#include "XMLTree.h"\r
#include <cstring>\r
\r
- // functions to convert tag positions to the corresponding tree node and viceversa. \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
// Current implementation corresponds to balanced-parentheses representation for\r
return (int)x;\r
}\r
\r
-// to prevent suprious "unused result" warnings\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
\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
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
{\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
ufread(&XML_Tree->ntagnames, sizeof(int), 1, fp);\r
- XML_Tree->TagName = (unsigned char **)malloc(XML_Tree->ntagnames*sizeof(unsigned char *));\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
\r
for (i=0; i<XML_Tree->ntagnames;i++) {\r
\r
- // Kim is it needed ?\r
+ // OJO Kim is it needed ?\r
int k = feof(fp);\r
- // fscanf chokes ont "\n" which is the case for the root element\r
- char * r = fgets(filenameaux,1023,fp);\r
- // int r = fscanf(fp, "<%s>\n",filenameaux);\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
-\r
- int len = strlen((const char*)filenameaux);\r
- XML_Tree->TagName[i] = (unsigned char *)calloc(len,sizeof(char));\r
-\r
- //XML_Tree->TagName[i] = (unsigned char *)malloc(sizeof(unsigned char)*(strlen((const char *)filenameaux)+1));\r
- //the - 1 removes the trailing \n\r
- strncpy((char *)XML_Tree->TagName[i], (const char *)filenameaux,len - 1);\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
- if (!(XML_Tree->indexing_empty_texts)) XML_Tree->EBVector = static_bitsequence_rrr02::load(fp);\r
+ \r
+ s_tree+=sizeof(bool)*4;\r
\r
+ if (!(XML_Tree->indexing_empty_texts)) XML_Tree->EBVector = static_bitsequence_rrr02::load(fp);\r
+ \r
+ s_tree+= XML_Tree->EBVector->size();\r
+ \r
// loads the tags\r
XML_Tree->Tags = static_sequence::load(fp);\r
\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
else\r
XML_Tree->Text = NULL;\r
\r
+ s_text = ftell(fp) - s_text;\r
fclose(fp);\r
+\r
+ std::cerr << "Tree part is " << s_tree/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
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
+ \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
\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 = (unsigned char **) malloc(2*sizeof(unsigned char*));\r
- if (!TagName){\r
- fprintf(stderr, "Error: not enough memory\n");\r
- return NULLT;\r
- }\r
+ TagName = (unsigned char **) umalloc(2*sizeof(unsigned char*));\r
+\r
+ TagName[0] = (unsigned char *) umalloc(4*sizeof(unsigned char));\r
\r
- TagName[0] = (unsigned char *) malloc(4*sizeof(unsigned char));\r
strcpy((char *) TagName[0], "<@>");\r
\r
- if (!TagName[0]){\r
- fprintf(stderr, "Error: not enough memory\n");\r
- return NULLT;\r
- }\r
-\r
- TagName[1] = (unsigned char *) malloc(4*sizeof(unsigned char));\r
- if (!TagName[1]){\r
- fprintf(stderr, "Error: not enough memory\n");\r
- return NULLT;\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 *)malloc(sizeof(unsigned int));\r
- if (!empty_texts_aux) {\r
- fprintf(stderr, "Error: not enough memory\n");\r
- return NULLT;\r
- }\r
- }\r
+ if (!indexing_empty_texts) \r
+ empty_texts_aux = (unsigned int *)umalloc(sizeof(unsigned int));\r
+ \r
+ \r
\r
Text = TextCollection::InitTextCollection((unsigned)sample_rate_text);\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
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
// 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
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
\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
\r
// transforms the tagname into a tag identifier. If the tag is new, we insert\r
if (strcmp((const char *)tagname,(const char *)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
+ 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 *)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
};\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
}\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
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
-\r
TagType id = XMLTree::GetTagId(tagname);\r
if (id == NULLT){\r
id = ntagnames;\r
ntagnames = ntagnames + 1; \r
- TagName = (unsigned char **) realloc(TagName,ntagnames*(sizeof(unsigned char*)));\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