Datasets:
Tasks:
Text Generation
Modalities:
Text
Sub-tasks:
document-retrieval
Size:
100K - 1M
ArXiv:
Tags:
code
License:
upload data config file
Browse files- repobench-c.py +169 -0
- utils.py +16 -0
repobench-c.py
ADDED
@@ -0,0 +1,169 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# Copyright 2020 The HuggingFace Datasets Authors and the current dataset script contributor.
|
2 |
+
#
|
3 |
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
4 |
+
# you may not use this file except in compliance with the License.
|
5 |
+
# You may obtain a copy of the License at
|
6 |
+
#
|
7 |
+
# http://www.apache.org/licenses/LICENSE-2.0
|
8 |
+
#
|
9 |
+
# Unless required by applicable law or agreed to in writing, software
|
10 |
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
11 |
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
12 |
+
# See the License for the specific language governing permissions and
|
13 |
+
# limitations under the License.
|
14 |
+
|
15 |
+
"""RepoBench: Benchmarking Repository-Level Code Auto-Completion Systems"""
|
16 |
+
|
17 |
+
import gzip
|
18 |
+
import pickle
|
19 |
+
import textwrap
|
20 |
+
import datasets
|
21 |
+
from utils import construct_prompt
|
22 |
+
|
23 |
+
_CITATION = """\
|
24 |
+
@misc{liu2023repobench,
|
25 |
+
title={RepoBench: Benchmarking Repository-Level Code Auto-Completion Systems},
|
26 |
+
author={Tianyang Liu and Canwen Xu and Julian McAuley},
|
27 |
+
year={2023},
|
28 |
+
eprint={2306.03091},
|
29 |
+
archivePrefix={arXiv},
|
30 |
+
primaryClass={cs.CL}
|
31 |
+
}
|
32 |
+
"""
|
33 |
+
|
34 |
+
_DESCRIPTION = """\
|
35 |
+
RepoBench is a dataset that benchmarks repository-level code auto-completion systems.
|
36 |
+
|
37 |
+
RepoBench-C denotes RepoBench for code completion,
|
38 |
+
which is subtask of RepoBench for next-line code prediction given both cross-file and in-file context.
|
39 |
+
"""
|
40 |
+
|
41 |
+
_HOMEPAGE = "https://github.com/Leolty/repobench"
|
42 |
+
|
43 |
+
_LICENSE = "Apache License 2.0"
|
44 |
+
|
45 |
+
_URLs = {
|
46 |
+
"python_cff": "https://raw.githubusercontent.com/Leolty/repobench/main/data/code_completion/python/cross_file_first.gz",
|
47 |
+
"python_cfr": "https://raw.githubusercontent.com/Leolty/repobench/main/data/code_completion/python/cross_file_random.gz",
|
48 |
+
"python_if": "https://raw.githubusercontent.com/Leolty/repobench/main/data/code_completion/python/in_file.gz",
|
49 |
+
"java_cff": "https://raw.githubusercontent.com/Leolty/repobench/main/data/code_completion/java/cross_file_first.gz",
|
50 |
+
"java_cfr": "https://raw.githubusercontent.com/Leolty/repobench/main/data/code_completion/java/cross_file_random.gz",
|
51 |
+
"java_if": "https://raw.githubusercontent.com/Leolty/repobench/main/data/code_completion/java/in_file.gz"
|
52 |
+
}
|
53 |
+
|
54 |
+
class RepoBenchR(datasets.GeneratorBasedBuilder):
|
55 |
+
"""RepoBench"""
|
56 |
+
|
57 |
+
VERSION = datasets.Version("1.0.0")
|
58 |
+
|
59 |
+
BUILDER_CONFIGS = [
|
60 |
+
datasets.BuilderConfig(
|
61 |
+
name="python_cff",
|
62 |
+
description=textwrap.dedent(
|
63 |
+
"""
|
64 |
+
cff: cross_file_first -> mask the the line that a cross-file module is first used
|
65 |
+
"""
|
66 |
+
)
|
67 |
+
),
|
68 |
+
datasets.BuilderConfig(
|
69 |
+
name="python_cfr",
|
70 |
+
description=textwrap.dedent(
|
71 |
+
"""
|
72 |
+
cfr: cross_file_random -> mask a random line that a cross-file module is used (not the first time)
|
73 |
+
"""
|
74 |
+
)
|
75 |
+
),
|
76 |
+
