#include "XMLTree.h"\r
+#include "basics.h"\r
#include <cstring>\r
\r
// functions to convert tag positions to the corresponding tree node and viceversa. \r
// the tree, and storing 2 tags per tree node (opening and closing tags).\r
\r
// tag position -> tree node\r
-inline treeNode tagpos2node(int t) {\r
- return (treeNode)t;\r
-}\r
+inline treeNode tagpos2node(int t) \r
+ {\r
+ return (treeNode)t;\r
+ }\r
\r
// tree node -> tag position\r
-inline int node2tagpos(treeNode x) {\r
- return (int)x;\r
-}\r
+inline int node2tagpos(treeNode x) \r
+ {\r
+ return (int)x;\r
+ }\r
\r
\r
-//KIM OJO to prevent suprious "unused result" warnings\r
+XMLTree::XMLTree(pb *par, uint npar, unsigned char **TN, uint ntagnm, uint *empty_texts_bmp, TagType *tags,\r
+ TextCollection *TC, vector<string> CT, bool indexing_empty_t, bool dis_tc)\r
+ {\r
+ // creates the data structure for the tree topology\r
+ Par = (bp *)umalloc(sizeof(bp));\r
+ bp_construct(Par, npar, par, OPT_DEGREE|0); \r
\r
-inline void ufread(void *ptr, size_t size, size_t nmemb, FILE *stream){\r
- size_t res;\r
- res = fread(ptr,size,nmemb,stream);\r
- if (res < nmemb)\r
- throw "ufread I/O error";\r
\r
- return;\r
-}\r
+ // creates structure for tags\r
\r
-inline void ufwrite(const void *ptr, size_t size, size_t nmemb, FILE *stream){\r
- size_t res;\r
- res = fwrite(ptr,size,nmemb,stream);\r
- if (res < nmemb)\r
- throw "ufwrite I/O error";\r
- return;\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
+ TagName = TN;\r
+ ntagnames = ntagnm;\r
+ uint max_tag = 0;\r
+ for(uint i=0;i<(uint)npar-1;i++)\r
+ max_tag = max(max_tag,tags[i]);\r
+ int ntagsize = 2*ntagnames + 2;\r
\r
-// OJO to fail cleanly while doing a realloc\r
-// if we can't realloc we are pretty much screwed anyway but\r
-// it makes the code clearer to not have a bunch of if (!ptr) { printf("..."); exit(1); };\r
-inline void * urealloc(void *ptr, size_t size){\r
+ static_bitsequence_builder *bmb = new static_bitsequence_builder_sdarray();\r
+ alphabet_mapper *am = new alphabet_mapper_none();\r
+ Tags = new static_sequence_bs((uint*)tags,npar,am,bmb);\r
+ \r
+ cout << "Tags test: " << Tags->test((uint*)tags,npar) << endl;\r
\r
- void * dest = realloc(ptr,size);\r
- //don't fail if we requested size 0\r
- if (dest == NULL && size > 0 )\r
- throw std::bad_alloc();\r
- return dest;\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;i++)\r
+ set_field(tags_fix,tags_blen,i,tags[i]);\r
\r
-}\r
+ delete bmb; \r
+ free(tags);\r
+ tags = NULL;\r
+ \r
+ Text = TC;\r
\r
-inline void * ucalloc(size_t nmemb, size_t size){\r
+ CachedText = CT;\r
\r
- void * dest = calloc(nmemb,size);\r
- //don't fail if we requested size 0\r
- if (dest == NULL && nmemb > 0 && size > 0 )\r
- throw std::bad_alloc();\r
- return dest;\r
+ // creates the data structure marking the non-empty texts (just in the case it is necessary)\r
+ indexing_empty_texts = indexing_empty_t;\r
+ if (!indexing_empty_t) {\r
+ EBVector = new static_bitsequence_rrr02((uint *)empty_texts_bmp,(ulong)npar,(uint)32);\r
+ free(empty_texts_bmp);\r
+ empty_texts_bmp = NULL;\r
+ }\r
\r
-}\r
+ TagArray = new TagArrayEntry[ntagnames];\r
+ for (uint i=0; i<ntagnames; i++) {\r
+ TagArray[i].first = NULL;\r
+ TagArray[i].last = NULL;\r
+ }\r
+ \r
+ disable_tc = dis_tc;\r
+ }\r
\r
-inline void * umalloc(size_t size){\r
- void * dest = malloc(size);\r
- if (dest == NULL && size > 0)\r
- throw std::bad_alloc();\r
- 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
+// ~XMLTree: frees memory of XML tree.\r
+XMLTree::~XMLTree() \r
+ { \r
+ int i;\r
+\r
+ destroyTree(Par);\r
+ free(Par); // frees the memory of struct Par\r
+ \r
+ for (i=0; i<ntagnames;i++) \r
+ free(TagName[i]);\r
+ \r
+ free(TagName);\r
+\r
+ if (!indexing_empty_texts) {\r
+ delete EBVector;\r
+ EBVector = NULL;\r
+ }\r
+\r
+ delete Tags;\r
+ Tags = NULL;\r
+\r
+ delete Text; \r
+ Text = NULL;\r
+\r
+ delete TagArray;\r
+ }\r
+\r
+\r
+void XMLTree::print_stats() \r
+ {\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
FILE *fp;\r
char filenameaux[1024];\r
int i;\r
\r
// stores the flags\r
ufwrite(&indexing_empty_texts, sizeof(bool), 1, fp);\r
- ufwrite(&initialized, sizeof(bool), 1, fp);\r
- ufwrite(&finished, sizeof(bool), 1, fp);\r
+ bool ignore = true;\r
+ ufwrite(&ignore, sizeof(bool),1,fp);\r
+ ufwrite(&ignore, sizeof(bool),1,fp);\r
ufwrite(&disable_tc, sizeof(bool),1,fp);\r
\r
if (!indexing_empty_texts) EBVector->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
+ 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
// a pointer to the loaded data structure\r
XMLTree *XMLTree::Load(unsigned char *filename, int sample_rate_text) \r
{\r
-\r
FILE *fp;\r
char buffer[1024];\r
XMLTree *XML_Tree;\r
\r
XML_Tree = new XMLTree();\r
\r
+\r
XML_Tree->Par = (bp *)umalloc(sizeof(bp));\r
\r
loadTree(XML_Tree->Par, fp); \r
\r
s_tree += sizeof(bp);\r
\r
+\r
// stores the table with tag names\r
ufread(&XML_Tree->ntagnames, sizeof(int), 1, fp);\r
- \r
+\r
s_tree += sizeof(int);\r
\r
XML_Tree->TagName = (unsigned char **)umalloc(XML_Tree->ntagnames*sizeof(unsigned char *));\r
