#include "XMLTree.h"\r
#include <cstring>\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
// without affecting the code so much.\r
return (int)x;\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
+}\r
+\r
+inline void ufwrite(const void *ptr, size_t size, size_t nmemb, FILE *stream){\r
+ size_t res;\r
+ res = fwrite(ptr,size,nmemb,stream);\r
+ if (res < nmemb)\r
+ throw "ufwrite I/O error";\r
+ return;\r
+}\r
+\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
+\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
+}\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
+\r
// Save: saves XML tree data structure to file. \r
void XMLTree::Save(unsigned char *filename) \r
{\r
saveTree(Par, fp);\r
\r
// stores the table with tag names\r
- fwrite(&ntagnames, sizeof(int), 1, fp);\r
+ ufwrite(&ntagnames, sizeof(int), 1, fp);\r
for (i=0; i<ntagnames;i++)\r
- fprintf(fp, "%s\n",TagName[i]);\r
+ fprintf(fp, "%s\n",TagName[i]);\r
+ \r
\r
// stores the flags\r
- fwrite(&indexing_empty_texts, sizeof(bool), 1, fp);\r
- fwrite(&initialized, sizeof(bool), 1, fp);\r
- fwrite(&finished, sizeof(bool), 1, fp);\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
\r
// stores the tags\r
Tags->save(fp);\r
+ ufwrite(&tags_blen,sizeof(uint),1,fp);\r
+ ufwrite(&tags_len,sizeof(uint),1,fp);\r
+ ufwrite(tags_fix,sizeof(uint),uint_len(tags_blen,tags_len),fp);\r
\r
// stores the texts \r
- Text->Save(fp);\r
-\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
+ };\r
fclose(fp);\r
\r
}\r
{\r
\r
FILE *fp;\r
- char filenameaux[1024];\r
+ char buffer[1024];\r
XMLTree *XML_Tree;\r
int i;\r
- \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(filenameaux, "%s.srx", filename);\r
- fp = fopen(filenameaux, "r");\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", filenameaux);\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
- XML_Tree->Par = (bp *)malloc(sizeof(bp));\r
+ XML_Tree->Par = (bp *)umalloc(sizeof(bp));\r
\r
loadTree(XML_Tree->Par, fp); \r
- \r
+\r
+ s_tree += sizeof(bp);\r
+\r
// stores the table with tag names\r
- fread(&XML_Tree->ntagnames, sizeof(int), 1, fp);\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
+ \r
+ s_tags += sizeof(unsigned char*)*XML_Tree->ntagnames;\r
\r
- XML_Tree->TagName = (unsigned char **)malloc(XML_Tree->ntagnames*sizeof(unsigned char *));\r
\r
for (i=0; i<XML_Tree->ntagnames;i++) {\r
- int k = feof(fp);\r
- fscanf(fp, "%s\n",filenameaux);\r
- XML_Tree->TagName[i] = (unsigned char *)malloc(sizeof(unsigned char)*(strlen((const char *)filenameaux)+1));\r
- strcpy((char *)XML_Tree->TagName[i], (const char *)filenameaux);\r
+ \r
+ // OJO Kim is it needed ?\r
+ int k = feof(fp);\r
+\r
+ \r
+ // fscanf chokes on "\n" which is the case for the root element\r
+ char * r = fgets(buffer,1023,fp);\r
+ // int r = fscanf(fp, "%s\n",buffer);\r
+ if (r==NULL)\r
+ throw "Cannot read tag list";\r
+\r
+ // 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
- fread(&(XML_Tree->indexing_empty_texts), sizeof(bool), 1, fp);\r
- fread(&(XML_Tree->initialized), sizeof(bool), 1, fp);\r
- fread(&(XML_Tree->finished), sizeof(bool), 1, fp);\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
\r
- if (!(XML_Tree->indexing_empty_texts)) XML_Tree->EBVector = static_bitsequence_rrr02::load(fp);\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
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
+\r
+ }\r
+ else {\r
+ XML_Tree->Text = NULL;\r
+ }\r
+ s_text = ftell(fp) - s_text;\r
+\r
+ \r
\r
- // loads the texts \r
- XML_Tree->Text->Load(fp,sample_rate_text);\r
\r
fclose(fp);\r
- \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
