+#include "basics.h"\r
#include "XMLTree.h"\r
-#include <cstring>\r
+#include "timings.h"\r
+#include <errno.h>\r
+using std::cout;\r
+using std::endl;\r
+using std::min;\r
+using std::string;\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
// Current implementation corresponds to balanced-parentheses representation for\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
+static int bits8 (int t ) {\r
+ int r = bits(t);\r
+ if (r <= 8)\r
+ return 8;\r
+ else if (r <= 16)\r
+ return 16;\r
+ else \r
+ return r;\r
}\r
\r
-// tree node -> tag position\r
-inline int node2tagpos(treeNode x) {\r
- return (int)x;\r
+\r
+\r
+static treeNode fast_sibling(bp* Par,treeNode x,TagType tag){\r
+\r
+ if (tag == PCDATA_TAG_ID){\r
+ x = x+2;\r
+ return fast_inspect(Par,x)==OP ? x : NULLT;\r
+ } else return fast_next_sibling(Par,x);\r
+\r
}\r
\r
\r
-//KIM OJO to prevent suprious "unused result" warnings\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
+inline uint get_field_no_power(uint *A, uint len, uint index) {\r
+ \r
+ register uint i=index*len/W, j=index*len-W*i;\r
+ return (j+len <= W) ? (A[i] << (W-j-len)) >> (W-len) : (A[i] >> j) | (A[i+1] << (WW-j-len)) >> (W-len);\r
+\r
}\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
+static uint fast_get_field(uint* A,int len, int idx)\r
+{\r
+ uint f1, f2;\r
+ switch (len) {\r
+ case 8:\r
+ return (uint) (((uchar*)A)[idx]);\r
+ case 16:\r
+ f2 = ((unsigned short*)A)[idx];\r
+ f1 = ((unsigned short*)A)[idx+1];\r
+ return (f1 << 16) + f2;\r
+ default:\r
+ return get_field_no_power (A,len,idx);\r
+ };\r
+\r
}\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
\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
\r
-}\r
\r
-inline void * ucalloc(size_t nmemb, size_t size){\r
+XMLTree::XMLTree( pb * const par, uint npar, vector<string> * const TN, TagIdMap * const tim, \r
+ uint *empty_texts_bmp, TagType *tags,\r
+ TextCollection * const TC, bool dis_tc,\r
+ TextCollectionBuilder::index_type_t _index_type )\r
+ {\r
+ buffer = 0;\r
+ print_stack = 0;\r
+ // creates the data structure for the tree topology\r
+ Par = (bp *)umalloc(sizeof(bp));\r
+ STARTTIMER();\r
+ bp_construct(Par, npar, (pb*) par, OPT_DEGREE|0);\r
+ STOPTIMER(Building);\r
+ PRINTTIME("Building parenthesis struct", Building);\r
+ STARTTIMER();\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
+ \r
+ // creates structure for tags\r
\r
-}\r
+ TagName = (vector<string>*)TN;\r
+ tIdMap = (TagIdMap *) tim;\r
+ \r
+ uint max_tag = TN->size() - 1;\r
+ \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
+ \r
+ //cout << "Tags test: " << Tags->test((uint*)tags,npar) << endl;\r
+\r
+ //Ensures that for small tag numbers, we are on an 8bit boundary.\r
+ //Makes tag access way faster with negligeable waste of space.\r
+ tags_blen = bits8(max_tag);\r
+ std::cerr << "Tags blen is " << tags_blen << "\n";\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
+ delete bmb; \r
+ free(tags);\r
+ tags = NULL;\r
+\r
+ STOPTIMER(Building);\r
+ PRINTTIME("Building Tag Structure", Building);\r
+ \r
+ Text = (TextCollection*) TC;\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
+ EBVector = new static_bitsequence_rrr02(empty_texts_bmp,npar,32);\r
+ //EBVector = new static_bitsequence_sdarray(empty_texts_bmp,npar);\r
+ free(empty_texts_bmp);\r
+ empty_texts_bmp = NULL;\r
\r
-// Save: saves XML tree data structure to file. \r
-void XMLTree::Save(unsigned char *filename) \r
+ \r
+ disable_tc = dis_tc;\r
+ text_index_type = _index_type; \r
+ std::cerr << "Number of distinct tags " << TagName->size() << "\n";\r
+ //std::cerr.flush();\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
+ delete tIdMap;\r
+ tIdMap = NULL;\r
+ \r
+ delete TagName;\r
+ TagName = NULL;\r
+ \r
+ delete Tags;\r
+ Tags = NULL;\r
+\r
+ delete Text; \r
+ Text = NULL;\r
+\r
+ delete EBVector;\r
+ EBVector = NULL;\r
+\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(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
