-#include "XMLTree.h"\r
+\r
#include "basics.h"\r
#include <cstring>\r
+#include "XMLTree.h"\r
\r
// functions to convert tag positions to the corresponding tree node and viceversa. \r
// These are implemented in order to be able to change the tree and Tags representations, \r
// tag position -> tree node\r
inline treeNode tagpos2node(int t) \r
{\r
- return (treeNode)t;\r
+ return (treeNode) t;\r
}\r
\r
// tree node -> tag position\r
inline int node2tagpos(treeNode x) \r
- {\r
- return (int)x;\r
- }\r
+{\r
+ return (int)x;\r
+}\r
+\r
+// returns NULLT if the test is true\r
+#define NULLT_IF(x) do { if (x) return NULLT; } while (0)\r
\r
\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
+XMLTree::XMLTree( pb * const par, uint npar, vector<string> * const TN, TagIdMap * const tim, uint *empty_texts_bmp, TagType *tags,\r
+ TextCollection * const TC, vector<string> * const CT, bool dis_tc)\r
+\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
-\r
+ bp_construct(Par, npar, (pb*) par, OPT_DEGREE|0); \r
+ \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
- 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
+ TagName = (vector<string>*)TN;\r
+ tIdMap = (TagIdMap *) tim;\r
+ \r
+ uint max_tag = TN->size() - 1;\r
+ \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
free(tags);\r
tags = NULL;\r
\r
- Text = TC;\r
-\r
- CachedText = CT;\r
+ Text = (TextCollection*) TC;\r
\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
+ CachedText = (vector<string>*) CT;\r
+ EBVector = new static_bitsequence_rrr02(empty_texts_bmp,npar,32);\r
+ free(empty_texts_bmp);\r
+ empty_texts_bmp = NULL;\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
+ \r
disable_tc = dis_tc;\r
}\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
+ delete tIdMap;\r
+ tIdMap = NULL;\r
+ \r
+ delete TagName;\r
+ TagName = NULL;\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
+ delete EBVector;\r
+ EBVector = NULL;\r
+\r
+\r
}\r
\r
\r
}\r
\r
// Save: saves XML tree data structure to file. \r
-void XMLTree::Save(unsigned char *filename) \r
+void XMLTree::Save(int fd) \r
{\r
FILE *fp;\r
char filenameaux[1024];\r
int i;\r
- \r
- sprintf(filenameaux, "%s.srx", filename);\r
- fp = fopen(filenameaux, "w");\r
- if (fp == NULL) {\r
- printf("Error: cannot create file %s to store the tree structure of XML collection\n", filenameaux);\r
- exit(1);\r
- } \r
- \r
+\r
+ fp = fdopen(fd, "wa");\r
// first stores the tree topology\r
saveTree(Par, fp);\r
- \r
+\r
// stores the table with tag names\r
- ufwrite(&ntagnames, sizeof(int), 1, fp);\r
- for (i=0; i<ntagnames;i++)\r
- fprintf(fp, "%s\n",TagName[i]);\r
- \r
- \r
- // stores the flags\r
- ufwrite(&indexing_empty_texts, 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
+ int ntags = TagName->size();\r
+\r
+ ufwrite(&ntags, sizeof(int), 1, fp);\r
+ for (i = 0; i<ntags;i++)\r
+ fprintf(fp, "%s\n",TagName->at(i).c_str());\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
+ // flags \r
+ ufwrite(&disable_tc, sizeof(bool),1,fp);\r
+ \r
+ //text positions\r
+ EBVector->save(fp);\r
+ \r
// stores the texts \r
if (!disable_tc) {\r
- Text->Save(fp);\r
- int st = CachedText.size();\r
+ int st = CachedText->size();\r
+\r
ufwrite(&st, sizeof(int),1,fp);\r
- for (int i = 0; i< CachedText.size(); i++){\r
- st = CachedText.at(i).size();\r
+ for (int i = 0; i< CachedText->size(); i++){\r
+ st = CachedText->at(i).size();\r
+\r
ufwrite(&st, sizeof(int),1,fp);\r
- ufwrite(CachedText.at(i).c_str(),sizeof(char),1+CachedText.at(i).size(),fp);\r
+\r
+ ufwrite(CachedText->at(i).c_str(),sizeof(char),1+CachedText->at(i).size(),fp);\r
+\r
};\r
+ \r
+ Text->Save(fp);\r
};\r
- fclose(fp);\r
+\r
\r
}\r
\r
\r
// Load: loads XML tree data structure from file. Returns\r
// a pointer to the loaded data structure\r
-XMLTree *XMLTree::Load(unsigned char *filename, int sample_rate_text) \r
+XMLTree *XMLTree::Load(int fd) \r
{\r
FILE *fp;\r
char buffer[1024];\r
XMLTree *XML_Tree;\r
int i;\r
- size_t s_tree = 0;\r
