sps commited on
Commit
b430bfb
1 Parent(s): 472f417

Refactor data load script with recent updates

Browse files
Files changed (1) hide show
  1. codequeries.py +53 -26
codequeries.py CHANGED
@@ -13,12 +13,10 @@
13
  # See the License for the specific language governing permissions and
14
  # limitations under the License.
15
 
16
- """The CodeQueries benchmark."""
17
 
18
 
19
  import json
20
- import os
21
-
22
  import datasets
23
 
24
  logger = datasets.logging.get_logger(__name__)
@@ -58,6 +56,8 @@ class CodequeriesConfig(datasets.BuilderConfig):
58
  features: `list[string]`, list of the features that will appear in the
59
  feature dict. Should not include "label".
60
  citation: `string`, citation for the data set.
 
 
61
  **kwargs: keyword arguments forwarded to super.
62
  """
63
  # Version history:
@@ -76,8 +76,9 @@ class Codequeries(datasets.GeneratorBasedBuilder):
76
  CodequeriesConfig(
77
  name="ideal",
78
  description=_IDEAL_DESCRIPTION,
79
- features=["query_name", "context_blocks", "answer_spans",
80
- "supporting_fact_spans", "code_file_path", "example_type",
 
81
  "subtokenized_input_sequence", "label_sequence"],
82
  citation=_CODEQUERIES_CITATION,
83
  data_url={
@@ -85,43 +86,47 @@ class Codequeries(datasets.GeneratorBasedBuilder):
85
  "dev": "ideal_val.json",
86
  "test": "ideal_test.json"
87
  },
88
- url="",
89
  ),
90
  CodequeriesConfig(
91
  name="prefix",
92
  description=_PREFIX_DESCRIPTION,
93
- features=["query_name", "answer_spans",
94
- "supporting_fact_spans", "code_file_path", "example_type",
 
95
  "subtokenized_input_sequence", "label_sequence"],
96
  citation=_CODEQUERIES_CITATION,
97
  data_url={
98
  "test": "prefix_test.json"
99
  },
100
- url="",
101
  ),
102
  CodequeriesConfig(
103
  name="file_ideal",
104
  description=_FILE_IDEAL_DESCRIPTION,
105
- features=["query_name", "context_blocks", "answer_spans",
106
- "supporting_fact_spans", "code_file_path", "example_type",
 
107
  "subtokenized_input_sequence", "label_sequence"],
108
  citation=_CODEQUERIES_CITATION,
109
  data_url={
110
  "test": "file_ideal_test.json"
111
  },
112
- url="",
113
  ),
114
  CodequeriesConfig(
115
  name="twostep",
116
  description=_TWOSTEP_DESCRIPTION,
117
- features=["query_name", "context_blocks", "answer_spans",
118
- "supporting_fact_spans", "code_file_path", "example_type",
119
- "subtokenized_input_sequence", "label_sequence"],
 
 
120
  citation=_CODEQUERIES_CITATION,
121
  data_url={
122
  "test": ["twostep_relevance/" + "twostep_relevance_test_" + str(i) + ".json" for i in range(0,10)]
123
  },
124
- url="",
125
  ),
126
  ]
127
 
@@ -130,11 +135,13 @@ class Codequeries(datasets.GeneratorBasedBuilder):
130
  def _info(self):
131
  features = {}
132
  features["query_name"] = datasets.Value("string")
 
133
  features["context_blocks"] = [
134
  {
135
  "content": datasets.Value("string"),
136
  "metadata": datasets.Value("string"),
137
- "header": datasets.Value("string")
 
138
  }
139
  ]
140
  features["answer_spans"] = [
@@ -155,11 +162,11 @@ class Codequeries(datasets.GeneratorBasedBuilder):
155
  'end_column': datasets.Value("int32")
156
  }
157
  ]
158
- features["code_file_path"] = datasets.Value("string")
159
- features["example_type"] = datasets.Value("int32")
160
  features["subtokenized_input_sequence"] = datasets.features.Sequence(datasets.Value("string"))
161
- features["label_sequence"] = datasets.features.Sequence(datasets.Value("int32"))
162
- features["relevance_label"] = datasets.Value("int32")
163
 
164
  return datasets.DatasetInfo(
165
  description=self.config.description,
@@ -170,7 +177,7 @@ class Codequeries(datasets.GeneratorBasedBuilder):
170
 
