Edward J. Schwartz commited on
Commit
65df539
1 Parent(s): f57ae1a

Add bylibrary

Browse files
Files changed (1) hide show
  1. oo-method-test-split.py +52 -5
oo-method-test-split.py CHANGED
@@ -2,11 +2,19 @@
2
 
3
  import datasets
4
 
 
 
 
 
5
  import pyarrow as pa
6
  import pyarrow.parquet as pq
7
 
8
  BASE_DATASET = "ejschwartz/oo-method-test"
9
 
 
 
 
 
10
  class OOMethodTestDataset(datasets.ArrowBasedBuilder):
11
 
12
  BUILDER_CONFIGS = [
@@ -24,6 +32,11 @@ class OOMethodTestDataset(datasets.ArrowBasedBuilder):
24
  name="byfuncname",
25
  version=datasets.Version("1.0.0"),
26
  description="Split by function name",
 
 
 
 
 
27
  )
28
 
29
  ]
@@ -35,25 +48,28 @@ class OOMethodTestDataset(datasets.ArrowBasedBuilder):
35
  return datasets.DatasetInfo()
36
 
37
  def _split_generators(self, dl_manager):
38
- ds = datasets.load_dataset(BASE_DATASET)
39
 
40
  #print(files)
41
  #print(downloaded_files)
42
 
 
 
 
43
  if self.config.name == "combined":
44
 
45
  return [
46
  datasets.SplitGenerator(
47
  name="combined",
48
  gen_kwargs={
49
- "ds": ds['combined'],
50
  },
51
  ),
52
  ]
53
 
54
  elif self.config.name == "byrow":
55
 
56
- ds = ds['combined'].train_test_split(test_size=0.1, seed=42)
57
  #print(ds)
58
 
59
  return [
@@ -74,8 +90,6 @@ class OOMethodTestDataset(datasets.ArrowBasedBuilder):
74
 
75
  elif self.config.name == "byfuncname":
76
 
77
- ds = ds['combined']
78
-
79
  unique_names = ds.unique('Name')
80
  nameds = datasets.Dataset.from_dict({'Name': unique_names})
81
 
@@ -100,6 +114,39 @@ class OOMethodTestDataset(datasets.ArrowBasedBuilder):
100
  ),
101
 
102
  ]
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
103
 
104
  else:
105
  assert False
 
2
 
3
  import datasets
4
 
5
+ import itertools
6
+
7
+ import os
8
+
9
  import pyarrow as pa
10
  import pyarrow.parquet as pq
11
 
12
  BASE_DATASET = "ejschwartz/oo-method-test"
13
 
14
+ def setexe(r):
15
+ r['Dirname'], r['Exename'] = os.path.split(r['Binary'])
16
+ return r
17
+
18
  class OOMethodTestDataset(datasets.ArrowBasedBuilder):
19
 
20
  BUILDER_CONFIGS = [
 
32
  name="byfuncname",
33
  version=datasets.Version("1.0.0"),
34
  description="Split by function name",
35
+ ),
36
+ datasets.BuilderConfig(
37
+ name="bylibrary",
38
+ version=datasets.Version("1.0.0"),
39
+ description="Split so that library functions (those appearing in >1 exe) are used for training, and non-library functions are used for testing",
40
  )
41
 
42
  ]
 
48
  return datasets.DatasetInfo()
49
 
50
  def _split_generators(self, dl_manager):
51
+ ds = datasets.load_dataset(BASE_DATASET)['combined']
52
 
53
  #print(files)
54
  #print(downloaded_files)
55
 
56
+
57
+ ds = ds.map(setexe, batched=False)
58
+
59
  if self.config.name == "combined":
60
 
61
  return [
62
  datasets.SplitGenerator(
63
  name="combined",
64
  gen_kwargs={
65
+ "ds": ds,
66
  },
67
  ),
68
  ]
69
 
70
  elif self.config.name == "byrow":
71
 
72
+ ds = ds.train_test_split(test_size=0.1, seed=42)
73
  #print(ds)
74
 
75
  return [
 
90
 
91
  elif self.config.name == "byfuncname":
92
 
 
 
93
  unique_names = ds.unique('Name')
94
  nameds = datasets.Dataset.from_dict({'Name': unique_names})
95
 
 
114
  ),
115
 
116
  ]
117
+
118
+ elif self.config.name == "bylibrary":
119
+ # A function (name) is a library function if it appears in more than one Exename
120
+
121
+ # this is (('func', 'oo.exe'): 123)
122
+ testcount = set(zip(ds['Name'], ds['Exename']))
123
+
124
+ # sorted pairs by function name
125
+ testcount = sorted(testcount, key=lambda x: x[0])
126
+
127
+ # group by function name
128
+ grouped = itertools.groupby(testcount, lambda t: t[0])
129
+
130
+ grouped = {k: [b for _,b in g] for k, g in grouped}
131
+
132
+ library_func_names = {f for f, exes in grouped.items() if len(exes) > 1}
133
+ nonlibrary_func_names = {f for f, exes in grouped.items() if len(exes) == 1}
134
+
135
+ return [
136
+ datasets.SplitGenerator(
137
+ name="train",
138
+ gen_kwargs={
139
+ "ds": ds.filter(lambda r: r['Name'] in library_func_names),
140
+ },
141
+ ),
142
+ datasets.SplitGenerator(
143
+ name="test",
144
+ gen_kwargs={
145
+ "ds": ds.filter(lambda r: r['Name'] in nonlibrary_func_names),
146
+ },
147
+ ),
148
+
149
+ ]
150
 
151
  else:
152
  assert False