Tags = NULL;\r
\r
//Text->~TextCollection();\r
- delete Text;\r
+ delete TextBuilder; \r
+ TextBuilder = NULL;\r
+ delete Text; \r
Text = NULL;\r
\r
initialized = false;\r
treeNode XMLTree::Root() \r
{\r
if (!finished) {\r
- fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
+ fprintf(stderr, "Root() : Error: data structure has not been constructed properly\n");\r
exit(1);\r
}\r
return root_node(Par);\r
fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
exit(1);\r
}\r
+ if (x == Root())\r
+ x = first_child(Par,x);\r
+ \r
\r
int s = x + 2*subtree_size(Par, x) - 1;\r
\r
fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
exit(1);\r
}\r
-\r
return postorder_rank(Par, x);\r
}\r
\r
fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
exit(1);\r
}\r
-\r
- return Tags->access(node2tagpos(x));\r
+ \r
+ return get_field(tags_fix,tags_blen,node2tagpos(x)); //Tags->access(node2tagpos(x));\r
}\r
\r
// DocIds(x): returns the range of text identifiers that descend from node x.\r
}\r
\r
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
fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
exit(1);\r
}\r
-\r
- return parent(Par, x);\r
+ if (x == Root())\r
+ return NULLT;\r
+ else\r
+ return parent(Par, x);\r
}\r
\r
// Child(x,i): returns the i-th child of node x, assuming it exists.\r
fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
exit(1);\r
}\r
-\r
+ if (x == Root() || x==NULLT)\r
+ return NULLT;\r
+ \r
return next_sibling(Par, x);\r
}\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 (Tags->access(node2tagpos(child)) == tag) { // current child is labeled with tag of interest\r
+ if (get_field(tags_fix,tags_blen,node2tagpos(child)) /*Tags->access(node2tagpos(child))*/ == tag) { // current child is labeled with tag of interest\r
i--;\r
if (i==0) return child; // we have seen i children of x tagged tag, this is the one we are looking for\r
}\r
\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
else return y;\r
}\r
\r
+treeNode XMLTree::TaggedDescOnly(treeNode x,TagType *desctags, unsigned int dtlen)\r
+{\r
+\r
+ treeNode res,y;\r
+ if (isleaf(Par,x))\r
+ return NULLT;\r
+ \r
+ res=NULLT;\r
+ for (unsigned int i = 0; i < dtlen; i ++ )\r
+ {\r
+ y = TaggedDesc(x,desctags[i]);\r
+ res = (res == NULLT) || (( res != NULLT) && (y =! NULLT) && y < res) ? y : res;\r
+ \r
+ };\r
+ \r
+ return res;\r
+ \r
+}\r
+\r
+\r
+treeNode XMLTree::TaggedBelow(treeNode x, TagType *childtags, unsigned int ctlen,\r
+ TagType *desctags, unsigned int dtlen)\r
+{\r
+ treeNode fs,y,res;\r
+ TagType tag;\r
+\r
+ if (isleaf(Par,x))\r
+ return NULLT;\r
+ \r
+ res = NULLT;\r
+ fs = first_child(Par,x);\r
+ while (fs != NULLT) {\r
+ tag = get_field(tags_fix,tags_blen,node2tagpos(fs));\r
+ \r
+ /* Check for first_child */\r
+ for (unsigned int i = 0; i < ctlen; i++) {\r
+ if (childtags[i] == tag)\r
+ return fs;\r
+ };\r
+ \r
+ for (unsigned int i = 0; i < dtlen; i++)\r
+ if (desctags[i] == tag)\r
+ return fs; \r
+ \r
+ /* check in the descendants */\r
+ res = NULLT;\r
+ for (unsigned int i = 0; i < dtlen; i ++ ){\r
+ /* maybe inline by hand */\r
+ y = TaggedDesc(fs,desctags[i]);\r
+ res = (res==NULLT || (y != NULLT) &&(y < res)) ? y : res; \r
+ };\r
+ if (res != NULLT)\r
+ return res;\r
+ \r
+ fs = next_sibling(Par,fs);\r
+ };\r
+ return res;\r
+ \r
+}\r
+treeNode XMLTree::TaggedFollOnly(treeNode x,TagType *folltags, unsigned int ftlen,treeNode root)\r
+{\r
+\r
+ treeNode res,y,lim;\r
+ lim = find_close(Par,root); \r
+ res=NULLT;\r
+ for (unsigned int i = 0; i < ftlen; i ++ )\r
+ {\r
+ y = TaggedFoll(x,folltags[i]);\r
