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
+ /// FIXME:UGLY tests!\r
+ uint * seq = new uint[XML_Tree->tags_len];\r
+ for(uint i=0;i<XML_Tree->tags_len;i++)\r
+ seq[i] = get_field(XML_Tree->tags_fix,XML_Tree->tags_blen,i);\r
+ cout << "Tags test: " << XML_Tree->Tags->test(seq,XML_Tree->tags_len) << endl;\r
+ delete [] seq;\r
+ /// End ugly tests\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
+ XML_Tree->Text = TextCollection::Load(fp,sample_rate_text);\r
int sst;\r
int st;\r
ufread(&sst, sizeof(int),1,fp);\r
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
if (!is_ancestor(Par, x, y)) return false;\r
return depth(Par, x) == (depth(Par, y) + 1);\r
}\r
-\r
+bool XMLTree::IsFirstChild(treeNode x){\r
+ return ((x != NULLT)&&(x==Root() || prev_sibling(Par,x) == NULLT));\r
+}\r
// NumChildren(x): number of children of node x. Constant time with the data structure\r
// of Sadakane.\r
int XMLTree::NumChildren(treeNode x) \r
}\r
\r
range r;\r
+ if (x == NULLT)\r
+ {\r
+ r.min = NULLT;\r
+ r.max = NULLT;\r
+ return r;\r
+ };\r
+ \r
+ \r
if (indexing_empty_texts) { // faster, no rank needed\r
r.min = x;\r
r.max = x+2*subtree_size(Par, x)-2;\r
return first_child(Par, x);\r
}\r
\r
+treeNode XMLTree::LastChild(treeNode x) \r
+{\r
+ if (x == Root() || isleaf(Par,x) || x == NULLT)\r
+ return x;\r
+ else\r
+ return find_open(Par,find_close(Par,parent(Par,x))-1);\r
+}\r
+\r
// NextSibling(x): returns the next sibling of node x, assuming it exists.\r
treeNode XMLTree::NextSibling(treeNode x) \r
{\r
fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
exit(1);\r
}\r
- if (x == Root())\r
+ if (x == Root() || x==NULLT)\r
return NULLT;\r
- \r
+ \r
return next_sibling(Par, x);\r
}\r
\r
exit(1);\r
}\r
\r
- int r, s;\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
+ int s = (int) Tags->select_next(tag,node2tagpos(x));\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
if (!is_ancestor(Par, x, y)) return NULLT; // the next node tagged tag (in preorder) is not within the subtree of x.\r
else return y;\r
}\r
\r
+treeNode XMLTree::TaggedDescOnly(treeNode x,TagType *desctags, unsigned int dtlen)\r
+{\r
+\r
+ treeNode res,y;\r
+ if (isleaf(Par,x))\r
+ return NULLT;\r
+ \r
+ res=NULLT;\r
+ for (unsigned int i = 0; i < dtlen; i ++ )\r
+ {\r
+ y = TaggedDesc(x,desctags[i]);\r
+ res = (res == NULLT) || (( res != NULLT) && (y =! NULLT) && y < res) ? y : res;\r
+ \r
+ };\r
+ \r
+ return res;\r
+ \r
+}\r
+\r
+\r
+treeNode XMLTree::TaggedBelow(treeNode x, TagType *childtags, unsigned int ctlen,\r
+ TagType *desctags, unsigned int dtlen)\r
+{\r
+ treeNode fs,y,res;\r
+ TagType tag;\r
+\r
+ if (isleaf(Par,x))\r
+ return NULLT;\r
+ \r
+ res = NULLT;\r
+ fs = first_child(Par,x);\r
+ while (fs != NULLT) {\r
+ tag = get_field(tags_fix,tags_blen,node2tagpos(fs));\r
+ \r
+ /* Check for first_child */\r
+ for (unsigned int i = 0; i < ctlen; i++) {\r
+ if (childtags[i] == tag)\r
+ return fs;\r
+ };\r
+ \r
+ for (unsigned int i = 0; i < dtlen; i++)\r
+ if (desctags[i] == tag)\r
+ return fs; \r
+ \r
+ /* check in the descendants */\r
+ res = NULLT;\r
+ for (unsigned int i = 0; i < dtlen; i ++ ){\r
+ /* maybe inline by hand */\r
+ y = TaggedDesc(fs,desctags[i]);\r
+ res = (res==NULLT || (y != NULLT) &&(y < res)) ? y : res; \r
+ };\r
+ if (res != NULLT)\r
+ return res;\r
+ \r
+ fs = next_sibling(Par,fs);\r
+ };\r
+ return res;\r
+ \r
+}\r
+treeNode XMLTree::TaggedFollOnly(treeNode x,TagType *folltags, unsigned int ftlen,treeNode root)\r
+{\r
+\r
+ treeNode res,y,lim;\r
+ lim = find_close(Par,root); \r
+ res=NULLT;\r
+ for (unsigned int i = 0; i < ftlen; i ++ )\r
+ {\r
+ y = TaggedFoll(x,folltags[i]);\r
+ res = (res == NULLT) || (( res != NULLT) && (y =! NULLT) && y < res) ? y : res;\r
+ \r
+ };\r
+ \r
+ return res < lim ? res : NULLT;\r
+ \r
+}\r
+\r
+treeNode XMLTree::TaggedDescOrFollOnly(treeNode x,TagType *folltags, unsigned int ftlen,treeNode root)\r
+{\r
+\r
+ treeNode res,y,lim;\r
+ //int r,s;\r
+ lim = find_close(Par,root); \r
+ res=NULLT;\r
+ for (unsigned int i = 0; i < ftlen; i ++ )\r
+ {\r
+\r
+ int s = (int) Tags->select_next(folltags[i],node2tagpos(x));\r
+ /*r = (int) Tags->rank(folltags[i], node2tagpos(x));\r
+ s = (int) Tags->select(folltags[i], r+1);*/\r
+ if (s == -1) \r
