Jouni's Incremental BWT integrated into TextCollection

[SXSI/XMLTree.git] / XMLTree.cpp

diff --git a/XMLTree.cpp b/XMLTree.cpp
index 9057e31..70cf4b7 100644 (file)
--- a/XMLTree.cpp
+++ b/XMLTree.cpp
@@ -223,8 +223,7 @@ XMLTree *XMLTree::Load(unsigned char *filename, int sample_rate_text)
 \r
     // loads the texts\r
     if (!XML_Tree->disable_tc){\r
-      XML_Tree->Text = TextCollection::InitTextCollection(sample_rate_text);\r
-      XML_Tree->Text->Load(fp,sample_rate_text);\r
+      XML_Tree->Text = TextCollection::Load(fp,sample_rate_text);\r
       int sst;\r
       int st;\r
       ufread(&sst, sizeof(int),1,fp);\r
@@ -284,7 +283,9 @@ XMLTree::~XMLTree()
     Tags = NULL;\r
 \r
     //Text->~TextCollection();\r
-    delete Text;\r
+    delete TextBuilder; \r
+    TextBuilder = NULL;\r
+    delete Text; \r
     Text = NULL;\r
 \r
     initialized = false;\r
@@ -1013,9 +1014,11 @@ int XMLTree::OpenDocument(bool empty_texts, int sample_rate_text,bool dtc)
     if (!indexing_empty_texts) \r
       empty_texts_aux = (unsigned int *)umalloc(sizeof(unsigned int));\r
        \r
-    \r
-    \r
-    Text = TextCollection::InitTextCollection((unsigned)sample_rate_text);\r
+    if (disable_tc)\r
+        TextBuilder = 0;\r
+    else \r
+        TextBuilder = new TextCollectionBuilder((unsigned)sample_rate_text);\r
+    Text = 0;\r
     \r
     return 1;  // indicates success in the initialization of the data structure\r
  }\r
@@ -1044,20 +1047,21 @@ int XMLTree::CloseDocument()
                uint max_tag = 0;\r
                for(uint i=0;i<(uint)npar-1;i++)\r
                        max_tag = max(max_tag,tags_aux[i]);\r
-               max_tag++;\r
-               tags_aux = (TagType *) urealloc(tags_aux, sizeof(TagType)*(npar + 1));\r
-               tags_aux[npar++] = max_tag;\r
+               //max_tag++;\r
+               //tags_aux = (TagType *) urealloc(tags_aux, sizeof(TagType)*(npar + 1));\r
+               //tags_aux[npar++] = max_tag;\r
     //int ntagsize = found_attributes ? 2*ntagnames-1 : 2*ntagnames - 2;\r
     int ntagsize = 2*ntagnames + 2;\r
 \r
     //static_bitsequence_builder * bmb = new static_bitsequence_builder_brw32(20);\r
     //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
-               static_bitsequence_builder * bmb = new static_bitsequence_builder_brw32(20);\r
+               static_bitsequence_builder * bmb = new static_bitsequence_builder_sdarray();\r
                alphabet_mapper *am = new alphabet_mapper_none();\r
-               wt_coder * wc = new wt_coder_huff((uint*)tags_aux,npar,am);\r
-               Tags = new static_sequence_wvtree((uint*)tags_aux,npar,wc ,bmb, am);\r
+               //wt_coder * wc = new wt_coder_huff((uint*)tags_aux,npar,am);\r
+               //Tags = new static_sequence_wvtree((uint*)tags_aux,npar,wc ,bmb, am);\r
     //Tags = new static_sequence_gmr((uint *) tags_aux, (uint) npar,ntagsize, bmb, ssb);\r
+               Tags = new static_sequence_bs((uint*)tags_aux,npar,am,bmb);\r
                \r
                cout << "Tags test: " << Tags->test((uint*)tags_aux,npar) << endl;\r
 \r
@@ -1072,10 +1076,16 @@ int XMLTree::CloseDocument()
     //delete ssb;\r
 \r
     \r
-               // makes the text collection static\r
+    // makes the text collection static\r
     if (!disable_tc)\r
-      Text->MakeStatic();\r
-    \r
+    {\r
+        assert(Text = 0);\r
+        assert(TextBuilder != 0);\r
+        Text = TextBuilder->InitTextCollection();\r
+        delete TextBuilder;\r
+        TextBuilder = 0;\r
+    }\r
+\r
     // creates the data structure marking the non-empty texts (just in the case it is necessary)\r
     if (!indexing_empty_texts)  {\r
        EBVector = new static_bitsequence_rrr02((uint *)empty_texts_aux,(ulong)npar,(uint)32);\r
@@ -1213,7 +1223,7 @@ int XMLTree::NewText(unsigned char *s)
               bitset(empty_texts_aux, npar-1);  // marks the non-empty text with a 1 in the bit vector\r
     }\r
     \r
-    Text->InsertText(s);\r
+    TextBuilder->InsertText(s);\r
     string cpps = (char*) s;\r
     CachedText.push_back(cpps); \r
     \r
@@ -1238,7 +1248,7 @@ int XMLTree::NewEmptyText()
        \r
        bitclean(empty_texts_aux, npar-1);  // marks the empty text with a 0 in the bit vector\r
     }\r
-    else Text->InsertText(&c); // we insert the empty text just in case we index all the texts\r
+    else TextBuilder->InsertText(&c); // we insert the empty text just in case we index all the texts\r
     \r
     return 1; // success    \r
  }\r
@@ -1252,7 +1262,7 @@ TagType XMLTree::GetTagId(unsigned char *tagname)
     // this should be changed for more efficient processing\r
     for (i=0; i<ntagnames; i++)\r
        if (strcmp((const char *)tagname,(const char *)TagName[i])==0) break; \r
-    if (i==ntagnames) return ntagnames; //(TagType)NULLT; // tagname does not exists in the table\r
+    if (i==ntagnames) return (TagType)-1; //ntagnames; //(TagType)NULLT; // tagname does not exists in the table\r
     else return i;\r
  }\r
 \r
@@ -1262,7 +1272,7 @@ TagType XMLTree::GetTagId(unsigned char *tagname)
 unsigned char *XMLTree::GetTagName(TagType tagid)\r
  {\r
     unsigned char *s;\r
-\r
+               if(tagid==(uint)-1) return NULL;\r
     if (tagid >= ntagnames) return NULL; // invalid tag identifier\r
     s = (unsigned char *)umalloc((strlen((const char *)TagName[tagid])+1)*sizeof(unsigned char));\r
     strcpy((char *)s, (const char *)TagName[tagid]);\r
@@ -1274,6 +1284,7 @@ unsigned char *XMLTree::GetTagName(TagType tagid)
 \r
 const unsigned char *XMLTree::GetTagNameByRef(TagType tagid)\r
  {\r
+               if(tagid==(uint)-1) return NULL;\r
     if (tagid >= ntagnames) return NULL; // invalid tag identifier\r
     return ((const unsigned char*)  TagName[tagid]);\r
  }\r