2 /******************************************************************************
\r
3 * Copyright (C) 2008 by Diego Arroyuelo *
\r
4 * Interface for the in-memory XQuery/XPath engine *
\r
6 * This program is free software; you can redistribute it and/or modify *
\r
7 * it under the terms of the GNU Lesser General Public License as published *
\r
8 * by the Free Software Foundation; either version 2 of the License, or *
\r
9 * (at your option) any later version. *
\r
11 * This program is distributed in the hope that it will be useful, *
\r
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of *
\r
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the *
\r
14 * GNU Lesser General Public License for more details. *
\r
16 * You should have received a copy of the GNU Lesser General Public License *
\r
17 * along with this program; if not, write to the *
\r
18 * Free Software Foundation, Inc., *
\r
19 * 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. *
\r
20 ******************************************************************************/
\r
24 #include "TextCollection/TextCollection.h"
\r
29 #include <static_bitsequence.h>
\r
30 #include <alphabet_mapper.h>
\r
31 #include <static_sequence.h>
\r
32 using SXSI::TextCollection;
\r
35 // this constant is used to efficiently compute the child operation in the tree
\r
40 #define PERM_SAMPLE 10
\r
43 #define bitset(e,p) ((e)[(p)/W] |= (1<<((p)%W)))
\r
44 // cleans bit p in e
\r
45 #define bitclean(e,p) ((e)[(p)/W] &= ~(1<<((p)%W)))
\r
48 typedef int treeNode;
\r
49 typedef int TagType;
\r
59 /** Balanced parentheses representation of the tree */
\r
62 /** Mapping from tag identifer to tag name */
\r
63 unsigned char **TagName;
\r
65 /** boolean flag indicating whether we are indexing empty texts or not */
\r
66 bool indexing_empty_texts;
\r
68 /** Bit vector indicating with a 1 the positions of the non-empty texts. */
\r
69 static_bitsequence_rrr02 *EBVector;
\r
71 /** Tag sequence represented with a data structure for rank and select */
\r
72 static_sequence *Tags;
\r
74 /** The texts in the XML document */
\r
75 TextCollection *Text;
\r
77 /** Flag indicating whether the whole data structure has been constructed or
\r
78 * not. If the value is true, you cannot add more texts, nodes, etc. */
\r
81 /** Flag indicating whether the construction of the data structure has been
\r
82 * initialized or not (by calling method OpenDocument()). If this is true,
\r
83 * you cannot insert new tags or texts. */
\r
86 /* the following components are used for construction purposes */
\r
92 unsigned int *empty_texts_aux;
\r
93 bool found_attributes;
\r
98 /** Data structure constructor */
\r
99 XMLTree() {finished = false; initialized = false;};
\r
101 /** Data structure destructor */
\r
104 /** root(): returns the tree root. */
\r
107 /** SubtreeSize(x): the number of nodes (and attributes) in the subtree of
\r
109 int SubtreeSize(treeNode x);
\r
111 /** SubtreeTags(x,tag): the number of occurrences of tag within the subtree
\r
113 int SubtreeTags(treeNode x, TagType tag);
\r
115 /** IsLeaf(x): returns whether node x is leaf or not. In the succinct
\r
116 * representation this is just a bit inspection. */
\r
117 bool IsLeaf(treeNode x);
\r
119 /** IsAncestor(x,y): returns whether node x is ancestor of node y. */
\r
120 bool IsAncestor(treeNode x, treeNode y);
\r
122 /** IsChild(x,y): returns whether node x is parent of node y. */
\r
123 bool IsChild(treeNode x, treeNode y);
\r
125 /** NumChildren(x): number of children of node x. Constant time with the
\r
126 * data structure of Sadakane. */
\r
127 int NumChildren(treeNode x);
\r
129 /** ChildNumber(x): returns i if node x is the i-th children of its
\r
131 inline int ChildNumber(treeNode x);
\r
133 /** Depth(x): depth of node x, a simple binary rank on the parentheses
\r
135 int Depth(treeNode x);
\r
137 /** Preorder(x): returns the preorder number of node x, just regarding tree
\r
138 * nodes (and not texts). */
\r
139 int Preorder(treeNode x);
\r
141 /** Postorder(x): returns the postorder number of node x, just regarding
\r
142 * tree nodes (and not texts). */
\r
143 int Postorder(treeNode x);
\r
145 /** Tag(x): returns the tag identifier of node x. */
\r
146 TagType Tag(treeNode x);
\r
148 /** DocIds(x): returns the range (i.e., a pair of integers) of document
\r
149 * identifiers that descend from node x. */
\r
150 range DocIds(treeNode x);
\r
152 /** Parent(x): returns the parent node of node x. */
\r
153 treeNode Parent(treeNode x);
\r
155 /** Child(x,i): returns the i-th child of node x, assuming it exists. */
\r
156 treeNode Child(treeNode x, int i);
\r
158 /** FirstChild(x): returns the first child of node x, assuming it exists.
