X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=xml-tree.cpp;h=82f00b0af86742bfd66720272e51c0fd9e650c26;hb=HEAD;hp=d31c1ce92f7209b51d2c052f6ed4d1f7562c576e;hpb=f01f30f68feacc70b6cb4367301f885a8407210b;p=SXSI%2FXMLTree.git

diff --git a/xml-tree.cpp b/xml-tree.cpp
index d31c1ce..82f00b0 100644
--- a/xml-tree.cpp
+++ b/xml-tree.cpp
@@ -95,14 +95,20 @@ xml_tree::xml_tree(std::vector<int32_t> *tags_,
   delete parbitmap;
 
   this->tag_ids = tag_ids;
+
   tag_names = new std::vector<std::string>();
   tag_names->resize(tag_ids->size());
+  this->attribute_ids = new std::unordered_set<xml_tree::tag_t>();
   std::unordered_map<std::string, tag_t>::iterator val;
-  //for(auto val : *(this->tag_ids))
-  //(*this->tag_names)[val.second] = val.first;
-  for(val = this->tag_ids->begin(); val != this->tag_ids->end(); ++val)
-    (*this->tag_names)[val->second] = val->first;
-
+  for(val = this->tag_ids->begin(); val != this->tag_ids->end(); ++val){
+    (*tag_names)[val->second] = val->first;
+    if (val->first.size() >= 3 &&
+        val->first[0] == '<' &&
+        val->first[1] == '@' &&
+        val->first[2] == '>'){
+      this->attribute_ids->insert(val->second);
+    };
+  }
 
   uint32_t max_tag = tag_names->size() - 1;
   bit_vector *tmp_bitmap = new bit_vector(npar, 1, 0);
@@ -166,6 +172,7 @@ xml_tree::~xml_tree()
   delete [] tag_seq;
   delete tag_names;
   delete tag_ids;
+  delete attribute_ids;
   if (text_collection) delete text_collection;
   if (text_positions) delete text_positions;
 }
@@ -283,6 +290,7 @@ xml_tree* xml_tree::load(int fd, char* name, bool load_tc, int sf)
   tree->par = loadTree(fp); //TODO use new api
   tree->tag_names = new std::vector<std::string>();
   tree->tag_ids = new std::unordered_map<std::string, xml_tree::tag_t>();
+  tree->attribute_ids = new std::unordered_set<xml_tree::tag_t>();
   std::string s;
   int ntags;
 
@@ -297,6 +305,9 @@ xml_tree* xml_tree::load(int fd, char* name, bool load_tc, int sf)
     tree->tag_names->push_back(s);
     tree->tag_ids->insert(std::make_pair(s,
                                          static_cast<xml_tree::tag_t>(i)));
+    if (s.size() >= 3 && s[0] == '<' && s[1] == '@' && s[2] == '>'){
+      tree->attribute_ids->insert(static_cast<xml_tree::tag_t>(i));
+    };
 
   };
 
@@ -306,7 +317,7 @@ xml_tree* xml_tree::load(int fd, char* name, bool load_tc, int sf)
 
   //tree->tags = static_sequence_bs::load(fp);
   ufread(&tree->bits_per_tag, sizeof(uint), 1, fp);
-  fprintf(stderr, "\nBits per tag: %u\n", tree->bits_per_tag);
+  //fprintf(stderr, "\nBits per tag: %u\n", tree->bits_per_tag);
   ufread(&tree->tag_seq_len, sizeof(uint), 1, fp);
   size_t size = uint_len(tree->bits_per_tag, tree->tag_seq_len);
   tree->tag_seq = new uint[size];
@@ -376,7 +387,7 @@ std::pair<int32_t, int32_t> xml_tree::text_id_range(xml_tree::node_t x) const
   else
     i = text_positions->rank1(x-1);
   j = text_positions->rank1(y);
-//  fprintf(stderr, "Rank of node %i is %i, rank of closing %i is %i\n", x, i, y, j);
+
   if (i == j)
     return std::make_pair(xml_tree::NIL, xml_tree::NIL);
   else
@@ -388,9 +399,9 @@ int32_t xml_tree::text_id(xml_tree::node_t x) const
   return (int32_t) text_positions->rank1(x) - 1;
 }
 
-unsigned char* xml_tree::get_text(int32_t id) const
+const char* xml_tree::get_text(int32_t id) const
 {
-  unsigned char * s = text_collection->GetText(id);
+  const char * s = reinterpret_cast<const char*>(text_collection->GetText(id));
   return s + (s[0] == 1);
 }
 
@@ -439,12 +450,10 @@ void xml_tree::uputc(const char c, int fd)
 
 const char * xml_tree::get_tag_name_by_ref(xml_tree::tag_t tagid) const
 {
-
-   unsigned char *s;
    if (tagid < 0 || tagid >= tag_names->size())
      return "<INVALID TAG>";
 
-   return (const char *) (*tag_names)[tagid].c_str();
+   return (*tag_names)[tagid].c_str();
 }
 
 xml_tree::tag_t xml_tree::register_tag(char *s)
@@ -573,11 +582,11 @@ void xml_tree::print(xml_tree::node_t x, int fd, bool no_text)
             label = tag(n);
           } else
             uputc('>', fd);
-        } else {
+          } else {
           uputs("/>", fd);
           n++;
           label = tag(n);
-        };
+          };
       };
     } else do {
       uputs("</", fd);