datasets.BuilderConfig(
|
77 |
+
name="python_if",
|
78 |
+
description=textwrap.dedent(
|
79 |
+
"""
|
80 |
+
if: in_file -> mask a random line with no cross-file module
|
81 |
+
"""
|
82 |
+
)
|
83 |
+
),
|
84 |
+
datasets.BuilderConfig(
|
85 |
+
name="java_cff",
|
86 |
+
description=textwrap.dedent(
|
87 |
+
"""
|
88 |
+
cff: cross_file_first -> mask the the line that a cross-file module is first used
|
89 |
+
"""
|
90 |
+
)
|
91 |
+
),
|
92 |
+
datasets.BuilderConfig(
|
93 |
+
name="java_cfr",
|
94 |
+
description=textwrap.dedent(
|
95 |
+
"""
|
96 |
+
cfr: cross_file_random -> mask a random line that a cross-file module is used (not the first time)
|
97 |
+
"""
|
98 |
+
)
|
99 |
+
),
|
100 |
+
datasets.BuilderConfig(
|
101 |
+
name="java_if",
|
102 |
+
description=textwrap.dedent(
|
103 |
+
"""
|
104 |
+
if: in_file -> mask a random line with no cross-file module
|
105 |
+
"""
|
106 |
+
)
|
107 |
+
)
|
108 |
+
]
|
109 |
+
|
110 |
+
def _info(self):
|
111 |
+
features = datasets.Features(
|
112 |
+
{
|
113 |
+
"repo_name": datasets.Value("string"),
|
114 |
+
"file_path": datasets.Value("string"),
|
115 |
+
"context": datasets.Sequence(datasets.Value("string")),
|
116 |
+
"import_statement": datasets.Value("string"),
|
117 |
+
"code": datasets.Value("string"),
|
118 |
+
"prompt": datasets.Value("string"),
|
119 |
+
"next_line": datasets.Value("string")
|
120 |
+
}
|
121 |
+
)
|
122 |
+
|
123 |
+
return datasets.DatasetInfo(
|
124 |
+
description=_DESCRIPTION,
|
125 |
+
features=features,
|
126 |
+
homepage=_HOMEPAGE,
|
127 |
+
license=_LICENSE,
|
128 |
+
citation=_CITATION,
|
129 |
+
)
|
130 |
+
|
131 |
+
def _split_generators(self, dl_manager):
|
132 |
+
"""Returns SplitGenerators."""
|
133 |
+
config_urls = _URLs[self.config.name]
|
134 |
+
data_dir = dl_manager.download(config_urls)
|
135 |
+
|
136 |
+
return [
|
137 |
+
datasets.SplitGenerator(
|
138 |
+
name=datasets.Split("train"),
|
139 |
+
gen_kwargs={"data_dir": data_dir, "split": "train"},
|
140 |
+
),
|
141 |
+
datasets.SplitGenerator(
|
142 |
+
name=datasets.Split("dev"),
|
143 |
+
gen_kwargs={"data_dir": data_dir, "split": "dev"},
|
144 |
+
),
|
145 |
+
datasets.SplitGenerator(
|
146 |
+
name=datasets.Split("test"),
|
147 |
+
gen_kwargs={"data_dir": data_dir, "split": "test"},
|
148 |
+
)
|
149 |
+
]
|
150 |
+
|
151 |
+
def _generate_examples(self, data_dir, split):
|
152 |
+
""" Yields examples. """
|
153 |
+
with gzip.open(data_dir, "rb") as f:
|
154 |
+
data = pickle.load(f)
|
155 |
+
|
156 |
+
for i, example in enumerate(data[split]):
|
157 |
+
|
158 |
+
prompt = construct_prompt(example, self.config.name.split("_")[0])
|
159 |
+
|
160 |
+
yield i, {
|
161 |
+
"repo_name": example["repo_name"],
|
162 |
+
"file_path": example["file_path"],
|
163 |
+
"context": example["context"],
|
164 |
+
"import_statement": example["import_statement"],
|
165 |
+
"code": example["code"],
|
166 |
+
"prompt": prompt,
|
167 |
+
"next_line": example["next_line"]
|
168 |
+
}
|
169 |
+
|
utils.py
ADDED
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
|
2 |
+
def construct_prompt(data_point:dict, language:str):
|
3 |
+
|
4 |
+
if language == "python":
|
5 |
+
path = f"# Path: {data_point['file_path']}"
|
6 |
+
|
7 |
+
elif language == "java":
|
8 |
+
path = f"// Path: {data_point['file_path']}"
|
9 |
+
|
10 |
+
prompt = f"""{data_point['context']}
|
11 |
+
{path}
|
12 |
+
{data_point['import_statement']}
|
13 |
+
|
14 |
+
{data_point['code']}"""
|
15 |
+
|
16 |
+
return prompt
|