- long s_text = 0;\r
- size_t s_tags = 0;\r
-\r
- // first load the tree topology\r
- sprintf(buffer, "%s.srx", filename);\r
- fp = fopen(buffer, "r");\r
- if (fp == NULL) {\r
- printf("Error: cannot open file %s to load the tree structure of XML collection\n", buffer);\r
- exit(1);\r
- } \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
+ fp = fdopen(fd, "r");\r
\r
- // stores the table with tag names\r
- ufread(&XML_Tree->ntagnames, sizeof(int), 1, fp);\r
-\r
- s_tree += sizeof(int);\r
+ XML_Tree = new XMLTree();\r
\r
- XML_Tree->TagName = (unsigned char **)umalloc(XML_Tree->ntagnames*sizeof(unsigned char *));\r
+ // Load the tree structure\r
+ XML_Tree->Par = (bp *)umalloc(sizeof(bp));\r
\r
- s_tags += sizeof(unsigned char*)*XML_Tree->ntagnames;\r
+ loadTree(XML_Tree->Par, fp); \r
\r
+ XML_Tree->TagName = new vector<string>();\r
+ XML_Tree->tIdMap = new std::unordered_map<string,int>();\r
+ \r
+ string s;\r
+ int ntags;\r
+ \r
+ // Load the tag names\r
+ ufread(&ntags, sizeof(int), 1, fp);\r
\r
- for (i=0; i<XML_Tree->ntagnames;i++) {\r
- char * r = fgets(buffer,1023,fp);\r
+ for (i=0; i<ntags;i++) {\r
+ char * r = fgets(buffer,1023,fp);\r
if (r==NULL)\r
throw "Cannot read tag list";\r
+ s = (const char*) buffer;\r
+ // remove the trailing \n\r
+ s.erase(s.size()-1); \r
+ XML_Tree->TagName->push_back(s);\r
+ XML_Tree->tIdMap->insert(std::make_pair(s,i));\r
+ \r
+ };\r
\r
- // strlen is actually the right size, since there is a trailing '\n'\r
- int len = strlen((const char*)buffer);\r
- XML_Tree->TagName[i] = (unsigned char *)ucalloc(len,sizeof(char));\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
- 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
- s_tree+=sizeof(bool)*4;\r
-\r
- if (!(XML_Tree->indexing_empty_texts)) XML_Tree->EBVector = static_bitsequence_rrr02::load(fp);\r
-\r
- s_tree+= XML_Tree->EBVector->size();\r
\r
- // loads the tags\r
+ // loads the tag structure\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
\r
+ // TODO ask francisco about this\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
+ 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
+ XML_Tree->Tags->test(seq,XML_Tree->tags_len);\r
delete [] seq;\r
/// End ugly tests\r
+ \r
\r
- s_text = ftell(fp);\r
+ // loads the flags\r
+ \r
+ ufread(&(XML_Tree->disable_tc), sizeof(bool), 1, fp);\r
+\r
+ XML_Tree->EBVector = static_bitsequence_rrr02::load(fp);\r
\r
+ // Not used \r
+ int sample_rate_text = 64;\r
// loads the texts\r
if (!XML_Tree->disable_tc){\r
- XML_Tree->Text = TextCollection::Load(fp,sample_rate_text);\r
+ XML_Tree->CachedText = new vector<string>;\r
int sst;\r
int st;\r
ufread(&sst, sizeof(int),1,fp);\r
+ \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
+ XML_Tree->CachedText->push_back(cppstr);\r
free(str);\r
- };\r
\r
+ };\r
+ XML_Tree->Text = TextCollection::Load(fp,sample_rate_text);\r
}\r
else XML_Tree->Text = NULL;\r
-\r
- s_text = ftell(fp) - s_text;\r
-\r
- fclose(fp);\r
-\r
- XML_Tree->print_stats();\r
+ \r
return XML_Tree;\r
}\r
\r
\r
// root(): returns the tree root.\r
-treeNode XMLTree::Root() \r
+inline treeNode XMLTree::Root() \r
{\r
- return root_node(Par);\r
+ return 0; //root_node(Par);\r
}\r
\r
// SubtreeSize(x): the number of nodes (and attributes) in the subtree of node x.\r
// Tag(x): returns the tag identifier of node x.\r
TagType XMLTree::Tag(treeNode x) \r
{\r
- return get_field(tags_fix,tags_blen,node2tagpos(x)); //Tags->access(node2tagpos(x));\r
+ return get_field(tags_fix,tags_blen,node2tagpos(x));\r
}\r
\r
// DocIds(x): returns the range of text identifiers that descend from node x.\r
return 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
+ \r
+ int min = EBVector->rank1(x-1); \r
+ int max = EBVector->rank1(x+2*subtree_size(Par, x)-2); \r
+ if (min==max) { // range is empty, no texts within the subtree of x\r
+ r.min = NULLT;\r
+ r.max = NULLT;\r
}\r
- else { // we are not indexing empty texts, we need rank\r
- int min = EBVector->rank1(x-1); \r
