.

git-svn-id: svn+ssh://idea.nguyen.vg/svn/sxsi/trunk/xpathcomp@803 3cdefd35-fc62-479d-8e8d-bae585ffb9ca

diff --git a/OCamlDriver.cpp b/OCamlDriver.cpp
index 21e5398..7a97e18 100644 (file)
--- a/OCamlDriver.cpp
+++ b/OCamlDriver.cpp
@@ -255,13 +255,14 @@ extern "C"  value caml_text_collection_count_lessthan(value tree,value str){
 }
 
 static value sort_alloc_array(std::vector<DocID> results, value resarray){
-    std::sort(results.begin(), results.end(), docId_comp);
+  std::sort(results.begin(), results.end(), docId_comp);
     size_t s = results.size();
     resarray = caml_alloc_tuple(s);
     for (size_t i = 0; i < s ;i++){
       caml_initialize(&Field(resarray,i),Val_int(results[i]));
     };
-    return resarray;
+    return resarray; 
+
 }
 
 /**

diff --git a/XMLDocShredder.cpp b/XMLDocShredder.cpp
index 7f7d408..cb70e19 100644 (file)
--- a/XMLDocShredder.cpp
+++ b/XMLDocShredder.cpp
@@ -102,8 +102,11 @@ void XMLDocShredder::processStartElement()
   // fetch element name; this will be the full qualified name
   ustring name = reader_->get_name();
   bool empty = false;
-  
-  tb->NewOpenTag(name);
+  size_t found = name.find_first_of(':');
+  if (found == ustring::npos)  
+    tb->NewOpenTag(name);
+  else
+    tb->NewOpenTag(name.substr(found+1,name.length() - found - 1));
   
   /* We must be really carefull here. calling process attributes moves
      the document pointer on the last attribute, hence calling reader_->is_empty

diff --git a/main.ml b/main.ml
index 0650b6b..f440b3a 100644 (file)
--- a/main.ml
+++ b/main.ml
@@ -54,6 +54,28 @@ let test_loop2 tree tag =
   Hashtbl.add f (hash 101) `Foo;
   g t' Tree.root
 
+let test_text doc = 
+  let _ = Printf.eprintf "Contains(bree)" in
+  let _ = time (Tree.test_contains doc) "bree" in
+  let _ = Printf.eprintf "Contains(brain)" in
+  let _ = time (Tree.test_contains doc) "brain" in
+  let _ = Printf.eprintf "Contains(brain)" in
+  let i = time (Tree.test_contains doc) "brain" in
+  let _ = Printf.eprintf "%i\nContains(Australia)" i in
+  let i = time (Tree.test_contains doc) "AUSTRALIA" in
+  let _ = Printf.eprintf "%i\n Contains(1930)" i in
+  let i = time (Tree.test_contains doc) "1930" in
+  let _ = Printf.eprintf "%i\n startswith(bar)" i in
+  let i = time (Tree.test_prefix doc) "bar" in
+  let _ = Printf.eprintf "%i\n endswith(LAND)" i in
+  let i = time (Tree.test_suffix doc) "LAND" in
+  let _ = Printf.eprintf "%i\n =(2001)" i in
+  let i = time (Tree.test_equals doc) "2001" in
+  let _ = Printf.eprintf "%i\n =(Nguyen)" i in
+  let i = time (Tree.test_equals doc) "Nguyen" in
+  Printf.eprintf "%i\n" i ;
+  ()
+
 type pointers
 external build_pointers : Tree.t -> pointers = "caml_build_pointers"
 external iter_pointers : pointers -> int = "caml_iter_pointers"
@@ -71,6 +93,7 @@ let main v query_string output =
          Ulexer.Loc.Exc_located ((x,y),e) -> Printf.eprintf "character %i-%i %s\n" x y (Printexc.to_string e);exit 1
       in
       let _ = Printf.eprintf "Number of nodes %i\n%!" (Tree.size v) in
+      let _ = test_text v in
 (*      let _ = Tree.stats v in
       let _ = Printf.eprintf "Timing first_child/next_sibling %!" in
       let _ = time (Tree.benchmark_fcns)  v in

diff --git a/tree.ml b/tree.ml
index a034636..0a1bdda 100644 (file)
--- a/tree.ml
+++ b/tree.ml
@@ -788,3 +788,7 @@ let stats t =
 
 
 
+let test_prefix t s = Array.length (text_prefix t.doc s)
+let test_suffix t s = Array.length (text_suffix t.doc s)
+let test_contains t s = Array.length (text_contains t.doc s) 
+let test_equals t s = Array.length (text_equals t.doc s)

diff --git a/tree.mli b/tree.mli
index 4a8f12c..280d97d 100644 (file)
--- a/tree.mli
+++ b/tree.mli
@@ -91,3 +91,8 @@ val benchmark_jump : t -> Tag.t -> unit
 val benchmark_fcns : t -> unit
 val benchmark_lcps : t -> unit
 val stats : t -> unit
+
+val test_suffix : t -> string -> int
+val test_prefix : t -> string -> int
+val test_equals : t -> string -> int
+val test_contains : t -> string -> int