\r
159 * Very fast in BP. */
\r
160 treeNode FirstChild(treeNode x);
\r
162 /** NextSibling(x): returns the next sibling of node x, assuming it
\r
164 treeNode NextSibling(treeNode x);
\r
166 /** PrevSibling(x): returns the previous sibling of node x, assuming it
\r
168 treeNode PrevSibling(treeNode x);
\r
170 /** TaggedChild(x,i,tag): returns the i-th child of node x tagged tag, or
\r
171 * NULLT if there is none. Because of the balanced-parentheses representation
\r
172 * of the tree, this operation is not supported efficiently, just iterating
\r
173 * among the children of node x until finding the desired child. */
\r
174 treeNode TaggedChild(treeNode x, int i, TagType tag);
\r
176 /** TaggedDesc(x,tag): returns the first node tagged tag with larger
\r
177 * preorder than x and within the subtree of x. Returns NULT if there
\r
179 treeNode TaggedDesc(treeNode x, TagType tag);
\r
181 /** TaggedPrec(x,tag): returns the first node tagged tag with smaller
\r
182 * preorder than x and not an ancestor of x. Returns NULLT if there
\r
184 treeNode TaggedPrec(treeNode x, TagType tag);
\r
186 /** TaggedFoll(x,tag): returns the first node tagged tag with larger
\r
187 * preorder than x and not in the subtree of x. Returns NULLT if there
\r
189 treeNode TaggedFoll(treeNode x, TagType tag);
\r
191 /** PrevText(x): returns the document identifier of the text to the left of
\r
192 * node x, or NULLT if x is the root node. */
\r
193 DocID PrevText(treeNode x);
\r
195 /** NextText(x): returns the document identifier of the text to the right of
\r
196 * node x, or NULLT if x is the root node. */
\r
197 DocID NextText(treeNode x);
\r
199 /** MyText(x): returns the document identifier of the text below node x, or
\r
200 * NULLT if x is not a leaf node. */
\r
201 DocID MyText(treeNode x);
\r
203 /** TextXMLId(d): returns the preorder of document with identifier d in the
\r
204 * tree consisting of all tree nodes and all text nodes. */
\r
205 int TextXMLId(DocID d);
\r
207 /** NodeXMLId(x): returns the preorder of node x in the tree consisting of
\r
208 * all tree nodes and all text nodes. */
\r
209 int NodeXMLId(treeNode x);
\r
211 /** ParentNode(d): returns the parent node of document identifier d. */
\r
212 treeNode ParentNode(DocID d);
\r
214 /** OpenDocument(empty_texts,sample_rate_text,dtc): initilizes the construction
\r
215 * of the data structure for the XML document. Parameter empty_texts
\r
216 * indicates whether we index empty texts in document or not. Parameter
\r
217 * sample_rate_text indicates the sampling rate for the text searching data
\r
218 * structures (small values get faster searching but a bigger space
\r
219 * requirement). dtc disable the use of the TextCollection
\r
220 * (i.e. everything is considered an empty text *)
\r
221 * Returns a non-zero value upon success, NULLT in case of
\r
223 int OpenDocument(bool empty_texts, int sample_rate_text, bool dtc);
\r
225 /** CloseDocument(): finishes the construction of the data structure for
\r
226 * the XML document. Tree and tags are represented in the final form,
\r
227 * dynamic data structures are made static, and the flag "finished" is set
\r
228 * to true. After that, the data structure can be queried. */
\r
229 int CloseDocument();
\r
231 /** NewOpenTag(tagname): indicates the event of finding a new opening tag
\r
232 * in the document. Tag name is given. Returns a non-zero value upon
\r
233 * success, and returns NULLT in case of error. */
\r
234 int NewOpenTag(unsigned char *tagname);
\r
236 /** NewClosingTag(tagname): indicates the event of finding a new closing tag
\r
237 * in the document. Tag name is given. Returns a non-zero value upon
\r
238 * success, and returns NULLT in case of error. */
\r
239 int NewClosingTag(unsigned char *tagname);
\r
241 /** NewText(s): indicates the event of finding a new (non-empty) text s in
\r
242 * the document. The new text is inserted within the text collection.
