-\r
#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
// 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
Tags->save(fp);\r
\r
// stores the texts \r
- //Text->Save(fp);\r
-\r
+ if (!disable_tc)\r
+ Text->Save(fp);\r
+ if (!disable_tc){\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+strlen(CachedText.at(i).c_str())),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_wvtree::load(fp);\r
+ XML_Tree->Tags = static_sequence::load(fp);\r
+ s_tree+= XML_Tree->Tags->size();\r
+\r
+ s_text = ftell(fp);\r
+\r
+ // loads the texts\r
+ if (!XML_Tree->disable_tc){\r
+ XML_Tree->Text = TextCollection::InitTextCollection(sample_rate_text);\r
+ XML_Tree->Text->Load(fp,sample_rate_text);\r
+ 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
- // loads the texts \r
- //XML_Tree->Text->Load(fp,sample_rate_text);\r
+ }\r
+ else\r
+ XML_Tree->Text = NULL;\r
+\r
+ s_text = ftell(fp) - s_text;\r
\r
- fclose(fp);\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 << "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
return XML_Tree;\r
}\r
\r
free(TagName);\r
\r
if (!indexing_empty_texts) {\r
- EBVector->~static_bitsequence_rrr02();\r
+ //EBVector->~static_bitsequence_rrr02();\r
delete EBVector;\r
EBVector = NULL;\r
}\r
\r
- Tags->~static_sequence_wvtree();\r
+ //Tags->~static_sequence_wvtree();\r
delete Tags;\r
Tags = NULL;\r
\r
//Text->~TextCollection();\r
- // delete Text;\r
- // Text = NULL;\r
+ delete Text;\r
+ Text = NULL;\r
\r
initialized = false;\r
finished = false;\r
// root(): returns the tree root.\r
treeNode XMLTree::Root() \r
{\r
+ if (!finished) {\r
+ fprintf(stderr, "Root() : Error: data structure has not been constructed properly\n");\r
+ exit(1);\r
+ }\r
return root_node(Par);\r
}\r
\r
// SubtreeSize(x): the number of nodes (and attributes) in the subtree of node x.\r
int XMLTree::SubtreeSize(treeNode x) \r
{\r
+ if (!finished) {\r
+ fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
+ exit(1);\r
+ }\r
return subtree_size(Par, x);\r
}\r
\r
// SubtreeTags(x,tag): the number of occurrences of tag within the subtree of node x.\r
int XMLTree::SubtreeTags(treeNode x, TagType tag) \r
{\r
+ if (!finished) {\r
+ fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
+ exit(1);\r
+ }\r
+ if (x == Root())\r
+ x = first_child(Par,x);\r
+ \r
+\r
int s = x + 2*subtree_size(Par, x) - 1;\r
\r
return Tags->rank(tag, s) - Tags->rank(tag, node2tagpos(x)-1);\r
// IsAncestor(x,y): returns whether node x is ancestor of node y.\r
bool XMLTree::IsAncestor(treeNode x, treeNode y) \r
{\r
+ if (!finished) {\r
+ fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
+ exit(1);\r
+ }\r
+\r
return is_ancestor(Par, x, y);\r
}\r
\r
// IsChild(x,y): returns whether node x is parent of node y.\r
bool XMLTree::IsChild(treeNode x, treeNode y) \r
{\r
+ if (!finished) {\r
+ fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
+ exit(1);\r
+ }\r
+\r
if (!is_ancestor(Par, x, y)) return false;\r
return depth(Par, x) == (depth(Par, y) + 1);\r
}\r
// of Sadakane.\r
int XMLTree::NumChildren(treeNode x) \r
{\r
+ if (!finished) {\r
+ fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
+ exit(1);\r
+ }\r
+\r
return degree(Par, x);\r
}\r
\r
// ChildNumber(x): returns i if node x is the i-th children of its parent.\r
int XMLTree::ChildNumber(treeNode x) \r
{\r
+ if (!finished) {\r
+ fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
+ exit(1);\r
+ }\r
+\r
return child_rank(Par, x);\r
}\r
\r
// Depth(x): depth of node x, a simple binary rank on the parentheses sequence.\r
int XMLTree::Depth(treeNode x) \r
{\r
+ if (!finished) {\r
+ fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
+ exit(1);\r
+ }\r
+\r
return depth(Par, x);\r
}\r
\r
// nodes (i.e., tags, it disregards the texts in the tree).\r
int XMLTree::Preorder(treeNode x) \r
{\r
+ if (!finished) {\r
+ fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
+ exit(1);\r
+ }\r
+\r
return preorder_rank(Par, x);\r
}\r
\r
// nodes (i.e., tags, it disregards the texts in the tree).\r
int XMLTree::Postorder(treeNode x) \r
{\r
+ if (!finished) {\r
+ fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
+ exit(1);\r
