X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=XMLTree.h;h=6d11d78de3e5c2fa624eeb937209f3c8a5f8b9e6;hb=f32808a35be7a1e62830a5972473178014fa44e5;hp=00ad143f5e8e6075300cbb96ffebeb7436d1a8c5;hpb=b2df171c52f1e6d35a8b131299e4a7f494520333;p=SXSI%2FXMLTree.git

diff --git a/XMLTree.h b/XMLTree.h
index 00ad143..6d11d78 100644
--- a/XMLTree.h
+++ b/XMLTree.h
@@ -20,23 +20,13 @@
 
 #ifndef XMLTREE_H_
 #define XMLTREE_H_
-extern "C" {
-#define CAML_NAME_SPACE
-#include <caml/mlvalues.h>
-#include <caml/custom.h>
-#define XMLTREE(x) ((XMLTree *)(* (XMLTree**) Data_custom_val(x)))
-  //#define XMLTREE(x) ((XMLTree*) (x))
-}
+
+
 #include <unordered_set>
 #include <unordered_map>
 #include <sstream>
 #include "TextCollection/TextCollectionBuilder.h"
 
-#include <cstdio>
-#include <cstdlib>
-#include <cstring>
-
-
 #undef W
 #undef WW
 #undef Wminusone
@@ -50,7 +40,6 @@ using SXSI::TextCollection;
 using SXSI::TextCollectionBuilder;
 
 
-
 // this constant is used to efficiently compute the child operation in the tree
 #define OPTD 10
 
@@ -96,7 +85,7 @@ typedef struct {
 
 
 typedef std::unordered_set<int> TagIdSet;
-typedef std::unordered_map<string,int> TagIdMap;
+typedef std::unordered_map<std::string,int> TagIdMap;
 typedef TagIdMap::const_iterator TagIdMapIT;
 
 #define REGISTER_TAG(v,h,t) do { (h)->insert(std::make_pair((t),(v)->size()));\
@@ -123,7 +112,7 @@ class XMLTree {
    bp *Par;
  
    /** Mapping from tag identifer to tag name */  
-   vector<string> *TagName;
+   std::vector<std::string> *TagName;
    TagIdMap * tIdMap;
   
    /** Bit vector indicating with a 1 the positions of the non-empty texts. */
@@ -141,13 +130,47 @@ class XMLTree {
    bool disable_tc;
    
    FILE* stream;
-   int   stream_fd;
+   int   stream_fd; 
+   std::string * buffer;
+   void myfputs(const char* s, FILE * fp){
+     buffer->append(s);
+     if (buffer->size() >= 100000){
+       fputs(buffer->c_str(),fp);
+       buffer->clear();
+     };
 
+   }
+   void myfputc(const char c, FILE*fp){
+     buffer->append(1,c);
+     if (buffer->size() >= 100000){
+       fputs(buffer->c_str(),fp);
+       buffer->clear();
+     };
+   }
+   void mybufferflush(FILE* fp){
+     fputs(buffer->c_str(), fp);
+     buffer->clear();
+   }
+
+   size_t myfprintf(const char* s, FILE * fp){
+     if (s == NULL)
+       return 0;
+     size_t i = buffer->size();
+     buffer->append(s);
+     size_t j = buffer->size();
+     if (buffer->size() >= 100000){
+       fputs(buffer->c_str(),fp);
+       buffer->clear();
+     };
+     return (j-i);
+   }
+
+   void PrintNode(treeNode n, int fd);
    /** Data structure constructors */
-   XMLTree(){;};
+   XMLTree(){ buffer = 0;};
 
    // non const pointer are freed by this method.
-  XMLTree( pb * const par, uint npar,  vector<string> * const TN,  TagIdMap * const tim, uint *empty_texts_bmp, TagType *tags,
+   XMLTree( pb * const par, uint npar,  std::vector<std::string> * const TN,  TagIdMap * const tim, uint *empty_texts_bmp, TagType *tags,
 	   TextCollection * const TC, bool dis_tc);
 
 public: 
@@ -237,7 +260,7 @@ public:
     *    if none.
     */
    treeNode FirstElement(treeNode x);
-   value CamlFirstElement(value x);
+
    /** LastChild(x): returns the last child of node x.  */
    treeNode LastChild(treeNode x);
    
@@ -250,7 +273,7 @@ public:
     *    if none.
     */
    treeNode NextElement(treeNode x);
-   value CamlNextElement(value x);
+
    /** PrevSibling(x): returns the previous sibling of node x, assuming it 
     * exists. */
 
@@ -439,7 +462,7 @@ public:
    uchar* GetText(DocID d) {
      
        uchar * s = Text->GetText(d);
-       return (s[0] == 1 ? (uchar*)"" : s);
+       return (s[0] == 1 ? (s+1) : s);
    }
 
    /** GetText(i, j): returns the texts corresponding to documents with
@@ -477,8 +500,6 @@ public:
 
 };
 
-extern "C" value caml_cpp_fast_first_element(value xmltree, value node);
-extern "C" value caml_cpp_fast_next_element(value xmltree, value node);