- ufwrite(&initialized, sizeof(bool), 1, fp);\r
- ufwrite(&finished, 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
+ 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
- ufwrite(&st, sizeof(int),1,fp);\r
- for (int i = 0; i< CachedText.size(); i++){\r
- st = CachedText.at(i).size();\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(&text_index_type, sizeof(TextCollectionBuilder::index_type_t), 1, fp);\r
+ \r
+ const char * pref;\r
+ switch (text_index_type){\r
+ case TextCollectionBuilder::index_type_default:\r
+ pref = "default_";\r
+ break;\r
+ case TextCollectionBuilder::index_type_swcsa:\r
+ pref = "swcsa_";\r
+ break;\r
+ case TextCollectionBuilder::index_type_rlcsa:\r
+ pref = "rlcsa_";\r
+ break;\r
};\r
- };\r
- fclose(fp);\r
+ \r
+ Text->Save(fp, pref);\r
+ \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, bool load_tc,int sample_factor) \r
{\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
+ buffer[1023] = '\0';\r
+\r
+ fp = fdopen(fd, "r");\r
\r
+ XML_Tree = new XMLTree();\r
+ STARTTIMER();\r
+ // Load the tree structure\r
XML_Tree->Par = (bp *)umalloc(sizeof(bp));\r
\r
loadTree(XML_Tree->Par, fp); \r
-\r
- s_tree += sizeof(bp);\r
-\r
- // stores the table with tag names\r
- ufread(&XML_Tree->ntagnames, sizeof(int), 1, fp);\r
- \r
- s_tree += sizeof(int);\r
-\r
- XML_Tree->TagName = (unsigned char **)umalloc(XML_Tree->ntagnames*sizeof(unsigned char *));\r
+ STOPTIMER(Loading);\r
+ PRINTTIME("Loading parenthesis struct", Loading);\r
+ STARTTIMER();\r
+\r
+ XML_Tree->TagName = new std::vector<std::string>();\r
+ XML_Tree->tIdMap = new std::unordered_map<std::string,int>();\r
+ std::string s;\r
+ int ntags;\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
+ // Load the tag names\r
+ ufread(&ntags, sizeof(int), 1, 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
+ for (i=0; i<ntags;i++) {\r
+ if (fgets(buffer,1022,fp) != buffer)\r
throw "Cannot read tag list";\r
+ s = 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
+ STOPTIMER(Loading);\r
+ PRINTTIME("Loading tag names struct", Loading);\r
+ STARTTIMER();\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
- // 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
- ufread(&(XML_Tree->disable_tc), sizeof(bool), 1, fp);\r
+ // loads the tag structure\r
+ XML_Tree->Tags = static_sequence::load(fp);\r
+ ufread(&XML_Tree->tags_blen,sizeof(uint),1,fp);\r
+ std::cerr << "tags_blen is "<< XML_Tree->tags_blen <<"\n"; \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
+\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
+ //XML_Tree->Tags->test(seq,XML_Tree->tags_len);\r
+ //delete [] seq;\r
+ /// End ugly tests\r
\r
- s_tree+=sizeof(bool)*4;\r
+ STOPTIMER(Loading);\r
+ std::cerr << (uint_len(XML_Tree->tags_blen,XML_Tree->tags_len)*sizeof(uint))/(1024*1024) << " MB for tag sequence" << std::endl;\r
+ PRINTTIME("Loading tag struct", Loading);\r
+ STARTTIMER();\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
+ // loads the flags\r
\r
- // loads the tags\r
- XML_Tree->Tags = static_sequence::load(fp);\r
- ufread(&XML_Tree->tags_blen,sizeof(uint),1,fp);\r
- ufread(&XML_Tree->tags_len,sizeof(uint),1,fp);\r
- XML_Tree->tags_fix = new uint[uint_len(XML_Tree->tags_blen,XML_Tree->tags_len)];\r
- ufread(XML_Tree->tags_fix,sizeof(uint),uint_len(XML_Tree->tags_blen,XML_Tree->tags_len),fp);\r
- s_tree+=2*sizeof(uint)+sizeof(uint)*uint_len(XML_Tree->tags_blen,XML_Tree->tags_len);\r
- s_tree+= XML_Tree->Tags->size();\r
-\r
- /// FIXME:UGLY tests!\r
- /*uint * seq = new uint[XML_Tree->tags_len];\r
- for(uint i=0;i<XML_Tree->tags_len;i++)\r
- seq[i] = get_field(XML_Tree->tags_fix,XML_Tree->tags_blen,i);\r
- cout << "Tags test: " << XML_Tree->Tags->test(seq,XML_Tree->tags_len) << endl;\r
- delete [] seq;*/\r
- /// End ugly tests\r
-\r
- s_text = ftell(fp);\r
-\r
- // loads the texts\r
- if (!XML_Tree->disable_tc){\r
