Spaces:
Sleeping
Sleeping
Commit
·
e06bf99
1
Parent(s):
5bc4016
Create data_loader.py
Browse files- utils/data_loader.py +28 -0
utils/data_loader.py
ADDED
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import pandas as pd
|
2 |
+
from glob import glob
|
3 |
+
|
4 |
+
def load_subs(path):
|
5 |
+
path = path + "/*.srt"
|
6 |
+
subs = glob(path)
|
7 |
+
episodes = []
|
8 |
+
scripts = []
|
9 |
+
|
10 |
+
for sub in subs:
|
11 |
+
with open(sub, "r", encoding="utf-8") as f:
|
12 |
+
lines = f.readlines()
|
13 |
+
cnt = 0
|
14 |
+
con = []
|
15 |
+
for line in lines:
|
16 |
+
line = line.strip()
|
17 |
+
if line.isnumeric() or "-->" in line:
|
18 |
+
cnt += 1
|
19 |
+
else:
|
20 |
+
con.append(line)
|
21 |
+
|
22 |
+
script = " ".join(con)
|
23 |
+
epno = int(sub.split("-")[1].strip()[-1])
|
24 |
+
episodes.append(epno)
|
25 |
+
scripts.append(script)
|
26 |
+
|
27 |
+
df = pd.DataFrame({"episode": episodes, "script": scripts})
|
28 |
+
return df
|