- Added an option to disable the TextCollection.

[SXSI/XMLTree.git] / XMLTree.cpp

diff --git a/XMLTree.cpp b/XMLTree.cpp
index 3c51b2f..ecc9a27 100644 (file)
--- a/XMLTree.cpp
+++ b/XMLTree.cpp
@@ -311,8 +311,10 @@ treeNode XMLTree::Parent(treeNode x)
        fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
        exit(1);\r
     }\r
-\r
-    return parent(Par, x);\r
+    if (x == Root())\r
+      return NULLT;\r
+    else\r
+      return parent(Par, x);\r
  }\r
 \r
 // Child(x,i): returns the i-th child of node x, assuming it exists.\r
@@ -568,14 +570,16 @@ treeNode XMLTree::ParentNode(DocID d)
 // OpenDocument(empty_texts): it starts the construction of the data structure for\r
 // the XML document. Parameter empty_texts indicates whether we index empty texts\r
 // in document or not. Returns a non-zero value upon success, NULLT in case of error.\r
-int XMLTree::OpenDocument(bool empty_texts, int sample_rate_text)\r
+int XMLTree::OpenDocument(bool empty_texts, int sample_rate_text,bool dtc)\r
  {\r
     initialized = true;\r
     finished = false;\r
+    found_attributes = false;\r
     npar = 0;\r
     parArraySize = 1;\r
-    ntagnames = 0;    \r
- \r
+    ntagnames = 2;    \r
+    disable_tc = dtc;\r
+    \r
     indexing_empty_texts = empty_texts;\r
     \r
     par_aux = (pb *)malloc(sizeof(pb)*parArraySize);\r
@@ -590,7 +594,28 @@ int XMLTree::OpenDocument(bool empty_texts, int sample_rate_text)
        return NULLT;\r
     }\r
     \r
-    TagName = NULL;\r
+    TagName = (unsigned char **) malloc(2*sizeof(unsigned char*));\r
+    if (!TagName){\r
+       fprintf(stderr, "Error: not enough memory\n");\r
+       return NULLT;\r
+    }\r
+\r
+    TagName[0] = (unsigned char *) malloc(4*sizeof(unsigned char));\r
+    strcpy((char *) TagName[0], "<@>");\r
+\r
+    if (!TagName[0]){\r
+      fprintf(stderr, "Error: not enough memory\n");\r
+      return NULLT;\r
+    }\r
+\r
+    TagName[1] = (unsigned char *) malloc(4*sizeof(unsigned char));\r
+    if (!TagName[1]){\r
+      fprintf(stderr, "Error: not enough memory\n");\r
+      return NULLT;\r
+    }\r
+\r
+    strcpy((char *) TagName[1], "<$>");\r
+\r
 \r
     if (!indexing_empty_texts) {\r
        empty_texts_aux = (unsigned int *)malloc(sizeof(unsigned int));\r
@@ -631,17 +656,29 @@ int XMLTree::CloseDocument()
     static_permutation_builder * pmb = new static_permutation_builder_mrrr(PERM_SAMPLE, bmb);\r
     static_sequence_builder * ssb = new static_sequence_builder_gmr_chunk(bmb, pmb);\r
 \r
-    Tags = new static_sequence_gmr((uint *) tags_aux, (uint) npar-1,2*ntagnames, bmb, ssb);\r
+\r
+    // If we found an attribute then "<@>" is present in the tree\r
+    // if we didn't then it is not. "<$>" is never present in the tree\r
+    int ntagsize = found_attributes ? 2*ntagnames-1 : 2*ntagnames - 2;\r
+\r
+    Tags = new static_sequence_gmr((uint *) tags_aux, (uint) npar-1,ntagsize, bmb, ssb);\r
     \r
     delete bmb;\r
     delete pmb;\r
     delete ssb;\r
     // makes the text collection static\r
-    Text->MakeStatic();\r
+    if (!disable_tc)\r
+      Text->MakeStatic();\r
     \r
     // creates the data structure marking the non-empty texts (just in the case it is necessary)\r
-    if (!indexing_empty_texts) \r
+    if (!indexing_empty_texts)  {\r
        EBVector = new static_bitsequence_rrr02((uint *)empty_texts_aux,(ulong)npar,(uint)32);\r
+       free (empty_texts_aux);\r
+       empty_texts_aux = NULL;\r
+    }\r
+   \r
+    free(tags_aux);\r
+    tags_aux = NULL;\r
 \r
     finished = true;\r
 \r
@@ -677,8 +714,13 @@ int XMLTree::NewOpenTag(unsigned char *tagname)
     // transforms the tagname into a tag identifier. If the tag is new, we insert\r
     // it in the table.\r
     for (i=0; i<ntagnames; i++)\r
-       if (strcmp((const char *)tagname,(const char *)TagName[i])==0) break;\r
+      if (strcmp((const char *)tagname,(const char *)TagName[i])==0) break;\r
  \r
+\r
+    // NewOpenTag("<@>") was called\r
+    if (i==0) \r
+      found_attributes=true;\r
+\r
     if (i==ntagnames) { // the tag is a new one, then we insert it\r
        TagName = (unsigned char **)realloc(TagName, sizeof(char *)*(ntagnames+1));\r
        \r
@@ -775,6 +817,11 @@ int XMLTree::NewText(unsigned char *s)
        return NULLT;\r
     }\r
 \r
+    if (disable_tc) {\r
+      XMLTree::NewEmptyText();\r
+      return 1;\r
+    };\r
+\r
     if (!indexing_empty_texts) {\r
        empty_texts_aux = (unsigned int *)realloc(empty_texts_aux, sizeof(pb)*(1+(npar-1)/(8*sizeof(pb))));\r
        if (!empty_texts_aux) {\r
@@ -844,4 +891,19 @@ unsigned char *XMLTree::GetTagName(TagType tagid)
  }\r
 \r
 \r
-\r
+TagType XMLTree::RegisterTag(unsigned char *tagname)\r
+{\r
+  if (!finished)\r
+    return NULLT;\r
+  \r
+\r
+  TagType id = XMLTree::GetTagId(tagname);\r
+  if (id == NULLT){\r
+    id = ntagnames;\r
+    ntagnames = ntagnames + 1;    \r
+    TagName = (unsigned char **) realloc(TagName,ntagnames*(sizeof(unsigned char*)));\r
+    strcpy((char*)TagName[id], (const char *)tagname);  \r
+  };\r
+\r
+  return id;\r
+}\r