- XML_Tree->Text = TextCollection::Load(fp,sample_rate_text);\r
- int sst;\r
- int st;\r
- ufread(&sst, sizeof(int),1,fp);\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
- free(str);\r
- };\r
+ ufread(&(XML_Tree->disable_tc), sizeof(bool), 1, fp);\r
+ if (load_tc) {\r
+ XML_Tree->EBVector = static_bitsequence_rrr02::load(fp);\r
\r
+ STOPTIMER(Loading);\r
+ PRINTTIME("Loading text bitvector struct", Loading);\r
+ STARTTIMER();\r
+ \r
+ // Not used \r
+ // loads the texts\r
+ if (!XML_Tree->disable_tc){\r
+ ufread(&(XML_Tree->text_index_type),\r
+ sizeof(TextCollectionBuilder::index_type_t), 1, fp);\r
+ const char * pref;\r
+ switch (!XML_Tree->text_index_type){\r
+ case TextCollectionBuilder::index_type_default:\r
+ pref = "default_";\r
+ break;\r
+ case TextCollectionBuilder::index_type_swcsa:\r
+ pref = "swcsa_";\r
+ break;\r
+ case TextCollectionBuilder::index_type_rlcsa:\r
+ pref = "rlcsa_";\r
+ break;\r
+ }; \r
+ XML_Tree->Text = TextCollection::Load(fp, pref, TextCollection::index_mode_default, sample_factor);\r
+\r
+ }\r
+ else XML_Tree->Text = NULL;\r
+ STOPTIMER(Loading);\r
+ PRINTTIME("Loading TextCollection", Loading);\r
+ STARTTIMER(); \r
}\r
else {\r
+ XML_Tree->EBVector = NULL;\r
XML_Tree->Text = NULL;\r
- }\r
- s_text = ftell(fp) - s_text;\r
-\r
+ XML_Tree->disable_tc = true;\r
+ };\r
\r
\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
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
+/*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
+*/\r
// SubtreeTags(x,tag): the number of occurrences of tag within the subtree of node x.\r
+/*\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
+ x = fast_first_child(Par,x);\r
\r
\r
int s = x + 2*subtree_size(Par, x) - 1;\r
\r
- return Tags->rank(tag, s) - Tags->rank(tag, node2tagpos(x)-1);\r
+ return (Tags->rank(tag, s) - Tags->rank(tag, node2tagpos(x)-1))+1;\r
+ }\r
+*/\r
+int XMLTree::SubtreeElements(treeNode x) \r
+ {\r
+ \r
+ int size = subtree_size(Par,x);\r
+ if (x == Root()){\r
+ x = fast_first_child(Par,x);\r
+ size = size - 1;\r
+ };\r
+\r
+ int s = x + 2*size - 1;\r
+ int ntext = Tags->rank(PCDATA_TAG_ID, s) - Tags->rank(PCDATA_TAG_ID, node2tagpos(x)-1);\r
+ size = size - ntext;\r
+ treeNode fin = fast_find_close(Par,x);\r
+ treeNode y = Tags->select_next(ATTRIBUTE_TAG_ID,node2tagpos(x));\r
+ while (y != NULLT && y < fin){\r
+ size -= SubtreeSize(y);\r
+ y = Tags->select_next(ATTRIBUTE_TAG_ID,node2tagpos(y));\r
+ };\r
+ return size; \r
}\r
\r
// IsLeaf(x): returns whether node x is leaf or not. In the succinct representation\r
// this is just a bit inspection.\r
bool XMLTree::IsLeaf(treeNode x) \r
{\r
- return isleaf(Par, x);\r
+ NULLT_IF(x==NULLT);\r
+ return fast_isleaf(Par, x);\r
} \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
+ return fast_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
+ if (!fast_is_ancestor(Par, x, y)) return false;\r
return depth(Par, x) == (depth(Par, y) + 1);\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
+/*\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
+ return fast_get_field(tags_fix,tags_blen,node2tagpos(x));\r
}\r
-\r
+*/\r
// DocIds(x): returns the range of text identifiers that descend from node x.\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
- \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
- 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
- }\r
- return r;\r
+ range r;\r
+ if (x == NULLT) {\r
+ r.min = NULLT;\r
+ r.max = NULLT;\r
+ return r;\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 { // 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
// Parent(x): returns the parent node of node x.\r
+/*\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
- return parent(Par, x);\r
- }\r
+ return parent(Par, x);;\r
+ }*/\r
\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
+}\r
\r
// FirstChild(x): returns the first child of node x, assuming it exists. Very fast in BP.\r
