Create nationalities.py
Browse files- nationalities.py +31 -0
nationalities.py
ADDED
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import tqdm
|
2 |
+
import re
|
3 |
+
import requests
|
4 |
+
from huggingface_hub import login
|
5 |
+
from datasets import Dataset
|
6 |
+
from datasets import load_dataset
|
7 |
+
from google.colab import userdata
|
8 |
+
|
9 |
+
login(userdata.get('HF_TOKEN'))
|
10 |
+
|
11 |
+
data = load_dataset("Ibrahemqasim/categories_en2ar", split="train")
|
12 |
+
countries = load_dataset("Ibrahemqasim/countries", split="train")
|
13 |
+
|
14 |
+
nats_url = "https://raw.githubusercontent.com/MrIbrahem/Nationalities/refs/heads/main/nats.json"
|
15 |
+
response = requests.get(nats_url)
|
16 |
+
nationalities = response.json()
|
17 |
+
|
18 |
+
data_list = nationalities["data"]
|
19 |
+
# ---
|
20 |
+
print("______________")
|
21 |
+
print(f"len of nationalities : {len(data_list)}.")
|
22 |
+
# ---
|
23 |
+
print("____________________________")
|
24 |
+
# ---
|
25 |
+
# إنشاء Dataset
|
26 |
+
dataset = Dataset.from_list(data_list)
|
27 |
+
|
28 |
+
# رفع Dataset إلى Hugging Face
|
29 |
+
dataset.push_to_hub(f"Ibrahemqasim/nationalities")
|
30 |
+
# ---
|
31 |
+
print(f"dataset: Ibrahemqasim/nationalities push_to_hub successfully!")
|