Add a test program for experimenting with lexicographic indices.

diff --git a/src/lexindex_stub.cpp b/src/lexindex_stub.cpp
new file mode 100644 (file)
index 0000000..a09216e
--- /dev/null
+++ b/src/lexindex_stub.cpp
@@ -0,0 +1,76 @@
+#include <cstring>
+#include <utility>
+#include <algorithm>
+#include <vector>
+#include "xml-tree.hpp"
+#include "utils_stub.hpp"
+
+
+//define a type for the lexicographic index
+
+class lex_index {
+public:
+  //The tag ID
+  xml_tree::tag_t tag;
+  //The text data
+  std::vector<std::pair<std::string, xml_tree::node_t> > data;
+};
+
+
+
+using namespace SXSI;
+
+
+static xml_tree*& XMLTREE(value v)
+{
+  return Obj_val<xml_tree*>(v);
+}
+
+static lex_index*& LEXINDEX(value v)
+{
+  return Obj_val<lex_index*>(v);
+}
+
+static xml_tree::node_t TREENODE(value i)
+{
+  return static_cast<xml_tree::node_t>(Int_val(i));
+}
+
+static xml_tree::tag_t TAG(value i)
+{
+  return static_cast<xml_tree::tag_t>(Int_val(i));
+}
+
+
+
+
+extern "C" value caml_build_lex_index(value vtree, value vtag)
+{
+  CAMLparam2(vtree, vtag);
+  CAMLlocal1(vindex);
+  vindex = sxsi_alloc_custom<lex_index*>();
+  xml_tree * tree = XMLTREE(vtree);
+  xml_tree::tag_t tag = TAG(vtag);
+
+  //Uncomment the following and comment the failwith line
+  //LEXINDEX(vindex) = ... return a lex_index* ....
+
+
+  caml_failwith("build_lex_index not implemented");
+
+
+  CAMLreturn (vindex);
+}
+
+extern "C" value caml_print_lex_index(value vindex)
+{
+  CAMLparam1(vindex);
+  lex_index* index = LEXINDEX(vindex);
+
+  //Print the index to the terminal
+
+  caml_failwith("print_lex_index not implemented");
+
+
+  CAMLreturn (Val_unit);
+}

diff --git a/src/lextest.ml b/src/lextest.ml
new file mode 100644 (file)
index 0000000..0c297ca
--- /dev/null
+++ b/src/lextest.ml
@@ -0,0 +1,35 @@
+(* Abstract type for the index, generated from C++ code *)
+
+type index
+
+external build_lex_index : Tree.t -> Tag.t -> index = "caml_build_lex_index"
+external print_lex_index : index -> unit = "caml_print_lex_index"
+
+
+let main () =
+  if Array.length Sys.argv != 3 then
+  let () = Printf.eprintf "Error: invalid argument" in exit 1
+  else
+  let file = Sys.argv.(1) in
+  (* Load the index or the xml file *)
+  Printf.printf "Loading document\n%!";
+  let document =
+    if Filename.check_suffix file ".srx"
+    then Tree.load file
+    else if Filename.check_suffix file ".xml" then Tree.parse_xml_uri file
+    else
+    let () = Printf.eprintf "Error: unrecognized extension" in exit 2
+  in
+  Printf.printf "Building lex index\n%!";
+  let index = build_lex_index document (Tag.tag Sys.argv.(2)) in
+  Printf.printf "Printing lex index\n%!";
+  print_lex_index index;
+  exit 0
+
+
+let () =
+  try
+    main ()
+  with
+    e -> Printf.eprintf "Fatal error: %s" (Printexc.to_string e); exit 3
+      

diff --git a/src/libcamlshredder.clib b/src/libcamlshredder.clib
index 7dc17cf..001a22c 100644 (file)
--- a/src/libcamlshredder.clib
+++ b/src/libcamlshredder.clib
@@ -3,3 +3,4 @@ bp_stub.o
 xml-tree-builder_stub.o
 xml-tree_stub.o
 common_stub.o
+lexindex_stub.o