+/*\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
+ NULLT_IF(x==NULLT);\r
+ return fast_first_child(Par, x);\r
}\r
-\r
+*/\r
+/*\r
+treeNode XMLTree::FirstElement(treeNode x) \r
+ {\r
+ NULLT_IF(x==NULLT);\r
+ x = fast_first_child(Par, x);\r
+ NULLT_IF(x == NULLT);\r
+ switch (Tag(x)){\r
+ \r
+ case PCDATA_TAG_ID:\r
+ x = x+2;\r
+ return (fast_inspect(Par,x)==OP)? x : NULLT;\r
+ \r
+ case ATTRIBUTE_TAG_ID: \r
+ x = fast_next_sibling(Par,x);\r
+ if (x != NULLT && Tag(x) == PCDATA_TAG_ID){\r
+ x = x+2;\r
+ return (fast_inspect(Par,x)==OP)? x : NULLT;\r
+ } \r
+ else return x; \r
+ default:\r
+ return x;\r
+ }\r
+ }\r
+*//*\r
+treeNode XMLTree::NextElement(treeNode x) \r
+{\r
+ NULLT_IF(x==NULLT);\r
+ x = fast_next_sibling(Par, x);\r
+ NULLT_IF(x == NULLT); \r
+ if (Tag(x) == PCDATA_TAG_ID){\r
+ x = x+2;\r
+ return (fast_inspect(Par,x)==OP)? x : NULLT;\r
+ }\r
+ else return x; \r
+ }*/\r
+\r
+// LastChild(x): returns the last child of node x.\r
+ /*treeNode XMLTree::LastChild(treeNode x)\r
+ {\r
+ NULLT_IF(x == NULLT || fast_isleaf(Par,x));\r
+ return find_open(Par, fast_find_close(Par, x)-1);\r
+ }\r
+ */\r
// NextSibling(x): returns the next sibling of node x, assuming it exists.\r
-treeNode XMLTree::NextSibling(treeNode x) \r
+/*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
- return next_sibling(Par, x);\r
+ NULLT_IF(x==NULLT || x == Root() );\r
+ x = fast_find_close(Par,x)+1;\r
+ return (fast_inspect(Par,x) == CP ? NULLT : x);\r
}\r
+*/\r
\r
// PrevSibling(x): returns the previous sibling of node x, assuming it exists.\r
-treeNode XMLTree::PrevSibling(treeNode x) \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
+ NULLT_IF(x==NULLT);\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
+*/\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
+/*\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
- }\r
- return NULLT; // no such child was found \r
+ NULLT_IF(x==NULLT || fast_isleaf(Par,x));\r
+ treeNode child; \r
+ child = fast_first_child(Par, x); // starts at first child of node x\r
+ if (Tag(child) == tag)\r
+ return child;\r
+ else\r
+ return TaggedFollowingSibling(child,tag);\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
- r = (int) Tags->rank(tag, node2tagpos(x));\r
- s = (int) Tags->select(tag, r+1);\r
- if (s == -1) return NULLT; // there is no such node\r
- y = tagpos2node(s); // transforms the tag position into a node position\r
- if (!is_ancestor(Par, x, y)) return NULLT; // the next node tagged tag (in preorder) is not within the subtree of x.\r
- else return y;\r
- }\r
-\r
-treeNode XMLTree::TaggedDescOnly(treeNode x,TagType *desctags, unsigned int dtlen)\r
+// TaggedSibling(x,tag): returns the first sibling of node x tagged tag, or NULLT if there is none.\r
+treeNode XMLTree::TaggedFollowingSibling(treeNode x, TagType tag)\r
{\r
-\r
- treeNode res,y;\r
- if (isleaf(Par,x))\r
- return NULLT;\r
- \r
- res=NULLT;\r
- for (unsigned int i = 0; i < dtlen; i ++ )\r
- {\r
- y = TaggedDesc(x,desctags[i]);\r
- res = (res == NULLT) || (( res != NULLT) && (y =! NULLT) && y < res) ? y : res;\r
- \r
- };\r
- \r
- return res;\r
- \r
+ NULLT_IF(x==NULLT);\r
+ treeNode sibling = fast_next_sibling(Par, x);\r
+ TagType ctag;\r
+ while (sibling != NULLT) {\r
+ ctag = Tag(sibling);\r
+ if (ctag == tag) // current sibling is labeled with tag of interest\r
+ return sibling; \r
+ sibling = fast_sibling(Par, sibling, ctag); // OK, let's try with the next sibling\r
+ }\r
+ return NULLT; // no such sibling was found \r
}\r
-\r
-\r
-treeNode XMLTree::TaggedBelow(treeNode x, TagType *childtags, unsigned int ctlen,\r
- TagType *desctags, unsigned int dtlen)\r
+*/\r
+treeNode XMLTree::SelectChild(treeNode x, TagIdSet *tags)\r
{\r
- treeNode fs,y,res;\r
- TagType tag;\r
-\r
- if (isleaf(Par,x))\r
- return NULLT;\r
\r
- res = NULLT;\r
- fs = first_child(Par,x);\r
- while (fs != NULLT) {\r