- int max = EBVector->rank1(x+2*subtree_size(Par, x)-2); \r
- if (min==max) { // range is empty, no texts within the subtree of x\r
- r.min = NULLT;\r
- r.max = NULLT;\r
- }\r
- else { // the range is non-empty, there are texts within the subtree of x\r
- r.min = min+1;\r
- r.max = max;\r
- }\r
+ else { // the range is non-empty, there are texts within the subtree of x\r
+ r.min = min+1;\r
+ r.max = max;\r
}\r
return r;\r
+\r
}\r
\r
// Parent(x): returns the parent node of node x.\r
{\r
if (i <= OPTD) return naive_child(Par, x, i);\r
else return child(Par, x, i);\r
- }\r
+}\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
- return first_child(Par, x);\r
+ NULLT_IF(x==NULLT);\r
+ return first_child(Par, x);\r
+ }\r
+\r
+treeNode XMLTree::FirstElement(treeNode x) \r
+ {\r
+ NULLT_IF(x==NULLT);\r
+ treeNode fc = first_child(Par, x);\r
+ //<$> is 2\r
+ return ((fc == NULLT || Tag(fc) != PCDATA_TAG_ID) ? fc : next_sibling(Par,fc));\r
+\r
}\r
\r
+treeNode XMLTree::NextElement(treeNode x) \r
+{\r
+ NULLT_IF(x==NULLT);\r
+ treeNode ns = next_sibling(Par, x);\r
+ return ((ns == NULLT || Tag(ns) != PCDATA_TAG_ID) ? ns : next_sibling(Par,ns));\r
+}\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
+ NULLT_IF(x==NULLT || x == Root() || isleaf(Par,x));\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 (x == Root() || x==NULLT)\r
- return NULLT;\r
- \r
- return next_sibling(Par, x);\r
+ NULLT_IF(x==NULLT || x == Root() );\r
+ return next_sibling(Par, x);\r
}\r
\r
// PrevSibling(x): returns the previous sibling of node x, assuming it exists.\r
treeNode XMLTree::PrevSibling(treeNode x) \r
{\r
- return prev_sibling(Par, x);\r
+ NULLT_IF(x==NULLT || x == Root());\r
+ return prev_sibling(Par, x);\r
}\r
\r
// TaggedChild(x,tag): returns the first child of node x tagged tag, or NULLT if there is none.\r
// efficiently, just iterating among the children of node x until finding the desired child.\r
treeNode XMLTree::TaggedChild(treeNode x, TagType tag) \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)) == 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
+ NULLT_IF(x==NULLT || isleaf(Par,x));\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
+ treeNode child; \r
+ child = first_child(Par, x); // starts at first child of node x\r
+ if (get_field(tags_fix,tags_blen,node2tagpos(child)) == tag)\r
+ return child;\r
+ else\r
+ return TaggedFollSibling(child,tag);\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
+treeNode XMLTree::TaggedFollSibling(treeNode x, TagType tag)\r
+{\r
+ NULLT_IF(x==NULLT);\r
+ treeNode sibling = next_sibling(Par, x); \r
+ while (sibling != NULLT) {\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
+treeNode XMLTree::SelectChild(treeNode x, std::unordered_set<int> *tags)\r
+{\r
+ \r
+ NULLT_IF(x==NULLT || isleaf(Par,x));\r
+ int i;\r
+ treeNode child = first_child(Par, x); \r
+ TagType t = get_field(tags_fix, tags_blen, node2tagpos(child));\r
+ std::unordered_set<int>::const_iterator tagit = tags->find(t);\r
+ if (tagit != tags->end()) return child; \r
+ return SelectFollSibling(child,tags);\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
+treeNode XMLTree::SelectFollSibling(treeNode x, std::unordered_set<int> *tags)\r
+{\r
+\r
+ NULLT_IF(x==NULLT);\r
+ int i;\r
+ TagType t;\r
+ treeNode sibling = next_sibling(Par, x);\r
+ std::unordered_set<int>::const_iterator tagit;\r
+ while (sibling != NULLT) {\r
+ t = get_field(tags_fix, tags_blen, node2tagpos(sibling));\r
+ tagit = tags->find(t);\r
+ if (tagit != tags->end()) return sibling;\r
+ sibling = next_sibling(Par, sibling);\r
+ }\r
+ return NULLT; \r
}\r
\r
\r
// the subtree of x. Returns NULLT if there is none.\r
treeNode XMLTree::TaggedDesc(treeNode x, TagType tag) \r
{\r
- treeNode y;\r
- if (isleaf(Par,x))\r
- return NULLT;\r
-\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
-\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
- 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