- \r
+\r
s_tags += sizeof(unsigned char*)*XML_Tree->ntagnames;\r
\r
\r
for (i=0; i<XML_Tree->ntagnames;i++) {\r
- \r
- // OJO Kim is it needed ?\r
- int k = feof(fp);\r
-\r
- \r
- // fscanf chokes on "\n" which is the case for the root element\r
char * r = fgets(buffer,1023,fp);\r
- // int r = fscanf(fp, "%s\n",buffer);\r
if (r==NULL)\r
throw "Cannot read tag list";\r
\r
strncpy((char *)XML_Tree->TagName[i], (const char *)buffer,len - 1);\r
s_tags+= len*sizeof(char);\r
}\r
+\r
\r
// loads the flags\r
\r
ufread(&(XML_Tree->indexing_empty_texts), sizeof(bool), 1, fp);\r
- ufread(&(XML_Tree->initialized), sizeof(bool), 1, fp);\r
- ufread(&(XML_Tree->finished), sizeof(bool), 1, fp);\r
+ bool ignore;\r
+ ufread(&ignore, sizeof(bool), 1, fp);\r
+ ufread(&ignore, sizeof(bool), 1, fp);\r
ufread(&(XML_Tree->disable_tc), sizeof(bool), 1, fp);\r
- \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
+\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
+ 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
- /// 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
+ /// 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
};\r
\r
}\r
- else {\r
- XML_Tree->Text = NULL;\r
- }\r
- s_text = ftell(fp) - s_text;\r
-\r
- \r
+ else XML_Tree->Text = NULL;\r
\r
+ s_text = ftell(fp) - s_text;\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+(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
- XML_Tree->print_stats();\r
+ XML_Tree->print_stats();\r
return XML_Tree;\r
}\r
\r
\r
-// ~XMLTree: frees memory of XML tree.\r
-XMLTree::~XMLTree() \r
- { \r
- int i;\r
-\r
- destroyTree(Par);\r
- free(Par); // frees the memory of struct Par\r
- \r
- for (i=0; i<ntagnames;i++) \r
- free(TagName[i]);\r
- \r
- free(TagName);\r
-\r
- if (!indexing_empty_texts) {\r
- //EBVector->~static_bitsequence_rrr02();\r
- delete EBVector;\r
- EBVector = NULL;\r
- }\r
-\r
- //Tags->~static_sequence_wvtree();\r
- delete Tags;\r
- Tags = NULL;\r
-\r
- //Text->~TextCollection();\r
- delete TextBuilder; \r
- TextBuilder = NULL;\r
- delete Text; \r
- Text = NULL;\r
-\r
- initialized = false;\r
- finished = false;\r
- }\r
-\r
// root(): returns the tree root.\r
treeNode XMLTree::Root() \r
{\r
- if (!finished) {\r
- fprintf(stderr, "Root() : Error: data structure has not been constructed properly\n");\r
- exit(1);\r
- }\r
return root_node(Par);\r
}\r
\r
// SubtreeSize(x): the number of nodes (and attributes) in the subtree of node x.\r
int XMLTree::SubtreeSize(treeNode x) \r
{\r
- if (!finished) {\r
- fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
- exit(1);\r
- }\r
return subtree_size(Par, x);\r
}\r
\r
// SubtreeTags(x,tag): the number of occurrences of tag within the subtree of node x.\r
int XMLTree::SubtreeTags(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
if (x == Root())\r
x = first_child(Par,x);\r
\r
// IsAncestor(x,y): returns whether node x is ancestor of node y.\r
bool XMLTree::IsAncestor(treeNode x, treeNode y) \r
{\r
- if (!finished) {\r
- fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
- exit(1);\r
- }\r
-\r
return is_ancestor(Par, x, y);\r
}\r
\r
// IsChild(x,y): returns whether node x is parent of node y.\r
bool XMLTree::IsChild(treeNode x, treeNode y) \r
{\r
- if (!finished) {\r
- fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
- exit(1);\r
- }\r
-\r
if (!is_ancestor(Par, x, y)) return false;\r
return depth(Par, x) == (depth(Par, y) + 1);\r
}\r
-bool XMLTree::IsFirstChild(treeNode x){\r
- return ((x != NULLT)&&(x==Root() || prev_sibling(Par,x) == NULLT));\r
-}\r
+\r
+// IsFirstChild(x): returns whether node x is the first child of its parent.\r
+bool XMLTree::IsFirstChild(treeNode x)\r
+ {\r
+ return ((x != NULLT)&&(x==Root() || prev_sibling(Par,x) == (treeNode)-1));\r
+ }\r
+\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
- if (!finished) {\r
- fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
- exit(1);\r
- }\r
-\r
return degree(Par, x);\r
}\r
\r
// ChildNumber(x): returns i if node x is the i-th children of its parent.\r
int XMLTree::ChildNumber(treeNode x) \r
{\r
- if (!finished) {\r
- fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
- exit(1);\r
- }\r
-\r
return child_rank(Par, x);\r
}\r
\r
// Depth(x): depth of node x, a simple binary rank on the parentheses sequence.\r
int XMLTree::Depth(treeNode x) \r
{\r
- if (!finished) {\r
- fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
- exit(1);\r
- }\r
-\r
return depth(Par, x);\r
}\r
\r
// nodes (i.e., tags, it disregards the texts in the tree).\r
int XMLTree::Preorder(treeNode x) \r
{\r
- if (!finished) {\r
- fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
- exit(1);\r
- }\r
-\r
return preorder_rank(Par, x);\r
}\r
\r
// nodes (i.e., tags, it disregards the texts in the tree).\r
int XMLTree::Postorder(treeNode x) \r
{\r
- if (!finished) {\r
- fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
- exit(1);\r
- }\r
return postorder_rank(Par, x);\r
}\r
\r
// Tag(x): returns the tag identifier of node x.\r
TagType XMLTree::Tag(treeNode x) \r
{\r
- if (!finished) {\r
- fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
- exit(1);\r
- }\r
- \r
return get_field(tags_fix,tags_blen,node2tagpos(x)); //Tags->access(node2tagpos(x));\r
}\r
\r
// returns {NULLT, NULLT} when there are no texts descending from x.\r
range XMLTree::DocIds(treeNode x) \r
{\r
- if (!finished) {\r
- fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
- exit(1);\r
- }\r
-\r
range r;\r
- if (x == NULLT)\r
- {\r
- r.min = NULLT;\r
- r.max = NULLT;\r
- return r;\r
- };\r
+ if (x == NULLT) {\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
// Parent(x): returns the parent node of node x.\r
treeNode XMLTree::Parent(treeNode x) \r
{\r
- if (!finished) {\r
- fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
- exit(1);\r
- }\r
if (x == Root())\r
return NULLT;\r
else\r
// Child(x,i): returns the i-th child of node x, assuming it exists.\r
treeNode XMLTree::Child(treeNode x, int i) \r
{\r
- if (!finished) {\r
- fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
- exit(1);\r
- }\r
-\r
if (i <= OPTD) return naive_child(Par, x, i);\r
else return child(Par, x, i);\r
}\r
// FirstChild(x): returns the first child of node x, assuming it exists. Very fast in BP.\r
treeNode XMLTree::FirstChild(treeNode x) \r
{\r
- if (!finished) {\r
- fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
- exit(1);\r
- }\r
-\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
+// LastChild(x): returns the last child of node x.\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
+ return find_open(Par, find_close(Par, x)-1);\r
+ }\r
+\r
\r
// NextSibling(x): returns the next sibling of node x, assuming it exists.\r
treeNode XMLTree::NextSibling(treeNode x) \r
{\r
- if (!finished) {\r
- fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
- exit(1);\r
- }\r
if (x == Root() || x==NULLT)\r
return NULLT;\r
\r
// PrevSibling(x): returns the previous sibling of node x, assuming it exists.\r
treeNode XMLTree::PrevSibling(treeNode x) \r
{\r
- if (!finished) {\r
- fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
- exit(1);\r
- }\r
-\r
return prev_sibling(Par, x);\r
}\r
\r
-// TaggedChild(x,i,tag): returns the i-th child of node x tagged tag, or NULLT if there is none.\r
+// TaggedChild(x,tag): returns the first child of node x tagged tag, or NULLT if there is none.\r
// Because of the balanced-parentheses representation of the tree, this operation is not supported\r
// efficiently, just iterating among the children of node x until finding the desired child.\r
-treeNode XMLTree::TaggedChild(treeNode x, int i, TagType tag) \r
+treeNode XMLTree::TaggedChild(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
treeNode child;\r
\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 (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
- child = next_sibling(Par, x); // OK, let's try with the next child\r
+ if (get_field(tags_fix,tags_blen,node2tagpos(child)) == tag) // current child is labeled with tag of interest\r
+ return child; \r
+ child = next_sibling(Par, child); // OK, let's try with the next child\r
}\r
return NULLT; // no such child was found \r
}\r
\r
+\r
+treeNode XMLTree::SelectChild(treeNode x, TagType *tags, int ntags)\r
+ {\r
+ int i;\r
+ treeNode child = first_child(Par, x);\r
+\r
+ while (child!=(treeNode)-1) {\r
+ TagType t = get_field(tags_fix, tags_blen, node2tagpos(child));\r
+ for (i=0; i<ntags; i++)\r
+ if (t==tags[i]) return child;\r
+ child = next_sibling(Par, child);\r
+ }\r
+ return NULLT;\r
+ }\r
+\r
+\r
+// TaggedSibling(x,tag): returns the first sibling of node x tagged tag, or NULLT if there is none.\r
+treeNode XMLTree::TaggedSibling(treeNode x, TagType tag)\r
+ {\r
+ treeNode sibling = next_sibling(Par, x);\r
+ \r
+ while (sibling!=(treeNode)-1) {\r
+ if (get_field(tags_fix,tags_blen,node2tagpos(sibling)) == tag) // current sibling is labeled with tag of interest\r
+ return sibling; \r
+ sibling = next_sibling(Par, sibling); // OK, let's try with the next sibling\r
+ }\r
+ return NULLT; // no such sibling was found \r
+ }\r
+\r
+\r
+treeNode XMLTree::SelectSibling(treeNode x, TagType *tags, int ntags)\r
+ {\r
+ int i;\r
+ treeNode sibling = next_sibling(Par, x);\r
+\r
+ while (sibling!=(treeNode)-1) {\r
+ TagType t = get_field(tags_fix, tags_blen, node2tagpos(sibling));\r
+ for (i=0; i<ntags; i++)\r
+ if (t==tags[i]) return sibling;\r
+ sibling = next_sibling(Par, sibling);\r
+ }\r
+ return NULLT; \r
+ }\r
+\r
+\r
// TaggedDesc(x,tag): returns the first node tagged tag with larger preorder than x and within\r
// the subtree of x. Returns NULLT if there is none.\r
treeNode XMLTree::TaggedDesc(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 (isleaf(Par,x))\r
return NULLT;\r
\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
+ int s = (int) Tags->select_next(tag,node2tagpos(x));\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
+treeNode XMLTree::SelectDesc(treeNode x, TagType *tags, int ntags)\r
+ {\r
+ int i;\r
+ treeNode min = NULLT;\r
+ treeNode fc = first_child(Par,x);\r
+ \r
+ for (i=0; i<ntags; i++) {\r
+ treeNode aux = TaggedDesc(x, tags[i]);\r