- tag = get_field(tags_fix,tags_blen,node2tagpos(fs));\r
- \r
- /* Check for first_child */\r
- for (unsigned int i = 0; i < ctlen; i++) {\r
- if (childtags[i] == tag)\r
- return fs;\r
- };\r
- \r
- for (unsigned int i = 0; i < dtlen; i++)\r
- if (desctags[i] == tag)\r
- return fs; \r
- \r
- /* check in the descendants */\r
- res = NULLT;\r
- for (unsigned int i = 0; i < dtlen; i ++ ){\r
- /* maybe inline by hand */\r
- y = TaggedDesc(fs,desctags[i]);\r
- res = (res==NULLT || (y != NULLT) &&(y < res)) ? y : res; \r
- };\r
- if (res != NULLT)\r
- return res;\r
- \r
- fs = next_sibling(Par,fs);\r
- };\r
- return res;\r
- \r
+ NULLT_IF(x==NULLT || fast_isleaf(Par,x));\r
+ int i;\r
+ treeNode child = fast_first_child(Par, x); \r
+ TagType t;\r
+ while (child != NULLT) {\r
+ t = Tag(child);\r
+ if (tags->find(t) != tags->end()) return child;\r
+ child = fast_sibling(Par, child,t);\r
+ }\r
+ return NULLT; \r
}\r
-treeNode XMLTree::TaggedFollOnly(treeNode x,TagType *folltags, unsigned int ftlen,treeNode root)\r
-{\r
\r
- treeNode res,y,lim;\r
- lim = find_close(Par,root); \r
- res=NULLT;\r
- for (unsigned int i = 0; i < ftlen; i ++ )\r
- {\r
- y = TaggedFoll(x,folltags[i]);\r
- res = (res == NULLT) || (( res != NULLT) && (y =! NULLT) && y < res) ? y : res;\r
- \r
- };\r
- \r
- return res < lim ? res : NULLT;\r
- \r
-}\r
\r
-treeNode XMLTree::TaggedDescOrFollOnly(treeNode x,TagType *folltags, unsigned int ftlen,treeNode root)\r
+treeNode XMLTree::SelectFollowingSibling(treeNode x, TagIdSet *tags)\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
- r = (int) Tags->rank(folltags[i], node2tagpos(x));\r
- s = (int) Tags->select(folltags[i], r+1);\r
- if (s == -1) \r
- y = NULLT; // there is no such node\r
- else {\r
- y = tagpos2node(s); \r
- if (y >= lim)\r
- y = NULLT;\r
- };\r
- res = (res == NULLT) || (( res != NULLT) && (y =! NULLT) && y < res) ? y : res;\r
- \r
- };\r
- \r
- return res < lim ? res : NULLT;\r
- \r
-}\r
+ NULLT_IF(x==NULLT);\r
+ int i;\r
+ TagType t;\r
+ treeNode sibling = fast_next_sibling(Par, x);\r
+ while (sibling != NULLT) {\r
+ t = Tag(sibling);\r
+ if (tags->find(t) != tags->end()) return sibling;\r
+ sibling = fast_sibling(Par, sibling,t);\r
+ }\r
+ return 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
+// TaggedDescendant(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
+/*\r
+treeNode XMLTree::TaggedDescendant(treeNode x, TagType tag) \r
{\r
- treeNode y,old_y,lim,res;\r
- TagType tag;\r
- if (x == NULLT || x == Root())\r
- return NULLT;\r
+ //NULLT_IF(x==NULLT || fast_isleaf(Par,x));\r
\r
+ int s = (int) Tags->select_next(tag,node2tagpos(x));\r
+ NULLT_IF (s == -1);\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
- };\r
- \r
- return res < lim ? res : NULLT;\r
+ treeNode y = tagpos2node(s); // transforms the tag position into a node position\r
\r
+ return (fast_is_ancestor(Par,x,y) ? y : NULLT);\r
+ }\r
+*/\r
+/*\r
+treeNode XMLTree::SelectDescendant(treeNode x, TagIdSet *tags)\r
+ {\r
+ NULLT_IF (x ==NULLT || fast_isleaf(Par,x));\r
+ int i;\r
+ treeNode min = NULLT;\r
+ treeNode fc = fast_first_child(Par,x);\r
+ treeNode aux;\r
+ TagIdSet::const_iterator tagit;\r
+ for (tagit = tags->begin(); tagit != tags->end(); tagit++) {\r
+ aux = TaggedDescendant(x, (TagType) *tagit);\r
+ if (aux == fc) return fc;\r
+ if (aux == NULLT) continue;\r
+ if ((min == NULLT) || (aux < min)) min = aux;\r
+ };\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
- {\r
- if (!finished) {\r
- fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
- exit(1);\r
- }\r
- \r
+treeNode XMLTree::TaggedPreceding(treeNode x, TagType tag) \r
+ { \r
int r, s;\r
treeNode node_s, root;\r
r = (int)Tags->rank(tag, node2tagpos(x)-1);\r
s = (int)Tags->select(tag, r);\r
root = root_node(Par);\r
node_s = tagpos2node(s);\r
- while (is_ancestor(Par, node_s, x) && (node_s!=root)) { // the one that we found is an ancestor of x\r
