Do-LSTMs-learn-Syntax/experiments/make_templates.py at master · MirunaPislar/Do-LSTMs-learn-Syntax · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
from templates import TenseAgreementTemplates, CoordinationTemplates
from terminals import TenseAgreementTerminals, CoordinationTerminals
import argparse
import os
import pandas as pd
import random

parser = argparse.ArgumentParser(
    description="Create targeted syntactic templates.")

parser.add_argument("--filename", type=str, default="data/my_template.tab",
                    help="Path where to save the templates.")
parser.add_argument("--type", type=str, default="tense",
                    help="Type of template to construct. "
                         "Choose between coord or tense.")
args = parser.parse_args()


class MakeAgreementTemplate:
    def __init__(self):
        if args.type == "coord":
            self.terminals = CoordinationTerminals().terminals
            self.rules = CoordinationTemplates().rules
        elif args.type == "tense":
            self.terminals = TenseAgreementTerminals().terminals
            self.rules = TenseAgreementTemplates().rules
        else:
            raise ValueError("Unknown template name %s. Please, choose"
                             " between coord or tense." % args.type)

    @staticmethod
    def switch_tense(words, preterms_idx):
        valid_verb_switches = {
            "presBe": {"sg": ["will", "did", "has"],
                       "pl": ["will", "did", "have"]},
            "past": {"sg": ["is", "was"],
                     "pl": ["are", "were"]},
            "future": {"sg": ["is"],
                       "pl": ["are"]}}

        new_words = []
        my_number = preterms_idx.split("_")[-1]
        my_tense = preterms_idx.split("_")[-2]
        valid_verbs = valid_verb_switches[my_tense][my_number]

        for word in words:
            splits = word.split()
            sampled_verb = random.choice(valid_verbs)
            if len(splits) > 1:
                new_words.append(" ".join(sampled_verb + splits[1:]))
            else:
                new_words.append(sampled_verb)
        return new_words

    @staticmethod
    def switch_number(words, preterms_idx):
        new_words = []
        is_verb = preterms_idx[-1] == "V"
        is_mod = len(preterms_idx) > 2 and preterms_idx[-3:] == "MOD"
        for word in words:
            splits = word.split()
            if splits[0] == "is":
                new_words.append(" ".join(["are"] + splits[1:]))
            elif splits[0] == "are":
                new_words.append(" ".join(["is"] + splits[1:]))
            elif splits[0] == "was":
                new_words.append(" ".join(["were"] + splits[1:]))
            elif splits[0] == "were":
                new_words.append(" ".join(["was"] + splits[1:]))
            elif splits[0] == "has":
                new_words.append(" ".join(["have"] + splits[1:]))
            elif splits[0] == "have":
                new_words.append(" ".join(["has"] + splits[1:]))
            elif is_mod:
                if len(splits) > 1:
                    if splits[0][-2:] == "es":
                        new_words.append(" ".join([splits[0][:-2]] + splits[1:]))
                    else:
                        new_words.append(" ".join([splits[0] + "es"] + splits[1:]))
                else:
                    if word[-2:] == "es":
                        new_words.append(word[:-2])
                    else:
                        new_words.append(word + "es")
            elif is_verb:
                if len(splits) > 1:
                    if splits[0][-1] == "s":
                        new_words.append(" ".join([splits[0][:-1]] + splits[1:]))
                    else:
                        new_words.append(" ".join([splits[0] + "s"] + splits[1:]))
                else:
                    if word[-1] == "s":
                        new_words.append(word[:-1])
                    else:
                        new_words.append(word + "s")
            elif word[-4:] == "self":
                new_words.append("themselves")
            else:
                new_words.append(word + "s")
        return new_words

    def switch_numbers(self, base_sent, variables, preterms):
        new_sent = base_sent[:]
        for idx in variables:
            if args.type == "coord":
                new_sent[idx] = self.switch_number(new_sent[idx], preterms[idx])
            elif args.type == "tense":
                new_sent[idx] = self.switch_tense(new_sent[idx], preterms[idx])
            else:
                raise ValueError("Unknown value for template type: %s."
                                 "Choose between tense and coord." % args.type)
        return new_sent

    def make_variable_sentences(self, preterms, match):
        base_sent = [self.terminals[p] for p in preterms]
        grammatical = base_sent[:]
        ungrammatical = self.switch_numbers(grammatical, match[1], preterms)
        return [grammatical, ungrammatical]


class MakeTestCase:
    def __init__(self, template, test_case):
        self.template = template
        self.test_case = test_case
        self.sent_templates = self.get_rules()

    def get_rules(self):
        sent_templates = {"pattern": [], "sent": [], "sent_alt": []}
        preterminals, templates = self.template.rules[self.test_case]
        if templates is not None:
            sentences = self.template.make_variable_sentences(
                preterminals, templates["match"])
            grammatical = list(self.expand_sent(sentences[0]))
            ungrammatical = list(self.expand_sent(sentences[1]))
            for i in range(len(grammatical)):
                sent_templates["pattern"].append(self.test_case)
                sent_templates["sent"].append(grammatical[i])
                sent_templates["sent_alt"].append(ungrammatical[i])
        return sent_templates

    def expand_sent(self, sent, partial="", switch_ds=False):
        if len(sent) == 1:
            for word in sent[0]:
                if switch_ds:
                    splits = partial.split(" ")
                    no = splits[0]
                    the = splits[3]
                    splits = partial.split()
                    partial_one = " ".join([x for x in splits[1:3]])
                    partial_two = " ".join([x for x in splits[4:]])
                    yield " ".join([the, partial_one, no, partial_two, word])
                elif word not in partial:
                    yield partial + word
                else:
                    yield "None"
        else:
            for word in sent[0]:
                for x in self.expand_sent(
                        sent=sent[1:],
                        partial=partial + word + " ",
                        switch_ds=switch_ds):
                    if x != "None":
                        yield x


def main():
    if os.path.isfile(args.filename):
        os.remove(args.filename)

    test_cases = MakeAgreementTemplate()
    for case in test_cases.rules.keys():
        sentences = MakeTestCase(test_cases.rules, case)
        df = pd.DataFrame.from_dict(sentences.sent_templates)
        if os.path.isfile(args.filename):
            df.to_csv(
                args.filename, mode="a", header=False, index=False, sep="\t")
        else:
            df.to_csv(
                args.filename, mode="a", header=True, index=False, sep="\t")


if __name__ == "__main__":
    main()