Added caching of TextCollection

[SXSI/XMLTree.git] / XMLTree.cpp

diff --git a/XMLTree.cpp b/XMLTree.cpp
index f155d3e..cdfe952 100644 (file)
--- a/XMLTree.cpp
+++ b/XMLTree.cpp
@@ -105,7 +105,15 @@ void XMLTree::Save(unsigned char *filename)
     // stores the texts   \r
     if (!disable_tc)\r
       Text->Save(fp);\r
-\r
+    if (!disable_tc){\r
+    int st = CachedText.size();\r
+    ufwrite(&st, sizeof(int),1,fp);\r
+    for (int i = 0; i< CachedText.size(); ++i){\r
+      st = CachedText.at(i).size();\r
+      ufwrite(&st, sizeof(int),1,fp);\r
+      ufwrite(CachedText.at(i).c_str(),sizeof(char),(1+strlen(CachedText.at(i).c_str())),fp);\r
+    };\r
+    };\r
     fclose(fp);\r
 \r
  }\r
@@ -192,11 +200,27 @@ XMLTree *XMLTree::Load(unsigned char *filename, int sample_rate_text)
     if (!XML_Tree->disable_tc){\r
       XML_Tree->Text = TextCollection::InitTextCollection(sample_rate_text);\r
       XML_Tree->Text->Load(fp,sample_rate_text);\r
+      int sst;\r
+      int st;\r
+      ufread(&sst, sizeof(int),1,fp);\r
+      for (int i=0;i<sst;i++){\r
+       ufread(&st, sizeof(int),1,fp);\r
+       char* str = (char*) malloc(sizeof(char)*st+1);\r
+       ufread(str,sizeof(char),st+1,fp);\r
+       string cppstr = str;\r
+       XML_Tree->CachedText.push_back(cppstr);\r
+       free(str);\r
+      };\r
+\r
     }\r
     else\r
       XML_Tree->Text = NULL;\r
 \r
     s_text = ftell(fp) - s_text;\r
+\r
+    \r
+\r
+\r
     fclose(fp);\r
 \r
     std::cerr << "Tree part is " << s_tree/1024 << " Kbytes,\n"\r
@@ -268,6 +292,9 @@ int XMLTree::SubtreeTags(treeNode x, TagType tag)
        fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
        exit(1);\r
     }\r
+    if (x == Root())\r
+      x = first_child(Par,x);\r
+    \r
 \r
     int s = x + 2*subtree_size(Par, x) - 1;\r
  \r
@@ -368,7 +395,7 @@ TagType XMLTree::Tag(treeNode x)
        fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
        exit(1);\r
     }\r
-\r
+    \r
     return Tags->access(node2tagpos(x));\r
  }\r
 \r
@@ -496,6 +523,9 @@ treeNode XMLTree::TaggedDesc(treeNode x, TagType tag)
 \r
     int r, s;\r
     treeNode y;\r
+    if (isleaf(Par,x))\r
+      return NULLT;\r
+\r
     r = (int) Tags->rank(tag, node2tagpos(x));\r
     s = (int) Tags->select(tag, r+1);\r
     if (s == -1) return NULLT; // there is no such node\r
@@ -504,6 +534,28 @@ treeNode XMLTree::TaggedDesc(treeNode x, TagType tag)
     else return y;\r
  }\r
 \r
+// TaggedNext(x,tag): returns the first node tagged tag with larger preorder than x \r
+// Returns NULLT if there is none.\r
+treeNode XMLTree::TaggedNext(treeNode x, TagType tag) \r
+ {\r
+    if (!finished) {\r
+       fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
+       exit(1);\r
+    }\r
+\r
+    int r, s;\r
+    treeNode y;\r
+    if (x==NULLT)\r
+      return NULLT;\r
+\r
+    r = (int) Tags->rank(tag, node2tagpos(x));\r
+    s = (int) Tags->select(tag, r+1);\r
+    if (s == -1) return NULLT; // there is no such node\r
+    y = tagpos2node(s); // transforms the tag position into a node position  \r
+    return (y<=x ? NULLT : y);\r
+ }\r
+\r
+\r
 // TaggedPrec(x,tag): returns the first node tagged tag with smaller preorder than x and not an\r
 // ancestor of x. Returns NULLT if there is none.\r
 treeNode XMLTree::TaggedPrec(treeNode x, TagType tag) \r
@@ -539,6 +591,9 @@ treeNode XMLTree::TaggedFoll(treeNode x, TagType tag)
     }\r
 \r
     int r, s;\r
+    if (x ==NULLT || x == Root()|| (next_sibling(Par,x) == -1 ))\r
+      return NULLT;\r
+\r
     r = (int) Tags->rank(tag, node2tagpos(next_sibling(Par, x))-1);\r
     s = (int) Tags->select(tag, r+1);  // select returns -1 in case that there is no r+1-th tag.\r
     if (s==-1) return NULLT;\r
@@ -811,7 +866,7 @@ int XMLTree::NewOpenTag(unsigned char *tagname)
     tags_aux[npar] = i; // inserts the new tag id within the preorder sequence of tags\r
     \r
     npar++;\r
-\r
+    \r
     return 1;\r
     \r
  }\r
@@ -884,6 +939,8 @@ int XMLTree::NewText(unsigned char *s)
     }\r
     \r
     Text->InsertText(s);\r
+    string cpps = (char*) s;\r
+    CachedText.push_back(cpps); \r
     \r
     return 1; // success\r
  }\r