Spaces:
Runtime error
Runtime error
jiarongqiu commited on
Commit ·
75bc718
1
Parent(s): 480d732
add database
Browse files- service/database.py +40 -0
service/database.py
ADDED
|
@@ -0,0 +1,40 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import os
|
| 2 |
+
import json
|
| 3 |
+
from pathlib import Path
|
| 4 |
+
from huggingface_hub import CommitScheduler
|
| 5 |
+
from datasets import load_dataset
|
| 6 |
+
from util.time import TimeUtil
|
| 7 |
+
from util.schema import Dir
|
| 8 |
+
|
| 9 |
+
|
| 10 |
+
class Database:
|
| 11 |
+
|
| 12 |
+
LOCAL = os.path.join(os.path.dirname(__file__),"..","data")
|
| 13 |
+
USERNAME = "JR818"
|
| 14 |
+
REPO = "Web3"
|
| 15 |
+
|
| 16 |
+
def __init__(self):
|
| 17 |
+
self.scheduler = CommitScheduler(
|
| 18 |
+
repo_id=self.REPO,
|
| 19 |
+
repo_type="dataset",
|
| 20 |
+
folder_path=self.LOCAL,
|
| 21 |
+
token=os.getenv("HF_TOKEN")
|
| 22 |
+
)
|
| 23 |
+
|
| 24 |
+
def load(self,dir:Dir):
|
| 25 |
+
dataset = load_dataset(self.USERNAME+'/'+self.REPO,data_dir=dir.path,ignore_verifications=True,split="train")
|
| 26 |
+
dataset = dataset.with_format("pandas").data.to_pandas()
|
| 27 |
+
return dataset
|
| 28 |
+
|
| 29 |
+
def write(self,data,dir:Dir):
|
| 30 |
+
path = Path(self.LOCAL,dir.path)
|
| 31 |
+
path.mkdir(parents=True, exist_ok=True)
|
| 32 |
+
path = path/f'{TimeUtil.get_date()}.json'
|
| 33 |
+
try:
|
| 34 |
+
data['datetime'] = TimeUtil.get_time()
|
| 35 |
+
data = json.dumps(data)+"\n"
|
| 36 |
+
with self.scheduler.lock:
|
| 37 |
+
with path.open("a") as f:
|
| 38 |
+
f.write(data)
|
| 39 |
+
except Exception as e:
|
| 40 |
+
print(e)
|