+ res = (res == NULLT) || (( res != NULLT) && (y =! NULLT) && y < res) ? y : res;\r
+ \r
+ };\r
+ \r
+ return res < lim ? res : NULLT;\r
+ \r
+}\r
+\r
+treeNode XMLTree::TaggedDescOrFollOnly(treeNode x,TagType *folltags, unsigned int ftlen,treeNode root)\r
+{\r
+\r
+ treeNode res,y,lim;\r
+ int r,s;\r
+ lim = find_close(Par,root); \r
+ res=NULLT;\r
+ for (unsigned int i = 0; i < ftlen; i ++ )\r
+ {\r
+\r
+ r = (int) Tags->rank(folltags[i], node2tagpos(x));\r
+ s = (int) Tags->select(folltags[i], r+1);\r
+ if (s == -1) \r
+ y = NULLT; // there is no such node\r
+ else {\r
+ y = tagpos2node(s); \r
+ if (y >= lim)\r
+ y = NULLT;\r
+ };\r
+ res = (res == NULLT) || (( res != NULLT) && (y =! NULLT) && y < res) ? y : res;\r
+ \r
+ };\r
+ \r
+ return res < lim ? res : NULLT;\r
+ \r
+}\r
+\r
+\r
+// TaggedNext(x,tag): returns the first node tagged tag with larger preorder than x \r
+// Returns NULLT if there is none.\r
+treeNode XMLTree::TaggedNext(treeNode x, TagType *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
+\r
+ lim = find_close(Par,root); \r
+\r
+ res = NULLT;\r
+ \r
+ y = next_sibling(Par,x);\r
+ while (y != NULLT) {\r
+ tag = get_field(tags_fix,tags_blen,node2tagpos(y));\r
+ for(unsigned int i = 0; i < ctlen;i++)\r
+ if (childtags[i] == tag)\r
+ return y;\r
+ \r
+ for(unsigned int i = 0; i < flen;i++)\r
+ if (folltags[i] == tag)\r
+ return y;\r
+\r
+ res = TaggedBelow(y,NULL,0,folltags,flen);\r
+ if (res != NULLT)\r
+ return res;\r
+ \r
+ y = next_sibling(Par,y);\r
+ };\r
+ //Found nothing in the following sibling of x.\r
+ res = NULLT;\r
+ for(unsigned int i = 0; i < flen;i++){\r
+ y = TaggedFoll(x,folltags[i]);\r
+ res = (y!= x && (res == NULLT || (y != NULLT && y < res)))? y : res;\r
+ };\r
+ \r
+ return res < lim ? res : NULLT;\r
+ \r
+ }\r
+\r
+\r
// 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
return NULLT; // there is no such node \r
}\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::TaggedFoll(treeNode x, TagType tag) \r
+treeNode XMLTree::TaggedFoll(treeNode x, TagType tag)\r
{\r
if (!finished) {\r
fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
}\r
\r
int r, s;\r
- r = (int) Tags->rank(tag, node2tagpos(next_sibling(Par, x))-1);\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
+ {\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
+\r
+\r
+// TaggedFollowingSibling(x,tag): returns the first node tagged tag with larger preorder than x and not in\r
+// the subtree of x. Returns NULLT if there is none.\r
+treeNode XMLTree::TaggedFollowingSibling(treeNode x, TagType tag) \r
+ {\r
+ if (!finished) {\r
+ fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
+ exit(1);\r
+ }\r
+\r
+ int r, s;\r
+ treeNode ns = next_sibling(Par,x);\r
+\r
+ if (x == NULLT || x == Root() || ns == -1)\r
+ return NULLT;\r
+\r
+ r = (int) Tags->rank(tag, node2tagpos(ns)-1);\r
+ s = (int) Tags->select(tag, r+1); // select returns -1 in case that there is no r+1-th tag.\r
+ if (s==-1) return NULLT;\r
+ else return tagpos2node(s);\r
+ }\r
+\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
+ 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
+ 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
fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
exit(1);\r
}\r
-\r
+ \r
+ if (d == NULLT)\r
+ return NULLT;\r
+ \r
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);\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
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)\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 = 0; \r
- \r
+ ntagnames = 4; \r
+ disable_tc = dtc;\r
+ \r
indexing_empty_texts = empty_texts;\r
\r