- 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
-\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
- \r
- return res;\r
-}\r
+ NULLT_IF(x==NULLT || isleaf(Par,x));\r
\r
+ int s = (int) Tags->select_next(tag,node2tagpos(x));\r
+ NULLT_IF (s == -1);\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
- return res < lim ? res : NULLT;\r
+ treeNode y = tagpos2node(s); // transforms the tag position into a node position\r
+ \r
+ return (is_ancestor(Par,x,y) ? y : NULLT);\r
}\r
\r
\r
-treeNode XMLTree::TaggedDescOrFollOnly(treeNode x,TagType *folltags, unsigned int ftlen,treeNode root)\r
+treeNode XMLTree::SelectDesc(treeNode x, std::unordered_set<int> *tags)\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
-// 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 *childtags, unsigned int ctlen,\r
- TagType *folltags, unsigned int flen,treeNode root)\r
- {\r
- treeNode y,old_y,lim,res;\r
- TagType tag;\r
- if (x == NULLT || x == Root())\r
- return NULLT;\r
-\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
+ NULLT_IF (x ==NULLT || isleaf(Par,x));\r
+ int i;\r
+ treeNode min = NULLT;\r
+ treeNode fc = first_child(Par,x);\r
+ treeNode aux;\r
+ std::unordered_set<int>::const_iterator tagit;\r
+ for (tagit = tags->begin(); tagit != tags->end(); tagit++) {\r
+ aux = TaggedDesc(x, (TagType) *tagit);\r
+ if (aux == fc) return fc;\r
+ if (aux == NULLT) continue;\r
+ if ((min == NULLT) || (aux < min)) min = aux;\r
};\r
- \r
- return res < lim ? res : NULLT; \r
+ return min;\r
}\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
// the subtree of x. Returns NULLT if there is none.\r
treeNode XMLTree::TaggedFoll(treeNode x, TagType tag)\r
{\r
- if (x ==NULLT || x == Root())\r
- return NULLT;\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
+ NULLT_IF (x ==NULLT || x == Root());\r
+ \r
+ return tagpos2node(Tags->select_next(tag,find_close(Par, x)));\r
+\r
} \r
\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
- 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
+ NULLT_IF (x == NULLT || x == Root());\r
+ \r
+ treeNode s = tagpos2node(Tags->select_next(tag, find_close(Par, x)));\r
+ \r
+ if (root == Root()) return s;\r
+ NULLT_IF (s == NULLT || s >= find_close(Par, root));\r
+ \r
+ return s;\r
+} \r
\r
-treeNode XMLTree::SelectFollBelow(treeNode x, TagType *tags, int ntags, treeNode ctx)\r
+/* Here we inline TaggedFoll to find the min globally, and only at the end\r
+ we check if the min is below the context node */\r
+treeNode XMLTree::SelectFollBelow(treeNode x, std::unordered_set<int> *tags, treeNode root)\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
+ NULLT_IF(x==NULLT || x==Root());\r
+ int i;\r
+ treeNode min = NULLT;\r
+ treeNode ns = next_sibling(Par, x);\r
+ treeNode aux;\r
+ std::unordered_set<int>::const_iterator tagit;\r
+ for (tagit = tags->begin(); tagit != tags->end(); tagit++) {\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
- treeNode ns = next_sibling(Par,x);\r
+ aux = tagpos2node(Tags->select_next(*tagit, find_close(Par, x)));\r
+ \r
+ // The next sibling of x is guaranteed to be below ctx\r
+ // and is the node with lowest preorder which is after ctx.\r
+ // if we find it, we return early;\r
+ \r
+ if (aux == ns ) return ns;\r
+ if (aux == NULLT) continue;\r
+ if ((min == NULLT) || (aux < min)) min = aux;\r
+ };\r
+ \r
+ // found the smallest node in preorder which is after x.\r
+ // if ctx is the root node, just return what we found.\r
\r
- if (x == NULLT || x == Root() || ns == -1)\r
- return NULLT;\r
+ if (root == Root()) return min;\r
+ // else check whether if is in below the ctx node\r
\r
- int s = (int) Tags->select_next(tag, node2tagpos(ns)-1);\r
- if (s==-1) return NULLT;\r
- else return tagpos2node(s);\r
+ NULLT_IF (min == NULLT || min >= find_close(Par, root));\r
+ \r
+ return min;\r
+ \r
}\r
\r
\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
-DocID XMLTree::PrevText(treeNode x) \r
- {\r
- if (x == Root()) return NULLT;\r