+ if (aux == fc) \r
+ return fc;\r
+ else \r
+ if ((min == (treeNode)NULLT) || (aux < min)) min = aux;\r
+ }\r
+ return min;\r
+ }\r
+\r
+\r
+treeNode XMLTree::TaggedDescOnly(treeNode x,TagType *desctags, unsigned int dtlen)\r
+ {\r
+ treeNode res, y;\r
+ if (isleaf(Par,x))\r
+ return NULLT;\r
\r
- return res;\r
+ res=NULLT;\r
+ for (unsigned int i = 0; i < dtlen; i ++ ) {\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
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
+ TagType *desctags, unsigned int dtlen) \r
+ {\r
+ treeNode fs,y,res;\r
+ TagType tag;\r
\r
- if (isleaf(Par,x))\r
- return NULLT;\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
+ 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
+ /* 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
+ 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
+ /* 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
+ fs = next_sibling(Par,fs);\r
+ }\r
\r
+ return res;\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
+treeNode XMLTree::TaggedFollOnly(treeNode x,TagType *folltags, unsigned int ftlen,treeNode root)\r
+ {\r
+ treeNode res,y,lim;\r
+ lim = find_close(Par,root); \r
+ res=NULLT;\r
+ \r
+ for (unsigned int i = 0; i < ftlen; i ++ ) {\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
-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
+treeNode XMLTree::TaggedDescOrFollOnly(treeNode x,TagType *folltags, unsigned int ftlen,treeNode root)\r
+ {\r
+ treeNode res,y,lim;\r
+ lim = find_close(Par,root); \r
+ res=NULLT;\r
\r
+ for (unsigned int i = 0; i < ftlen; i++) {\r
+ int s = (int) Tags->select_next(folltags[i],node2tagpos(x));\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
+ return res < lim ? res : NULLT;\r
}\r
\r
\r
if (x == NULLT || x == Root())\r
return NULLT;\r
\r
-\r
lim = find_close(Par,root); \r
\r
res = NULLT;\r
res = (y!= x && (res == NULLT || (y != NULLT && y < res)))? y : res;\r
};\r
\r
- return res < lim ? res : NULLT;\r
- \r
+ return res < lim ? res : NULLT; \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
- if (!finished) {\r
- fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
- exit(1);\r
- }\r
- \r
+ { \r
int r, s;\r
treeNode node_s, root;\r
r = (int)Tags->rank(tag, node2tagpos(x)-1);\r
// the subtree of x. Returns NULLT if there is none.\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
- 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
+ \r
+ int s = (int) Tags->select_next(tag,find_close(Par, x)); \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
+// TaggedFollBelow(x,tag,root): returns the first node tagged tag with larger preorder than x \r
+// and not in the subtree of x. Returns NULLT if there is none.\r
treeNode XMLTree::TaggedFollBelow(treeNode x, TagType tag, treeNode root)\r
{\r
-\r
- if (x == NULLT || x == Root()) return NULLT;\r
- treeNode s = (treeNode) Tags->select_next(tag,find_close(Par,x));\r
- /*int r = (int) Tags->rank(tag, find_close(Par, x));\r
- int s = (int) Tags->select(tag, r+1); */\r
- if (root == Root())\r
- return s;\r
- \r
- if (s == NULLT || s >= find_close(Par,root)) return NULLT;\r
- return s;\r
+ if (x == NULLT || x == Root())\r
+ return NULLT;\r
+ \r
+ treeNode s = (treeNode) Tags->select_next(tag, find_close(Par, x));\r
+ \r
+ if (root == Root()) \r
+ return s;\r
+ if (s == NULLT || s >= find_close(Par, root)) return NULLT;\r
+ return s;\r
} \r
\r
\r
+treeNode XMLTree::SelectFollBelow(treeNode x, TagType *tags, int ntags, treeNode ctx)\r
+ {\r
+ int i;\r
+ treeNode min = NULLT;\r
+ treeNode fc = first_child(Par, x);\r
+ \r
+ for (i=0; i<ntags; i++) {\r
+ treeNode aux = TaggedFollBelow(x, tags[i], ctx);\r
+ if (aux == fc)\r
+ return fc;\r
+ else\r
+ if ((min == NULLT) || (aux < min)) min = aux;\r
+ }\r
+ return min; \r
+ }\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
- exit(1);\r
- }\r
-\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
- 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
+ int s = (int) Tags->select_next(tag, node2tagpos(ns)-1);\r
if (s==-1) return NULLT;\r
else return tagpos2node(s);\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
+ { \r
if (x == NULLT || x == Root())\r
return NULLT;\r
\r
// Assumes Doc ids start from 0.\r
DocID XMLTree::PrevText(treeNode x) \r
{\r
- if (!finished) {\r
- fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
- exit(1);\r
- }\r
-\r
if (x == Root()) return NULLT;\r
if (indexing_empty_texts) // faster, no rank needed\r
return (DocID)x-1;\r
// of node x, or NULLT if x is the root node. Assumes Doc ids start from 0.\r
DocID XMLTree::NextText(treeNode x) \r
{\r
- if (!finished) {\r
- fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
- exit(1);\r
- }\r
-\r
if (x == Root()) return NULLT;\r
if (indexing_empty_texts) // faster, no rank needed\r
return (DocID)x+2*subtree_size(Par, x)-1;\r
// ids start from 0.\r
DocID XMLTree::MyText(treeNode x) \r
{\r
- if (!finished) {\r
- fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
- exit(1);\r
- }\r
-\r
if (!IsLeaf(x)) return NULLT;\r
if (indexing_empty_texts) // faster, no rank needed\r
