Mercurial Hosting > luan
comparison src/luan/lib/queryparser/SaneQueryParser.java @ 1344:dc2af9d5463b
move queryparser to lib
author | Franklin Schmidt <fschmidt@gmail.com> |
---|---|
date | Mon, 25 Feb 2019 07:00:55 -0700 |
parents | src/luan/modules/lucene/queryparser/SaneQueryParser.java@88b5b81cad4a |
children | 77f2d091f17f |
comparison
equal
deleted
inserted
replaced
1343:7d9a1f8894b0 | 1344:dc2af9d5463b |
---|---|
1 package luan.lib.queryparser; | |
2 | |
3 import java.util.List; | |
4 import java.util.ArrayList; | |
5 import java.util.regex.Pattern; | |
6 import org.apache.lucene.search.Query; | |
7 import org.apache.lucene.search.MatchAllDocsQuery; | |
8 import org.apache.lucene.search.BooleanClause; | |
9 import org.apache.lucene.search.BooleanQuery; | |
10 import org.apache.lucene.search.Sort; | |
11 import org.apache.lucene.search.SortField; | |
12 import luan.lib.parser.Parser; | |
13 import luan.lib.parser.ParseException; | |
14 | |
15 | |
16 public class SaneQueryParser { | |
17 | |
18 public static Query parseQuery(FieldParser fieldParser,String query) throws ParseException { | |
19 return new SaneQueryParser(fieldParser,query).parseQuery(); | |
20 } | |
21 | |
22 private static Pattern specialChar = Pattern.compile("[ \\t\\r\\n\":\\[\\]{}^+\\-(),?*\\\\]"); | |
23 | |
24 public static String literal(String s) { | |
25 return specialChar.matcher(s).replaceAll("\\\\$0"); | |
26 } | |
27 | |
28 public static Sort parseSort(FieldParser fieldParser,String sort) throws ParseException { | |
29 return new SaneQueryParser(fieldParser,sort).parseSort(); | |
30 } | |
31 | |
32 | |
33 private static final String NOT_IN_TERM = " \t\r\n\":[]{}^+-()"; | |
34 private static final String NOT_IN_FIELD = NOT_IN_TERM + ","; | |
35 private final FieldParser fieldParser; | |
36 private final Parser parser; | |
37 | |
38 private SaneQueryParser(FieldParser fieldParser,String query) { | |
39 this.fieldParser = fieldParser; | |
40 this.parser = new Parser(query); | |
41 parser.begin(); | |
42 } | |
43 | |
44 ParseException exception(String msg) { | |
45 parser.failure(); | |
46 return new ParseException(parser,msg); | |
47 } | |
48 | |
49 ParseException exception(Exception cause) { | |
50 parser.failure(); | |
51 return new ParseException(parser,cause); | |
52 } | |
53 | |
54 private Query parseQuery() throws ParseException { | |
55 Spaces(); | |
56 BooleanQuery bq = new BooleanQuery(); | |
57 while( !parser.endOfInput() ) { | |
58 bq.add( Term(null) ); | |
59 } | |
60 BooleanClause[] clauses = bq.getClauses(); | |
61 switch( clauses.length ) { | |
62 case 0: | |
63 return new MatchAllDocsQuery(); | |
64 case 1: | |
65 { | |
66 BooleanClause bc = clauses[0]; | |
67 if( bc.getOccur() != BooleanClause.Occur.MUST_NOT ) | |
68 return bc.getQuery(); | |
69 } | |
70 default: | |
71 return bq; | |
72 } | |
73 } | |
74 | |
75 private BooleanClause Term(String defaultField) throws ParseException { | |
76 BooleanClause.Occur occur; | |
77 if( parser.match('+') ) { | |
78 occur = BooleanClause.Occur.MUST; | |
79 Spaces(); | |
80 } else if( parser.match('-') ) { | |
81 occur = BooleanClause.Occur.MUST_NOT; | |
82 Spaces(); | |
83 } else { | |
84 occur = BooleanClause.Occur.SHOULD; | |
85 } | |
86 String field = QueryField(); | |
87 if( field == null ) | |
88 field = defaultField; | |
89 Query query = NestedTerm(field); | |
90 if( query == null ) | |
91 query = RangeTerm(field); | |
92 if( query == null ) { | |
93 parser.begin(); | |
94 String match = SimpleTerm(); | |
95 query = fieldParser.getQuery(this,field,match); | |
96 parser.success(); | |
97 } | |
98 if( parser.match('^') ) { | |
99 Spaces(); | |
100 int start = parser.begin(); | |
101 try { | |
102 while( parser.anyOf("0123456789.") ); | |
103 String match = parser.textFrom(start); | |
104 float boost = Float.parseFloat(match); | |
105 query.setBoost(boost); | |
106 } catch(NumberFormatException e) { | |
107 throw exception(e); | |
108 } | |
109 parser.success(); | |
110 Spaces(); | |
111 } | |
112 BooleanClause bc = new BooleanClause(query,occur); | |
113 return bc; | |
114 } | |
115 | |
116 private Query NestedTerm(String field) throws ParseException { | |
117 parser.begin(); | |
118 if( !parser.match('(') ) | |
119 return parser.failure(null); | |