- if (indexing_empty_texts) // faster, no rank needed\r
- return (DocID)x-1;\r
- else { // we are not indexing empty texts, rank is needed\r
- if (EBVector->access(x-1) == 0) \r
- return (DocID)NULLT; // there is no text to the left of node (text is empty)\r
- else\r
- return (DocID)EBVector->rank1(x-1)-1; //-1 because document ids start from 0\r
- }\r
- }\r
-\r
-// NextText(x): returns the document identifier of the text to the right\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 (x == Root()) return NULLT;\r
- if (indexing_empty_texts) // faster, no rank needed\r
- return (DocID)x+2*subtree_size(Par, x)-1;\r
- else { // we are not indexing empty texts, rank is needed\r
- int p = x+2*subtree_size(Par, x)-1;\r
- if (EBVector->access(p) == 0) // there is no text to the right of node\r
- return (DocID)NULLT;\r
- else\r
- return (DocID)EBVector->rank1(p)-1; //-1 because document ids start from 0\r
- }\r
- }\r
\r
// MyText(x): returns the document identifier of the text below node x, \r
// or NULLT if x is not a leaf node or the text is empty. Assumes Doc \r
// ids start from 0.\r
DocID XMLTree::MyText(treeNode x) \r
- {\r
- if (!IsLeaf(x)) return NULLT;\r
- if (indexing_empty_texts) // faster, no rank needed\r
- return (DocID)x;\r
- else { // we are not indexing empty texts, rank is needed\r
- if (EBVector->access(x) == 0) // there is no text below node x\r
- return (DocID)NULLT;\r
- else\r
- return (DocID)EBVector->rank1(x)-1; //-1 because document ids start from 0\r
- } \r
- }\r
+{\r
+ TagType tag = Tag(x);\r
+ // seems faster than testing EBVector->access(x);\r
+\r
+ if (tag == PCDATA_TAG_ID || tag == ATTRIBUTE_DATA_TAG_ID)\r
+ return (DocID) (EBVector->rank1(x)-1); //-1 because document ids start from 0\r
+ else \r
+ return (DocID) NULLT;\r
+ \r
+}\r
\r
// TextXMLId(d): returns the preorder of document with identifier d in the tree consisting of\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 (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
- int s = EBVector->select1(d+1);\r
- return rank_open(Par, s) + d + 1; // +1 because root has preorder 1\r
- }\r
+ int s = EBVector->select1(d+1);\r
+ return rank_open(Par, s) + d + 1; // +1 because root has preorder 1\r
+ \r
}\r
\r
// NodeXMLId(x): returns the preorder of node x in the tree consisting \r
// preorder 0;\r
int XMLTree::NodeXMLId(treeNode x) \r
{\r
- if (indexing_empty_texts)\r
- return x - 1 + rank_open(Par, x);\r
- else {\r
- if (x == Root()) return 1; // root node has preorder 1\r
- else\r
- return rank_open(Par, x) + EBVector->rank1(x-1);\r
- }\r
+ if (x == Root()) return 1; // root node has preorder 1\r
+ return rank_open(Par, x) + EBVector->rank1(x-1);\r
}\r
\r
// ParentNode(d): returns the parent node of document identifier d.\r
treeNode XMLTree::ParentNode(DocID d) \r
{ \r
- if (d == NULLT)\r
- return NULLT;\r
- \r
- int s;\r
- // OJO : Kim : I added the d+1. before that, else branch was \r
- // EBVector->select1(d)\r
- // and gave wrong results (I'm really poking a bear with a stick here).\r
- if (indexing_empty_texts) s = d;\r
- else s = EBVector->select1(d+1);\r
- \r
- if (inspect(Par,s) == CP) // is a closing parenthesis\r
- return parent(Par, find_open(Par, s));\r
- else // is an opening parenthesis\r
- return (treeNode)s;\r
- \r
- }\r
-treeNode XMLTree::PrevNode(DocID d) \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
+ NULLT_IF (d == NULLT); \r
+ return (treeNode) EBVector->select1(d+1); \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
- int i;\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 (TagType)-1; //ntagnames; //(TagType)NULLT; // tagname does not exists in the table\r
- else return i;\r
+ \r
+ string s = (char *) tagname;\r
+ TagIdMapIT it = tIdMap->find(s); \r
+ return (TagType) ((it != tIdMap->end()) ? it->second : -1);\r
+ \r
}\r
\r
\r
unsigned char *XMLTree::GetTagName(TagType tagid)\r
{\r
unsigned char *s;\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
- return s;\r