return (DocID)x;\r
// all tree nodes and all text nodes. Assumes that the tree root has preorder 1.\r
int XMLTree::TextXMLId(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 (indexing_empty_texts) \r
return d + rank_open(Par, d)+1; // +1 because root has preorder 1\r
else { // slower, needs rank and select\r
// preorder 0;\r
int XMLTree::NodeXMLId(treeNode x) \r
{\r
- if (!finished) {\r
- fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
- exit(1);\r
- }\r
-\r
if (indexing_empty_texts)\r
return x - 1 + rank_open(Par, x);\r
else {\r
\r
// ParentNode(d): returns the parent node of document identifier d.\r
treeNode XMLTree::ParentNode(DocID d) \r
- {\r
- if (!finished) {\r
- fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
- exit(1);\r
- }\r
- \r
+ { \r
if (d == NULLT)\r
return NULLT;\r
\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
+ { \r
if (d == NULLT)\r
return NULLT;\r
\r
}\r
\r
\r
-// OpenDocument(empty_texts): it starts the construction of the data structure for\r
-// the XML document. Parameter empty_texts indicates whether we index empty texts\r
-// in document or not. Returns a non-zero value upon success, NULLT in case of error.\r
-int XMLTree::OpenDocument(bool empty_texts, int sample_rate_text,bool dtc)\r
- {\r
- initialized = true;\r
- finished = false;\r
- found_attributes = false;\r
- npar = 0;\r
- parArraySize = 1;\r
- ntagnames = 4; \r
- disable_tc = dtc;\r
- \r
- indexing_empty_texts = empty_texts;\r
- \r
- par_aux = (pb *)umalloc(sizeof(pb)*parArraySize);\r
- \r
- tags_aux = (TagType *) umalloc(sizeof(TagType));\r
- \r
- TagName = (unsigned char **) umalloc(4*sizeof(unsigned char*));\r
-\r
- TagName[0] = (unsigned char *) umalloc(4*sizeof(unsigned char));\r
-\r
- strcpy((char *) TagName[0], "<@>");\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], "/<$>");\r
-\r
-\r
- if (!indexing_empty_texts) \r
- empty_texts_aux = (unsigned int *)umalloc(sizeof(unsigned int));\r
- \r
- if (disable_tc)\r
- TextBuilder = 0;\r
- else \r
- TextBuilder = new TextCollectionBuilder((unsigned)sample_rate_text);\r
- Text = 0;\r
- \r
- return 1; // indicates success in the initialization of the data structure\r
- }\r
-\r
-// CloseDocument(): it finishes the construction of the data structure for the XML\r
-// document. Tree and tags are represented in the final form, dynamic data \r
-// structures are made static, and the flag "finished" is set to true. After that, \r
-// the data structure can be queried.\r
-int XMLTree::CloseDocument()\r
- {\r
- if (!initialized) { // data structure has not been initialized properly\r
- fprintf(stderr, "Error: data structure has not been initialized properly (by calling method OpenDocument)\n");\r
- return NULLT;\r
- }\r
- \r
- // closing parenthesis for the tree root\r
- par_aux = (pb *)urealloc(par_aux, sizeof(pb)*(1+npar/(8*sizeof(pb))));\r
- \r
- // creates the data structure for the tree topology\r
- Par = (bp *)umalloc(sizeof(bp));\r
- bp_construct(Par, npar, par_aux, OPT_DEGREE|0); \r
- // creates structure for tags\r
-\r
- // If we found an attribute then "<@>" is present in the tree\r
- // if we didn't then it is not. "<$>" 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
- //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_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,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;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
- {\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
- free (empty_texts_aux);\r
- empty_texts_aux = NULL;\r
- }\r
- \r
- // OJO was leaked before, found by valgrind\r
- free(tags_aux);\r
-\r
- tags_aux = NULL;\r
-\r
- finished = true;\r
- print_stats();\r
-\r
- return 1; // indicates success in the inicialization\r
- }\r
-\r
-\r
-// NewOpenTag(tagname): indicates the event of finding a new opening tag in the document.\r
-// Tag name is given. Returns a non-zero value upon success, and returns NULLT\r
-// in case of failing when trying to insert the new tag.\r
-int XMLTree::NewOpenTag(unsigned char *tagname)\r
- {\r
- int i;\r
-\r
- if (!initialized) { // data structure has not been initialized properly\r
- fprintf(stderr, "Error: you cannot insert a new opening tag without first calling method OpenDocument first\n");\r
- return NULLT;\r
- }\r
- \r
- // inserts a new opening parentheses in the bit sequence\r
- if (sizeof(pb)*8*parArraySize == npar) { // no space left for the new parenthesis\r
- par_aux = (pb *)urealloc(par_aux, sizeof(pb)*2*parArraySize);\r
- parArraySize *= 2;\r
- }\r
- \r
- setbit(par_aux,npar,OP); // marks a new opening 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
- \r
-\r
- // NewOpenTag("<@>") was called\r
- if (i==0) \r
- found_attributes=true;\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
- if (!TagName) {\r
- fprintf(stderr, "Error: not enough memory\n");\r
- return NULLT;\r
- }\r
- \r
- ntagnames++;\r
- TagName[i] = (unsigned char *)umalloc(sizeof(unsigned char)*(strlen((const char *)tagname)+1));\r
- strcpy((char *)TagName[i], (const char *)tagname);\r
- } \r
