handler : expat_parser -> (string -> unit) -> unit =
+  "expat_XML_SetDefaultHandler"
+external reset_default_handler : expat_parser -> unit =
+  "expat_XML_ResetDefaultHandler"
+
 (* some general parser query calls *)
 external get_current_byte_index : expat_parser -> int = 
   "expat_XML_GetCurrentByteIndex"

expat/expat.mli

@@ -5,7 +5,7 @@
 (* LICENCE for details.                                                *)
 (***********************************************************************)
 
-(* $Id: expat.mli,v 1.1 2003/05/31 10:32:42 cvscast Exp $ *)
+(* $Id: expat.mli,v 1.2 2003/06/03 20:45:50 cvscast Exp $ *)
 
 (** The Ocaml Expat library provides an interface to the Expat XML Parser.
 
@@ -40,6 +40,9 @@ val parser_create_ns : encoding:string -> separator:char -> expat_parser
 (** Let the parser parse a chunk of an XML document. 
     @raise Expat_error error *)
 val parse : expat_parser -> string -> unit
+
+(** Let the parser parse a chunk of an XML document in a substring
+    @raise Expat_error error *)
 val parse_sub : expat_parser -> string -> int -> int -> unit
 
 (** Inform the parser that the entire document has been parsed.  *) 
@@ -79,6 +82,19 @@ val reset_processing_instruction_handler : expat_parser -> unit
 val set_comment_handler : expat_parser -> (string -> unit) -> unit
 val reset_comment_handler : expat_parser -> unit
 
+(** {6 CData Section handler setting and resetting *)
+
+val set_start_cdata_handler : expat_parser -> (unit -> unit) -> unit
+val reset_start_cdata_handler : expat_parser -> unit
+
+val set_end_cdata_handler : expat_parser -> (unit -> unit) -> unit
+val reset_end_cdata_handler : expat_parser -> unit
+
+(** {6 Default Handler setting and resetting *)
+
+val set_default_handler : expat_parser -> (string -> unit) -> unit
+val reset_default_handler : expat_parser -> unit
+
 (** {5 Parse Position Functions} *)
 
 val get_current_byte_index : expat_parser -> int

expat/expat_stubs.c

@@ -5,7 +5,7 @@
 /* LICENCE for details.                                                */
 /***********************************************************************/
 
-/* $Id: expat_stubs.c,v 1.2 2003/06/03 20:36:11 cvscast Exp $ */
+/* $Id: expat_stubs.c,v 1.3 2003/06/03 20:45:50 cvscast Exp $ */
 
 /* Stub code to interface Ocaml with Expat */
 
@@ -26,13 +26,16 @@
  * handler tuple which is registered as global root. Handlers for
  * new functions should go here.
  */
-#define NUM_HANDLERS 5
+#define NUM_HANDLERS 8
 enum expat_handler {
   EXPAT_START_ELEMENT_HANDLER,
   EXPAT_END_ELEMENT_HANDLER,
   EXPAT_CHARACTER_DATA_HANDLER,
   EXPAT_PROCESSING_INSTRUCTION_HANDLER,
-  EXPAT_COMMENT_HANDLER
+  EXPAT_COMMENT_HANDLER,
+  EXPAT_START_CDATA_HANDLER,
+  EXPAT_END_CDATA_HANDLER,
+  EXPAT_DEFAULT_HANDLER
 };
 
 static void 
@@ -42,7 +45,6 @@ xml_parser_finalize(value parser)
   value *handlers = XML_GetUserData(xml_parser);
 
   remove_global_root(handlers);
-  stat_free(handlers);
 
   XML_ParserFree(xml_parser);  
 }
@@ -92,6 +94,7 @@ static value
 create_ocaml_expat_parser(XML_Parser xml_parser)
 {
   CAMLparam0();
+
   CAMLlocal1(parser);
   int i;
   value *handlers;
@@ -106,8 +109,12 @@ create_ocaml_expat_parser(XML_Parser xml_parser)
    * handlers and register it as global root. 
    */
   handlers = stat_alloc(sizeof *handlers);
-  *handlers = Val_unit;
+  *handlers = Val_unit; 
   register_global_root(handlers);
+
+  /*
+   * Create a tuple which will hold the handlers.
+   */
   *handlers = alloc_tuple(NUM_HANDLERS);
   for(i = 0; i < NUM_HANDLERS; i++) {
     Field(*handlers, i) = Val_unit;
@@ -228,26 +235,31 @@ expat_XML_Parse(value parser, value string)
   CAMLreturn (Val_unit);
 }
 
+/*
+ * A sub-string variant of the parse call
+ */
 CAMLprim value
-expat_XML_Parse_sub(value parser, value string, value pos, value len)
+expat_XML_ParseSub(value parser, value string, value pos, value len)
 {
-  CAMLparam2(parser, string);
+  CAMLparam4(parser, string, pos, len);
   XML_Parser xml_parser =  XML_Parser_val(parser);
-
-  if(!XML_Parse(xml_parser, String_val(string)+Int_val(pos), Int_val(len), 0)) {
+  
+  if(!XML_Parse(xml_parser, 
+		String_val(string) + Int_val(pos), Int_val(len), 0)) {
     expat_error(XML_GetErrorCode(xml_parser));
   }
-  
+
   CAMLreturn (Val_unit);
 }
 
