Various fixes, mainly tagged_sibling/select_sibling.

diff --git a/xml-tree-inc.hpp b/xml-tree-inc.hpp
index 6ce1724..cd8db71 100644 (file)
--- a/xml-tree-inc.hpp
+++ b/xml-tree-inc.hpp
@@ -5,6 +5,19 @@
 #ifndef XML_TREE_INC_HPP_
 #define XML_TREE_INC_HPP_
 
+#include <cstdio>
+
+#if 0
+#define ASSERT_NODE(orig, res) do {                                     \
+    if (res < -1 || res >= par->n|| (res != -1 && res < orig))          \
+    fprintf(stderr,                                                     \
+      "Assertion failure: original node %i, result %i, line %i\n",      \
+            orig, res, __LINE__);                                       \
+  } while (0)
+#else
+#define ASSERT_NODE(orig, res)
+#endif
+
 inline uint32_t xml_tree::size() const
 {
   return tag_seq_len / 2;
@@ -86,7 +99,9 @@ inline xml_tree::node_t xml_tree::parent(xml_tree::node_t x) const
 
 inline xml_tree::node_t xml_tree::first_child(node_t x) const
 {
-  return bp_first_child(this->par, x);
+  xml_tree::node_t result = bp_first_child(this->par, x);
+  ASSERT_NODE(x, result);
+  return result;
 }
 
 inline xml_tree::node_t xml_tree::last_child(xml_tree::node_t x) const
@@ -99,7 +114,9 @@ inline xml_tree::node_t xml_tree::last_child(xml_tree::node_t x) const
 
 inline xml_tree::node_t xml_tree::next_sibling(xml_tree::node_t x) const
 {
-  return bp_next_sibling(this->par, x);
+  xml_tree::node_t result = bp_next_sibling(this->par, x);
+  ASSERT_NODE(x, result);
+  return result;
 }
 
 inline xml_tree::node_t xml_tree::prev_sibling(xml_tree::node_t x) const
@@ -119,6 +136,8 @@ inline xml_tree::node_t xml_tree::first_element(xml_tree::node_t x) const
   case PCDATA_OPEN_TAG_ID:
     n = n + 2;
     return bp_inspect(this->par, n) ? n : xml_tree::NIL;
+  default:
+    return n;
   };
 }
 
@@ -161,17 +180,29 @@ inline xml_tree::node_t xml_tree::tagged_child(xml_tree::node_t x,
                                                xml_tree::tag_t t) const
 {
   xml_tree::node_t c = first_child(x);
+  xml_tree::node_t result;
   if (is_nil(c) || tag(c) == t)
     return c;
   else
-    tagged_sibling(c, t);
+    return tagged_sibling(c, t);
+  /*  ASSERT_NODE(x, result);
+      return result;*/
 }
 
 inline xml_tree::node_t xml_tree::tagged_sibling(xml_tree::node_t x,
                                                  xml_tree::tag_t t) const
 {
   xml_tree::node_t sibling = next_sibling(x);
-  while(!is_nil(sibling) && tag(sibling) != t) sibling = next_sibling(sibling);
+  xml_tree::tag_t stag;
+  while (sibling  != xml_tree::NIL) {
+    stag = tag(sibling);
+    if (stag == t) {
+      ASSERT_NODE(x, sibling);
+      return  sibling;
+    }
+    sibling = next_sibling(sibling);
+  };
+  ASSERT_NODE(x, sibling);
   return sibling;
 }
 

diff --git a/xml-tree.cpp b/xml-tree.cpp
index 6d33878..758c1b4 100644 (file)
--- a/xml-tree.cpp
+++ b/xml-tree.cpp
@@ -13,6 +13,10 @@ extern "C" {
 
 using namespace SXSI;
 
+const xml_tree::node_t xml_tree::NIL;
+const xml_tree::node_t xml_tree::ROOT;
+
+
 const xml_tree::tag_t xml_tree::NIL_TAG_ID;
 const char* xml_tree::NIL_TAG = "<!NIL!>";
 const xml_tree::tag_t xml_tree::DOCUMENT_OPEN_TAG_ID;
@@ -26,6 +30,7 @@ const char* xml_tree::ATTRIBUTE_DATA_OPEN_TAG = "<@$>";
 const xml_tree::tag_t xml_tree::CLOSE_TAG_ID;
 const char* xml_tree::CLOSE_TAG = "</>";
 