+ if ( tagid < 0 || tagid >= TagName->size())\r
+ return (unsigned char *) "<INVALID TAG>";\r
+ strcpy((char *)s, TagName->at(tagid).c_str());\r
+ \r
+ return (s == NULL ? (unsigned char*) "<INVALID TAG>" : s);\r
}\r
\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
+ unsigned char *s;\r
+ if ( tagid < 0 || tagid >= TagName->size())\r
+ return (unsigned char *) "<INVALID TAG>";\r
+ \r
+ return (const unsigned char *) TagName->at(tagid).c_str();\r
+ \r
}\r
\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
+ string s = (char *) tagname; \r
+ REGISTER_TAG(TagName,tIdMap,s);\r
+ \r
+ };\r
+ \r
return id;\r
}\r
\r
\r
#ifndef XMLTREE_H_\r
#define XMLTREE_H_\r
+#include <unordered_set>\r
+#include <unordered_map>\r
#include "TextCollection/TextCollectionBuilder.h"\r
#include <stdio.h>\r
#include <stdlib.h>\r
#undef Wminusone\r
\r
#include "bp.h"\r
-//#include "basics.h"\r
+\r
#include <static_bitsequence.h>\r
#include <alphabet_mapper.h>\r
#include <static_sequence.h>\r
int max;\r
} range;\r
\r
-typedef struct nd {\r
- uint position;\r
- struct nd *next;\r
-} ListNode;\r
+// Encoding of the XML Document :\r
+// The following TAGs and IDs are fixed, "" is the tag of the root.\r
+// a TextNode is represented by a leaf <<$>></<$>> The DocId in the TextCollection\r
+// of that leaf is kept in a bit sequence.\r
+// a TextNode below an attribute is likewise represented by a leaf <<@$>><</@$>>\r
+// An element <e a1="v1" a2="v2" ... an="vn" > ...</e> the representation is:\r
+// <e><<@>> <<@>a1> <<$@>>DocID(v1)</<$@>></<@>a1> ... </<@>> .... </e>\r
+// Hence the attributes (if any) are always below the first child of their element,\r
+// as the children of a fake node <@>.\r
\r
-typedef struct {\r
- ListNode *first;\r
- ListNode *last;\r
-} TagArrayEntry;\r
+\r
+#define DOCUMENT_OPEN_TAG ""\r
+#define DOCUMENT_TAG_ID 0\r
+#define ATTRIBUTE_OPEN_TAG "<@>"\r
+#define ATTRIBUTE_TAG_ID 1\r
+#define PCDATA_OPEN_TAG "<$>"\r
+#define PCDATA_TAG_ID 2\r
+#define ATTRIBUTE_DATA_OPEN_TAG "<@$>"\r
+#define ATTRIBUTE_DATA_TAG_ID 3\r
+#define DOCUMENT_CLOSE_TAG "/"\r
+#define ATTRIBUTE_CLOSE_TAG "/<@>"\r
+#define PCDATA_CLOSE_TAG "/<$>"\r
+#define ATTRIBUTE_DATA_CLOSE_TAG "/<@$>"\r
+\r
+\r
+\r
+typedef std::unordered_map<string,int> TagIdMap;\r
+typedef TagIdMap::const_iterator TagIdMapIT;\r
+\r
+#define REGISTER_TAG(v,h,t) do { (h)->insert(std::make_pair((t),(v)->size()));\\r
+ (v)->push_back(t); } while (false)\r
+\r
+\r
+class XMLTreeBuilder;\r
\r
class XMLTree {\r
+\r
+ // Only the builder can access the constructor\r
+ friend class XMLTreeBuilder;\r
+\r
+ private:\r
/** Balanced parentheses representation of the tree */\r
bp *Par;\r
\r
/** Mapping from tag identifer to tag name */ \r
- unsigned char **TagName;\r
- uint ntagnames;\r
+ vector<string> *TagName;\r
+ TagIdMap * tIdMap;\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 *EBVector; \r
\r
TextCollection *Text;\r
\r
/** The texts in the XML document (cached for faster display) */\r
- vector<string> CachedText;\r
- \r
- TagArrayEntry *TagArray;\r
+ vector<string> *CachedText;\r
\r
// Allows to disable the TextCollection for benchmarkin purposes\r
bool disable_tc;\r
\r
-public:\r
+\r
/** Data structure constructors */\r
- XMLTree() {;};\r
+ XMLTree(){;};\r
\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
+ // non const pointer are freed by this method.\r
+ XMLTree( pb * const par, uint npar, vector<string> * const TN, TagIdMap * const tim, uint *empty_texts_bmp, TagType *tags,\r
+ TextCollection * const TC, vector<string> * const CT, bool dis_tc);\r
+\r
+public: \r
/** Data structure destructor */\r
~XMLTree();\r
\r
/** FirstChild(x): returns the first child of node x, assuming it exists. \r
* Very fast in BP. */\r
treeNode FirstChild(treeNode x);\r
- \r
+ treeNode FirstElement(treeNode x);\r
+\r
/** LastChild(x): returns the last child of node x. */\r
treeNode LastChild(treeNode x);\r
\r
/** NextSibling(x): returns the next sibling of node x, assuming it \r
