|
1
|
|
|
import functools |
|
2
|
|
|
from itertools import combinations |
|
3
|
|
|
|
|
4
|
|
|
from coalib.misc.StringConverter import StringConverter |
|
5
|
|
|
from coalib.results.HiddenResult import HiddenResult |
|
6
|
|
|
from coalib.settings.Setting import typed_ordered_dict, path_list |
|
7
|
|
|
from coalib.collecting.Collectors import collect_dirs |
|
8
|
|
|
from coalib.bears.GlobalBear import GlobalBear |
|
9
|
|
|
from bears.codeclone_detection.ClangCountVectorCreator import ( |
|
10
|
|
|
ClangCountVectorCreator) |
|
11
|
|
|
from bears.codeclone_detection.ClangCountingConditions import condition_dict |
|
12
|
|
|
from bears.codeclone_detection.CloneDetectionRoutines import ( |
|
13
|
|
|
compare_functions, |
|
14
|
|
|
get_count_matrices) |
|
15
|
|
|
|
|
16
|
|
|
|
|
17
|
|
|
# counting_condition_dict is a function object generated by typed_dict. This |
|
18
|
|
|
# function takes a setting and creates a dictionary out of it while it |
|
19
|
|
|
# converts all keys to counting condition function objects (via the |
|
20
|
|
|
# condition_dict) and all values to floats while unset values default to 1. |
|
21
|
|
|
counting_condition_dict = typed_ordered_dict( |
|
22
|
|
|
lambda setting: condition_dict[str(setting).lower()], |
|
23
|
|
|
float, |
|
24
|
|
|
1) |
|
25
|
|
|
|
|
26
|
|
|
default_cc_dict = counting_condition_dict(StringConverter( |
|
27
|
|
|
""" |
|
28
|
|
|
used: 0, |
|
29
|
|
|
returned: 1.4, |
|
30
|
|
|
is_condition: 0, |
|
31
|
|
|
in_condition: 1.4, |
|
32
|
|
|
in_second_level_condition: 1.4, |
|
33
|
|
|
in_third_level_condition: 1.0, |
|
34
|
|
|
is_assignee: 0, |
|
35
|
|
|
is_assigner: 0.6, |
|
36
|
|
|
loop_content: 0, |
|
37
|
|
|
second_level_loop_content, |
|
38
|
|
|
third_level_loop_content, |
|
39
|
|
|
is_param: 2, |
|
40
|
|
|
is_called: 1.4, |
|
41
|
|
|
is_call_param: 0.0, |
|
42
|
|
|
in_sum: 2.0, |
|
43
|
|
|
in_product: 0, |
|
44
|
|
|
in_binary_operation, |
|
45
|
|
|
member_accessed""")) |
|
46
|
|
|
|
|
47
|
|
|
|
|
48
|
|
|
def get_difference(function_pair, |
|
49
|
|
|
count_matrices, |
|
50
|
|
|
average_calculation, |
|
51
|
|
|
poly_postprocessing, |
|
52
|
|
|
exp_postprocessing): |
|
53
|
|
|
""" |
|
54
|
|
|
Retrieves the difference between two functions using the munkres algorithm. |
|
55
|
|
|
|
|
56
|
|
|
:param function_pair: A tuple containing both indices for the |
|
57
|
|
|
count_matrices dictionary. |
|
58
|
|
|
:param count_matrices: A dictionary holding CMs. |
|
59
|
|
|
:param average_calculation: If set to true the difference calculation |
|
60
|
|
|
function will take the average of all variable |
|
61
|
|
|
differences as the difference, else it will |
|
62
|
|
|
normalize the function as a whole and thus |
|
63
|
|
|
weighting in variables dependent on their size. |
|
64
|
|
|
:param poly_postprocessing: If set to true, the difference value of big |
|
65
|
|
|
function pairs will be reduced using a |
|
66
|
|
|
polynomial approach. |
|
67
|
|
|
:param exp_postprocessing: If set to true, the difference value of big |
|
68
|
|
|
function pairs will be reduced using an |
|
69
|
|
|
exponential approach. |
|
70
|
|
|
:return: A tuple containing both function ids and their |
|
71
|
|
|
difference. |
|
72
|
|
|
""" |
|
73
|
|
|
function_1, function_2 = function_pair |
|
74
|
|
|
return (function_1, |
|
75
|
|
|
function_2, |
|
76
|
|
|
compare_functions(count_matrices[function_1], |
|
77
|
|
|
count_matrices[function_2], |
|
78
|
|
|
average_calculation, |
|
79
|
|
|
poly_postprocessing, |
|
80
|
|
|
exp_postprocessing)) |
|
81
|
|
|
|
|
82
|
|
|
|
|
83
|
|
|
class ClangFunctionDifferenceBear(GlobalBear): |
|
84
|
|
|
def run(self, |
|
85
|
|
|
counting_conditions: counting_condition_dict=default_cc_dict, |
|
86
|
|
|
average_calculation: bool=False, |
|
87
|
|
|