120 BooleanQuery bq = new BooleanQuery(); | |
121 while( !parser.match(')') ) { | |
122 if( parser.endOfInput() ) | |
123 throw exception("unclosed parentheses"); | |
124 bq.add( Term(field) ); | |
125 } | |
126 Spaces(); | |
127 BooleanClause[] clauses = bq.getClauses(); | |
128 switch( clauses.length ) { | |
129 case 0: | |
130 throw exception("empty parentheses"); | |
131 case 1: | |
132 { | |
133 BooleanClause bc = clauses[0]; | |
134 if( bc.getOccur() != BooleanClause.Occur.MUST_NOT ) | |
135 return parser.success(bc.getQuery()); | |
136 } | |
137 default: | |
138 return parser.success(bq); | |
139 } | |
140 } | |
141 | |
142 private Query RangeTerm(String field) throws ParseException { | |
143 parser.begin(); | |
144 if( !parser.anyOf("[{") ) | |
145 return parser.failure(null); | |
146 boolean includeMin = parser.lastChar() == '['; | |
147 Spaces(); | |
148 String minQuery = SimpleTerm(); | |
149 TO(); | |
150 String maxQuery = SimpleTerm(); | |
151 if( !parser.anyOf("]}") ) | |
152 throw exception("unclosed range"); | |
153 boolean includeMax = parser.lastChar() == ']'; | |
154 Spaces(); | |
155 Query query = fieldParser.getRangeQuery(this,field,minQuery,maxQuery,includeMin,includeMax); | |
156 return parser.success(query); | |
157 } | |
158 | |
159 private void TO() throws ParseException { | |
160 parser.begin(); | |
161 if( !(parser.match("TO") && Space()) ) | |
162 throw exception("'TO' expected"); | |
163 Spaces(); | |
164 parser.success(); | |
165 } | |
166 | |
167 private String SimpleTerm() throws ParseException { | |
168 parser.begin(); | |
169 String match; | |
170 if( parser.match('"') ) { | |
171 int start = parser.currentIndex() - 1; | |
172 while( !parser.match('"') ) { | |
173 if( parser.endOfInput() ) | |
174 throw exception("unclosed quotes"); | |
175 parser.anyChar(); | |
176 checkEscape(); | |
177 } | |
178 match = parser.textFrom(start); | |
179 Spaces(); | |
180 } else { | |
181 match = Unquoted(NOT_IN_TERM); | |
182 } | |
183 if( match.length() == 0 ) | |
184 throw exception("invalid input"); | |
185 return parser.success(match); | |
186 } | |
187 | |
188 private String QueryField() throws ParseException { | |
189 parser.begin(); | |
190 String match = Field(); | |
191 if( match==null || !parser.match(':') ) | |
192 return parser.failure((String)null); | |
193 Spaces(); | |
194 return parser.success(match); | |
195 } | |
196 | |
197 private String Field() throws ParseException { | |
198 parser.begin(); | |
199 String match = Unquoted(NOT_IN_FIELD); | |
200 if( match.length()==0 ) | |
201 return parser.failure((String)null); | |
202 match = StringFieldParser.escape(this,match); | |
203 return parser.success(match); | |
204 } | |
205 | |
206 private String Unquoted(String exclude) throws ParseException { | |
207 int start = parser.begin(); | |
208 while( parser.noneOf(exclude) ) { | |
209 checkEscape(); | |
210 } | |
211 String match = parser.textFrom(start); | |
212 Spaces(); | |
213 return parser.success(match); | |
214 } | |
215 | |
216 private void checkEscape() { | |
217 if( parser.lastChar() == '\\' ) | |
218 parser.anyChar(); | |
219 } | |
220 | |
221 private void Spaces() { | |
222 while( Space() ); | |
223 } | |
224 | |
225 private boolean Space() { | |
226 return parser.anyOf(" \t\r\n"); | |
227 } | |
228 | |
229 | |
230 // sort | |
231 | |
232 private Sort parseSort() throws ParseException { | |
233 Spaces(); | |
234 if( parser.endOfInput() ) | |
235 return null; | |
236 List<SortField> list = new ArrayList<SortField>(); | |
237 list.add( SortField() ); | |
238 while( !parser.endOfInput() ) { | |
239 parser.begin(); | |
240 if( !parser.match(',') ) | |
241 throw exception("',' expected"); | |
242 Spaces(); | |
243 parser.success(); | |
244 list.add( SortField() ); | |
245 } | |
246 return new Sort(list.toArray(new SortField[0])); | |
247 } | |
248 | |
249 private SortField SortField() throws ParseException { | |
250 parser.begin(); | |
251 String field = Field(); | |
252 if( field==null ) | |
253 throw exception("invalid input"); | |
254 boolean reverse = !parser.matchIgnoreCase("asc") && parser.matchIgnoreCase("desc"); | |
255 Spaces(); | |
256 SortField sf = fieldParser.getSortField(this,field,reverse); | |
257 return parser.success(sf); | |
258 } | |
259 | |
260 } |