- par_aux = (pb *)malloc(sizeof(pb)*parArraySize);\r
- if (!par_aux) {\r
- fprintf(stderr, "Error: not enough memory\n");\r
- return NULLT;\r
- }\r
+ par_aux = (pb *)umalloc(sizeof(pb)*parArraySize);\r
\r
- tags_aux = (TagType *) malloc(sizeof(TagType));\r
- if (!tags_aux) {\r
- fprintf(stderr, "Error: not enough memory\n");\r
- return NULLT;\r
- }\r
+ tags_aux = (TagType *) umalloc(sizeof(TagType));\r
\r
- TagName = NULL;\r
+ TagName = (unsigned char **) umalloc(4*sizeof(unsigned char*));\r
\r
- if (!indexing_empty_texts) {\r
- empty_texts_aux = (unsigned int *)malloc(sizeof(unsigned int));\r
- if (!empty_texts_aux) {\r
- fprintf(stderr, "Error: not enough memory\n");\r
- return NULLT;\r
- }\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], "<$>");\r
\r
- Text = TextCollection::InitTextCollection((unsigned)sample_rate_text);\r
+ //OJO need to put these in the table too.\r
+ TagName[2] = (unsigned char *) umalloc(5*sizeof(unsigned char));\r
+\r
+ strcpy((char *) TagName[2], "/<@>");\r
+\r
+ TagName[3] = (unsigned char *) umalloc(5*sizeof(unsigned char));\r
+\r
+ strcpy((char *) TagName[3], "/<$>");\r
+\r
+\r
+ if (!indexing_empty_texts) \r
+ empty_texts_aux = (unsigned int *)umalloc(sizeof(unsigned int));\r
+ \r
+ if (disable_tc)\r
+ TextBuilder = 0;\r
+ else \r
+ TextBuilder = new TextCollectionBuilder((unsigned)sample_rate_text);\r
+ Text = 0;\r
\r
return 1; // indicates success in the initialization of the data structure\r
}\r
}\r
\r
// closing parenthesis for the tree root\r
- par_aux = (pb *)realloc(par_aux, sizeof(pb)*(1+npar/(8*sizeof(pb))));\r
- if (!par_aux) {\r
- fprintf(stderr, "Error: not enough memory\n");\r
- return NULLT; \r
- }\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 *)malloc(sizeof(bp));\r
+ Par = (bp *)umalloc(sizeof(bp));\r
bp_construct(Par, npar, par_aux, OPT_DEGREE|0); \r
// creates structure for tags\r
- static_bitsequence_builder * bmb = new static_bitsequence_builder_brw32(20);\r
- static_permutation_builder * pmb = new static_permutation_builder_mrrr(PERM_SAMPLE, bmb);\r
- static_sequence_builder * ssb = new static_sequence_builder_gmr_chunk(bmb, pmb);\r
\r
- Tags = new static_sequence_gmr((uint *) tags_aux, (uint) npar-1,2*ntagnames, bmb, ssb);\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
- // makes the text collection static\r
- Text->MakeStatic();\r
+ //delete pmb;\r
+ //delete ssb;\r
+\r
\r
+ // makes the text collection static\r
+ if (!disable_tc)\r
+ {\r
+ assert(Text == 0);\r
+ assert(TextBuilder != 0);\r
+ Text = TextBuilder->InitTextCollection();\r
+ delete TextBuilder;\r
+ TextBuilder = 0;\r
+ }\r
+\r
// creates the data structure marking the non-empty texts (just in the case it is necessary)\r
- if (!indexing_empty_texts) \r
+ 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
// 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 *)realloc(par_aux, sizeof(pb)*2*parArraySize);\r
+ par_aux = (pb *)urealloc(par_aux, sizeof(pb)*2*parArraySize);\r
parArraySize *= 2;\r
}\r
\r
- if (!par_aux) {\r
- fprintf(stderr, "Error: not enough memory\n");\r
- return NULLT; \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
+ 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 **)realloc(TagName, sizeof(char *)*(ntagnames+1));\r
+ TagName = (unsigned char **)urealloc(TagName, sizeof(char *)*(ntagnames+1));\r
\r
if (!TagName) {\r
fprintf(stderr, "Error: not enough memory\n");\r
}\r
\r
ntagnames++;\r
- TagName[i] = (unsigned char *)malloc(sizeof(unsigned char)*(strlen((const char *)tagname)+1));\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 *) realloc(tags_aux, sizeof(TagType)*(npar + 1));\r