+
 static int bits8 (int t ) {
   int r = bits(t);
   if (r <= 8)
@@ -78,6 +83,7 @@ xml_tree::xml_tree(std::vector<int32_t> *tags,
 
   size_t npar = parbitmap->size();
   parbitmap->pack();
+
   par = bp_construct(npar,
                      parbitmap->get_vector_ptr(),
                      OPT_DEGREE);
@@ -92,7 +98,8 @@ xml_tree::xml_tree(std::vector<int32_t> *tags,
   uint32_t max_tag = tag_names->size() - 1;
   static_bitsequence_builder *bmb = new static_bitsequence_builder_sdarray();
   alphabet_mapper *am = new alphabet_mapper_none();
-  this->tags = new static_sequence_bs((uint32_t *) &tags[0], npar, am, bmb);
+
+  this->tags = new static_sequence_bs((uint32_t*)&((*tags)[0]), npar, am, bmb);
   bits_per_tag = bits8(max_tag);
   tag_seq_len = npar;
   tag_seq = new uint32_t[uint_len(bits_per_tag, tag_seq_len)];
@@ -110,10 +117,15 @@ xml_tree::xml_tree(std::vector<int32_t> *tags,
     text_positions = new static_bitsequence_rrr02(textbm,
                                                   npar,
                                                   32);
-    delete [] textbm;
+    //delete [] textbm;
+    delete textbitmap;
 
     this->text_index_type = idx_type;
+    fprintf(stderr, "Before!\n");
+    fflush(stderr);
     text_collection = tc_builder->InitTextCollection();
+    fprintf(stderr, "After!\n");
+    fflush(stderr);
     delete tc_builder;
   };
 
@@ -177,15 +189,18 @@ xml_tree::select_descendant(xml_tree::node_t x,
   return min;
 }
 
-
 xml_tree::node_t
 xml_tree::select_sibling(xml_tree::node_t x,
                          std::unordered_set<tag_t> *tags) const
 {
   xml_tree::node_t sibling = next_sibling(x);
-  while(!is_nil(sibling) && tags->find(tag(sibling)) == tags->end())
+  xml_tree::tag_t t;
+  while(!is_nil(sibling)) {
+    t = tag(sibling);
+    if (tags->find(t) != tags->end()) return sibling;
     sibling = next_sibling(sibling);
-  return (sibling);
+  };
+  return sibling;
 }
 
 xml_tree::node_t
@@ -225,10 +240,12 @@ void xml_tree::save(int fd, char* s)
   ufwrite(tag_seq, sizeof(uint), uint_len(bits_per_tag, tag_seq_len), fp);
 
   bool disable_tc = text_collection == 0 || text_positions == 0;
-  ufwrite(&disable_tc, sizeof(bool),1,fp);
 
-  text_positions->save(fp);
+  ufwrite(&disable_tc, sizeof(bool),1,fp);
+  fprintf(stderr, "whoot\n");
+  fflush(stderr);
   if (!disable_tc) {
+    text_positions->save(fp);
     ufwrite(&text_index_type,
             sizeof(TextCollectionBuilder::index_type_t),
             1, fp);
@@ -253,8 +270,7 @@ void xml_tree::save(int fd, char* s)
   fclose(fp);
 
 }
-//static xml_tree* load(char*, bool, int);
-//  void print(int, node_t, bool no_text=false);
+
 xml_tree* xml_tree::load(int fd, char* name, bool load_tc, int sf)
 {
   FILE *fp;
@@ -263,7 +279,6 @@ xml_tree* xml_tree::load(int fd, char* name, bool load_tc, int sf)
   int i;
   buffer[1023] = '\0';
   fp = fdopen(fd, "r");
-
   xml_tree *tree = new xml_tree();
 
   tree->par = loadTree(fp); //TODO use new api
@@ -576,5 +591,5 @@ void xml_tree::print(int fd, xml_tree::node_t x, bool no_text)
       text_collection->DeleteText(orig_text - 1);
     else
       text_collection->DeleteText(orig_text);
-  
+
 }