Mercurial Hosting > luan

comparison lucene/src/luan/modules/lucene/Lucene.luan @ 544:c5a93767cc5c

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
lucene overhaul, untested
author Franklin Schmidt <fschmidt@gmail.com>
date Fri, 12 Jun 2015 19:11:44 -0600
parents 7170a0c414d5
children ddcd4296107a
comparison
equal deleted inserted replaced
543:9767da72545b 544:c5a93767cc5c
1 java() 1 java()
2 local Luan = require "luan:Luan" 2 local Luan = require "luan:Luan"
3 local pairs = Luan.pairs
4 local ipairs = Luan.ipairs
5 local type = Luan.type
6 local number_type = Luan.number_type
7 local error = Luan.error 3 local error = Luan.error
8 local Number = require "luan:Number"
9 local number_type = Number.number_type
10 local Table = require "luan:Table"
11 local unpack = Table.unpack
12 local LuceneIndex = require "java:luan.modules.lucene.LuceneIndex" 4 local LuceneIndex = require "java:luan.modules.lucene.LuceneIndex"
13 local Term = require "java:org.apache.lucene.index.Term" 5 local NumberFieldParser = require "java:sane.lucene.queryparser.NumberFieldParser"
14 local TermQuery = require "java:org.apache.lucene.search.TermQuery" 6
15 local TermRangeQuery = require "java:org.apache.lucene.search.TermRangeQuery"
16 local NumericRangeQuery = require "java:org.apache.lucene.search.NumericRangeQuery"
17 local MatchAllDocsQuery = require "java:org.apache.lucene.search.MatchAllDocsQuery"
18 local BooleanQuery = require "java:org.apache.lucene.search.BooleanQuery"
19 local BooleanClause = require "java:org.apache.lucene.search.BooleanClause"
20 local BytesRef = require "java:org.apache.lucene.util.BytesRef"
21 local NumericUtils = require "java:org.apache.lucene.util.NumericUtils"
22 local Sort = require "java:org.apache.lucene.search.Sort"
23 local SortField = require "java:org.apache.lucene.search.SortField"
24 7
25 local M = {} 8 local M = {}
9
10 M.type = {
11 string = LuceneIndex.STRING_FIELD_PARSER;
12 integer = NumberFieldParser.INT;
13 long = NumberFieldParser.LONG;
14 double = NumberFieldParser.DOUBLE;
15 }
26 16
27 function M.index(indexDir) 17 function M.index(indexDir)
28 local index = {} 18 local index = {}
29 local java_index = LuceneIndex.new(indexDir) 19 local java_index = LuceneIndex.new(indexDir)
30 index.fields = java_index.fields.newTable() 20 index.indexed_fields = java_index.indexedFieldsMeta.newTable()
31 index.to_string = java_index.to_string 21 index.to_string = java_index.to_string
32 index.backup = java_index.backup 22 index.backup = java_index.backup
33 index.Writer = java_index.Writer 23 index.Writer = java_index.Writer
34 index.Searcher = java_index.Searcher 24 index.Searcher = java_index.Searcher
35 index.delete_all = java_index.delete_all 25 index.delete_all = java_index.delete_all
36 index.map_field_name = java_index.map_field_name
37 index.close = java_index.close 26 index.close = java_index.close
38 27
39 function index.save_document(doc) 28 function index.save_document(doc)
40 index.Writer( function(writer) 29 index.Writer( function(writer)
41 writer.save_document(doc) 30 writer.save_document(doc)
68 local _, _, total_hits = searcher.search(query,0) 57 local _, _, total_hits = searcher.search(query,0)
69 return total_hits 58 return total_hits
70 end ) 59 end )
71 end 60 end
72 61
73
74
75 local queryTbl = {}
76 index.query = queryTbl
77
78 queryTbl.parse = java_index.parse
79
80 queryTbl.all_docs = MatchAllDocsQuery.new()
81
82 function queryTbl.term(field,value)
83 if value==nil and type(field)=="table" then
84 local iter = pairs(field)
85 field, value = iter()
86 iter()==nil or error "only one term allowed"
87 end
88 field and value or error "missing term"
89 field = index.map_field_name(field)
90 local tp = type(value)
91 if tp == "string" then
92 return TermQuery.new(Term.new(field,value))
93 end
94 if tp == "number" then
95 local br = BytesRef.new()
96 tp = number_type(value)
97 if tp == "integer" then
98 NumericUtils.intToPrefixCoded(value,0,br)
99 elseif tp == "long" then
100 NumericUtils.longToPrefixCoded(value,0,br)
101 elseif tp == "double" then
102 value = NumericUtils.doubleToSortableLong(value)
103 NumericUtils.longToPrefixCoded(value,0,br)
104 else
105 error("invalid term value type: "..tp)
106 end
107 return TermQuery.new(Term.new(field,br))
108 end
109 error("invalid term value type: "..tp)
110 end
111
112 function queryTbl.range(field,min,max,include_min,include_max)
113 if include_min == nil then include_min = true end
114 if include_max == nil then include_max = true end
115 field = index.map_field_name(field)
116 local tp = type(min)
117 tp == type(max) or error "min and max types don't match"
118 if tp == "string" then
119 return TermRangeQuery.newStringRange(field,min,max,include_min,include_max)
120 end
121 if tp == "number" then
122 tp = number_type(min)
123 tp == number_type(max) or error "min and max number types don't match"
124 if tp == "integer" then
125 return NumericRangeQuery.newIntRange(field,min,max,include_min,include_max)
126 end
127 if tp == "long" then
128 return NumericRangeQuery.newLongRange(field,min,max,include_min,include_max)
129 end
130 if tp == "double" then
131 return NumericRangeQuery.newDoubleRange(field,min,max,include_min,include_max)
132 end
133 error("invalid term value type: "..tp)
134 end
135 error("invalid term value type: "..tp)
136 end
137
138 function queryTbl.boolean(t)
139 local boolean_query = BooleanQuery.new()
140 for query, occur_string in pairs(t) do
141 local occur = BooleanClause.Occur.valueOf( occur_string.upper() )
142 boolean_query.add( query, occur )
143 end
144 return boolean_query
145 end
146
147 -- and list
148 function queryTbl.all(t)
149 local bt = {}
150 for key, query in pairs(t) do
151 if type(key)=="string" then
152 query = queryTbl.term{[key]=query}
153 end
154 bt[query] = "MUST"
155 end
156 return queryTbl.boolean(bt)
157 end
158
159 -- or list
160 function queryTbl.any(t)
161 local bt = {}
162 for key, query in pairs(t) do
163 if type(key)=="string" then
164 query = queryTbl.term{[key]=query}
165 end
166 bt[query] = "SHOULD"
167 end
168 return queryTbl.boolean(bt)
169 end
170
171
172 function queryTbl.sort(fields)
173 #fields > 0 or error "list of sort fields expected"
174 local a = {}
175 for _, f in ipairs(fields) do
176 f.field or error "missing sort field"
177 f.type or error "missing sort type"
178 local field = index.map_field_name(f.field)
179 local type = SortField.Type.valueOf( f.type.upper() )
180 local reverse = f.reverse or false
181 a[#a+1] = SortField.new(field,type,reverse)
182 end
183 return Sort.new(unpack(a))
184 end
185
186
187 return index 62 return index
188 end 63 end
189 64
190 return M 65 return M