|
1
by Antonio Terceiro
Import upstream version 0.11.6 |
1 |
require File.dirname(__FILE__) + "/../../test_helper" |
2 |
||
3 |
class QueryParserTest < Test::Unit::TestCase |
|
4 |
include Ferret::Analysis |
|
5 |
||
6 |
def test_strings() |
|
7 |
parser = Ferret::QueryParser.new(:default_field => "xxx", |
|
8 |
:fields => ["xxx", "field", "f1", "f2"], |
|
9 |
:tokenized_fields => ["xxx", "f1", "f2"]) |
|
10 |
pairs = [ |
|
11 |
['', ''], |
|
12 |
['*:word', 'word field:word f1:word f2:word'], |
|
13 |
['word', 'word'], |
|
14 |
['field:word', 'field:word'], |
|
15 |
['"word1 word2 word#"', '"word1 word2 word"'], |
|
16 |
['"word1 %%% word3"', '"word1 <> word3"~1'], |
|
17 |
['field:"one two three"', 'field:"one two three"'], |
|
18 |
['field:"one %%% three"', 'field:"one %%% three"'], |
|
19 |
['f1:"one %%% three"', 'f1:"one <> three"~1'], |
|
20 |
['field:"one <> three"', 'field:"one <> three"'], |
|
21 |
['field:"one <> three <>"', 'field:"one <> three"'], |
|
22 |
['field:"one <> <> <> three <>"', 'field:"one <> <> <> three"'], |
|
23 |
['field:"one <> 222 <> three|four|five <>"', 'field:"one <> 222 <> three|four|five"'], |
|
24 |
['field:"on1|tw2 THREE|four|five six|seven"', 'field:"on1|tw2 THREE|four|five six|seven"'], |
|
25 |
['field:"testing|trucks"', 'field:"testing|trucks"'], |
|
26 |
['[aaa bbb]', '[aaa bbb]'], |
|
27 |
['{aaa bbb]', '{aaa bbb]'], |
|
28 |
['field:[aaa bbb}', 'field:[aaa bbb}'], |
|
29 |
['{aaa bbb}', '{aaa bbb}'], |
|
30 |
['{aaa>', '{aaa>'], |
|
31 |
['[aaa>', '[aaa>'], |
|
32 |
['field:<a\ aa}', 'field:<a aa}'], |
|
33 |
['<aaa]', '<aaa]'], |
|
34 |
['>aaa', '{aaa>'], |
|
35 |
['>=aaa', '[aaa>'], |
|
36 |
['<aaa', '<aaa}'], |
|
37 |
['[A>', '[a>'], |
|
38 |
['field:<=aaa', 'field:<aaa]'], |
|
39 |
['REQ one REQ two', '+one +two'], |
|
40 |
['REQ one two', '+one two'], |
|
41 |
['one REQ two', 'one +two'], |
|
42 |
['+one +two', '+one +two'], |
|
43 |
['+one two', '+one two'], |
|
44 |
['one +two', 'one +two'], |
|
45 |
['-one -two', '-one -two'], |
|
46 |
['-one two', '-one two'], |
|
47 |
['one -two', 'one -two'], |
|
48 |
['!one !two', '-one -two'], |
|
49 |
['!one two', '-one two'], |
|
50 |
['one !two', 'one -two'], |
|
51 |
['NOT one NOT two', '-one -two'], |
|
52 |
['NOT one two', '-one two'], |
|
53 |
['one NOT two', 'one -two'], |
|
54 |
['NOT two', '-two +*'], |
|
55 |
['one two', 'one two'], |
|
56 |
['one OR two', 'one two'], |
|
57 |
['one AND two', '+one +two'], |
|
58 |
['one two AND three', 'one two +three'], |
|
59 |
['one two OR three', 'one two three'], |
|
60 |
['one (two AND three)', 'one (+two +three)'], |
|
61 |
['one AND (two OR three)', '+one +(two three)'], |
|
62 |
['field:(one AND (two OR three))', '+field:one +(field:two field:three)'], |
|
63 |
['one AND (two OR [aaa vvv})', '+one +(two [aaa vvv})'], |
|
64 |
['one AND (f1:two OR f2:three) AND four', '+one +(f1:two f2:three) +four'], |
|
65 |
['one^1.23', 'one^1.23'], |
|
66 |
['(one AND two)^100.23', '(+one +two)^100.23'], |
|
67 |
['field:(one AND two)^100.23', '(+field:one +field:two)^100.23'], |
|
68 |
['field:(one AND [aaa bbb]^23.3)^100.23', '(+field:one +field:[aaa bbb]^23.3)^100.23'], |
|
69 |
['(REQ field:"one two three")^23', 'field:"one two three"^23.0'], |
|
70 |
['asdf~0.2', 'asdf~0.2'], |
|
71 |
['field:asdf~0.2', 'field:asdf~0.2'], |
|
