X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=XMLTree.cpp;h=a37263a16c30577984fcbf5537d1ca382bed6a9f;hb=b6f779e236950176bab71fce980ac1685e8661c0;hp=4ddec0170a5186dc02c43e98ab59e525c89380c2;hpb=dd9992dcd5366f37820c24ed7cddf24ecbc0d549;p=SXSI%2FXMLTree.git

diff --git a/XMLTree.cpp b/XMLTree.cpp
index 4ddec01..a37263a 100644
--- a/XMLTree.cpp
+++ b/XMLTree.cpp
@@ -105,7 +105,15 @@ void XMLTree::Save(unsigned char *filename)
     // stores the texts   
     if (!disable_tc)
       Text->Save(fp);
-
+    if (!disable_tc){
+    int st = CachedText.size();
+    ufwrite(&st, sizeof(int),1,fp);
+    for (int i = 0; i< CachedText.size(); ++i){
+      st = CachedText.at(i).size();
+      ufwrite(&st, sizeof(int),1,fp);
+      ufwrite(CachedText.at(i).c_str(),sizeof(char),(1+strlen(CachedText.at(i).c_str())),fp);
+    };
+    };
     fclose(fp);
 
  }
@@ -192,11 +200,27 @@ XMLTree *XMLTree::Load(unsigned char *filename, int sample_rate_text)
     if (!XML_Tree->disable_tc){
       XML_Tree->Text = TextCollection::InitTextCollection(sample_rate_text);
       XML_Tree->Text->Load(fp,sample_rate_text);
+      int sst;
+      int st;
+      ufread(&sst, sizeof(int),1,fp);
+      for (int i=0;i<sst;i++){
+	ufread(&st, sizeof(int),1,fp);
+	char* str = (char*) malloc(sizeof(char)*st+1);
+	ufread(str,sizeof(char),st+1,fp);
+	string cppstr = str;
+	XML_Tree->CachedText.push_back(cppstr);
+	free(str);
+      };
+
     }
     else
       XML_Tree->Text = NULL;
 
     s_text = ftell(fp) - s_text;
+
+    
+
+
     fclose(fp);
 
     std::cerr << "Tree part is " << s_tree/1024 << " Kbytes,\n"
@@ -557,9 +581,30 @@ treeNode XMLTree::TaggedPrec(treeNode x, TagType tag)
     return NULLT; // there is no such node 
  }
 
+
 // TaggedFoll(x,tag): returns the first node tagged tag with larger preorder than x and not in
 // the subtree of x. Returns NULLT if there is none.
-treeNode XMLTree::TaggedFoll(treeNode x, TagType tag) 
+treeNode XMLTree::TaggedFoll(treeNode x, TagType tag)
+ {
+    if (!finished) {
+       fprintf(stderr, "Error: data structure has not been constructed properly\n");
+       exit(1);
+    }
+
+    int r, s;
+    if (x ==NULLT || x == Root())
+       return NULLT;
+                   
+    r = (int) Tags->rank(tag, find_close(Par, x));
+    s = (int) Tags->select(tag, r+1);  // select returns -1 in case that there is no r+1-th tag.
+    if (s==-1) return NULLT;
+    else return tagpos2node(s);
+ } 
+
+
+// TaggedFollowingSibling(x,tag): returns the first node tagged tag with larger preorder than x and not in
+// the subtree of x. Returns NULLT if there is none.
+treeNode XMLTree::TaggedFollowingSibling(treeNode x, TagType tag) 
  {
     if (!finished) {
        fprintf(stderr, "Error: data structure has not been constructed properly\n");
@@ -567,15 +612,39 @@ treeNode XMLTree::TaggedFoll(treeNode x, TagType tag)
     }
 
     int r, s;
-    if (x ==NULLT || x == Root()|| (next_sibling(Par,x) == -1 ))
+    treeNode ns = next_sibling(Par,x);
+
+    if (x == NULLT || x == Root() || ns == -1)
       return NULLT;
 
-    r = (int) Tags->rank(tag, node2tagpos(next_sibling(Par, x))-1);
+    r = (int) Tags->rank(tag, node2tagpos(ns)-1);
     s = (int) Tags->select(tag, r+1);  // select returns -1 in case that there is no r+1-th tag.
     if (s==-1) return NULLT;
     else return tagpos2node(s);
  }
 
+
+// TaggedAncestor(x, tag): returns the closest ancestor of x tagged tag. Return
+// NULLT is there is none.
+treeNode XMLTree::TaggedAncestor(treeNode x, TagType tag)
+ {
+    if (!finished) {
+       fprintf(stderr, "Error: data structure has not been constructed properly\n");
+       exit(1);
+    }
+    
+    if (x == NULLT || x == Root())
+       return NULLT;
+    
+    treeNode s = parent(Par, x), r = Root();
+    while (s != r) {
+       if (Tags->access(node2tagpos(s)) == tag) return s;
+       s = parent(Par, s);
+    }
+    return NULLT;
+ }
+
+
 // PrevText(x): returns the document identifier of the text to the left 
 // of node x, or NULLT if x is the root node or the text is empty.
 // Assumes Doc ids start from 0.
@@ -842,7 +911,7 @@ int XMLTree::NewOpenTag(unsigned char *tagname)
     tags_aux[npar] = i; // inserts the new tag id within the preorder sequence of tags
     
     npar++;
-
+    
     return 1;
     
  }
@@ -866,17 +935,16 @@ int XMLTree::NewClosingTag(unsigned char *tagname)
        parArraySize *= 2;
     }
     
-    setbit(par_aux,npar,CP);  // marks a new closing opening parenthesis
+    setbit(par_aux,npar,CP);  // marks a new closing parenthesis
 
     // transforms the tagname into a tag identifier. If the tag is new, we insert
     // it in the table.
     for (i=0; i<ntagnames; i++)
-       if (strcmp((const char *)tagname,(const char *)TagName[i])==0) break;
+       if ((strcmp((const char *)tagname,(const char *)(TagName[i]+1))==0) && (TagName[i][0]=='/')) break;
  
     if (i==ntagnames) { // the tag is a new one, then we insert it
        TagName = (unsigned char **)urealloc(TagName, sizeof(char *)*(ntagnames+1));
        
-       
        ntagnames++;
        TagName[i] = (unsigned char *)umalloc(sizeof(char)*(strlen((const char *)tagname)+2));
        TagName[i][0] = '/';
@@ -915,6 +983,8 @@ int XMLTree::NewText(unsigned char *s)
     }
     
     Text->InsertText(s);
+    string cpps = (char*) s;
+    CachedText.push_back(cpps); 
     
     return 1; // success
  }