171
  def _split_generators(self, dl_manager):
172
  dl_dir = dl_manager.download_and_extract(self.config.data_url)
173
- print(dl_dir)
174
  if self.config.name in ["prefix", "file_ideal", "twostep"]:
175
  return [
176
  datasets.SplitGenerator(
@@ -209,7 +216,7 @@ class Codequeries(datasets.GeneratorBasedBuilder):
209
  def _generate_examples(self, filepath, split):
210
  if self.config.name in ["prefix", "file_ideal", "twostep"]:
211
  assert split == datasets.Split.TEST
212
- logger.info("generating examples from = %s", filepath)
213
 
214
  if self.config.name == "twostep":
215
  key = 0
@@ -221,15 +228,35 @@ class Codequeries(datasets.GeneratorBasedBuilder):
221
  instance_key = str(key) + "_" + row["query_name"] + "_" + row["code_file_path"]
222
  yield instance_key, {
223
  "query_name": row["query_name"],
 
224
  "context_blocks": row["context_blocks"],
225
  "answer_spans": row["answer_spans"],
226
  "supporting_fact_spans": row["supporting_fact_spans"],
227
- "code_file_path": row["code_file_path"],
228
  "example_type": row["example_type"],
 
229
  "subtokenized_input_sequence": row["subtokenized_input_sequence"],
 
230
  "relevance_label": row["relevance_label"],
231
  }
232
  key += 1
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
233
  else:
234
  with open(filepath, encoding="utf-8") as f:
235
  key = 0
@@ -239,13 +266,13 @@ class Codequeries(datasets.GeneratorBasedBuilder):
239
  instance_key = str(key) + "_" + row["query_name"] + "_" + row["code_file_path"]
240
  yield instance_key, {
241
  "query_name": row["query_name"],
 
242
  "context_blocks": row["context_blocks"],
243
  "answer_spans": row["answer_spans"],
244
  "supporting_fact_spans": row["supporting_fact_spans"],
245
- "code_file_path": row["code_file_path"],
246
  "example_type": row["example_type"],
 
247
  "subtokenized_input_sequence": row["subtokenized_input_sequence"],
248
  "label_sequence": row["label_sequence"],
249
  }
250
  key += 1
251
-
 
13
  # See the License for the specific language governing permissions and
14
  # limitations under the License.
15
 
16
+ """CodeQueries: The CodeQueries benchmark dataset."""
17
 
18
 
19
  import json
 
 
20
  import datasets
21
 
22
  logger = datasets.logging.get_logger(__name__)
 
56
  features: `list[string]`, list of the features that will appear in the
57
  feature dict. Should not include "label".
58
  citation: `string`, citation for the data set.
59
+ data_url: `string`, relative data path in repo
60
+ url: `string`, link to dataset info page
61
  **kwargs: keyword arguments forwarded to super.
62
  """
63
  # Version history:
 
76
  CodequeriesConfig(
77
  name="ideal",
78
  description=_IDEAL_DESCRIPTION,
79
+ features=["query_name", "code_file_path", "context_blocks",
80
+ "answer_spans", "supporting_fact_spans",
81
+ "example_type", "single_hop",
82
  "subtokenized_input_sequence", "label_sequence"],
83
  citation=_CODEQUERIES_CITATION,
84
  data_url={
 
86
  "dev": "ideal_val.json",
87
  "test": "ideal_test.json"
88
  },
89
+ url="https://huggingface.co/datasets/thepurpleowl/codequeries",
90
  ),
91
  CodequeriesConfig(
92
  name="prefix",
93
  description=_PREFIX_DESCRIPTION,
94
+ features=["query_name", "code_file_path",
95
+ "answer_spans", "supporting_fact_spans",
96
+ "example_type", "single_hop",
97
  "subtokenized_input_sequence", "label_sequence"],
98
  citation=_CODEQUERIES_CITATION,
99
  data_url={
100
  "test": "prefix_test.json"
101
  },
102
+ url="https://huggingface.co/datasets/thepurpleowl/codequeries",
103
  ),
104
  CodequeriesConfig(
105
  name="file_ideal",
106
  description=_FILE_IDEAL_DESCRIPTION,
107
+ features=["query_name", "code_file_path", "context_blocks",
108
+ "answer_spans", "supporting_fact_spans",
109
+ "example_type", "single_hop",
110
  "subtokenized_input_sequence", "label_sequence"],
111
  citation=_CODEQUERIES_CITATION,
112
  data_url={
113
  "test": "file_ideal_test.json"
114
  },
115
+ url="https://huggingface.co/datasets/thepurpleowl/codequeries",
116
  ),
117
  CodequeriesConfig(
118
  name="twostep",
119
  description=_TWOSTEP_DESCRIPTION,
120
+ features=["query_name", "code_file_path", "context_blocks",
121
+ "answer_spans", "supporting_fact_spans",
122
+ "example_type", "single_hop",
123
+ "subtokenized_input_sequence", "label_sequence",
124
+ "relevance_label"],
125
  citation=_CODEQUERIES_CITATION,
126
  data_url={
127
  "test": ["twostep_relevance/" + "twostep_relevance_test_" + str(i) + ".json" for i in range(0,10)]
128
  },
129
+ url="https://huggingface.co/datasets/thepurpleowl/codequeries",
130
  ),
131
  ]
132
 