72 |
['asdf~0.2^100.0', 'asdf~0.2^100.0'], |
|
73 |
['field:asdf~0.2^0.1', 'field:asdf~0.2^0.1'], |
|
74 |
['field:"asdf <> asdf|asdf"~4', 'field:"asdf <> asdf|asdf"~4'], |
|
75 |
['"one two three four five"~5', '"one two three four five"~5'], |
|
76 |
['ab?de', 'ab?de'], |
|
77 |
['ab*de', 'ab*de'], |
|
78 |
['asdf?*?asd*dsf?asfd*asdf?', 'asdf?*?asd*dsf?asfd*asdf?'], |
|
79 |
['field:a* AND field:(b*)', '+field:a* +field:b*'], |
|
80 |
['field:abc~ AND field:(b*)', '+field:abc~ +field:b*'], |
|
81 |
['asdf?*?asd*dsf?asfd*asdf?^20.0', 'asdf?*?asd*dsf?asfd*asdf?^20.0'], |
|
82 |
||
83 |
['*:xxx', 'xxx field:xxx f1:xxx f2:xxx'], |
|
84 |
['f1|f2:xxx', 'f1:xxx f2:xxx'], |
|
85 |
||
86 |
['*:asd~0.2', 'asd~0.2 field:asd~0.2 f1:asd~0.2 f2:asd~0.2'], |
|
87 |
['f1|f2:asd~0.2', 'f1:asd~0.2 f2:asd~0.2'], |
|
88 |
||
89 |
['*:a?d*^20.0', '(a?d* field:a?d* f1:a?d* f2:a?d*)^20.0'], |
|
90 |
['f1|f2:a?d*^20.0', '(f1:a?d* f2:a?d*)^20.0'], |
|
91 |
||
92 |
['*:"asdf <> xxx|yyy"', '"asdf <> xxx|yyy" field:"asdf <> xxx|yyy" f1:"asdf <> xxx|yyy" f2:"asdf <> xxx|yyy"'], |
|
93 |
['f1|f2:"asdf <> xxx|yyy"', 'f1:"asdf <> xxx|yyy" f2:"asdf <> xxx|yyy"'], |
|
94 |
['f1|f2:"asdf <> do|yyy"', 'f1:"asdf <> yyy" f2:"asdf <> yyy"'], |
|
95 |
['f1|f2:"do|cat"', 'f1:cat f2:cat'], |
|
96 |
||
97 |
['*:[bbb xxx]', '[bbb xxx] field:[bbb xxx] f1:[bbb xxx] f2:[bbb xxx]'], |
|
98 |
['f1|f2:[bbb xxx]', 'f1:[bbb xxx] f2:[bbb xxx]'], |
|
99 |
||
100 |
['*:(xxx AND bbb)', '+(xxx field:xxx f1:xxx f2:xxx) +(bbb field:bbb f1:bbb f2:bbb)'], |
|
101 |
['f1|f2:(xxx AND bbb)', '+(f1:xxx f2:xxx) +(f1:bbb f2:bbb)'], |
|
102 |
['asdf?*?asd*dsf?asfd*asdf?^20.0', 'asdf?*?asd*dsf?asfd*asdf?^20.0'], |
|
103 |
['"onewordphrase"', 'onewordphrase'], |
|
104 |
["who'd", "who'd"] |
|
105 |
]
|
|
106 |
||
107 |
pairs.each do |query_str, expected| |
|
108 |
assert_equal(expected, parser.parse(query_str).to_s("xxx")) |
|
109 |
end
|
|
110 |
end
|
|
111 |
||
112 |
def test_qp_with_standard_analyzer() |
|
113 |
parser = Ferret::QueryParser.new(:default_field => "xxx", |
|
114 |
:fields => ["xxx", "key"], |
|
115 |
:analyzer => StandardAnalyzer.new) |
|
116 |
pairs = [ |
|
117 |
['key:1234', 'key:1234'], |
|
118 |
['key:(1234 and Dave)', 'key:1234 key:dave'], |
|
119 |
['key:(1234)', 'key:1234'], |
|
120 |
['and the but they with', ''] |
|
121 |
]
|
|
122 |
||
123 |
pairs.each do |query_str, expected| |
|
124 |
assert_equal(expected, parser.parse(query_str).to_s("xxx")) |
|
125 |
end
|
|
126 |
||
127 |
end
|
|
128 |
||
129 |
def test_qp_changing_fields() |
|
130 |
parser = Ferret::QueryParser.new(:default_field => "xxx", |
|
131 |
:fields => ["xxx", "key"], |
|
132 |
:analyzer => WhiteSpaceAnalyzer.new) |
|
133 |
assert_equal('word key:word', parser.parse("*:word").to_s("xxx")) |
|
134 |
||
135 |
parser.fields = ["xxx", "one", "two", "three"] |
|
136 |
assert_equal('word one:word two:word three:word', |
|
137 |
parser.parse("*:word").to_s("xxx")) |
|
138 |
assert_equal('three:word four:word', |
|
139 |
parser.parse("three:word four:word").to_s("xxx")) |
|
140 |
end
|
|
141 |
||
142 |
def test_qp_allow_any_field() |
|
143 |
parser = Ferret::QueryParser.new(:default_field => "xxx", |
|
144 |
:fields => ["xxx", "key"], |
|
145 |
:analyzer => WhiteSpaceAnalyzer.new, |
|
146 |
:validate_fields => true) |
|
147 |
||
148 |
