Datasets:
Tasks:
Text Generation
Modalities:
Text
Sub-tasks:
language-modeling
Languages:
code
Size:
100K - 1M
ArXiv:
License:
Create new file
Browse files- xlcost_conversion.py +91 -0
xlcost_conversion.py
ADDED
|
@@ -0,0 +1,91 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from tqdm import tqdm
|
| 2 |
+
from datasets import Dataset
|
| 3 |
+
|
| 4 |
+
"""to run inside XLCOST_DATA folder after downloading XLCost data from this repo https://github.com/reddy-lab-code-research/XLCoST"""
|
| 5 |
+
|
| 6 |
+
|
| 7 |
+
class Example(object):
|
| 8 |
+
"""A single training/test example."""
|
| 9 |
+
def __init__(self,
|
| 10 |
+
idx,
|
| 11 |
+
source,
|
| 12 |
+
target,
|
| 13 |
+
):
|
| 14 |
+
self.idx = idx
|
| 15 |
+
self.source = source
|
| 16 |
+
self.target = target
|
| 17 |
+
|
| 18 |
+
def read_examples(filename):
|
| 19 |
+
"""Read examples from filename."""
|
| 20 |
+
examples=[]
|
| 21 |
+
assert len(filename.split(','))==2
|
| 22 |
+
src_filename = filename.split(',')[0]
|
| 23 |
+
trg_filename = filename.split(',')[1]
|
| 24 |
+
idx = 0
|
| 25 |
+
with open(src_filename) as f1,open(trg_filename) as f2:
|
| 26 |
+
for line1,line2 in zip(f1,f2):
|
| 27 |
+
examples.append(
|
| 28 |
+
Example(
|
| 29 |
+
idx = idx,
|
| 30 |
+
source=line1.strip(),
|
| 31 |
+
target=line2.strip(),
|
| 32 |
+
)
|
| 33 |
+
)
|
| 34 |
+
idx+=1
|
| 35 |
+
return examples
|
| 36 |
+
|
| 37 |
+
def create_data(filename):
|
| 38 |
+
examples = read_examples(filename)
|
| 39 |
+
text = []
|
| 40 |
+
code = []
|
| 41 |
+
print(len(examples))
|
| 42 |
+
for i in tqdm(range(len(examples))):
|
| 43 |
+
text.append(examples[i].source)
|
| 44 |
+
code.append(examples[i].target)
|
| 45 |
+
data = {"text": text, "code": code}
|
| 46 |
+
data = Dataset.from_dict(data)
|
| 47 |
+
return data
|
| 48 |
+
|
| 49 |
+
if __name__ == "__main__":
|
| 50 |
+
#clone xlcost-text-to-code hub repo
|
| 51 |
+
LANG = ["Python", "C", "C#", "Java", "PHP", "Javascript", "C++"]
|
| 52 |
+
EXTENSION = ["py", "c", "cs", "java", "php", "js", "cpp"]
|
| 53 |
+
|
| 54 |
+
for i in range(len(LANG)):
|
| 55 |
+
# for each language this saves train test and validation subsets for both snippet and program levels
|
| 56 |
+
lang = LANG[i]
|
| 57 |
+
ext = EXTENSION[i]
|
| 58 |
+
print(f"language: {lang}")
|
| 59 |
+
if lang == "C#":
|
| 60 |
+
path_snippet = f"Csharp-snippet-level"
|
| 61 |
+
path_program = f"Csharp-program-level"
|
| 62 |
+
else:
|
| 63 |
+
path_snippet = f"{lang}-snippet-level"
|
| 64 |
+
path_program = f"{lang}-program-level"
|
| 65 |
+
|
| 66 |
+
train_filename = f"generation/pair_data_tok_1_comment/{lang}-comment/train-{lang}-comment-tok.txt,generation/pair_data_tok_1_comment/{lang}-comment/train-{lang}-comment-tok.{ext}"
|
| 67 |
+
valid_filename = f"generation/pair_data_tok_1_comment/{lang}-comment/val-{lang}-comment-tok.txt,generation/pair_data_tok_1_comment/{lang}-comment/val-{lang}-comment-tok.{ext}"
|
| 68 |
+
test_filename = f"generation/pair_data_tok_1_comment/{lang}-comment/test-{lang}-comment-tok.txt,generation/pair_data_tok_1_comment/{lang}-comment/test-{lang}-comment-tok.{ext}"
|
| 69 |
+
|
| 70 |
+
train = create_data(train_filename)
|
| 71 |
+
valid = create_data(valid_filename)
|
| 72 |
+
test = create_data(test_filename)
|
| 73 |
+
|
| 74 |
+
train.to_json(f"xlcost-text-to-code/data/{path_snippet}/train.json", lines=True)
|
| 75 |
+
valid.to_json(f"xlcost-text-to-code/data/{path_snippet}/valid.json", lines=True)
|
| 76 |
+
test.to_json(f"xlcost-text-to-code/data/{path_snippet}/test.json", lines=True)
|
| 77 |
+
|
| 78 |
+
train_filename = f"generation/pair_data_tok_full_desc_comment/{lang}-desc/train-{lang}-desc-tok.txt,generation/pair_data_tok_full_desc_comment/{lang}-desc/train-{lang}-desc-tok.{ext}"
|
| 79 |
+
valid_filename = f"generation/pair_data_tok_full_desc_comment/{lang}-desc/val-{lang}-desc-tok.txt,generation/pair_data_tok_full_desc_comment/{lang}-desc/val-{lang}-desc-tok.{ext}"
|
| 80 |
+
test_filename = f"generation/pair_data_tok_full_desc_comment/{lang}-desc/test-{lang}-desc-tok.txt,generation/pair_data_tok_full_desc_comment/{lang}-desc/test-{lang}-desc-tok.{ext}"
|
| 81 |
+
|
| 82 |
+
train = create_data(train_filename)
|
| 83 |
+
valid = create_data(valid_filename)
|
| 84 |
+
test = create_data(test_filename)
|
| 85 |
+
|
| 86 |
+
train.to_json(f"xlcost-text-to-code/data/{path_program}/train.json", lines=True)
|
| 87 |
+
valid.to_json(f"xlcost-text-to-code/data/{path_program}/valid.json", lines=True)
|
| 88 |
+
test.to_json(f"xlcost-text-to-code/data/{path_program}/test.json", lines=True)
|
| 89 |
+
|
| 90 |
+
#push to hub the folder xlcost (containing data/ and xlcost.py dataset builder script)
|
| 91 |
+
|