return dest;\r
}\r
\r
+void XMLTree::print_stats() {\r
+ uint total_space = Tags->size()+sizeof(static_sequence*);\r
+ total_space += sizeof(uint*)+sizeof(uint)*(2+uint_len(tags_blen,tags_len));\r
+ cout << "Space usage for XMLTree:" << endl\r
+ << " - tags static_sequence: " << Tags->size()+sizeof(static_sequence*) << endl\r
+ << " - tags access array: " << sizeof(uint*)+sizeof(uint)*(2+uint_len(tags_blen,tags_len)) << endl\r
+ << " ... add Diego structures ... " << endl\r
+ << " *total* " << total_space << endl;\r
+}\r
+\r
// Save: saves XML tree data structure to file. \r
void XMLTree::Save(unsigned char *filename) \r
{\r
\r
// stores the tags\r
Tags->save(fp);\r
+ ufwrite(&tags_blen,sizeof(uint),1,fp);\r
+ ufwrite(&tags_len,sizeof(uint),1,fp);\r
+ ufwrite(tags_fix,sizeof(uint),uint_len(tags_blen,tags_len),fp);\r
\r
// stores the texts \r
- if (!disable_tc)\r
+ if (!disable_tc) {\r
Text->Save(fp);\r
-\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+CachedText.at(i).size(),fp);\r
+ };\r
+ };\r
fclose(fp);\r
\r
}\r
\r
// loads the tags\r
XML_Tree->Tags = static_sequence::load(fp);\r
+ ufread(&XML_Tree->tags_blen,sizeof(uint),1,fp);\r
+ ufread(&XML_Tree->tags_len,sizeof(uint),1,fp);\r
+ XML_Tree->tags_fix = new uint[uint_len(XML_Tree->tags_blen,XML_Tree->tags_len)];\r
+ ufread(XML_Tree->tags_fix,sizeof(uint),uint_len(XML_Tree->tags_blen,XML_Tree->tags_len),fp);\r
+ s_tree+=2*sizeof(uint)+sizeof(uint)*uint_len(XML_Tree->tags_blen,XML_Tree->tags_len);\r
s_tree+= XML_Tree->Tags->size();\r
\r
s_text = ftell(fp);\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
+ else {\r
XML_Tree->Text = NULL;\r
-\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
+ /*std::cerr << "Tree part is " << s_tree/1024 << " Kbytes,\n"\r
+ << "with node->tagid part " << XML_Tree->Tags->size()/1024+(uint_len(XML_Tree->tags_blen,XML_Tree->tags_len)*sizeof(uint))/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
+ << "Text collection is " << s_text/1024 << " Kbytes \n";*/\r
+ XML_Tree->print_stats();\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
- return Tags->access(node2tagpos(x));\r
+ \r
+ return get_field(tags_fix,tags_blen,node2tagpos(x)); //Tags->access(node2tagpos(x));\r
}\r
\r
// DocIds(x): returns the range of text identifiers that descend from node x.\r
child = first_child(Par, x); // starts at first child of node x\r
if (child==(treeNode)-1) return NULLT; // node x is a leaf, there is no such child\r
while (child!=(treeNode)-1) {\r
- if (Tags->access(node2tagpos(child)) == tag) { // current child is labeled with tag of interest\r
+ if (get_field(tags_fix,tags_blen,node2tagpos(child)) /*Tags->access(node2tagpos(child))*/ == tag) { // current child is labeled with tag of interest\r
i--;\r
if (i==0) return child; // we have seen i children of x tagged tag, this is the one we are looking for\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
return NULLT; // there is no such node \r
}\r
\r
+\r
// TaggedFoll(x,tag): returns the first node tagged tag with larger preorder than x and not in\r
// the subtree of x. Returns NULLT if there is none.\r
-treeNode XMLTree::TaggedFoll(treeNode x, TagType tag) \r
+treeNode XMLTree::TaggedFoll(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
+ if (x ==NULLT || x == Root())\r
+ return NULLT;\r
+ \r
+ r = (int) Tags->rank(tag, find_close(Par, x));\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
+ else return tagpos2node(s);\r
+ } \r
+\r
+\r
+// TaggedFollowingSibling(x,tag): returns the first node tagged tag with larger preorder than x and not in\r
+// the subtree of x. Returns NULLT if there is none.\r
+treeNode XMLTree::TaggedFollowingSibling(treeNode x, TagType tag) \r
{\r
if (!finished) {\r
fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
}\r
\r
int r, s;\r