+ while (fast_is_ancestor(Par, node_s, x) && (node_s!=root)) { // the one that we found is an ancestor of x\r
r--;\r
if (r==0) return NULLT; // there is no such node\r
s = (int)Tags->select(tag, r); // we should use select_prev instead when provided\r
\r
// TaggedFoll(x,tag): returns the first node tagged tag with larger preorder than x and not in\r
// the subtree of x. Returns NULLT if there is none.\r
-treeNode XMLTree::TaggedFoll(treeNode x, TagType tag)\r
+treeNode XMLTree::TaggedFollowing(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
+ NULLT_IF (x ==NULLT || x == Root()); \r
+ return tagpos2node(Tags->select_next(tag,fast_find_close(Par, x)));\r
\r
- int r, s;\r
- if (x ==NULLT || x == Root())\r
- return NULLT;\r
- \r
- r = (int) Tags->rank(tag, find_close(Par, x));\r
- s = (int) Tags->select(tag, r+1); // select returns -1 in case that there is no r+1-th tag.\r
- if (s==-1) return NULLT;\r
- else return tagpos2node(s);\r
} \r
\r
-// TaggedFoll(x,tag): returns the first node tagged tag with larger preorder than x and not in\r
-// the subtree of x. Returns NULLT if there is none.\r
-treeNode XMLTree::TaggedFollBelow(treeNode x, TagType tag, treeNode root)\r
+// 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
+/*\r
+treeNode XMLTree::TaggedFollowingBelow(treeNode x, TagType tag, treeNode ancestor)\r
+{\r
+ // NULLT_IF (x == NULLT || x == Root() || x == ancestor); \r
+\r
+ //Special optimisation, test for the following sibling first\r
+ treeNode close = fast_find_close(Par, x);\r
+ treeNode s = tagpos2node(Tags->select_next(tag, close));\r
+ \r
+ if (ancestor == Root() || s==NULLT || s < fast_find_close(Par,ancestor)) return s;\r
+ else return NULLT;\r
+} \r
+*/\r
+ /*\r
+treeNode XMLTree::TaggedFollowingBefore(treeNode x, TagType tag, treeNode closing)\r
+{\r
+\r
+ NULLT_IF (x == NULLT || x == Root());\r
+ \r
+ treeNode s = tagpos2node(Tags->select_next(tag, fast_find_close(Par, x))); \r
+ NULLT_IF (s == NULLT || s >= closing);\r
+ \r
+ return s;\r
+} \r
+ */\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::SelectFollowingBelow(treeNode x, TagIdSet *tags, treeNode ancestor)\r
{\r
\r
- int r, s;\r
- int lim = node2tagpos(find_close(Par,root));\r
- if (x ==NULLT || x == Root())\r
- return NULLT;\r
- \r
- r = (int) Tags->rank(tag,find_close(Par,x));\r
- s = (int) Tags->select(tag, r+1); // select returns -1 in case that there is no r+1-th tag.\r
- if (s==-1 || s >= lim) \r
- return NULLT;\r
- else \r
- return tagpos2node(s);\r
- } \r
+ NULLT_IF(x==NULLT || x==Root());\r
\r
+ treeNode close = fast_find_close(Par,x);\r
+ treeNode ns = close+1;\r
+ if ( (fast_inspect(Par,ns) == OP) && (tags->find(Tag(ns)) != tags->end()))\r
+ return ns;\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
+ int i;\r
+ treeNode min = NULLT;\r
+ treeNode aux;\r
+ \r
+\r
+ TagIdSet::const_iterator tagit;\r
+ for (tagit = tags->begin(); tagit != tags->end(); ++tagit) {\r
+\r
+ aux = tagpos2node(Tags->select_next(*tagit, close));\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 (ancestor == Root() || min == NULLT || min < fast_find_close(Par, ancestor)) return min;\r
+ else return NULLT;\r
+ \r
+ }\r
+/*\r
+treeNode XMLTree::SelectFollowingBefore(treeNode x, TagIdSet *tags, treeNode ancestor_closing)\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
+ NULLT_IF(x==NULLT || x==Root());\r
\r
- if (x == NULLT || x == Root() || ns == -1)\r
- return NULLT;\r
+ treeNode close = fast_find_close(Par,x);\r
+ treeNode ns = close+1;\r
+ if ( (fast_inspect(Par,ns) == OP) && (tags->find(Tag(ns)) != tags->end()))\r
+ return ns;\r
+\r
+ int i;\r
+ treeNode min = NULLT;\r
+ treeNode aux;\r
+ \r
\r
- r = (int) Tags->rank(tag, node2tagpos(ns)-1);\r
- s = (int) Tags->select(tag, r+1); // select returns -1 in case that there is no r+1-th tag.\r
- if (s==-1) return NULLT;\r
- else return tagpos2node(s);\r
+ TagIdSet::const_iterator tagit;\r
+ for (tagit = tags->begin(); tagit != tags->end(); ++tagit) {\r