- if (!tags_aux) {\r
- fprintf(stderr, "Error: not enough memory\n");\r
- return NULLT;\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
+ \r
return 1;\r
\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 *)realloc(par_aux, sizeof(pb)*2*parArraySize);\r
+ par_aux = (pb *)urealloc(par_aux, sizeof(pb)*2*parArraySize);\r
parArraySize *= 2;\r
}\r
\r
- if (!par_aux) {\r
- fprintf(stderr, "Error: not enough memory\n");\r
- return NULLT; \r
- }\r
- setbit(par_aux,npar,CP); // marks a new closing opening parenthesis\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])==0) break;\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 **)realloc(TagName, sizeof(char *)*(ntagnames+1));\r
- \r
- if (!TagName) {\r
- fprintf(stderr, "Error: not enough memory\n");\r
- return NULLT;\r
- }\r
+ TagName = (unsigned char **)urealloc(TagName, sizeof(char *)*(ntagnames+1));\r
\r
ntagnames++;\r
- TagName[i] = (unsigned char *)malloc(sizeof(char)*(strlen((const char *)tagname)+2));\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 *)realloc(tags_aux, sizeof(TagType)*(npar + 1));\r
-\r
- if (!tags_aux) {\r
- fprintf(stderr, "Error: not enough memory\n");\r
- return NULLT;\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
return NULLT;\r
}\r
\r
+ if (disable_tc) {\r
+ XMLTree::NewEmptyText();\r
+ return 1;\r
+ };\r
+\r
if (!indexing_empty_texts) {\r
- empty_texts_aux = (unsigned int *)realloc(empty_texts_aux, sizeof(pb)*(1+(npar-1)/(8*sizeof(pb))));\r
- if (!empty_texts_aux) {\r
- fprintf(stderr, "Error: not enough memory\n");\r
- return NULLT;\r
- }\r
- \r
- bitset(empty_texts_aux, npar-1); // marks the non-empty text with a 1 in the bit vector\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
- Text->InsertText(s);\r
+ TextBuilder->InsertText(s);\r
+ string cpps = (char*) s;\r
+ CachedText.push_back(cpps); \r
\r
return 1; // success\r
}\r
}\r
\r
if (!indexing_empty_texts) {\r
- empty_texts_aux = (unsigned int *)realloc(empty_texts_aux, sizeof(pb)*(1+(npar-1)/(8*sizeof(pb))));\r
- if (!empty_texts_aux) {\r
- fprintf(stderr, "Error: not enough memory\n");\r
- return NULLT;\r
- }\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 Text->InsertText(&c); // we insert the empty text just in case we index all the texts\r
+ else TextBuilder->InsertText(&c); // we insert the empty text just in case we index all the texts\r
\r
return 1; // success \r
}\r
// this should be changed for more efficient processing\r
for (i=0; i<ntagnames; i++)\r
if (strcmp((const char *)tagname,(const char *)TagName[i])==0) break; \r
- if (i==ntagnames) return (TagType)NULLT; // tagname does not exists in the table\r
+ if (i==ntagnames) return (TagType)-1; //ntagnames; //(TagType)NULLT; // tagname does not exists in the table\r
else return i;\r
}\r
\r
unsigned char *XMLTree::GetTagName(TagType tagid)\r
{\r
unsigned char *s;\r
-\r
+ if(tagid==(uint)-1) return NULL;\r
if (tagid >= ntagnames) return NULL; // invalid tag identifier\r
- s = (unsigned char *)malloc((strlen((const char *)TagName[tagid])+1)*sizeof(unsigned char));\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
+\r
+\r
\r
+TagType XMLTree::RegisterTag(unsigned char *tagname)\r
+{\r
+ if (!finished)\r
+ return NULLT;\r
+ \r
+ TagType id = XMLTree::GetTagId(tagname);\r
+ if (id == NULLT){\r
+ id = ntagnames;\r
+ ntagnames = ntagnames + 1; \r
+ TagName = (unsigned char **) urealloc(TagName,ntagnames*(sizeof(unsigned char*)));\r
+ TagName[id] = (unsigned char *) umalloc(sizeof(unsigned char)*strlen( (const char*) tagname)+1);\r
+ strcpy((char*)TagName[id], (const char *)tagname); \r
+ };\r
+\r
+ return id;\r
+}\r