\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
+ 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
\r
\r
disable_tc = dis_tc;\r
- stream = NULL;\r
- stream_fd = 0;\r
+ text_index_type = _index_type; \r
std::cerr << "Number of distinct tags " << TagName->size() << "\n";\r
//std::cerr.flush();\r
}\r
\r
delete EBVector;\r
EBVector = NULL;\r
- if (stream != NULL){\r
- fclose(stream);\r
- stream = NULL;\r
- stream_fd = 0;\r
- };\r
\r
}\r
\r
}\r
\r
// Save: saves XML tree data structure to file. \r
-void XMLTree::Save(int fd, char *filename) \r
+void XMLTree::Save(int fd) \r
{\r
FILE *fp;\r
char filenameaux[1024];\r
\r
// stores the texts \r
if (!disable_tc) {\r
- Text->Save(fp, filename);\r
- };\r
- }\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
+ Text->Save(fp, pref);\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(int fd, char *filename, bool load_tc,int sample_factor) \r
+XMLTree *XMLTree::Load(int fd, bool load_tc,int sample_factor) \r
{\r
\r
FILE *fp;\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
- //XML_Tree->EBVector = static_bitsequence_sdarray::load(fp);\r
-\r
- STOPTIMER(Loading);\r
- PRINTTIME("Loading text bitvector struct", Loading);\r
- STARTTIMER();\r
+ XML_Tree->EBVector = static_bitsequence_rrr02::load(fp);\r
\r
- // Not used \r
- // loads the texts\r
- if (!XML_Tree->disable_tc){\r
- XML_Tree->Text = TextCollection::Load(fp, filename, TextCollection::index_mode_default, sample_factor);\r
- }\r
- else XML_Tree->Text = NULL;\r
- STOPTIMER(Loading);\r
- PRINTTIME("Loading TextCollection", Loading);\r
- STARTTIMER(); \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
XML_Tree->disable_tc = true;\r
};\r
-\r
- XML_Tree->stream = NULL;\r
- XML_Tree->stream_fd = 0;\r
\r
+\r
return XML_Tree;\r
}\r
\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
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 (x == Root())\r
\r
return (Tags->rank(tag, s) - Tags->rank(tag, node2tagpos(x)-1))+1;\r
}\r
+*/\r
int XMLTree::SubtreeElements(treeNode x) \r
{\r
\r
}*/\r
\r
// LastChild(x): returns the last child of node x.\r
-treeNode XMLTree::LastChild(treeNode 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
+ */\r
// NextSibling(x): returns the next sibling of node x, assuming it exists.\r
/*treeNode XMLTree::NextSibling(treeNode 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
NULLT_IF(x==NULLT);\r
return prev_sibling(Par, x);\r
}\r
-\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
return (fast_is_ancestor(Par,x,y) ? y : NULLT);\r
}\r
*/\r
-\r
+/*\r
treeNode XMLTree::SelectDescendant(treeNode x, TagIdSet *tags)\r
{\r
NULLT_IF (x ==NULLT || fast_isleaf(Par,x));\r
return min;\r
}\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
else return NULLT;\r
\r
}\r
-\r
+/*\r
treeNode XMLTree::SelectFollowingBefore(treeNode x, TagIdSet *tags, treeNode ancestor_closing)\r
{\r
\r
else return NULLT;\r
\r
}\r
-\r
+*/\r
/*\r
treeNode XMLTree::SelectFollowingBefore(treeNode x, TagIdSet *tags, treeNode closing)\r
{\r
// seems faster than testing EBVector->access(x);\r
\r
if (tag == PCDATA_TAG_ID || tag == ATTRIBUTE_DATA_TAG_ID)\r
- //if (EBVector->access(x))\r
- return (DocID) (EBVector->rank1(x)-1); //-1 because document ids start from 0\r
- else \r
+ return (DocID) (EBVector->rank1(x)-1); \r
+ else \r
return (DocID) NULLT;\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
