File size: 6,791 Bytes
768d31a
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
# Copyright (c) Microsoft Corporation. 
# Licensed under the MIT license.
import numpy as np
import os
from tree_sitter import Language, Parser

from src.evaluator.CodeBLEU.parser import DFG_python, DFG_java, DFG_ruby, DFG_go, DFG_php, DFG_javascript, DFG_csharp
from src.evaluator.CodeBLEU.parser import (remove_comments_and_docstrings,
                                           tree_to_token_index,
                                           index_to_code_token)

root_dir = os.path.dirname(__file__)

dfg_function = {
    'c': DFG_java,
    'python': DFG_python,
    'java': DFG_java,
    'ruby': DFG_ruby,
    'go': DFG_go,
    'php': DFG_php,
    'javascript': DFG_javascript,
    'c_sharp': DFG_csharp,
}


def my_dataflow_match(references, candidates, lang):
    LANGUAGE = Language(root_dir + '/parser/languages.so', lang)
    parser = Parser()
    parser.set_language(LANGUAGE)
    parser = [parser, dfg_function[lang]]
    match_count = 0
    total_count = 0
    candidate_scores = []
    for i in range(len(candidates)):
        scores = []
        references_sample = references[i]
        candidate = candidates[i]
        for reference in references_sample:
            try:
                candidate = remove_comments_and_docstrings(candidate, 'java')
            except:
                pass
            try:
                reference = remove_comments_and_docstrings(reference, 'java')
            except:
                pass

            cand_dfg = get_data_flow(candidate, parser)
            ref_dfg = get_data_flow(reference, parser)

            normalized_cand_dfg = normalize_dataflow(cand_dfg)
            normalized_ref_dfg = normalize_dataflow(ref_dfg)

            if len(normalized_ref_dfg) > 0:
                total_count += len(normalized_ref_dfg)
                current_match_count = 0
                for dataflow in normalized_ref_dfg:
                    if dataflow in normalized_cand_dfg:
                        match_count += 1
                        normalized_cand_dfg.remove(dataflow)
                        current_match_count += 1
                scores.append(float(current_match_count) / len(normalized_ref_dfg))
            else:
                scores.append(0.)
        candidate_scores.append(max(scores) if len(scores) > 0 else 0.)
    return np.mean(candidate_scores) if len(candidates) > 0 else 0.


def calc_dataflow_match(references, candidate, lang):
    return corpus_dataflow_match([references], [candidate], lang)


def corpus_dataflow_match(references, candidates, lang):
    LANGUAGE = Language(root_dir + '/parser/languages.so', lang)
    parser = Parser()
    parser.set_language(LANGUAGE)
    parser = [parser, dfg_function[lang]]
    match_count = 0
    total_count = 0
    scores = []
    for i in range(len(candidates)):
        references_sample = references[i]
        candidate = candidates[i]
        for reference in references_sample:
            try:
                candidate = remove_comments_and_docstrings(candidate, 'java')
            except:
                pass
            try:
                reference = remove_comments_and_docstrings(reference, 'java')
            except:
                pass

            cand_dfg = get_data_flow(candidate, parser)
            ref_dfg = get_data_flow(reference, parser)

            normalized_cand_dfg = normalize_dataflow(cand_dfg)
            normalized_ref_dfg = normalize_dataflow(ref_dfg)

            if len(normalized_ref_dfg) > 0:
                total_count += len(normalized_ref_dfg)
                current_match_count = 0
                for dataflow in normalized_ref_dfg:
                    if dataflow in normalized_cand_dfg:
                        match_count += 1
                        normalized_cand_dfg.remove(dataflow)
                        current_match_count += 1
                scores.append(float(current_match_count) / len(normalized_ref_dfg))
            else:
                scores.append(0.)
    if total_count == 0:
        print(
            "WARNING: There is no reference data-flows extracted from the whole corpus, "
            "and the data-flow match score degenerates to 0. Please consider ignoring this score."
        )
        return 0
    score = match_count / total_count
    return score


def get_data_flow(code, parser):
    try:
        tree = parser[0].parse(bytes(code, 'utf8'))
        root_node = tree.root_node
        tokens_index = tree_to_token_index(root_node)
        code = code.split('\n')
        code_tokens = [index_to_code_token(x, code) for x in tokens_index]
        index_to_code = {}
        for idx, (index, code) in enumerate(zip(tokens_index, code_tokens)):
            index_to_code[index] = (idx, code)
        try:
            DFG, _ = parser[1](root_node, index_to_code, {})
        except:
            DFG = []
        DFG = sorted(DFG, key=lambda x: x[1])
        indexs = set()
        for d in DFG:
            if len(d[-1]) != 0:
                indexs.add(d[1])
            for x in d[-1]:
                indexs.add(x)
        new_DFG = []
        for d in DFG:
            if d[1] in indexs:
                new_DFG.append(d)
        codes = code_tokens
        dfg = new_DFG
    except:
        codes = code.split()
        dfg = []
    # merge nodes
    dic = {}
    for d in dfg:
        if d[1] not in dic:
            dic[d[1]] = d
        else:
            dic[d[1]] = (d[0], d[1], d[2], list(set(dic[d[1]][3] + d[3])), list(set(dic[d[1]][4] + d[4])))
    DFG = []
    for d in dic:
        DFG.append(dic[d])
    dfg = DFG
    return dfg


def normalize_dataflow_item(dataflow_item):
    var_name = dataflow_item[0]
    var_pos = dataflow_item[1]
    relationship = dataflow_item[2]
    par_vars_name_list = dataflow_item[3]
    par_vars_pos_list = dataflow_item[4]

    var_names = list(set(par_vars_name_list + [var_name]))
    norm_names = {}
    for i in range(len(var_names)):
        norm_names[var_names[i]] = 'var_' + str(i)

    norm_var_name = norm_names[var_name]
    relationship = dataflow_item[2]
    norm_par_vars_name_list = [norm_names[x] for x in par_vars_name_list]

    return (norm_var_name, relationship, norm_par_vars_name_list)


def normalize_dataflow(dataflow):
    var_dict = {}
    i = 0
    normalized_dataflow = []
    for item in dataflow:
        var_name = item[0]
        relationship = item[2]
        par_vars_name_list = item[3]
        for name in par_vars_name_list:
            if name not in var_dict:
                var_dict[name] = 'var_' + str(i)
                i += 1
        if var_name not in var_dict:
            var_dict[var_name] = 'var_' + str(i)
            i += 1
        normalized_dataflow.append((var_dict[var_name], relationship, [var_dict[x] for x in par_vars_name_list]))
    return normalized_dataflow