\r
#ifndef XMLTREE_H_\r
#define XMLTREE_H_\r
-#include "TextCollection/TextCollection.h"\r
+#include "TextCollection/TextCollectionBuilder.h"\r
#include <stdio.h>\r
#include <stdlib.h>\r
#include <cstring>\r
#include <alphabet_mapper.h>\r
#include <static_sequence.h>\r
using SXSI::TextCollection;\r
+using SXSI::TextCollectionBuilder;\r
\r
\r
// this constant is used to efficiently compute the child operation in the tree\r
\r
/** Tag sequence represented with a data structure for rank and select */\r
static_sequence *Tags;\r
+ uint * tags_fix;\r
+ uint tags_blen, tags_len;\r
\r
/** The texts in the XML document */\r
+ TextCollectionBuilder *TextBuilder;\r
TextCollection *Text;\r
+\r
/** The texts in the XML document (cached for faster display) */\r
vector<string> CachedText;\r
\r
bool disable_tc;\r
\r
public:\r
+ void print_stats();\r
\r
/** Data structure constructor */\r
- XMLTree() {finished = false; initialized = false;}; \r
+ XMLTree() {finished = false; initialized = false; Text = 0; TextBuilder = 0; }; \r
\r
/** Data structure destructor */\r
~XMLTree();\r
\r
/** IsChild(x,y): returns whether node x is parent of node y. */\r
bool IsChild(treeNode x, treeNode y);\r
+\r
+ /** IsChild(x,y): returns whether node x is the first child of its parent */\r
+ bool IsFirstChild(treeNode x);\r
\r
/** NumChildren(x): number of children of node x. Constant time with the \r
* data structure of Sadakane. */\r
* Very fast in BP. */\r
treeNode FirstChild(treeNode x);\r
\r
+ /** LastChild(x): returns the last child of node x. \r
+ * Implemented by Kim naively. */\r
+ treeNode LastChild(treeNode x);\r
+\r
/** NextSibling(x): returns the next sibling of node x, assuming it \r
* exists. */\r
+\r
treeNode NextSibling(treeNode x);\r
\r
/** PrevSibling(x): returns the previous sibling of node x, assuming it \r
* is none. */\r
treeNode TaggedDesc(treeNode x, TagType tag);\r
\r
- /** TaggedNext(x,tag): returns the first node tagged tag with larger \r
- * preorder than x. Returns NULT if there is none. */\r
- treeNode TaggedNext(treeNode x, TagType tag);\r
+\r
+ treeNode TaggedBelow(treeNode x, TagType *childtags, unsigned int ctlen,\r
+ TagType *desctags, unsigned int dtlen);\r
+ \r
+ treeNode TaggedNext(treeNode x, TagType *childtags, unsigned int ctlen,\r
+ TagType *folltags, unsigned int flen,treeNode root);\r
+\r
+ treeNode TaggedDescOnly(treeNode x, TagType *desctags, unsigned int dtlen);\r
+ \r
+ treeNode TaggedDescOrFollOnly(treeNode x, TagType *folltags, unsigned int flen,\r
+ treeNode root);\r
+\r
+ treeNode TaggedFollOnly(treeNode x, TagType *folltags, unsigned int flen,\r
+ treeNode root);\r
+ \r
\r
/** TaggedPrec(x,tag): returns the first node tagged tag with smaller \r
* preorder than x and not an ancestor of x. Returns NULLT if there \r
* preorder than x and not in the subtree of x. Returns NULLT if there \r
* is none. */\r
treeNode TaggedFoll(treeNode x, TagType tag);\r
+\r
+ treeNode TaggedFollBelow(treeNode x, TagType tag,treeNode root); \r
\r
/** TaggedFollowingSibling(x,tag) */\r
treeNode TaggedFollowingSibling(treeNode x, TagType tag);\r
\r
/** ParentNode(d): returns the parent node of document identifier d. */\r
treeNode ParentNode(DocID d);\r
+ treeNode PrevNode(DocID d);\r
\r
/** OpenDocument(empty_texts,sample_rate_text,dtc): initilizes the construction\r
* of the data structure for the XML document. Parameter empty_texts \r
TagType RegisterTag(unsigned char *tagname);\r
\r
bool EmptyText(DocID i) {\r
- return Text->EmptyText(i);\r
+ return Text->EmptyText(i);\r
}\r
/** Prefix(s): search for texts prefixed by string s. */\r
TextCollection::document_result Prefix(uchar const *s) {\r
bool IsLessThan(uchar const *s) {\r
return Text->IsLessThan(s);\r
}\r
+ \r
+ /** Count(s): Global counting */\r
+ unsigned Count(uchar const *s) {\r
+ return Text->Count(s);\r
+ }\r
\r
/** CountPrefix(s): counting version of Prefix(s). */\r
unsigned CountPrefix(uchar const *s) {\r
/** GetText(d): returns the text corresponding to document with\r
* id d. */\r
uchar* GetText(DocID d) {\r
- return Text->GetText(d);\r
+ return Text->GetText(d);\r
}\r
\r
uchar* GetCachedText(DocID d) {\r
- uchar * str = (uchar*) calloc(sizeof(char),(CachedText.at(d).size() + 1)); \r
+ uchar * str = (uchar*) calloc(sizeof(char),(CachedText.at(d).size() + 1));\r
+ strcpy((char*) str,(const char*) CachedText.at(d).c_str());\r
return (uchar*) (str);\r
}\r
\r