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
\r
void XMLTree::Print(int fd,treeNode x, bool no_text){\r
\r
- int newfd = dup(fd);\r
- stream = fdopen(newfd,"wa");\r
- if (stream == 0){\r
- perror(NULL);\r
- return;\r
+ if (buffer == 0) { \r
+ buffer = new string(BUFFER_ALLOC, 0);\r
+ buffer->clear();\r
+ print_stack = new std::vector<string *>();\r
+ print_stack->reserve(256);\r
};\r
\r
- if (buffer == 0)\r
- buffer = new string();\r
-\r
- FILE* fp = stream;\r
treeNode fin = fast_find_close(Par,x);\r
treeNode n = x;\r
TagType tag = Tag(n);\r
- uchar * tagstr;\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
+ if (first_att == NULLT) \r
+ first_idx = first_text;\r
else if (first_text == NULLT)\r
- first_idx = first_att;\r
+ first_idx = first_att;\r
else\r
- first_idx = min(first_att,first_text);\r
+ first_idx = min(first_att,first_text);\r
\r
uchar * current_text=NULL;\r
+\r
if (first_idx != NULLT)\r
- current_text = GetText(MyText(first_idx));\r
+ current_text = GetText(MyTextUnsafe(first_idx));\r
+ \r
size_t read = 0;\r
- std::vector<uchar*> st;\r
- while (n <= fin){\r
- if (fast_inspect(Par,n)){\r
- if (tag == PCDATA_TAG_ID ) { \r
-\r
- if (no_text)\r
- myfputs("<$/>",fp);\r
- else{\r
- read = myfprintf((const char*) current_text, fp);\r
- current_text += (read + 1);\r
- };\r
- n+=2; // skip closing $\r
- tag = Tag(n);\r
- \r
- }\r
- else {\r
- myfputc('<',fp);\r
- tagstr = (uchar*) GetTagNameByRef(tag);\r
- myfputs((const char*) tagstr ,fp);\r
- n++;\r
- if (fast_inspect(Par,n)) {\r
- st.push_back(tagstr);\r
- tag = Tag(n);\r
- if (tag == ATTRIBUTE_TAG_ID){\r
- n++;\r
- if (no_text) myfputs("><@@>",fp);\r
- while (fast_inspect(Par,n)){\r
- if (no_text) {\r
- myfputc('<',fp);\r
- myfputs((const char*) &(GetTagNameByRef(Tag(n))[3]),fp);\r
- myfputc('>',fp);\r
- myfputs("<$@/></",fp);\r
- myfputs((const char*) &(GetTagNameByRef(Tag(n))[3]),fp);\r
- myfputc('>',fp);\r
- n+= 4;\r
- }\r
- else {\r
- myfputc(' ',fp);\r
- myfputs((const char*) &(GetTagNameByRef(Tag(n))[3]),fp);\r
- n++;\r
- myfputs("=\"",fp);\r
- read = myfprintf((const char*) current_text,fp);\r
- current_text += (read + 1);\r
- myfputc('"',fp);\r
- n+=3;\r
- }\r
- };\r
- if (no_text) \r
- myfputs("</@@>",fp);\r
- else myfputc('>',fp);\r
- n++;\r
- tag=Tag(n);\r
- }\r
+\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
- myfputc('>',fp);\r
+ read = _dprintf((const char*) current_text, fd);\r
+ current_text += (read + 1);\r
};\r
- }\r
- else {// <foo /> tag\r
- myfputs("/>",fp);\r
+ n+=2; // skip closing $\r
+ tag = Tag(n);\r
+ \r
+ } else {\r
+\r
+ _dputc('<',fd); \r
+ _dput_str((*TagName)[tag], fd);\r
n++;\r
- tag=Tag(n); \r
- }; \r
- };\r
- }\r
- else\r
- do {\r
- myfputs("</",fp);\r
- myfputs((const char*)st.back(),fp);\r
- myfputc('>', fp);\r
- st.pop_back();\r
- n++;\r
- }while (!fast_inspect(Par,n) && !st.empty());\r
- tag=Tag(n);\r
- };\r
- myfputc('\n',fp);\r
- mybufferflush(fp);\r
- //fflush(fp);\r
- fclose(fp);\r
+ if (fast_inspect(Par,n)) {\r
+ print_stack->push_back(&((*TagName)[tag]));\r
+ tag = Tag(n);\r
+ if (tag == ATTRIBUTE_TAG_ID){\r
+ n++;\r
+ if (no_text) _dputs("><@@>",fd);\r
+\r
+ while (fast_inspect(Par,n)){\r
+ if (no_text) {\r
+ _dputc('<', fd);\r
+ _dputs((const char*) &(GetTagNameByRef(Tag(n))[3]), fd);\r
+ _dputc('>', fd);\r
+ _dputs("<$@/></", 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