poly_postprocessing: bool=True, |
|
88
|
|
|
exp_postprocessing: bool=False, |
|
89
|
|
|
extra_include_paths: path_list=()): |
|
90
|
|
|
''' |
|
91
|
|
|
Retrieves similarities for code clone detection. Those can be reused in |
|
92
|
|
|
another bear to produce results. |
|
93
|
|
|
|
|
94
|
|
|
Postprocessing may be done because small functions are less likely to |
|
95
|
|
|
be clones at the same difference value than big functions which may |
|
96
|
|
|
provide a better refactoring opportunity for the user. |
|
97
|
|
|
|
|
98
|
|
|
:param counting_conditions: A comma seperated list of counting |
|
99
|
|
|
conditions. Possible values are: used, |
|
100
|
|
|
returned, is_condition, in_condition, |
|
101
|
|
|
in_second_level_condition, |
|
102
|
|
|
in_third_level_condition, is_assignee, |
|
103
|
|
|
is_assigner, loop_content, |
|
104
|
|
|
second_level_loop_content, |
|
105
|
|
|
third_level_loop_content, is_param, |
|
106
|
|
|
in_sum, in_product, in_binary_operation, |
|
107
|
|
|
member_accessed. |
|
108
|
|
|
Weightings can be assigned to each |
|
109
|
|
|
condition due to providing a dict |
|
110
|
|
|
value, i.e. having used weighted in |
|
111
|
|
|
half as much as other conditions would |
|
112
|
|
|
simply be: "used: 0.5, is_assignee". |
|
113
|
|
|
Weightings default to 1 if unset. |
|
114
|
|
|
:param average_calculation: If set to true the difference calculation |
|
115
|
|
|
function will take the average of all |
|
116
|
|
|
variable differences as the difference, |
|
117
|
|
|
else it will normalize the function as a |
|
118
|
|
|
whole and thus weighting in variables |
|
119
|
|
|
dependent on their size. |
|
120
|
|
|
:param poly_postprocessing: If set to true, the difference value of big |
|
121
|
|
|
function pairs will be reduced using a |
|
122
|
|
|
polynomial approach. |
|
123
|
|
|
:param extra_include_paths: A list containing additional include paths. |
|
124
|
|
|
:param exp_postprocessing: If set to true, the difference value of big |
|
125
|
|
|
function pairs will be reduced using an |
|
126
|
|
|
exponential approach. |
|
127
|
|
|
''' |
|
128
|
|
|
self.debug("Using the following counting conditions:") |
|
129
|
|
|
for key, val in counting_conditions.items(): |
|
130
|
|
|
self.debug(" *", key.__name__, "(weighting: {})".format(val)) |
|
131
|
|
|
|
|
132
|
|
|
self.debug("Creating count matrices...") |
|
133
|
|
|
count_matrices = get_count_matrices( |
|
134
|
|
|
ClangCountVectorCreator(list(counting_conditions.keys()), |
|
135
|
|
|
list(counting_conditions.values())), |
|
136
|
|
|
list(self.file_dict.keys()), |
|
137
|
|
|
lambda prog: self.debug("{:2.4f}%...".format(prog)), |
|
138
|
|
|
self.section["files"].origin, |
|
139
|
|
|
collect_dirs(extra_include_paths)) |
|
140
|
|
|
|
|
141
|
|
|
self.debug("Calculating differences...") |
|
142
|
|
|
|
|
143
|
|
|
differences = [] |
|
144
|
|
|
function_count = len(count_matrices) |
|
145
|
|
|
# Thats n over 2, hardcoded to simplify calculation |
|
146
|
|
|
combination_length = function_count * (function_count-1) / 2 |
|
147
|
|
|
partial_get_difference = functools.partial( |
|
148
|
|
|
get_difference, |
|
149
|
|
|
count_matrices=count_matrices, |
|
150
|
|
|
average_calculation=average_calculation, |
|
151
|
|
|
poly_postprocessing=poly_postprocessing, |
|
152
|
|
|
exp_postprocessing=exp_postprocessing) |
|
153
|
|
|
|
|
154
|
|
|
for i, elem in enumerate( |
|
155
|
|
|
map(partial_get_difference, |
|
156
|
|
|
[(f1, f2) for f1, f2 in combinations(count_matrices, 2)])): |
|
157
|
|
|
if i % 50 == 0: |
|
158
|
|
|
self.debug("{:2.4f}%...".format(100*i/combination_length)) |
|
159
|
|
|
differences.append(elem) |
|
160
|
|
|
|
|
161
|
|
|
yield HiddenResult(self, differences) |
|
162
|
|
|
yield HiddenResult(self, count_matrices) |
|
163
|
|
|
|
coala-analyzer /
coala
