.

git-svn-id: svn+ssh://idea.nguyen.vg/svn/sxsi/trunk/xpathcomp@21 3cdefd35-fc62-479d-8e8d-bae585ffb9ca

diff --git a/OCamlDriver.cpp b/OCamlDriver.cpp
index b0e57d9..1898d5a 100644 (file)
--- a/OCamlDriver.cpp
+++ b/OCamlDriver.cpp
@@ -85,9 +85,14 @@ extern "C" CAMLprim value caml_xml_tree_text_collection(value tree){
 
 extern "C" CAMLprim value caml_xml_tree_first_child(value tree, value id){
   CAMLparam2(tree,id);
-  CAMLlocal1(res);
   CAMLreturn(Val_int (XMLTREE(tree)->FirstChild(TREENODEVAL(id))));
 }
+
+extern "C" CAMLprim value caml_xml_tree_is_leaf(value tree, value id){
+  CAMLparam2(tree,id);
+  CAMLreturn(Val_bool (XMLTREE(tree)->IsLeaf(TREENODEVAL(id))));
+}
+
 extern "C" CAMLprim value caml_xml_tree_next_sibling(value tree, value id){
   CAMLparam2(tree,id);
   CAMLreturn(Val_int (XMLTREE(tree)->NextSibling(TREENODEVAL(id))));

diff --git a/SXSIStorageInterface.cpp b/SXSIStorageInterface.cpp
index bd21cd0..14c1d70 100644 (file)
--- a/SXSIStorageInterface.cpp
+++ b/SXSIStorageInterface.cpp
@@ -24,22 +24,29 @@ SXSIStorageInterface::~SXSIStorageInterface()
 
 void SXSIStorageInterface::newChild(string name)
 { 
+  cerr << "New child " << name << "\n";
   tree->NewOpenTag((unsigned char*) name.c_str());
 }
 
 
 void SXSIStorageInterface::newText(string text)
 {
-   if (text.empty())
+  if (text.empty()){
+    cerr << "Calling newEmptyText()\n";
     tree->NewEmptyText();
-  else
+  }
+  else {
+    cerr << "Calling newText(" << text <<")\n";
     tree->NewText((unsigned char*) text.c_str());
- }
+    
+  };
+}
        
 
 
 void SXSIStorageInterface::nodeFinished(string name)
 {  
+    cerr << "Node Finished child " << name << "\n";
    tree->NewClosingTag((unsigned char*) name.c_str());
  }           
              

diff --git a/XMLDocShredder.cpp b/XMLDocShredder.cpp
index 722c291..3daaf45 100644 (file)
--- a/XMLDocShredder.cpp
+++ b/XMLDocShredder.cpp
@@ -106,8 +106,7 @@ void XMLDocShredder::processStartElement()
        
        if (empty){
            storageIfc_->newText("");  //myText
-           storageIfc_->nodeFinished(name);
-           storageIfc_->newText("");  //nextText
+           storageIfc_->nodeFinished(name);       
        };
 
 
@@ -131,7 +130,6 @@ void XMLDocShredder::processPCDATA()
          
        if (reader_->has_value())
        {         
-         storageIfc_->newChild("<$>");
          storageIfc_->newText(reader_->get_value());
          last_text = true;
        }
@@ -167,24 +165,22 @@ void XMLDocShredder::processAttributes()
                 
                else
                {
-                       storageIfc_->newChild(name);
-                       storageIfc_->newChild("<$>");
-                       storageIfc_->newText(value);
-                       storageIfc_->nodeFinished("<$>");
+                 storageIfc_->newText(""); //prevText
+                 storageIfc_->newChild(name);
+                 storageIfc_->newText(value);
+                 storageIfc_->nodeFinished(name);
                }
        }
        while (reader_->move_to_next_attribute());
+       storageIfc_->newText(""); //nextText
        storageIfc_->nodeFinished("<@>");
 }
 
 void XMLDocShredder::processSignificantWhitespace()
 {
-       ustring value = reader_->get_value();
-       
+       ustring value = reader_->get_value();   
        // each significant whitespace sequence constructs a text node
-       storageIfc_->newChild("<$>");
        storageIfc_->newText(value);
-
        
 }
 
@@ -199,6 +195,7 @@ void XMLDocShredder::processEndDocument()
 {
        /* tell the storage interface that document parsing has finished, and structures
         * can now be written to disk. */
+  storageIfc_->newText("");
   storageIfc_->nodeFinished("");
   storageIfc_->parsingFinished();      
 }
@@ -239,10 +236,8 @@ void XMLDocShredder::processCDATASection()
         */
   
        ustring value = reader_->get_value();
-       storageIfc_->newChild("<$>");
        storageIfc_->newText(value);
        last_text = true;
-       //      storageIfc_->nodeFinished();
 