assert_equal('key:word', |
|
149 |
parser.parse("key:word song:word").to_s("xxx")) |
|
150 |
assert_equal('word key:word', parser.parse("*:word").to_s("xxx")) |
|
151 |
||
152 |
||
153 |
parser = Ferret::QueryParser.new(:default_field => "xxx", |
|
154 |
:fields => ["xxx", "key"], |
|
155 |
:analyzer => WhiteSpaceAnalyzer.new) |
|
156 |
||
157 |
assert_equal('key:word song:word', |
|
158 |
parser.parse("key:word song:word").to_s("xxx")) |
|
159 |
assert_equal('word key:word', parser.parse("*:word").to_s("xxx")) |
|
160 |
end
|
|
161 |
||
162 |
def do_test_query_parse_exception_raised(str) |
|
163 |
parser = Ferret::QueryParser.new(:default_field => "xxx", |
|
164 |
:fields => ["f1", "f2", "f3"], |
|
165 |
:handle_parse_errors => false) |
|
166 |
assert_raise(Ferret::QueryParser::QueryParseException, |
|
167 |
str + " should have failed") do |
|
168 |
parser.parse(str) |
|
169 |
end
|
|
170 |
end
|
|
171 |
||
172 |
def test_or_default |
|
173 |
parser = Ferret::QueryParser.new(:default_field => :*, |
|
174 |
:fields => [:x, :y], |
|
175 |
:or_default => false, |
|
176 |
:analyzer => StandardAnalyzer.new) |
|
177 |
pairs = [ |
|
178 |
['word', 'x:word y:word'], |
|
179 |
['word1 word2', '+(x:word1 y:word1) +(x:word2 y:word2)'] |
|
180 |
]
|
|
181 |
||
182 |
pairs.each do |query_str, expected| |
|
183 |
assert_equal(expected, parser.parse(query_str).to_s("")) |
|
184 |
end
|
|
185 |
end
|
|
186 |
||
187 |
def test_prefix_query |
|
188 |
parser = Ferret::QueryParser.new(:default_field => "xxx", |
|
189 |
:fields => ["xxx"], |
|
190 |
:analyzer => StandardAnalyzer.new) |
|
191 |
assert_equal(Ferret::Search::PrefixQuery, parser.parse("asdg*").class) |
|
192 |
assert_equal(Ferret::Search::WildcardQuery, parser.parse("a?dg*").class) |
|
193 |
assert_equal(Ferret::Search::WildcardQuery, parser.parse("a*dg*").class) |
|
194 |
assert_equal(Ferret::Search::WildcardQuery, parser.parse("adg*c").class) |
|
195 |
end
|
|
196 |
||
197 |
def test_bad_queries |
|
198 |
parser = Ferret::QueryParser.new(:default_field => "xxx", |
|
199 |
:fields => ["f1", "f2"]) |
|
200 |
||
201 |
pairs = [ |
|
202 |
['::*word', 'word'], |
|
203 |
['::*&)(*^&*(', ''], |
|
204 |
['::*&one)(*two(*&"', '"one two"~1'], |
|
205 |
[':', ''], |
|
206 |
['[, ]', ''], |
|
207 |
['{, }', ''], |
|
208 |
['!', ''], |
|
209 |
['+', ''], |
|
210 |
['~', ''], |
|
211 |
['^', ''], |
|
212 |
['-', ''], |
|
213 |
['|', ''], |
|
214 |
['<, >', ''], |
|
215 |
['=', ''], |
|
216 |
['<script>', 'script'] |
|
217 |
]
|
|
218 |
||
219 |
pairs.each do |query_str, expected| |
|
220 |
do_test_query_parse_exception_raised(query_str) |
|
221 |
assert_equal(expected, parser.parse(query_str).to_s("xxx")) |
|
222 |
end
|
|
223 |
end
|
|
224 |
||
225 |
def test_use_keywords_switch |
|
226 |
analyzer = LetterAnalyzer.new |
|
227 |
parser = Ferret::QueryParser.new(:analyzer => analyzer, |
|
228 |
:default_field => "xxx") |
|
229 |
assert_equal("+www (+xxx +yyy) -zzz", |
|
230 |
parser.parse("REQ www (xxx AND yyy) OR NOT zzz").to_s("xxx")) |
|
231 |
||
232 |
parser = Ferret::QueryParser.new(:analyzer => analyzer, |
|
233 |
:default_field => "xxx", |
|
234 |
:use_keywords => false) |
|
235 |
assert_equal("req www (xxx and yyy) or not zzz", |
|
236 |
parser.parse("REQ www (xxx AND yyy) OR NOT zzz").to_s("xxx")) |
|
237 |
end
|
|
238 |
end
|