+
 /*
  * The final call
  */
 CAMLprim value
 expat_XML_Final(value parser)
 {
-  CAMLparam1(parser);
+  CAMLparam1(parser); 
   XML_Parser xml_parser =  XML_Parser_val(parser);
 
   if(!XML_Parse(xml_parser, NULL, 0, 1)) {
@@ -504,3 +516,136 @@ expat_XML_ResetCommentHandler(value parser)
   CAMLparam1(parser);
   CAMLreturn (set_comment_handler(parser, NULL, Val_unit));
 }
+
+/*
+ * Start CData handler, setting and resetting
+ */
+static void
+start_cdata_handler(void *user_data) 
+{
+  CAMLparam0();
+  value *handlers = user_data;
+  
+  callback(Field(*handlers, EXPAT_START_CDATA_HANDLER), Val_unit);
+
+  CAMLreturn0;
+}
+
+static value
+set_start_cdata_handler(value parser,
+			XML_StartCdataSectionHandler c_handler, 
+			value ocaml_handler)
+{
+  CAMLparam2(parser, ocaml_handler);
+  XML_Parser xml_parser = XML_Parser_val(parser);
+  value *handlers = XML_GetUserData(xml_parser);
+  
+  Store_field(*handlers, EXPAT_START_CDATA_HANDLER, ocaml_handler);
+  XML_SetStartCdataSectionHandler(xml_parser, c_handler);
+  
+  CAMLreturn (Val_unit);
+}
+
+CAMLprim value
+expat_XML_SetStartCDataHandler(value parser, value handler)
+{
+  CAMLparam2(parser, handler);
+  CAMLreturn (set_start_cdata_handler(parser, start_cdata_handler, handler));
+}
+
+CAMLprim value
+expat_XML_ResetStartCDataHandler(value parser)
+{
+  CAMLparam1(parser);
+  CAMLreturn (set_start_cdata_handler(parser, NULL, Val_unit));
+}
+
+/*
+ * End CData handler, setting and resetting
+ */
+static void
+end_cdata_handler(void *user_data) 
+{
+  CAMLparam0();
+  value *handlers = user_data;
+
+  callback(Field(*handlers, EXPAT_END_CDATA_HANDLER), Val_unit);
+
+  CAMLreturn0;
+}
+
+static value
+set_end_cdata_handler(value parser,
+		      XML_EndCdataSectionHandler c_handler, 
+		      value ocaml_handler)
+{
+  CAMLparam2(parser, ocaml_handler);
+  XML_Parser xml_parser = XML_Parser_val(parser);
+  value *handlers = XML_GetUserData(xml_parser);
+
+  Store_field(*handlers, EXPAT_END_CDATA_HANDLER, ocaml_handler);
+  XML_SetEndCdataSectionHandler(xml_parser, c_handler);
+  
+  CAMLreturn (Val_unit);
+}
+
+CAMLprim value
+expat_XML_SetEndCDataHandler(value parser, value handler)
+{
+  CAMLparam2(parser, handler);
+  CAMLreturn (set_end_cdata_handler(parser, end_cdata_handler, handler));
+}
+
+CAMLprim value
+expat_XML_ResetEndCDataHandler(value parser)
+{
+  CAMLparam1(parser);
+  CAMLreturn (set_end_cdata_handler(parser, NULL, Val_unit));
+}
+
+
+/*
+ * Default handler, setting and resetting
+ */
+static void
+default_handler(void *user_data, const char *data, int len) 
+{
+  CAMLparam0();
+  CAMLlocal1(d);
+  value *handlers = user_data;
+
+  d = alloc_string(len);
+  memmove(String_val(d), data, len);
+  callback(Field(*handlers, EXPAT_DEFAULT_HANDLER), d);
+
+  CAMLreturn0;
+}
+
+static value
+set_default_handler(value parser,
+		    XML_DefaultHandler c_handler, 
+		    value ocaml_handler)
+{
+  CAMLparam2(parser, ocaml_handler);
+  XML_Parser xml_parser = XML_Parser_val(parser);
+  value *handlers = XML_GetUserData(xml_parser);
+
+  Store_field(*handlers, EXPAT_DEFAULT_HANDLER, ocaml_handler);
+  XML_SetDefaultHandler(xml_parser, c_handler);
+  
+  CAMLreturn (Val_unit);
+}
+
+CAMLprim value
+expat_XML_SetDefaultHandler(value parser, value handler)
+{
+  CAMLparam2(parser, handler);
+  CAMLreturn (set_default_handler(parser, default_handler, handler));
+}
+
+CAMLprim value
+expat_XML_ResetDefaultHandler(value parser)
+{
+  CAMLparam1(parser);
+  CAMLreturn (set_default_handler(parser, NULL, Val_unit));
+}

runtime/load_xml.ml

@@ -91,12 +91,14 @@ let load_expat s =
       txt.pos <- 0; 
       create_elt accu !stack);
   Expat.set_character_data_handler p (add_string txt);
+(*  Gc.full_major (); *)
+(*   Gc.compact (); *)
   let ic = open_in s in
   let rec loop () =
     let n = input ic buf 0 buflen in
     if (n > 0) then 
-(*(Expat.parse p (String.sub buf 0 n); loop ())*)
- (Expat.parse_sub p buf 0 n; loop ()) 
+(*(Expat.parse p (String.sub buf 0 n); loop ()) *)
+      (Expat.parse_sub p buf 0 n; loop ()) 
   in
   try 
     loop();