X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=XMLTree.cpp;h=e17c89d90358731f2cbad205b9c5a4534f96af76;hb=4bd816265ae88b4e1631634260384baa559f8967;hp=62ca3fd7d0554c378187eeb1d60e588f83cdb712;hpb=ac53090f1525b3ef9cbde0b51da5fbb238a06744;p=SXSI%2FXMLTree.git diff --git a/XMLTree.cpp b/XMLTree.cpp index 62ca3fd..e17c89d 100644 --- a/XMLTree.cpp +++ b/XMLTree.cpp @@ -101,6 +101,9 @@ void XMLTree::Save(unsigned char *filename) // stores the tags Tags->save(fp); + ufwrite(&tags_blen,sizeof(uint),1,fp); + ufwrite(&tags_len,sizeof(uint),1,fp); + ufwrite(tags_fix,sizeof(uint),uint_len(tags_blen,tags_len),fp); // stores the texts if (!disable_tc) { @@ -191,6 +194,11 @@ XMLTree *XMLTree::Load(unsigned char *filename, int sample_rate_text) // loads the tags XML_Tree->Tags = static_sequence::load(fp); + ufread(&XML_Tree->tags_blen,sizeof(uint),1,fp); + ufread(&XML_Tree->tags_len,sizeof(uint),1,fp); + XML_Tree->tags_fix = new uint[uint_len(XML_Tree->tags_blen,XML_Tree->tags_len)]; + ufread(XML_Tree->tags_fix,sizeof(uint),uint_len(XML_Tree->tags_blen,XML_Tree->tags_len),fp); + s_tree+=2*sizeof(uint)+sizeof(uint)*uint_len(XML_Tree->tags_blen,XML_Tree->tags_len); s_tree+= XML_Tree->Tags->size(); s_text = ftell(fp); @@ -223,7 +231,7 @@ XMLTree *XMLTree::Load(unsigned char *filename, int sample_rate_text) fclose(fp); std::cerr << "Tree part is " << s_tree/1024 << " Kbytes,\n" - << "with node->tagid part " << XML_Tree->Tags->size()/1024 << "Kbytes \n" + << "with node->tagid part " << XML_Tree->Tags->size()/1024+(uint_len(XML_Tree->tags_blen,XML_Tree->tags_len)*sizeof(uint))/1024 << "Kbytes \n" << "size of Tag part : " << XML_Tree->Tags->length () << " elements\n" << "sizof(unsigned int)* " << XML_Tree->Tags->length () << " = " << sizeof(unsigned int) * XML_Tree->Tags->length () / 1024 << " Kbytes\n" @@ -395,7 +403,7 @@ TagType XMLTree::Tag(treeNode x) exit(1); } - return Tags->access(node2tagpos(x)); + return get_field(tags_fix,tags_blen,node2tagpos(x)); //Tags->access(node2tagpos(x)); } // DocIds(x): returns the range of text identifiers that descend from node x. @@ -502,7 +510,7 @@ treeNode XMLTree::TaggedChild(treeNode x, int i, TagType tag) child = first_child(Par, x); // starts at first child of node x if (child==(treeNode)-1) return NULLT; // node x is a leaf, there is no such child while (child!=(treeNode)-1) { - if (Tags->access(node2tagpos(child)) == tag) { // current child is labeled with tag of interest + if (get_field(tags_fix,tags_blen,node2tagpos(child)) /*Tags->access(node2tagpos(child))*/ == tag) { // current child is labeled with tag of interest i--; if (i==0) return child; // we have seen i children of x tagged tag, this is the one we are looking for } @@ -637,7 +645,7 @@ treeNode XMLTree::TaggedAncestor(treeNode x, TagType tag) treeNode s = parent(Par, x), r = Root(); while (s != r) { - if (Tags->access(node2tagpos(s)) == tag) return s; + if (get_field(tags_fix,tags_blen,node2tagpos(s)) /*Tags->access(node2tagpos(s))*/ == tag) return s; s = parent(Par, s); } return NULLT; @@ -836,20 +844,32 @@ int XMLTree::CloseDocument() Par = (bp *)umalloc(sizeof(bp)); bp_construct(Par, npar, par_aux, OPT_DEGREE|0); // creates structure for tags - static_bitsequence_builder * bmb = new static_bitsequence_builder_brw32(20); - static_permutation_builder * pmb = new static_permutation_builder_mrrr(PERM_SAMPLE, bmb); - static_sequence_builder * ssb = new static_sequence_builder_gmr_chunk(bmb, pmb); + //static_bitsequence_builder * bmb = new static_bitsequence_builder_brw32(20); + //static_permutation_builder * pmb = new static_permutation_builder_mrrr(PERM_SAMPLE, bmb); + //static_sequence_builder * ssb = new static_sequence_builder_gmr_chunk(bmb, pmb); // If we found an attribute then "<@>" is present in the tree // if we didn't then it is not. "<$>" is never present in the tree int ntagsize = found_attributes ? 2*ntagnames-1 : 2*ntagnames - 2; - Tags = new static_sequence_gmr((uint *) tags_aux, (uint) npar-1,ntagsize, bmb, ssb); + static_bitsequence_builder * bmb = new static_bitsequence_builder_brw32(20); + alphabet_mapper *am = new alphabet_mapper_none(); + wt_coder * wc = new wt_coder_huff((uint*)tags_aux,npar-1,am); + Tags = new static_sequence_wvtree((uint*)tags_aux,npar-1,wc ,bmb, am); + //Tags = new static_sequence_gmr((uint *) tags_aux, (uint) npar-1,ntagsize, bmb, ssb); + uint max_tag = 0; + for(uint i=0;i<(uint)npar-1;i++) + max_tag = max(max_tag,tags_aux[i]); + tags_blen = bits(max_tag); + tags_len = (uint)npar-1; + tags_fix = new uint[uint_len(tags_blen,tags_len)]; + for(uint i=0;i<(uint)npar-1;i++) + set_field(tags_fix,tags_blen,i,tags_aux[i]); delete bmb; - delete pmb; - delete ssb; + //delete pmb; + //delete ssb; // makes the text collection static if (!disable_tc) Text->MakeStatic();