- tags_aux = (TagType *) urealloc(tags_aux, sizeof(TagType)*(npar + 1));\r
-\r
- tags_aux[npar] = i; // inserts the new tag id within the preorder sequence of tags\r
- \r
- npar++;\r
- \r
- return 1;\r
- \r
- }\r
-\r
-\r
-// NewClosingTag(tagname): indicates the event of finding a new closing tag in the document.\r
-// Tag name is given. Returns a non-zero value upon success, and returns NULLT\r
-// in case of failing when trying to insert the new tag.\r
-int XMLTree::NewClosingTag(unsigned char *tagname)\r
- {\r
- int i;\r
-\r
- if (!initialized) { // data structure has not been initialized properly\r
- fprintf(stderr, "Error: you cannot insert a new closing tag without first calling method OpenDocument first\n");\r
- return NULLT;\r
- }\r
- \r
- // inserts a new closing parentheses in the bit sequence\r
- if (sizeof(pb)*8*parArraySize == npar) { // no space left for the new parenthesis\r
- par_aux = (pb *)urealloc(par_aux, sizeof(pb)*2*parArraySize);\r
- parArraySize *= 2;\r
- }\r
- \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]+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
- ntagnames++;\r
- TagName[i] = (unsigned char *)umalloc(sizeof(char)*(strlen((const char *)tagname)+2));\r
- TagName[i][0] = '/';\r
- strcpy((char *)&(TagName[i][1]), (const char *)tagname);\r
- } \r
-\r
- tags_aux = (TagType *)urealloc(tags_aux, sizeof(TagType)*(npar + 1));\r
-\r
- tags_aux[npar] = i; // inserts the new tag id within the preorder sequence of tags\r
- \r
- npar++;\r
-\r
- return 1; // success\r
- \r
- }\r
-\r
-\r
-// NewText(s): indicates the event of finding a new (non-empty) text s in the document.\r
-// The new text is inserted within the text collection. Returns a non-zero value upon\r
-// success, NULLT in case of error.\r
-int XMLTree::NewText(unsigned char *s)\r
- {\r
- if (!initialized) { // data structure has not been initialized properly\r
- fprintf(stderr, "Error: you cannot insert a new text without first calling method OpenDocument first\n");\r
- return NULLT;\r
- }\r
-\r
- if (disable_tc) {\r
- XMLTree::NewEmptyText();\r
- return 1;\r
- };\r
-\r
- if (!indexing_empty_texts) {\r
- empty_texts_aux = (unsigned int *)urealloc(empty_texts_aux, sizeof(pb)*(1+(npar-1)/(8*sizeof(pb))));\r
- bitset(empty_texts_aux, npar-1); // marks the non-empty text with a 1 in the bit vector\r
- }\r
- \r
- TextBuilder->InsertText(s);\r
- string cpps = (char*) s;\r
- CachedText.push_back(cpps); \r
- \r
- return 1; // success\r
- }\r
-\r
-// NewEmptyText(): indicates the event of finding a new empty text in the document.\r
-// In case of indexing empty and non-empty texts, we insert the empty texts into the\r
-// text collection. In case of indexing only non-empty texts, it just indicates an\r
-// empty text in the bit vector of empty texts. Returns a non-zero value upon\r
-// success, NULLT in case of error.\r
-int XMLTree::NewEmptyText() \r
- {\r
- unsigned char c = 0;\r
- if (!initialized) { // data structure has not been initialized properly\r
- fprintf(stderr, "Error: you cannot insert a new empty text without first calling method OpenDocument first\n");\r
- return NULLT;\r
- }\r
-\r
- if (!indexing_empty_texts) {\r
- empty_texts_aux = (unsigned int *)urealloc(empty_texts_aux, sizeof(pb)*(1+(npar-1)/(8*sizeof(pb))));\r
- \r
- bitclean(empty_texts_aux, npar-1); // marks the empty text with a 0 in the bit vector\r
- }\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
-\r
-\r
// GetTagId: returns the tag identifier corresponding to a given tag name.\r
// Returns NULLT in case that the tag name does not exists.\r
TagType XMLTree::GetTagId(unsigned char *tagname)\r
}\r
\r
\r
-//KIM : OJO need the two following methods\r
-\r
const unsigned char *XMLTree::GetTagNameByRef(TagType tagid)\r
{\r
- if(tagid==(uint)-1) return NULL;\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
\r
\r
TagType XMLTree::RegisterTag(unsigned char *tagname)\r
-{\r
- if (!finished)\r
- return NULLT;\r
- \r
- TagType id = XMLTree::GetTagId(tagname);\r
- if (id == NULLT){\r
- id = ntagnames;\r
- ntagnames = ntagnames + 1; \r
- TagName = (unsigned char **) urealloc(TagName,ntagnames*(sizeof(unsigned char*)));\r
- TagName[id] = (unsigned char *) umalloc(sizeof(unsigned char)*strlen( (const char*) tagname)+1);\r
- strcpy((char*)TagName[id], (const char *)tagname); \r
- };\r
-\r
- return id;\r
-}\r
+ { \r
+ TagType id = XMLTree::GetTagId(tagname);\r
+ if (id == NULLT) {\r
+ id = ntagnames;\r
+ ntagnames = ntagnames + 1; \r
+ TagName = (unsigned char **) urealloc(TagName,ntagnames*(sizeof(unsigned char*)));\r
+ TagName[id] = (unsigned char *) umalloc(sizeof(unsigned char)*strlen( (const char*) tagname)+1);\r
+ strcpy((char*)TagName[id], (const char *)tagname); \r
+ }\r
+\r
+ return id;\r
+ }\r
+\r
+\r
* along with this program; if not, write to the *\r
* Free Software Foundation, Inc., *\r
* 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. *\r
- ******************************************************************************/ \r
+ ******************************************************************************/\r