+ y = NULLT; // there is no such node\r
+ else {\r
+ y = tagpos2node(s); \r
+ if (y >= lim)\r
+ y = NULLT;\r
+ };\r
+ res = (res == NULLT) || (( res != NULLT) && (y =! NULLT) && y < res) ? y : res;\r
+ \r
+ };\r
+ \r
+ return res < lim ? res : NULLT;\r
+ \r
+}\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
+treeNode XMLTree::TaggedNext(treeNode x, TagType *childtags, unsigned int ctlen,\r
+ TagType *folltags, unsigned int flen,treeNode root)\r
{\r
- if (!finished) {\r
- fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
- exit(1);\r
- }\r
+ treeNode y,old_y,lim,res;\r
+ TagType tag;\r
+ if (x == NULLT || x == Root())\r
+ return NULLT;\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
+ lim = find_close(Par,root); \r
+\r
+ res = NULLT;\r
+ \r
+ y = next_sibling(Par,x);\r
+ while (y != NULLT) {\r
+ tag = get_field(tags_fix,tags_blen,node2tagpos(y));\r
+ for(unsigned int i = 0; i < ctlen;i++)\r
+ if (childtags[i] == tag)\r
+ return y;\r
+ \r
+ for(unsigned int i = 0; i < flen;i++)\r
+ if (folltags[i] == tag)\r
+ return y;\r
+\r
+ res = TaggedBelow(y,NULL,0,folltags,flen);\r
+ if (res != NULLT)\r
+ return res;\r
+ \r
+ y = next_sibling(Par,y);\r
+ };\r
+ //Found nothing in the following sibling of x.\r
+ res = NULLT;\r
+ for(unsigned int i = 0; i < flen;i++){\r
+ y = TaggedFoll(x,folltags[i]);\r
+ res = (y!= x && (res == NULLT || (y != NULLT && y < res)))? y : res;\r
+ };\r
+ \r
+ return res < lim ? res : NULLT;\r
+ \r
}\r
\r
\r
exit(1);\r
}\r
\r
- int r, s;\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
+ \r
+ int s = (int) Tags->select_next(tag,find_close(Par,x));\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
+// 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::TaggedFollBelow(treeNode x, TagType tag, treeNode root)\r
+ {\r
+\r
+ //int r, s;\r
+ int lim = node2tagpos(find_close(Par,root));\r
+ if (x ==NULLT || x == Root())\r
+ return NULLT;\r
+ \r
+ int s = (int) Tags->select_next(tag,find_close(Par,x));\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 || s >= lim) \r
+ return NULLT;\r
+ else \r
+ 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
exit(1);\r
}\r
\r
- int r, s;\r
+ //int r, s;\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(ns)-1);\r
- s = (int) Tags->select(tag, r+1); // select returns -1 in case that there is no r+1-th tag.\r
+ int s = (int) Tags->select_next(tag,node2tagpos(ns)-1);\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
return (treeNode)s;\r
\r
}\r
+treeNode XMLTree::PrevNode(DocID d) \r
+ {\r
+ if (!finished) {\r
+ fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
+ exit(1);\r
+ }\r
+ \r
+ if (d == NULLT)\r
+ return NULLT;\r
+ \r
+ int s;\r
+ \r
+ if (indexing_empty_texts) s = d;\r
+ else s = EBVector->select1(d+1);\r
+ if (s == -1)\r
+ return NULLT;\r
+ \r
+ if (inspect(Par,s) == CP) // is a closing parenthesis\r
+ return find_open(Par, s);\r
+ else // is an opening parenthesis\r
+ return NULLT;\r
+ \r
+ }\r
\r
\r
// OpenDocument(empty_texts): it starts the construction of the data structure for\r
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
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
-\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
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
- int ntagsize = found_attributes ? 2*ntagnames-1 : 2*ntagnames - 2;\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_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_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
- //Tags = new static_sequence_gmr((uint *) tags_aux, (uint) npar-1,ntagsize, bmb, ssb);\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
tags_blen = bits(max_tag);\r
tags_len = (uint)npar;\r
tags_fix = new uint[uint_len(tags_blen,tags_len)];\r
- for(uint i=0;i<(uint)npar-1;i++)\r
+ for(uint i=0;i<(uint)npar;i++)\r
set_field(tags_fix,tags_blen,i,tags_aux[i]);\r
\r
delete bmb;\r
//delete pmb;\r
//delete ssb;\r
+\r
+ \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
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
\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
// 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
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
\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
projects / SXSI / XMLTree.git / blobdiff