verify_dataset.py

import os
os.environ["HF_HUB_OFFLINE"] = "0"
os.environ["HF_DATASETS_OFFLINE"] = "0"
os.environ["REQUESTS_CA_BUNDLE"] = ""
os.environ["CURL_CA_BUNDLE"] = ""
import requests, urllib3
urllib3.disable_warnings()
orig = requests.Session.request
def patched(self, method, url, **kwargs):
    kwargs["verify"] = False
    return orig(self, method, url, **kwargs)
requests.Session.request = patched

from datasets import load_dataset

TOKEN = os.environ.get("HF_TOKEN", "")
ds = load_dataset("Shomi28/prompt-injection-dataset", token=TOKEN)

print("LIVE DATASET — Shomi28/prompt-injection-dataset")
print("=" * 55)
for split in ds:
    inj  = sum(1 for x in ds[split] if x["label"] == 1)
    safe = sum(1 for x in ds[split] if x["label"] == 0)
    print(f"  {split:12s}: {len(ds[split]):>5,} rows  |  safe={safe:<4}  injection={inj}")

print()
print("Sample — injection:")
sample_inj = next(x for x in ds["train"] if x["label"] == 1)
print(f"  {sample_inj['text']}")

print()
print("Sample — safe:")
sample_safe = next(x for x in ds["train"] if x["label"] == 0)
print(f"  {sample_safe['text']}")

print()
print("Dataset URL: https://huggingface.co/datasets/Shomi28/prompt-injection-dataset")