\r
243 * Returns a non-zero value upon success, NULLT in case of error. */
\r
244 int NewText(unsigned char *s);
\r
246 /** NewEmptyText(): indicates the event of finding a new empty text in the
\r
247 * document. In case of indexing empty and non-empty texts, we insert the
\r
248 * empty texts into the text collection. In case of indexing only non-empty
\r
249 * texts, it just indicates an empty text in the bit vector of empty texts.
\r
250 * Returns a non-zero value upon success, NULLT in case of error. */
\r
251 int NewEmptyText();
\r
253 /** GetTagId(tagname): returns the tag identifier corresponding to a given
\r
254 * tag name. Returns NULLT in case that the tag name does not exists. */
\r
255 TagType GetTagId(unsigned char *tagname);
\r
257 /** GetTagName(tagid): returns the tag name of a given tag identifier.
\r
258 * Returns NULL in case that the tag identifier is not valid.*/
\r
259 unsigned char *GetTagName(TagType tagid);
\r
262 TagType RegisterTag(unsigned char *tagname);
\r
264 bool EmptyText(DocID i) {
\r
265 return Text->EmptyText(i);
\r
267 /** Prefix(s): search for texts prefixed by string s. */
\r
268 TextCollection::document_result Prefix(uchar const *s) {
\r
269 return Text->Prefix(s);
\r
272 /** Suffix(s): search for texts having string s as a suffix. */
\r
273 TextCollection::document_result Suffix(uchar const *s) {
\r
274 return Text->Suffix(s);
\r
277 /** Equal(s): search for texts equal to string s. */
\r
278 TextCollection::document_result Equal(uchar const *s) {
\r
279 return Text->Equal(s);
\r
282 /** Contains(s): search for texts containing string s. */
\r
283 TextCollection::document_result Contains(uchar const *s) {
\r
284 return Text->Contains(s);
\r
287 /** LessThan(s): returns document identifiers for the texts that
\r
288 * are lexicographically smaller than string s. */
\r
289 TextCollection::document_result LessThan(uchar const *s) {
\r
290 return Text->LessThan(s);
\r
293 /** IsPrefix(x): returns true if there is a text prefixed by string s. */
\r
294 bool IsPrefix(uchar const *s) {
\r
295 return Text->IsPrefix(s);
\r
298 /** IsSuffix(s): returns true if there is a text having string s as a
\r
300 bool IsSuffix(uchar const *s) {
\r
301 return Text->IsSuffix(s);
\r
304 /** IsEqual(s): returns true if there is a text that equals given
\r
306 bool IsEqual(uchar const *s) {
\r
307 return Text->IsEqual(s);
\r
310 /** IsContains(s): returns true if there is a text containing string s. */
\r
311 bool IsContains(uchar const *s) {
\r
312 return Text->IsContains(s);
\r
315 /** IsLessThan(s): returns true if there is at least a text that is
\r
316 * lexicographically smaller than string s. */
\r
317 bool IsLessThan(uchar const *s) {
\r
318 return Text->IsLessThan(s);
\r
321 /** CountPrefix(s): counting version of Prefix(s). */
\r
322 unsigned CountPrefix(uchar const *s) {
\r
323 return Text->CountPrefix(s);
\r
326 /** CountSuffix(s): counting version of Suffix(s). */
\r
327 unsigned CountSuffix(uchar const *s) {
\r
328 return Text->CountSuffix(s);
\r
331 /** CountEqual(s): counting version of Equal(s). */
\r
332 unsigned CountEqual(uchar const *s) {
\r
333 return Text->CountEqual(s);
\r
336 /** CountContains(s): counting version of Contains(s). */
\r
337 unsigned CountContains(uchar const *s) {
\r
338 return Text->CountContains(s);
\r
341 /** CountLessThan(s): counting version of LessThan(s). */
\r
342 unsigned CountLessThan(uchar const *s) {
\r
343 return CountLessThan(s);
\r
346 /** GetText(d): returns the text corresponding to document with
\r
348 uchar* GetText(DocID d) {
\r
349 return Text->GetText(d);
\r
352 TextCollection *getTextCollection() {
\r
355 /** Save: saves XML tree data structure to file. */
\r
356 void Save(unsigned char *filename);
\r
358 /** Load: loads XML tree data structure from file. sample_rate_text
\r
359 * indicates the sample rate for the text search data structure. */
\r
360 static XMLTree *Load(unsigned char *filename, int sample_rate_text);
\r