* exists. */\r
treeNode NextSibling(treeNode x);\r
+ treeNode NextElement(treeNode x);\r
\r
/** PrevSibling(x): returns the previous sibling of node x, assuming it \r
* exists. */\r
* among the children of node x until finding the desired child. */\r
treeNode TaggedChild(treeNode x, TagType tag);\r
\r
- treeNode SelectChild(treeNode x, TagType *tags, int ntags);\r
+ treeNode SelectChild(treeNode x, std::unordered_set<int> * tags);\r
\r
- /** TaggedSibling(x,tag): returns the first sibling of node x tagged tag, or \r
+ /** TaggedFollSibling(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
+ treeNode TaggedFollSibling(treeNode x, TagType tag);\r
\r
- treeNode SelectSibling(treeNode x, TagType *tags, int ntags);\r
+ treeNode SelectFollSibling(treeNode x, std::unordered_set<int> * tags);\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
+ treeNode SelectDesc(treeNode x, std::unordered_set<int> * tags);\r
\r
- treeNode TaggedBelow(treeNode x, TagType *childtags, unsigned int ctlen,\r
- TagType *desctags, unsigned int dtlen);\r
- \r
- treeNode TaggedNext(treeNode x, TagType *childtags, unsigned int ctlen,\r
- TagType *folltags, unsigned int flen,treeNode root);\r
-\r
- treeNode TaggedDescOnly(treeNode x, TagType *desctags, unsigned int dtlen);\r
- \r
- treeNode TaggedDescOrFollOnly(treeNode x, TagType *folltags, unsigned int flen,\r
- treeNode root);\r
-\r
- treeNode TaggedFollOnly(treeNode x, TagType *folltags, unsigned int flen,\r
- treeNode root);\r
- \r
\r
/** TaggedPrec(x,tag): returns the first node tagged tag with smaller \r
* preorder than x and not an ancestor of x. Returns NULLT if there \r
\r
treeNode TaggedFollBelow(treeNode x, TagType tag,treeNode root); \r
\r
- treeNode SelectFollBelow(treeNode x, TagType *tags, int ntags, treeNode ctx);\r
+ treeNode SelectFollBelow(treeNode x, std::unordered_set<int> * tags, treeNode root);\r
\r
/** TaggedFollowingSibling(x,tag) */\r
treeNode TaggedFollowingSibling(treeNode x, TagType tag);\r
/** GetText(d): returns the text corresponding to document with\r
* id d. */\r
uchar* GetText(DocID d) {\r
- return Text->GetText(d);\r
+ uchar * s = Text->GetText(d);\r
+ return (s[0] == 1 ? (uchar*)"" : s);\r
}\r
\r
uchar* GetCachedText(DocID d) {\r
- uchar * str = (uchar*) calloc(sizeof(char),(CachedText.at(d).size() + 1));\r
- strcpy((char*) str,(const char*) CachedText.at(d).c_str());\r
+ uchar * str = (uchar*) calloc(sizeof(char),(CachedText->at(d).size() + 1));\r
+ strcpy((char*) str,(const char*) CachedText->at(d).c_str());\r
return (uchar*) (str);\r
}\r
\r
}\r
\r
/** Save: saves XML tree data structure to file. */\r
- void Save(unsigned char *filename);\r
+ void Save(int fd);\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
+ static XMLTree *Load(int fd); \r
\r
void insertTag(TagType tag, uint position);\r
\r
-\r
-#include "XMLTreeBuilder.h"\r
#include "basics.h"\r
+#include "XMLTreeBuilder.h"\r
+\r
+\r
+XMLTreeBuilder::~XMLTreeBuilder(){\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 XMLTreeBuilder::OpenDocument(bool empty_texts, int sample_rate_text, bool dtc)\r
{\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
+ \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
- TagName[0] = (unsigned char *) umalloc(4*sizeof(unsigned char));\r
- strcpy((char *) TagName[0], "<@>");\r
- TagName[1] = (unsigned char *) umalloc(4*sizeof(unsigned char));\r
- strcpy((char *) TagName[1], "<$>");\r
- TagName[2] = (unsigned char *) umalloc(5*sizeof(unsigned char));\r
- strcpy((char *) TagName[2], "/<@>");\r
- TagName[3] = (unsigned char *) umalloc(5*sizeof(unsigned char));\r
- strcpy((char *) TagName[3], "/<$>");\r
-\r
- if (!indexing_empty_texts) \r
- empty_texts_aux = (unsigned int *)umalloc(sizeof(unsigned int));\r
- \r
+ TagName = new vector<string>();\r
+ tIdMap = new std::unordered_map<string,int>();\r
+\r
+ REGISTER_TAG(TagName,tIdMap,DOCUMENT_OPEN_TAG);\r
+ REGISTER_TAG(TagName,tIdMap,ATTRIBUTE_OPEN_TAG);\r
+ REGISTER_TAG(TagName,tIdMap,PCDATA_OPEN_TAG);\r