+\r
+ aux = tagpos2node(Tags->select_next(*tagit, close));\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 (ancestor_closing == Root() || min == NULLT || min < ancestor_closing) return min;\r
+ else return NULLT;\r
+ \r
}\r
+*/\r
+/*\r
+treeNode XMLTree::SelectFollowingBefore(treeNode x, TagIdSet *tags, treeNode closing)\r
+ {\r
+\r
+ NULLT_IF(x==NULLT || x==Root());\r
+ int i;\r
+ treeNode min = NULLT;\r
+ treeNode ns = fast_next_sibling(Par, x);\r
+ treeNode close = ns - 1;\r
+ treeNode aux;\r
+ TagIdSet::const_iterator tagit;\r
+ for (tagit = tags->begin(); tagit != tags->end(); tagit++) {\r
+\r
+ aux = tagpos2node(Tags->select_next(*tagit, close));\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
+ NULLT_IF (min == NULLT || min >= closing);\r
+ \r
+ return min;\r
+ \r
+ }\r
+*/\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
treeNode s = parent(Par, x), r = Root();\r
while (s != r) {\r
- if (get_field(tags_fix,tags_blen,node2tagpos(s)) /*Tags->access(node2tagpos(s))*/ == tag) return s;\r
+ if (Tag(s) == tag) return s;\r
s = parent(Par, s);\r
}\r
return NULLT;\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 (!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
- 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 (!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
- 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 (!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
- 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); \r
+ else \r
+ return (DocID) NULLT;\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::MyTextUnsafe(treeNode x) \r
+{\r
+ return (DocID) (EBVector->rank1(x)-1); //-1 because document ids start from 0\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 (!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
- int s = EBVector->select1(d+1);\r
- return rank_open(Par, s) + d + 1; // +1 because root has preorder 1\r
- }\r
+ NULLT_IF(d == NULLT);\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 (!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
- if (x == Root()) return 1; // root node has preorder 1\r
- else\r
- return rank_open(Par, x) + EBVector->rank1(x-1);\r
- }\r
+ NULLT_IF(x == NULLT);\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 (!finished) {\r
- fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
- exit(1);\r
- }\r
- \r
- if (d == NULLT)\r
- return NULLT;\r
- \r
- int s;\r
- // 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 (!finished) {\r
- fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
- exit(1);\r
- }\r
- \r
- if (d == NULLT)\r
- return NULLT;\r
- \r
- int s;\r
- \r
- if (indexing_empty_texts) s = d;\r
- else s = EBVector->select1(d+1);\r
- if (s == -1)\r
- return NULLT;\r
- \r
- if (inspect(Par,s) == CP) // is a closing parenthesis\r
- return find_open(Par, s);\r
- else // is an opening parenthesis\r
- return NULLT;\r
- \r
- }\r
-\r
-\r
-// OpenDocument(empty_texts): it starts the construction of the data structure for\r
-// 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
+ NULLT_IF (d == NULLT); \r
+ return (treeNode) EBVector->select1(d+1); \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
+// 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
- 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
+ string s = (char *) tagname;\r
+ TagIdMapIT it = tIdMap->find(s); \r
+ return (TagType) ((it != tIdMap->end()) ? it->second : -1);\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
+// GetTagName(tagid): returns the tag name of a given tag identifier.\r
+// Returns NULL in case that the tag identifier is not valid.\r
+unsigned char *XMLTree::GetTagName(TagType tagid)\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
+ unsigned char *s;\r
+ if ( tagid < 0 || tagid >= TagName->size())\r
+ return (unsigned char *) "<INVALID TAG>";\r
+ strcpy((char *)s, (*TagName)[tagid].c_str());\r
\r