+ }\r
return postorder_rank(Par, x);\r
}\r
\r
// Tag(x): returns the tag identifier of node x.\r
TagType XMLTree::Tag(treeNode x) \r
{\r
+ if (!finished) {\r
+ fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
+ exit(1);\r
+ }\r
+ \r
return Tags->access(node2tagpos(x));\r
}\r
\r
// returns {NULLT, NULLT} when there are no texts descending from x.\r
range XMLTree::DocIds(treeNode x) \r
{\r
+ if (!finished) {\r
+ fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
+ exit(1);\r
+ }\r
+\r
range r;\r
if (indexing_empty_texts) { // faster, no rank needed\r
r.min = x;\r
// Parent(x): returns the parent node of node x.\r
treeNode XMLTree::Parent(treeNode x) \r
{\r
- return parent(Par, x);\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
\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
+{\r
+ if (!finished) {\r
+ fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
+ exit(1);\r
+ }\r
+\r
if (i <= OPTD) return naive_child(Par, x, i);\r
else return child(Par, x, i);\r
}\r
// FirstChild(x): returns the first child of node x, assuming it exists. Very fast in BP.\r
treeNode XMLTree::FirstChild(treeNode x) \r
{\r
+ if (!finished) {\r
+ fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
+ exit(1);\r
+ }\r
+\r
return first_child(Par, x);\r
}\r
\r
// NextSibling(x): returns the next sibling of node x, assuming it exists.\r
treeNode XMLTree::NextSibling(treeNode x) \r
{\r
+ if (!finished) {\r
+ fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
+ exit(1);\r
+ }\r
+ if (x == Root())\r
+ return NULLT;\r
+ \r
return next_sibling(Par, x);\r
}\r
\r
// PrevSibling(x): returns the previous sibling of node x, assuming it exists.\r
treeNode XMLTree::PrevSibling(treeNode x) \r
{\r
+ if (!finished) {\r
+ fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
+ exit(1);\r
+ }\r
+\r
return prev_sibling(Par, x);\r
}\r
\r
// 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
+ 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
// 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
else return y;\r
}\r
\r
+// TaggedNext(x,tag): returns the first node tagged tag with larger preorder than x \r
+// Returns NULLT if there is none.\r
+treeNode XMLTree::TaggedNext(treeNode x, TagType tag) \r
+ {\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 (x==NULLT)\r
+ return NULLT;\r
+\r
+ r = (int) Tags->rank(tag, node2tagpos(x));\r
+ s = (int) Tags->select(tag, r+1);\r
+ if (s == -1) return NULLT; // there is no such node\r
+ y = tagpos2node(s); // transforms the tag position into a node position \r
+ return (y<=x ? NULLT : y);\r
+ }\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
int r, s;\r
treeNode node_s, root;\r
r = (int)Tags->rank(tag, node2tagpos(x)-1);\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
+ exit(1);\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
+\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 (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
DocID XMLTree::PrevText(treeNode x) \r
{\r
+ if (!finished) {\r
+ fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
+ exit(1);\r
+ }\r
+\r
if (x == Root()) return NULLT;\r
if (indexing_empty_texts) // faster, no rank needed\r
return (DocID)x-1;\r
// of node x, or NULLT if x is the root node. Assumes Doc ids start from 0.\r
DocID XMLTree::NextText(treeNode x) \r
{\r
+ if (!finished) {\r
+ fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
+ exit(1);\r
+ }\r
+\r
if (x == Root()) return NULLT;\r
if (indexing_empty_texts) // faster, no rank needed\r
return (DocID)x+2*subtree_size(Par, x)-1;\r
// ids start from 0.\r
DocID XMLTree::MyText(treeNode x) \r
{\r
+ if (!finished) {\r
+ fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
+ exit(1);\r
+ }\r
+\r
if (!IsLeaf(x)) return NULLT;\r
if (indexing_empty_texts) // faster, no rank needed\r
return (DocID)x;\r
// all tree nodes and all text nodes. Assumes that the tree root has preorder 1.\r
int XMLTree::TextXMLId(DocID d) \r
{\r
+ if (!finished) {\r
+ fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
+ exit(1);\r
+ }\r
+\r
if (indexing_empty_texts) \r
return d + rank_open(Par, d)+1; // +1 because root has preorder 1\r