\r
#ifndef XMLTREE_H_\r
#define XMLTREE_H_\r
#include <stdlib.h>\r
#include <cstring>\r
\r
-//KIM : OJO\r
-//clash between TextCollection/Tools.h and libcds/includes/basics.h\r
+\r
#undef W\r
#undef WW\r
#undef Wminusone\r
\r
#include "bp.h"\r
+//#include "basics.h"\r
#include <static_bitsequence.h>\r
#include <alphabet_mapper.h>\r
#include <static_sequence.h>\r
\r
#define PERM_SAMPLE 10\r
\r
- // sets bit p in e\r
-#define bitset(e,p) ((e)[(p)/W] |= (1<<((p)%W)))\r
- // cleans bit p in e\r
-#define bitclean(e,p) ((e)[(p)/W] &= ~(1<<((p)%W)))\r
-\r
\r
typedef int treeNode;\r
typedef int TagType; \r
int max;\r
} range;\r
\r
+typedef struct nd {\r
+ uint position;\r
+ struct nd *next;\r
+} ListNode;\r
\r
-//KIM : OJO\r
-// I know this class implements the working draft that we have but the logics seem flawed here...\r
-// We should have two classes. One XMLTreeBuilder and one XMLTree.\r
-// XMLTreeBuilder would have OpenDocument, NewOpenTag,... and CloseDocument would return an XMLTree\r
-// XMLTree would have only an initialized structure. If find it really ugly to check (!finished) or (!initialized)\r
-// in every function (FirstChild....).\r
+typedef struct {\r
+ ListNode *first;\r
+ ListNode *last;\r
+} TagArrayEntry;\r
\r
class XMLTree {\r
/** Balanced parentheses representation of the tree */\r
\r
/** Mapping from tag identifer to tag name */ \r
unsigned char **TagName;\r
+ uint ntagnames;\r
\r
/** boolean flag indicating whether we are indexing empty texts or not */\r
bool indexing_empty_texts; \r
\r
/** Bit vector indicating with a 1 the positions of the non-empty texts. */\r
- static_bitsequence_rrr02 *EBVector; \r
+ static_bitsequence *EBVector; \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
+ uint * tags_fix;\r
+ uint tags_blen, tags_len;\r
\r
/** The texts in the XML document */\r
- TextCollectionBuilder *TextBuilder;\r
TextCollection *Text;\r
\r
/** The texts in the XML document (cached for faster display) */\r
vector<string> CachedText;\r
\r
- /** Flag indicating whether the whole data structure has been constructed or \r
- * not. If the value is true, you cannot add more texts, nodes, etc. */\r
- bool finished;\r
-\r
- /** Flag indicating whether the construction of the data structure has been\r
- * initialized or not (by calling method OpenDocument()). If this is true,\r
- * you cannot insert new tags or texts. */\r
- bool initialized;\r
- \r
- /* the following components are used for construction purposes */\r
- pb *par_aux;\r
- TagType *tags_aux;\r
- int npar;\r
- int parArraySize;\r
- int ntagnames;\r
- unsigned int *empty_texts_aux;\r
-\r
- // KIM : OJO\r
- // I added those two. The TagName array should always contains two special tags\r
- // <@> for attributes and <$> for PCDATA.\r
- // <$> can never be in a document (since we handle the text differently)\r
- // but <@> can be returned by the parser. This boolean is needed for the construction\r
- // of the Tag bitmap to know if <@> must be taken into account or not\r
- bool found_attributes;\r
-\r
- // KIM : OJO\r
+ TagArrayEntry *TagArray;\r
+\r
// Allows to disable the TextCollection for benchmarkin purposes\r
bool disable_tc;\r
\r
public:\r
- void print_stats();\r
+ /** Data structure constructors */\r
+ XMLTree() {;};\r
\r
- /** Data structure constructor */\r
- XMLTree() {finished = false; initialized = false; Text = 0; TextBuilder = 0; }; \r
+ XMLTree(pb *par, uint npar, unsigned char **TN, uint ntagnames, uint *empty_texts_bmp, TagType *tags,\r
+ TextCollection *TC, vector<string> CT, bool indexing_empty_t, bool dis_tc);\r
\r
/** Data structure destructor */\r
~XMLTree();\r
/** IsChild(x,y): returns whether node x is parent of node y. */\r
bool IsChild(treeNode x, treeNode y);\r
\r
- /** IsChild(x,y): returns whether node x is the first child of its parent */\r
+ /** IsFirstChild(x): returns whether node x is the first child of its parent. */\r
bool IsFirstChild(treeNode x);\r
- \r
+\r
/** NumChildren(x): number of children of node x. Constant time with the \r
* data structure of Sadakane. */\r
int NumChildren(treeNode x);\r
* Very fast in BP. */\r
treeNode FirstChild(treeNode x);\r
\r
- /** LastChild(x): returns the last child of node x. \r
- * Implemented by Kim naively. */\r
+ /** LastChild(x): returns the last child of node x. */\r
treeNode LastChild(treeNode x);\r
-\r
+ \r
/** NextSibling(x): returns the next sibling of node x, assuming it \r
* exists. */\r
-\r
treeNode NextSibling(treeNode x);\r
\r
/** PrevSibling(x): returns the previous sibling of node x, assuming it \r
* exists. */\r
treeNode PrevSibling(treeNode x);\r
\r
- /** TaggedChild(x,i,tag): returns the i-th child of node x tagged tag, or \r
+ /** TaggedChild(x,tag): returns the first child of node x tagged tag, or \r
* NULLT if there is none. Because of the balanced-parentheses representation \r
* of the tree, this operation is not supported efficiently, just iterating \r
* among the children of node x until finding the desired child. */\r
- treeNode TaggedChild(treeNode x, int i, TagType tag);\r