+ REGISTER_TAG(TagName,tIdMap,ATTRIBUTE_DATA_OPEN_TAG);\r
+ REGISTER_TAG(TagName,tIdMap,DOCUMENT_CLOSE_TAG);\r
+ REGISTER_TAG(TagName,tIdMap,ATTRIBUTE_CLOSE_TAG);\r
+ REGISTER_TAG(TagName,tIdMap,PCDATA_CLOSE_TAG);\r
+ REGISTER_TAG(TagName,tIdMap,ATTRIBUTE_DATA_CLOSE_TAG);\r
+\r
+\r
+ CachedText = new vector<string>;\r
if (disable_tc)\r
TextBuilder = 0;\r
else \r
- TextBuilder = new TextCollectionBuilder((unsigned)sample_rate_text);\r
+ TextBuilder = new TextCollectionBuilder((unsigned)sample_rate_text);\r
Text = 0;\r
- \r
+ empty_texts_aux = (unsigned int *)ucalloc(sizeof(unsigned int),1);\r
+ eta_size = sizeof(unsigned int);\r
return 1; // indicates success in the initialization of the data structure\r
}\r
\r
// the data structure can be queried.\r
XMLTree *XMLTreeBuilder::CloseDocument()\r
{ \r
- // closing parenthesis for the tree root\r
- par_aux = (pb *)urealloc(par_aux, sizeof(pb)*(1+npar/(8*sizeof(pb))));\r
- setbit(par_aux, npar, CP);\r
- npar++;\r
+ //closing parenthesis for the tree root\r
+ //par_aux = (pb *)urealloc(par_aux, sizeof(pb)*(1+npar/(8*sizeof(pb))));\r
+ //setbit(par_aux, npar, CP);\r
+ //npar++;\r
\r
// makes the text collection static\r
if (!disable_tc) {\r
delete TextBuilder;\r
TextBuilder = 0;\r
}\r
-\r
- XMLTree *T = new XMLTree(par_aux, npar, TagName, ntagnames, empty_texts_aux, tags_aux, \r
- Text, CachedText, indexing_empty_texts, disable_tc);\r
+ \r
+ XMLTree *T = new XMLTree(par_aux,\r
+ npar, \r
+ TagName,\r
+ tIdMap,\r
+ empty_texts_aux, // freed by the constructor\r
+ tags_aux, //freed by the constructor\r
+ Text,\r
+ CachedText,\r
+ disable_tc);\r
return T; \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 XMLTreeBuilder::NewOpenTag(unsigned char *tagname)\r
+int XMLTreeBuilder::NewOpenTag(string tagname)\r
{\r
int i;\r
\r
}\r
\r
setbit(par_aux,npar,OP); // marks a new opening parenthesis\r
+ \r
+ TagIdMapIT tag_id = tIdMap->find(tagname);\r
+ \r
+ if (tag_id == tIdMap->end()){\r
+ REGISTER_TAG(TagName,tIdMap,tagname);\r
+ i = TagName->size() - 1;\r
+ }\r
+ else\r
+ i = tag_id->second;\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
+ if (tagname.compare(PCDATA_OPEN_TAG) == 0 ||\r
+ tagname.compare(ATTRIBUTE_DATA_OPEN_TAG) == 0){\r
+ };\r
+ \r
tags_aux = (TagType *) urealloc(tags_aux, sizeof(TagType)*(npar + 1));\r
-\r
+ \r
tags_aux[npar] = i; // inserts the new tag id within the preorder sequence of tags\r
\r
npar++;\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 XMLTreeBuilder::NewClosingTag(unsigned char *tagname)\r
+int XMLTreeBuilder::NewClosingTag(string tagname)\r
{\r
int i;\r
\r
}\r
\r
setbit(par_aux,npar,CP); // marks a new closing parenthesis\r
+ \r
+ tagname.insert(0,"/");\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
+ TagIdMapIT tag_id = tIdMap->find(tagname); \r
+\r
+ if (tag_id == tIdMap->end()){\r
+ REGISTER_TAG(TagName,tIdMap,tagname);\r
+ i = TagName->size() - 1;\r
+ }\r
+ else\r
+ i = tag_id->second;\r
\r
tags_aux = (TagType *)urealloc(tags_aux, sizeof(TagType)*(npar + 1));\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 XMLTreeBuilder::NewText(unsigned char *s)\r
+int XMLTreeBuilder::NewText(string text)\r
{\r
- if (disable_tc) {\r
- XMLTreeBuilder::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
+ if (!disable_tc) {\r
+ if (text.empty())\r
+ TextBuilder->InsertText((uchar *)"\001");\r
+ else\r
+ TextBuilder->InsertText((uchar *) text.c_str());\r
+ };\r
+\r
+ CachedText->push_back(text); \r
+ int n_eta_size = sizeof(uint)*(1+(npar-1)/(8*sizeof(uint)));\r
+ //see basics.h, recalloc resizes and sets the new area to 0.\r
+ \r
+ empty_texts_aux = (uint *)urecalloc(empty_texts_aux,eta_size,n_eta_size);\r
+ eta_size = n_eta_size;\r
+ bitset(empty_texts_aux, npar-1); // marks the non-empty text with a 1 in the bit vector\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 XMLTreeBuilder::NewEmptyText() \r
- {\r
- unsigned char c = 0;\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