Initial revision

[helm.git] / helm / interface / cicParser.ml

diff --git a/helm/interface/cicParser.ml b/helm/interface/cicParser.ml
new file mode 100644 (file)
index 0000000..ec8c5ef
--- /dev/null
+++ b/helm/interface/cicParser.ml
@@ -0,0 +1,69 @@
+(******************************************************************************)
+(*                                                                            *)
+(*                               PROJECT HELM                                 *)
+(*                                                                            *)
+(*                Claudio Sacerdoti Coen <sacerdot@cs.unibo.it>               *)
+(*                                 24/01/2000                                 *)
+(*                                                                            *)
+(* This is the main (top level) module of a parser for cic objects from xml   *)
+(* files to the internal representation. It uses the modules cicParser2       *)
+(* (objects level) and cicParser3 (terms level)                               *)
+(*                                                                            *)
+(******************************************************************************)
+
+exception Warnings;;
+
+class warner =
+  object 
+    method warn w =
+      print_endline ("WARNING: " ^ w) ;
+      (raise Warnings : unit)
+  end
+;;
+
+exception EmptyUri;;
+
+(* given an uri u it returns the list of tokens of the base uri of u *)
+(* e.g.: token_of_uri "cic:/a/b/c/d.xml" returns ["a" ; "b" ; "c"]   *)
+let tokens_of_uri uri =
+ let uri' = UriManager.string_of_uri uri in
+ let rec chop_list =
+  function
+     [] -> raise EmptyUri
+   | he::[fn] -> [he]
+   | he::tl -> he::(chop_list tl)
+ in
+  let trimmed_uri = Str.replace_first (Str.regexp "cic:") "" uri' in
+   let list_of_tokens = Str.split (Str.regexp "/") trimmed_uri in
+    chop_list list_of_tokens
+;;
+
+(* given the filename of an xml file of a cic object it returns its internal *)
+(* representation. process_annotations is true if the annotations do really  *)
+(* matter                                                                    *)
+let term_of_xml filename uri process_annotations =
+ let module Y = Pxp_yacc in
+  try 
+    let d =
+      (* sets the current base uri to resolve relative URIs *)
+      CicParser3.current_sp := tokens_of_uri uri ;
+      CicParser3.current_uri := uri ;
+      CicParser3.process_annotations := process_annotations ;
+      CicParser3.ids_to_targets :=
+       if process_annotations then Some (Hashtbl.create 500) else None ;
+      let config = {Y.default_config with Y.warner = new warner} in
+      Y.parse_document_entity config
+(*PXP       (Y.ExtID (Pxp_types.System filename,
+         new Pxp_reader.resolve_as_file ~url_of_id ()))
+*)     (PxpUriResolver.from_file filename)
+       CicParser3.domspec
+    in
+     let ids_to_targets = !CicParser3.ids_to_targets in
+      let res = (CicParser2.get_term d#root, ids_to_targets) in
+       CicParser3.ids_to_targets := None ; (* let's help the GC *)
+       res
+  with
+   e ->
+     print_endline (Pxp_types.string_of_exn e) ;
+     raise e
+;;