andstor commited on
Commit
f90a46e
1 Parent(s): 80dd4fa

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +76 -0
README.md CHANGED
@@ -39,4 +39,80 @@ configs:
39
  path: data/test-*
40
  - split: validation
41
  path: data/validation-*
 
 
 
 
 
 
 
 
42
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
39
  path: data/test-*
40
  - split: validation
41
  path: data/validation-*
42
+ task_categories:
43
+ - text-generation
44
+ language:
45
+ - en
46
+ tags:
47
+ - unit test
48
+ - java
49
+ - code
50
  ---
51
+
52
+ ## Dataset Description
53
+ Microsoft created the methods2test dataset, consisting of Java Junit test cases with its corresponding focal methods.
54
+ It contains 780k pairs of JUnit test cases and focal methods which were extracted from a total of 91K
55
+ Java open source project hosted on GitHub.
56
+
57
+ This is an alternative version of the methods2test dataset. It provides convenient access to the different context levels based on the raw source code (e.g. newlines are preserved). The test cases and associated classes are also made available.
58
+
59
+ The mapping between test case and focal methods are based heuristics rules and Java developer's best practice.
60
+
61
+ More information could be found here:
62
+ - [methods2test Github repo](https://github.com/microsoft/methods2test)
63
+ - [Methods2Test: A dataset of focal methods mapped to test cases](https://arxiv.org/pdf/2203.12776.pdf)
64
+
65
+ ## Dataset Schema
66
+ ```
67
+ t: <TEST_CASE>
68
+ t_tc: <TEST_CASE> <TEST_CLASS_NAME>
69
+ fm: <FOCAL_METHOD>
70
+ fm_fc: <FOCAL_CLASS_NAME> <FOCAL_METHOD>
71
+ fm_fc_c: <FOCAL_CLASS_NAME> <FOCAL_METHOD> <CONTRSUCTORS>
72
+ fm_fc_c_m: <FOCAL_CLASS_NAME> <FOCAL_METHOD> <CONTRSUCTORS> <METHOD_SIGNATURES>
73
+ fm_fc_c_m_f: <FOCAL_CLASS_NAME> <FOCAL_METHOD> <CONTRSUCTORS> <METHOD_SIGNATURES> <FIELDS>
74
+ ```
75
+
76
+ ## Focal Context
77
+ - fm: this representation incorporates exclusively the source
78
+ code of the focal method. Intuitively, this contains the most
79
+ important information for generating accurate test cases for
80
+ the given method.
81
+ - fm+fc: this representations adds the focal class name, which
82
+ can provide meaningful semantic information to the model.
83
+ - fm+fc+c: this representation adds the signatures of the constructor methods of the focal class. The idea behind this
84
+ augmentation is that the test case may require instantiating
85
+ an object of the focal class in order to properly test the focal
86
+ method.
87
+ - fm+fc+c+m: this representation adds the signatures of the
88
+ other public methods in the focal class. The rationale which
89
+ motivated this inclusion is that the test case may need to
90
+ invoke other auxiliary methods within the class (e.g., getters,
91
+ setters) to set up or tear down the testing environment.
92
+ - fm+fc+c+m+f : this representation adds the public fields of
93
+ the focal class. The motivation is that test cases may need to
94
+ inspect the status of the public fields to properly test a focal
95
+ method.
96
+
97
+ ![image/png](https://huggingface.co/datasets/andstor/methods2test/resolve/main/figure-1-focal-context.png)
98
+
99
+ The different levels of focal contexts are the following:
100
+ ```
101
+ T: test case
102
+ T_TC: test case + test class name
103
+ FM: focal method
104
+ FM_FC: focal method + focal class name
105
+ FM_FC_C: focal method + focal class name + constructor signatures
106
+ FM_FC_C_M: focal method + focal class name + constructor signatures + public method signatures
107
+ FM_FC_C_M_F: focal method + focal class name + constructor signatures + public method signatures + public fields
108
+ ```
109
+
110
+ ## Limitations
111
+ The original authors validate the heuristics by inspecting a
112
+ statistically significant sample (confidence level of 95% within 10%
113
+ margin of error) of 97 samples from the training set. Two authors
114
+ independently evaluated the sample, then met to discuss the disagreements. We found that 90.72% of the samples have a correct
115
+ link between the test case and the corresponding focal method
116
+
117
+ ## Contribution
118
+ All thanks to the original authors.