 
135
  def _info(self):
136
  features = {}
137
  features["query_name"] = datasets.Value("string")
138
+ features["code_file_path"] = datasets.Value("string")
139
  features["context_blocks"] = [
140
  {
141
  "content": datasets.Value("string"),
142
  "metadata": datasets.Value("string"),
143
+ "header": datasets.Value("string"),
144
+ "index": datasets.Value("int32")
145
  }
146
  ]
147
  features["answer_spans"] = [
 
162
  'end_column': datasets.Value("int32")
163
  }
164
  ]
165
+ features["example_type"] = datasets.Value("int8")
166
+ features["single_hop"] = datasets.Value("bool")
167
  features["subtokenized_input_sequence"] = datasets.features.Sequence(datasets.Value("string"))
168
+ features["label_sequence"] = datasets.features.Sequence(datasets.Value("int8"))
169
+ features["relevance_label"] = datasets.Value("int8")
170
 
171
  return datasets.DatasetInfo(
172
  description=self.config.description,
 
177
 
178
  def _split_generators(self, dl_manager):
179
  dl_dir = dl_manager.download_and_extract(self.config.data_url)
180
+
181
  if self.config.name in ["prefix", "file_ideal", "twostep"]:
182
  return [
183
  datasets.SplitGenerator(
 
216
  def _generate_examples(self, filepath, split):
217
  if self.config.name in ["prefix", "file_ideal", "twostep"]:
218
  assert split == datasets.Split.TEST
219
+ logger.info("Generating examples from = %s", filepath)
220
 
221
  if self.config.name == "twostep":
222
  key = 0
 
228
  instance_key = str(key) + "_" + row["query_name"] + "_" + row["code_file_path"]
229
  yield instance_key, {
230
  "query_name": row["query_name"],
231
+ "code_file_path": row["code_file_path"],
232
  "context_blocks": row["context_blocks"],
233
  "answer_spans": row["answer_spans"],
234
  "supporting_fact_spans": row["supporting_fact_spans"],
 
235
  "example_type": row["example_type"],
236
+ "single_hop": row["single_hop"],
237
  "subtokenized_input_sequence": row["subtokenized_input_sequence"],
238
+ "label_sequence": row["label_sequence"],
239
  "relevance_label": row["relevance_label"],
240
  }
241
  key += 1
242
+ elif self.config.name == "prefix":
243
+ with open(filepath, encoding="utf-8") as f:
244
+ key = 0
245
+ for line in f:
246
+ row = json.loads(line)
247
+
248
+ instance_key = str(key) + "_" + row["query_name"] + "_" + row["code_file_path"]
249
+ yield instance_key, {
250
+ "query_name": row["query_name"],
251
+ "code_file_path": row["code_file_path"],
252
+ "answer_spans": row["answer_spans"],
253
+ "supporting_fact_spans": row["supporting_fact_spans"],
254
+ "example_type": row["example_type"],
255
+ "single_hop": row["single_hop"],
256
+ "subtokenized_input_sequence": row["subtokenized_input_sequence"],
257
+ "label_sequence": row["label_sequence"],
258
+ }
259
+ key += 1
260
  else:
261
  with open(filepath, encoding="utf-8") as f:
262
  key = 0
 
266
  instance_key = str(key) + "_" + row["query_name"] + "_" + row["code_file_path"]
267
  yield instance_key, {
268
  "query_name": row["query_name"],
269
+ "code_file_path": row["code_file_path"],
270
  "context_blocks": row["context_blocks"],
271
  "answer_spans": row["answer_spans"],
272
  "supporting_fact_spans": row["supporting_fact_spans"],
 
273
  "example_type": row["example_type"],
274
+ "single_hop": row["single_hop"],
275
  "subtokenized_input_sequence": row["subtokenized_input_sequence"],
276
  "label_sequence": row["label_sequence"],
277
  }
278
  key += 1