- if (x == Root() || (next_sibling(Par,x) == -1 ))\r
+ treeNode ns = next_sibling(Par,x);\r
+\r
+ if (x == NULLT || x == Root() || ns == -1)\r
return NULLT;\r
- \r
- r = (int) Tags->rank(tag, node2tagpos(next_sibling(Par, x))-1);\r
+\r
+ r = (int) Tags->rank(tag, node2tagpos(ns)-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
else return tagpos2node(s);\r
}\r
\r
+\r
+// TaggedAncestor(x, tag): returns the closest ancestor of x tagged tag. Return\r
+// NULLT is there is none.\r
+treeNode XMLTree::TaggedAncestor(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
+ if (x == NULLT || x == Root())\r
+ return NULLT;\r
+ \r
+ treeNode s = parent(Par, x), r = Root();\r
+ while (s != r) {\r
+ if (get_field(tags_fix,tags_blen,node2tagpos(s)) /*Tags->access(node2tagpos(s))*/ == tag) return s;\r
+ s = parent(Par, s);\r
+ }\r
+ return NULLT;\r
+ }\r
+\r
+\r
// PrevText(x): returns the document identifier of the text to the left \r
// of node x, or NULLT if x is the root node or the text is empty.\r
// Assumes Doc ids start from 0.\r
found_attributes = false;\r
npar = 0;\r
parArraySize = 1;\r
- ntagnames = 2; \r
+ ntagnames = 4; \r
disable_tc = dtc;\r
\r
indexing_empty_texts = empty_texts;\r
\r
tags_aux = (TagType *) umalloc(sizeof(TagType));\r
\r
- TagName = (unsigned char **) umalloc(2*sizeof(unsigned char*));\r
+ TagName = (unsigned char **) umalloc(4*sizeof(unsigned char*));\r
\r
TagName[0] = (unsigned char *) umalloc(4*sizeof(unsigned char));\r
\r
TagName[1] = (unsigned char *) umalloc(4*sizeof(unsigned char));\r
\r
strcpy((char *) TagName[1], "<$>");\r
+ \r
+ //OJO need to put these in the table too.\r
+ TagName[2] = (unsigned char *) umalloc(5*sizeof(unsigned char));\r
+\r
+ strcpy((char *) TagName[2], "/<@>");\r
+\r
+ TagName[3] = (unsigned char *) umalloc(5*sizeof(unsigned char));\r
+\r
+ strcpy((char *) TagName[3], "/<$>");\r
\r
\r
if (!indexing_empty_texts) \r
Par = (bp *)umalloc(sizeof(bp));\r
bp_construct(Par, npar, par_aux, OPT_DEGREE|0); \r
// creates structure for tags\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_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
\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
+ static_bitsequence_builder * bmb = new static_bitsequence_builder_brw32(20);\r
+ alphabet_mapper *am = new alphabet_mapper_none();\r
+ wt_coder * wc = new wt_coder_huff((uint*)tags_aux,npar-1,am);\r
+ Tags = new static_sequence_wvtree((uint*)tags_aux,npar-1,wc ,bmb, am);\r
+ //Tags = new static_sequence_gmr((uint *) tags_aux, (uint) npar-1,ntagsize, bmb, ssb);\r
+ uint max_tag = 0;\r
+ for(uint i=0;i<(uint)npar-1;i++)\r
+ max_tag = max(max_tag,tags_aux[i]);\r
+ tags_blen = bits(max_tag);\r
+ tags_len = (uint)npar-1;\r
+ tags_fix = new uint[uint_len(tags_blen,tags_len)];\r
+ for(uint i=0;i<(uint)npar-1;i++)\r
+ set_field(tags_fix,tags_blen,i,tags_aux[i]);\r
\r
delete bmb;\r
- delete pmb;\r
- delete ssb;\r
+ //delete pmb;\r
+ //delete ssb;\r
// makes the text collection static\r
if (!disable_tc)\r
Text->MakeStatic();\r
tags_aux = NULL;\r
\r
finished = true;\r
+ print_stats();\r
\r
return 1; // indicates success in the inicialization\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
parArraySize *= 2;\r
}\r
\r
- setbit(par_aux,npar,CP); // marks a new closing opening parenthesis\r
+ setbit(par_aux,npar,CP); // marks a new closing parenthesis\r
\r
// 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]+1))==0) && (TagName[i][0]=='/')) break;\r
\r
if (i==ntagnames) { // the tag is a new one, then we insert it\r
TagName = (unsigned char **)urealloc(TagName, sizeof(char *)*(ntagnames+1));\r
\r
- \r
ntagnames++;\r
TagName[i] = (unsigned char *)umalloc(sizeof(char)*(strlen((const char *)tagname)+2));\r
TagName[i][0] = '/';\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