+ return (s == NULL ? (unsigned char*) "<INVALID TAG>" : s);\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
+const unsigned char *XMLTree::GetTagNameByRef(TagType tagid)\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
+ unsigned char *s;\r
+ if ( tagid < 0 || tagid >= TagName->size())\r
+ return (unsigned char *) "<INVALID TAG>";\r
+ \r
+ return (const unsigned char *) (*TagName)[tagid].c_str();\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
+TagType XMLTree::RegisterTag(unsigned char *tagname)\r
+ { \r
+ TagType id = XMLTree::GetTagId(tagname);\r
+ if (id == NULLT) {\r
+ string s = (char *) tagname; \r
+ REGISTER_TAG(TagName,tIdMap,s); \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
+ return id;\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
-\r
-\r
-// GetTagName(tagid): returns the tag name of a given tag identifier.\r
-// Returns NULL in case that the tag identifier is not valid.\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
- }\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 >= ntagnames) return NULL; // invalid tag identifier\r
- return ((const unsigned char*) TagName[tagid]);\r
- }\r
+treeNode XMLTree::Closing(treeNode x) {\r
+ return fast_find_close(Par,x); \r
+}\r
+bool XMLTree::IsOpen(treeNode x) { return fast_inspect(Par,x); }\r
\r
+//WARNING this uses directly the underlying implementation for plain text\r
\r
\r
-TagType XMLTree::RegisterTag(unsigned char *tagname)\r
-{\r
- if (!finished)\r
- return NULLT;\r
+void XMLTree::Print(int fd,treeNode x, bool no_text){\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
+ if (buffer == 0) { \r
+ buffer = new string(BUFFER_ALLOC, 0);\r
+ print_stack = new std::vector<string *>();\r
+ print_stack->reserve(256);\r
};\r
+ treeNode fin = fast_find_close(Par,x);\r
+ treeNode n = x;\r
+ TagType tag = Tag(n);\r
+ \r
+ range r = DocIds(x);\r
+ treeNode first_idx;\r
+ treeNode first_text = (tag == PCDATA_TAG_ID ? x : ParentNode(r.min-1));\r
+ treeNode first_att = NULLT;\r
+ \r
+ if (first_att == NULLT) \r
+ first_idx = first_text;\r
+ else if (first_text == NULLT)\r
+ first_idx = first_att;\r
+ else\r
+ first_idx = min(first_att,first_text);\r
+ \r
+ uchar * current_text=NULL;\r
\r
- return id;\r
+ if (first_idx != NULLT)\r
+ current_text = GetText(MyTextUnsafe(first_idx));\r
+ \r
+ size_t read = 0;\r
+ while (n <= fin){\r
+ if (fast_inspect(Par,n)){\r
+ if (tag == PCDATA_TAG_ID) { \r
+ \r
+ if (no_text)\r
+ _dputs("<$/>", fd);\r
+ else {\r
+ read = _dprintf((const char*) current_text, fd);\r
+ current_text += (read + 1);\r
+ };\r
+ n+=2; // skip closing $\r
+ tag = Tag(n);\r
+ \r
+ } else {\r
+\r
+ _dputc('<',fd); \r
+ _dput_str((*TagName)[tag], fd);\r
+ n++;\r
+ if (fast_inspect(Par,n)) {\r
+ print_stack->push_back(&((*TagName)[tag]));\r
+ tag = Tag(n);\r
+ if (tag == ATTRIBUTE_TAG_ID){\r
+ n++;\r
+ if (no_text) _dputs("><@@>",fd);\r
+\r
+ while (fast_inspect(Par,n)){\r
+ if (no_text) {\r
+ _dputc('<', fd);\r
+ _dputs((const char*) &(GetTagNameByRef(Tag(n))[3]), fd);\r
+ _dputc('>', fd);\r
+ _dputs("<$@/></", fd);\r
+ _dputs((const char*) &(GetTagNameByRef(Tag(n))[3]), fd);\r
+ _dputc('>', fd);\r
+ n+= 4;\r
+ } else {\r
+ _dputc(' ', fd);\r
+ _dputs((const char*) &(GetTagNameByRef(Tag(n))[3]), fd);\r
+ n++;\r
+ _dputs("=\"", fd);\r
+ read = _dprintf((const char*) current_text, fd);\r
+ current_text += (read + 1);\r
+ _dputc('"', fd);\r
+ n+=3;\r
+ }\r
+ };\r
+ if (no_text) _dputs("</@@>", fd);\r
+ else _dputc('>', fd);\r
+ n++;\r
+ tag=Tag(n);\r
+ \r
+ } else \r
+ _dputc('>', fd);\r
+ \r
+ } else {// <foo /> tag\r
+ _dputs("/>", fd);\r
+ n++;\r
+ tag=Tag(n); \r
+ }; \r
+ };\r
+ } else do {\r
+ _dputs("</", fd);\r
+ _dput_str(*(print_stack->back()), fd);\r
+ _dputc('>', fd);\r
+ print_stack->pop_back();\r
+ n++;\r
+ } while (!(fast_inspect(Par,n) || print_stack->empty()));\r
+ tag = Tag(n);\r
+ };\r
+ _dputc('\n', fd);\r
+ //_flush(fd);\r
}\r
projects / SXSI / XMLTree.git / blobdiff