Ugly patch

git-svn-id: svn+ssh://idea.nguyen.vg/svn/sxsi/trunk/XMLTree@203 3cdefd35-fc62-479d-8e8d-bae585ffb9ca

diff --git a/XMLTree.cpp b/XMLTree.cpp
index 62ca3fd..e17c89d 100644 (file)
--- a/XMLTree.cpp
+++ b/XMLTree.cpp
@@ -101,6 +101,9 @@ void XMLTree::Save(unsigned char *filename)
     \r
     // stores the tags\r
     Tags->save(fp);\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
 \r
     // stores the texts   \r
     if (!disable_tc) {\r


@@ -191,6 +194,11 @@ XMLTree *XMLTree::Load(unsigned char *filename, int sample_rate_text)
     \r
     // loads the tags\r
     XML_Tree->Tags = static_sequence::load(fp);\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
     s_tree+= XML_Tree->Tags->size();\r
 \r
     s_text = ftell(fp);\r


@@ -223,7 +231,7 @@ XMLTree *XMLTree::Load(unsigned char *filename, int sample_rate_text)
     fclose(fp);\r
 \r
     std::cerr << "Tree part is " << s_tree/1024 << " Kbytes,\n"\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
+             << "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
              << "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


@@ -395,7 +403,7 @@ TagType XMLTree::Tag(treeNode x)
        exit(1);\r
     }\r
     \r
        exit(1);\r
     }\r
     \r

-    return Tags->access(node2tagpos(x));\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
  }\r
 \r
 // DocIds(x): returns the range of text identifiers that descend from node x.\r


@@ -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\r
     if (child==(treeNode)-1) return NULLT; // node x is a leaf, there is no such child\r
     while (child!=(treeNode)-1) {\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
           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


@@ -637,7 +645,7 @@ treeNode XMLTree::TaggedAncestor(treeNode x, TagType tag)
     \r
     treeNode s = parent(Par, x), r = Root();\r
     while (s != r) {\r
     \r
     treeNode s = parent(Par, x), r = Root();\r
     while (s != r) {\r

-       if (Tags->access(node2tagpos(s)) == tag) return s;\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
        s = parent(Par, s);\r
     }\r
     return NULLT;\r


@@ -836,20 +844,32 @@ int XMLTree::CloseDocument()
     Par = (bp *)umalloc(sizeof(bp));\r
     bp_construct(Par, npar, par_aux, OPT_DEGREE|0);    \r
     // creates structure for tags\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. "<$>" is never present in the tree\r
     int ntagsize = found_attributes ? 2*ntagnames-1 : 2*ntagnames - 2;\r
 \r
 \r
 \r
     // If we found an attribute then "<@>" is present in the tree\r
     // if we didn't then it is not. "<$>" 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
     \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
     // makes the text collection static\r
     if (!disable_tc)\r
       Text->MakeStatic();\r

diff --git a/XMLTree.h b/XMLTree.h
index b6aa410..3bfb7a4 100644 (file)
--- a/XMLTree.h
+++ b/XMLTree.h
@@ -84,6 +84,8 @@ class XMLTree {
                      \r
    /** Tag sequence represented with a data structure for rank and select */\r
    static_sequence *Tags;\r
                      \r
    /** Tag sequence represented with a data structure for rank and select */\r
    static_sequence *Tags;\r

+        uint * tags_fix;\r
+        uint tags_blen, tags_len;\r

 \r
    /** The texts in the XML document */\r
    TextCollection *Text;\r
 \r
    /** The texts in the XML document */\r
    TextCollection *Text;\r