// 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
char buffer[1024];\r
XMLTree *XML_Tree;\r
int i;\r
- \r
+ size_t s_tree = 0;\r
+ long s_text = 0;\r
+ size_t s_tags = 0;\r
+\r
// first load the tree topology\r
sprintf(buffer, "%s.srx", filename);\r
fp = fopen(buffer, "r");\r
XML_Tree->Par = (bp *)umalloc(sizeof(bp));\r
\r
loadTree(XML_Tree->Par, fp); \r
- \r
+\r
+ s_tree += sizeof(bp);\r
+\r
// stores the table with tag names\r
ufread(&XML_Tree->ntagnames, sizeof(int), 1, fp);\r
+ \r
+ s_tree += sizeof(int);\r
+\r
XML_Tree->TagName = (unsigned char **)umalloc(XML_Tree->ntagnames*sizeof(unsigned char *));\r
+ \r
+ s_tags += sizeof(unsigned char*)*XML_Tree->ntagnames;\r
+\r
\r
for (i=0; i<XML_Tree->ntagnames;i++) {\r
\r
int len = strlen((const char*)buffer);\r
XML_Tree->TagName[i] = (unsigned char *)ucalloc(len,sizeof(char));\r
strncpy((char *)XML_Tree->TagName[i], (const char *)buffer,len - 1);\r
+ s_tags+= len*sizeof(char);\r
}\r
\r
// loads the flags\r
ufread(&(XML_Tree->initialized), sizeof(bool), 1, fp);\r
ufread(&(XML_Tree->finished), sizeof(bool), 1, fp);\r
ufread(&(XML_Tree->disable_tc), sizeof(bool), 1, fp);\r
+ \r
+ s_tree+=sizeof(bool)*4;\r
\r
if (!(XML_Tree->indexing_empty_texts)) XML_Tree->EBVector = static_bitsequence_rrr02::load(fp);\r
-\r
+ \r
+ s_tree+= XML_Tree->EBVector->size();\r
+ \r
// loads the tags\r
XML_Tree->Tags = static_sequence::load(fp);\r
+ s_tree+= XML_Tree->Tags->size();\r
+\r
+ s_text = ftell(fp);\r
\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
+ 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
+ << "with node->tagid part " << XML_Tree->Tags->size()/1024 << "Kbytes \n"\r
+ << "size of Tag part : " << XML_Tree->Tags->length () << " elements\n"\r
+ << "sizof(unsigned int)* " << XML_Tree->Tags->length () << " = " << \r
+ sizeof(unsigned int) * XML_Tree->Tags->length () / 1024 << " Kbytes\n"\r
+ << "Tag part is " << s_tags/1024 << " Kbytes,\n"\r
+ << "Text collection is " << s_text/1024 << " Kbytes \n";\r
return XML_Tree;\r
}\r
\r
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
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
\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
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
}\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
if (d == NULLT)\r
return NULLT;\r
\r
- int s = d;\r
- // Kim : I added the d+1. before that, else was select1(d)\r
- // and gave wrong results but I'm really poking a dead bear with a stick here.\r
+ int s;\r
+ // OJO : Kim : I added the d+1. before that, else branch was \r
+ // EBVector->select1(d)\r
+ // and gave wrong results (I'm really poking a bear with a stick here).\r
if (indexing_empty_texts) s = d;\r
else s = EBVector->select1(d+1);\r
\r
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
}\r
\r
Text->InsertText(s);\r
+ string cpps = (char*) s;\r
+ CachedText.push_back(cpps); \r
\r
return 1; // success\r
}\r
projects / SXSI / XMLTree.git / blobdiff