+ treeNode TaggedChild(treeNode x, TagType tag);\r
\r
+ treeNode SelectChild(treeNode x, TagType *tags, int ntags);\r
+\r
+ /** TaggedSibling(x,tag): returns the first sibling of node x tagged tag, or \r
+ * NULLT if there is none. */\r
+ treeNode TaggedSibling(treeNode x, TagType tag);\r
+ \r
+ treeNode SelectSibling(treeNode x, TagType *tags, int ntags);\r
+\r
/** TaggedDesc(x,tag): returns the first node tagged tag with larger \r
* preorder than x and within the subtree of x. Returns NULT if there \r
* is none. */\r
treeNode TaggedDesc(treeNode x, TagType tag);\r
\r
+ treeNode SelectDesc(treeNode x, TagType *tags, int ntags);\r
\r
treeNode TaggedBelow(treeNode x, TagType *childtags, unsigned int ctlen,\r
TagType *desctags, unsigned int dtlen);\r
treeNode TaggedFoll(treeNode x, TagType tag);\r
\r
treeNode TaggedFollBelow(treeNode x, TagType tag,treeNode root); \r
- \r
+ \r
+ treeNode SelectFollBelow(treeNode x, TagType *tags, int ntags, treeNode ctx);\r
+\r
/** TaggedFollowingSibling(x,tag) */\r
treeNode TaggedFollowingSibling(treeNode x, TagType tag);\r
\r
\r
/** ParentNode(d): returns the parent node of document identifier d. */\r
treeNode ParentNode(DocID d);\r
+ \r
treeNode PrevNode(DocID d);\r
\r
- /** OpenDocument(empty_texts,sample_rate_text,dtc): initilizes the construction\r
- * of the data structure for the XML document. Parameter empty_texts \r
- * indicates whether we index empty texts in document or not. Parameter \r
- * sample_rate_text indicates the sampling rate for the text searching data\r
- * structures (small values get faster searching but a bigger space \r
- * requirement). dtc disable the use of the TextCollection\r
- * (i.e. everything is considered an empty text *)\r
- * Returns a non-zero value upon success, NULLT in case of \r
- * error. */\r
-\r
- int OpenDocument(bool empty_texts, int sample_rate_text, bool dtc);\r
-\r
- /** CloseDocument(): finishes the construction of the data structure for \r
- * the XML document. Tree and tags are represented in the final form, \r
- * dynamic data structures are made static, and the flag "finished" is set \r
- * to true. After that, the data structure can be queried. */\r
- int CloseDocument();\r
-\r
- /** NewOpenTag(tagname): indicates the event of finding a new opening tag \r
- * in the document. Tag name is given. Returns a non-zero value upon \r
- * success, and returns NULLT in case of error. */\r
- int NewOpenTag(unsigned char *tagname);\r
- \r
- /** NewClosingTag(tagname): indicates the event of finding a new closing tag\r
- * in the document. Tag name is given. Returns a non-zero value upon \r
- * success, and returns NULLT in case of error. */\r
- int NewClosingTag(unsigned char *tagname);\r
- \r
- /** NewText(s): indicates the event of finding a new (non-empty) text s in \r
- * the document. The new text is inserted within the text collection. \r
- * Returns a non-zero value upon success, NULLT in case of error. */\r
- int NewText(unsigned char *s);\r
-\r
- /** NewEmptyText(): indicates the event of finding a new empty text in the \r
- * document. In case of indexing empty and non-empty texts, we insert the \r
- * empty texts into the text collection. In case of indexing only non-empty\r
- * texts, it just indicates an empty text in the bit vector of empty texts. \r
- * Returns a non-zero value upon success, NULLT in case of error. */\r
- int NewEmptyText();\r
-\r
/** GetTagId(tagname): returns the tag identifier corresponding to a given \r
* tag name. Returns NULLT in case that the tag name does not exists. */\r
TagType GetTagId(unsigned char *tagname);\r
* Returns NULL in case that the tag identifier is not valid.*/\r
unsigned char *GetTagName(TagType tagid);\r
\r
-\r
- // OJO\r
/** GetTagName(tagid): returns the tag name of a given tag identifier. \r
* The result is just a reference and should not be freed by the caller.\r
*/\r
const unsigned char *GetTagNameByRef(TagType tagid);\r
\r
- //OJO\r
/** RegisterTag adds a new tag to the tag collection this is needed\r
* if the query contains a tag which is not in the document, we need\r
* to give this new tag a fresh id and store it somewhere. A logical\r
bool EmptyText(DocID i) {\r
return Text->EmptyText(i);\r
}\r
+\r
/** Prefix(s): search for texts prefixed by string s. */\r
TextCollection::document_result Prefix(uchar const *s) {\r
return Text->Prefix(s);\r
\r
/** CountLessThan(s): counting version of LessThan(s). */\r
unsigned CountLessThan(uchar const *s) {\r
- return CountLessThan(s);\r
+ return Text->CountLessThan(s);\r
}\r
\r
/** GetText(d): returns the text corresponding to document with\r
TextCollection *getTextCollection() {\r
return Text;\r
}\r
+ \r
/** Save: saves XML tree data structure to file. */\r
void Save(unsigned char *filename);\r
\r
/** Load: loads XML tree data structure from file. sample_rate_text \r
* indicates the sample rate for the text search data structure. */\r
static XMLTree *Load(unsigned char *filename, int sample_rate_text); \r
+\r
+ void insertTag(TagType tag, uint position);\r
+ \r
+ void print_stats();\r
};\r
#endif\r
+\r