00001 /*========================================================================== 00002 * Copyright (c) 2004 University of Massachusetts. All Rights Reserved. 00003 * 00004 * Use of the Lemur Toolkit for Language Modeling and Information Retrieval 00005 * is subject to the terms of the software license set forth in the LICENSE 00006 * file included with this software, and also available at 00007 * http://www.lemurproject.org/license.html 00008 * 00009 *========================================================================== 00010 */ 00011 00012 00013 // 00014 // OrderedWindowNode 00015 // 00016 // 26 January 2004 -- tds 00017 // 00018 00019 #ifndef INDRI_ORDEREDWINDOWNODE_HPP 00020 #define INDRI_ORDEREDWINDOWNODE_HPP 00021 00022 #include "indri/ListIteratorNode.hpp" 00023 #include <vector> 00024 #include <indri/greedy_vector> 00025 #include <assert.h> 00026 00027 // 00028 // Expression: 00029 // #od<number>( <word>+ ) | #<number>( <word>+ ) 00030 // Examples: 00031 // #2( word1 word2 ) 00032 // #od2( word1 word2 ) 00033 // 00034 // Semantics: 00035 // Means "find these words, in order, in the text, where no two 00036 // adjacent words in the query are further apart than <number>-1". 00037 // 00038 // Examples: 00039 // #2(george bush) matches "george w. bush" but not "george herbert walker bush" 00040 // #2(george w bush) matches "george this w. that bush" but not "george w. this that bush" 00041 // 00042 namespace indri 00043 { 00044 namespace infnet 00045 { 00046 00047 class OrderedWindowNode : public ListIteratorNode { 00048 private: 00049 struct extents_pointer { 00050 indri::utility::greedy_vector<indri::index::Extent>::const_iterator iter; 00051 indri::utility::greedy_vector<indri::index::Extent>::const_iterator end; 00052 }; 00053 int _windowSize; 00054 std::vector<ListIteratorNode*> _children; 00055 indri::utility::greedy_vector<indri::index::Extent> _extents; 00056 std::vector<extents_pointer> _pointers; 00057 std::string _name; 00058 00059 public: 00060 OrderedWindowNode( const std::string& name, const std::vector<ListIteratorNode*>& children ); 00061 OrderedWindowNode( const std::string& name, const std::vector<ListIteratorNode*>& children, int windowSize ); 00062 00063 lemur::api::DOCID_T nextCandidateDocument(); 00064 void indexChanged( indri::index::Index& index ); 00065 00066 void prepare( lemur::api::DOCID_T documentID ); 00067 const indri::utility::greedy_vector<indri::index::Extent>& extents(); 00068 const std::string& getName() const ; 00069 void annotate( class Annotator& annotator, lemur::api::DOCID_T documentID, indri::index::Extent &extent ); 00070 }; 00071 } 00072 } 00073 00074 #endif // INDRI_ORDEREDWINDOWNODE_HPP 00075