clause_extractor 0.0.7 → 0.0.8
Sign up to get free protection for your applications and to get access to all the features.
- data/lib/clause_extractor.rb +12 -26
- metadata +1 -1
data/lib/clause_extractor.rb
CHANGED
|
@@ -6,20 +6,18 @@ class ClauseExtractor
|
|
|
6
6
|
have_has = "(have|has|haven't|hasn't)"
|
|
7
7
|
contractions = "it's|he's|she's|[a-z]{1,4}'ve"
|
|
8
8
|
|
|
9
|
-
|
|
10
|
-
|
|
11
9
|
@tense_regexes = {
|
|
12
10
|
|
|
13
11
|
'third' => {
|
|
14
12
|
"simple present"
|
|
15
13
|
=> [
|
|
16
|
-
/\b(he|she|it)\s+search(s)?\b/i,
|
|
17
|
-
/\bsearch(s)?\s+(it|them|him|her|me|you|us)\b/i
|
|
14
|
+
/\b(he|she|it)\s+search(s)?\b/i, #he arrives
|
|
15
|
+
/\bsearch(s)?\s+(it|them|him|her|me|you|us)\b/i #adapts it
|
|
18
16
|
]
|
|
19
17
|
},
|
|
20
18
|
'infinitive' => {
|
|
21
19
|
"simple present" => [
|
|
22
|
-
/\b((I|you|they|we|to)\s+)*+search\b/i,
|
|
20
|
+
/\b((I|you|they|we|to)\s+)*+search\b/i, #arrive
|
|
23
21
|
],
|
|
24
22
|
|
|
25
23
|
"subjunctive future" => [
|
|
@@ -35,39 +33,28 @@ class ClauseExtractor
|
|
|
35
33
|
"going to-future" => [ /\b(#{pronouns}\s+)*(am|are|i'm|[a-z]{1,4}'re|[a-z]{1,4}'s)\s+(not\s+)*going\s+to\s+search/i], #they are going to cry
|
|
36
34
|
},
|
|
37
35
|
'gerund' => {
|
|
38
|
-
"conditional perfect progressive" => [/\b(#{pronouns}\s+)*would\s+(not\s+)*have\s+(not\s+)*been\s+search/i],
|
|
36
|
+
"conditional perfect progressive" => [/\b(#{pronouns}\s+)*would\s+(not\s+)*have\s+(not\s+)*been\s+search/i], #I would have been searching
|
|
39
37
|
"present perfect progressive" => [/\b(#{pronouns}\s+)*([a-z]{1,4}'ve|have|has)(n't)*\s+(#{pronouns}\s+)*(not\s+)*(#{present_perfect}\s+)*been\s+search/i], #I have been searching
|
|
40
38
|
"conditional progressive" => [/\b(#{pronouns}\s+)*(would|[a-z]{1,4}'d)\s+(not\s+)*be\s+search/i], #I would be searching (I'd)
|
|
41
39
|
"future progressive" => [
|
|
42
|
-
|
|
43
|
-
|
|
44
|
-
],
|
|
40
|
+
/\b((#{pronouns})\s+)*(will|[a-z]{1,4}'ll)\s+(not\s+)*be\s+search/i,
|
|
41
|
+
/\bwill\s+(#{pronouns}\s+)(not\s+)*be\s+search/i,
|
|
42
|
+
], #I will be searching
|
|
45
43
|
"past progressive" => [/\b(#{pronouns}\s+)*(was|were)(n't)*\s+(not\s+)*search/i], #I was searching
|
|
46
44
|
|
|
47
45
|
"present progressive" => [/\b(#{pronouns}\s+)*((am|are|is|i'm|\b[a-z]{1,4}'re|\b[a-z]{1,4}'s)\s+)*(not\s+)*search/i], #I'm rising
|
|
48
|
-
|
|
49
46
|
},
|
|
50
47
|
"past-participle" => {
|
|
51
48
|
"conditional perfect" => [/\b(#{pronouns}\s+)*(would|[a-z]{1,4}'d)\s+(not\s+)*have\s+(not\s+)*search/i], #I would not search
|
|
52
|
-
"future perfect" => [/\b(#{pronouns}\s+)*(will|[a-z]{1,4}'ll)\s+have\s+search/i],
|
|
49
|
+
"future perfect" => [/\b(#{pronouns}\s+)*(will|[a-z]{1,4}'ll)\s+have\s+search/i], #I'll have arisen
|
|
53
50
|
"past perfect" => [/\b(#{pronouns}\s+)*(had|[a-z]{1,4}'d)\s+(#{pronouns}\s+)*(not\s+)*((#{present_perfect})\s+)*search/i], #I had arisen
|
|
54
51
|
"present perfect" => [/\b(#{pronouns}\s+)*#{have_has}\s+(#{pronouns}\s+)*(not\s+)*((just|already|ever)\s+)*search/], #Have you seen
|
|
55
|
-
"subjunctive past" => [/\bif\s+(i|you|he|she|it|they|we)\s+search/i],
|
|
56
|
-
"simple past" => [/\b#{pronouns}\s+search/i]
|
|
52
|
+
"subjunctive past" => [/\bif\s+(i|you|he|she|it|they|we)\s+search/i], #if I arose
|
|
53
|
+
"simple past" => [/\b#{pronouns}\s+search/i] #you chose
|
|
57
54
|
},
|
|
58
|
-
|
|
59
|
-
|
|
60
|
-
|
|
61
|
-
|
|
62
|
-
|
|
63
|
-
|
|
64
|
-
|
|
65
|
-
# # "present perfect" => [/^\s*search\b/i], #arisen
|
|
66
|
-
# # "simple past" => [/^\s*search\b/i] #arose
|
|
55
|
+
#"present perfect" => [/^\s*search\b/i], #arisen
|
|
56
|
+
#"simple past" => [/^\s*search\b/i] #arose
|
|
67
57
|
}
|
|
68
|
-
|
|
69
|
-
|
|
70
|
-
|
|
71
58
|
def self.get_match_start_index(verb, match, index)
|
|
72
59
|
#get start position of last occurence of verb in match
|
|
73
60
|
verb_index_in_match = match.index /#{verb}(?!.*#{verb})/i
|
|
@@ -139,7 +126,6 @@ class ClauseExtractor
|
|
|
139
126
|
if ranges.each.select{|r| r.include?(lo) || r.include?(hi)}.size == 0
|
|
140
127
|
ranges << range
|
|
141
128
|
end
|
|
142
|
-
|
|
143
129
|
ranges
|
|
144
130
|
end
|
|
145
131
|
end
|