 }
 

diff --git a/main.ml b/main.ml
index 10fd915..88363e0 100644 (file)
--- a/main.ml
+++ b/main.ml
@@ -24,7 +24,8 @@ let total_time () =  List.fold_left (+.) 0. !l;;
 let test filename query output = 
   Printf.eprintf "Parsing document : %!";
   let v = time Tree.Binary.parse_xml_uri filename in
-  Tree.Binary.print_xml_fast stderr v
+    Tree.dump v;
+    Tree.Binary.print_xml_fast stderr v
 
 
 let main filename query output =

diff --git a/tests/test.xml b/tests/test.xml
index d83dfc0..87340f4 100644 (file)
--- a/tests/test.xml
+++ b/tests/test.xml
@@ -1,11 +1 @@
-<!DOCTYPE a SYSTEM "a.dtd" >
-<a id="3">
-  <c/>
-  <b>
-    <c/>
-    <d/>
-    <c/>
-  </b>
-  <b/>
-  <a/>
-</a>
+<a>YYYYYYYYYYYYY<b>AAAAAAAAA</b>XXXXXXXXXXXX</a>

diff --git a/tree.ml b/tree.ml
index 53872b3..219ee4b 100644 (file)
--- a/tree.ml
+++ b/tree.ml
@@ -240,10 +240,17 @@ struct
     (* Todo *)
     external nullt : unit -> [`Text ] node = "caml_xml_tree_nullt"
     let nil = nullt ()
-    external get_text : t -> [`Text] node -> string = "caml_text_collection_get_text"
-    let is_empty _ (n : [`Text] node) = equal nil n
+    external get_text1 : t -> [`Text] node -> string = "caml_text_collection_get_text"
+
+    let get_text t n = Printf.printf "@@@@@@%i\n%!" (Obj.magic n);
+      if equal nil n then "" 
+      else  get_text1 t n
+
+    let is_empty t (n : [`Text] node) = (get_text t n) = ""
+
+  end
+
 
-    end
   module Tree = 
   struct
 
@@ -271,17 +278,42 @@ struct
     let is_last t n = equal nil (next_sibling t n)
     
     external prev_text : t -> [`Tree] node -> [`Text ] node = "caml_xml_tree_prev_text"
+    let prev_text t id = Printf.eprintf "Calling PrevText for node %i with result" (Obj.magic id);
+      let did = if is_nil id then Text.nil else prev_text t id
+      in Printf.eprintf " %i!!!\n%!" (Obj.magic did); did
+         
+
+
     external my_text : t -> [`Tree] node -> [`Text ] node = "caml_xml_tree_my_text"
     external next_text : t -> [`Tree] node -> [`Text ] node = "caml_xml_tree_next_text"
-      
+
+    let next_text t id = Printf.eprintf "Calling NextText for node %i with result" (Obj.magic id);
+      let did = if is_nil id then Text.nil else next_text t id
+      in Printf.eprintf " %i!!!\n%!" (Obj.magic did); did
+
     external text_xml_id : t -> [`Text ] node -> int = "caml_xml_tree_text_xml_id"
     external node_xml_id : t -> [`Tree ] node -> int = "caml_xml_tree_node_xml_id"
-            
+      
+
+    let print_skel t =
+      let rec aux id = 
+       if (is_nil id)
+       then Printf.eprintf "#"
+       else 
+         begin
+           Printf.eprintf "%s(" (Tag.to_string (tag t id));
+           aux(first_child t id);
+           Printf.eprintf ",\n";
+           aux(next_sibling t id);
+           Printf.eprintf ")\n";
+         end
+      in
+       aux (root t)
   end
       
       
-  module Binary : BINARY = struct
-    
+  module Binary  = struct
+
     type node_content = 
        [ `Node of [`Tree ] node 
        | `String of [`Text ] node * [`Tree ] node ]
@@ -296,7 +328,7 @@ struct
     type t = { doc : doc;
               node : descr }
        
-
+    let dump { doc=t } = Tree.print_skel t       
     open Tree                 
     let node_of_t t = { doc= t; node= Node(`Node (root t)) }
 
@@ -319,7 +351,7 @@ struct
     let equal a b = (compare a b) == 0
 
     let string t = match t.node with
-      | String i -> Text.get_text (text_collection t.doc) i
+      | String i ->  Text.get_text (text_collection t.doc) i
       | _ -> assert false
          
     let norm (n : [`Tree ] node ) =  if is_nil n then Nil else Node (`Node n)
@@ -327,6 +359,7 @@ struct
     let descr t = t.node
 
     let first_child n = 
+      Printf.eprintf "first_child!\n%!";
       let node' = 
        match n.node with
          | Nil | String _ -> failwith "first_child"
@@ -342,6 +375,7 @@ struct
        { n with node = node'}
          
     let next_sibling n = 
+      Printf.eprintf "next_sibling!\n%!";
       let node' =
        match n.node with
          | Nil | String _  -> failwith "next_sibling"
@@ -385,6 +419,7 @@ struct
              output_string outc  tg;
              ( match l.node with
                    Nil -> output_string outc  "/>"
+                 | String _ -> assert false
                  | Node(_) when Tag.equal (tag l) Tag.attribute -> 
                      (loop_attributes (left l);
                       match (right l).node with
@@ -397,8 +432,8 @@ struct
                             output_char outc '>' )
                  | _ ->
                      output_char outc  '>'; 
-                     loop (left l);            
-                     output_string outc  "</";
+                     loop l;
+                     output_string outc "</";
                      output_string outc tg;
                      output_char outc '>'
              );if print_right then loop r
@@ -413,8 +448,13 @@ struct
        | _ -> ()
       in
        loop ~print_right:false t
-         
+
+
+
   end
-    
+
 end
+
+
+let dump = XML.Binary.dump
 include XML

diff --git a/tree.mli b/tree.mli
index 6bc38b5..060e086 100644 (file)
--- a/tree.mli
+++ b/tree.mli
@@ -24,3 +24,5 @@ sig
 end
 
 module Binary : BINARY
+
+val dump : Binary.t -> unit