else { // slower, needs rank and select\r
// preorder 0;\r
int XMLTree::NodeXMLId(treeNode x) \r
{\r
+ if (!finished) {\r
+ fprintf(stderr, "Error: data structure has not been constructed properly\n");\r
+ exit(1);\r
+ }\r
+\r
if (indexing_empty_texts)\r
return x - 1 + rank_open(Par, x);\r
else {\r
// 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);\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
// 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
- ntagnames = 0;\r
+ parArraySize = 1;\r
+ ntagnames = 2; \r
+ disable_tc = dtc;\r
\r
indexing_empty_texts = empty_texts;\r
\r
- par_aux = (pb *)malloc(sizeof(pb));\r
- if (!par_aux) {\r
- fprintf(stderr, "Error: not enough memory\n");\r
- return NULLT;\r
- }\r
- setbit(par_aux,npar,OP); // marks a new opening parenthesis for the tree root\r
- npar++;\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
- 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 = (unsigned char **) umalloc(2*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], "<$>");\r
+\r
+\r
+ if (!indexing_empty_texts) \r
+ empty_texts_aux = (unsigned int *)umalloc(sizeof(unsigned int));\r
+ \r
\r
- //Text = TextCollection::InitTextCollection(sample_rate_text);\r
+ \r
+ Text = TextCollection::InitTextCollection((unsigned)sample_rate_text);\r
\r
return 1; // indicates success in the initialization of the data structure\r
}\r
}\r
\r
// closing parenthesis for the tree root\r
- par_aux = (pb *)realloc(par_aux, sizeof(pb)*(1+npar/(8*sizeof(pb))));\r
- if (!par_aux) {\r
- fprintf(stderr, "Error: not enough memory\n");\r
- return NULLT; \r
- }\r
- setbit(par_aux,npar,CP); \r
- npar++;\r
+ par_aux = (pb *)urealloc(par_aux, sizeof(pb)*(1+npar/(8*sizeof(pb))));\r
\r
// creates the data structure for the tree topology\r
- Par = (bp *)malloc(sizeof(bp)); \r
+ Par = (bp *)umalloc(sizeof(bp));\r
bp_construct(Par, npar, par_aux, OPT_DEGREE|0); \r
// creates structure for tags\r
- alphabet_mapper * am = new alphabet_mapper_none();\r
- static_bitsequence_builder * bmb = new static_bitsequence_builder_rrr02(32); \r
- wt_coder * wtc = new wt_coder_huff((uint *)tags_aux,npar-1,am);\r
- Tags = new static_sequence_wvtree((uint *) tags_aux, (uint) npar-1, wtc, bmb, am);\r
+ static_bitsequence_builder * bmb = new static_bitsequence_builder_brw32(20);\r
+ static_permutation_builder * pmb = new static_permutation_builder_mrrr(PERM_SAMPLE, bmb);\r
+ static_sequence_builder * ssb = new static_sequence_builder_gmr_chunk(bmb, pmb);\r
+\r
+\r
+ // If we found an attribute then "<@>" is present in the tree\r
+ // if we didn't then it is not. "<$>" is never present in the tree\r
+ int ntagsize = found_attributes ? 2*ntagnames-1 : 2*ntagnames - 2;\r
\r
+ Tags = new static_sequence_gmr((uint *) tags_aux, (uint) npar-1,ntagsize, bmb, ssb);\r
+ \r
+ delete bmb;\r
+ delete pmb;\r
+ delete ssb;\r
// makes the text collection static\r
- //Text->MakeStatic();\r
+ if (!disable_tc)\r
+ Text->MakeStatic();\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
\r
}\r
\r
// inserts a new opening parentheses in the bit sequence\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
+ 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
+ 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
- \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
npar++;\r
-\r
+ \r
return 1;\r
\r
}\r
}\r
\r
// inserts a new closing parentheses in the bit sequence\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
+ 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
- setbit(par_aux,npar,CP); // marks a new closing opening parenthesis\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])==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)+1));\r
- strcpy((char *)TagName[i], (const char *)tagname);\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
+ Text->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 Text->InsertText(&c); // we insert the empty text just in case we index all the texts\r
\r
return 1; // success \r
}\r
unsigned char *s;\r
\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 >= 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