File size: 2,811 Bytes
7046435 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 |
from pprint import pprint
import tempfile
from pathlib import Path
import polars as pl
from huggingface_hub import hf_hub_url, list_repo_files
from tqdm import tqdm
import base64
from datasets import Dataset
from huggingface_hub import login
login(new_session=False) # Will prompt for your token or use cached token
cache_dir = Path(tempfile.gettempdir()) / "allenai_c4"
cache_dir.mkdir(exist_ok=True)
def cache_name(url: str) -> str:
return base64.urlsafe_b64encode(url.encode()).decode().rstrip("=") + ".parquet"
def cache_path(url: str, cache_dir=cache_dir) -> Path:
return cache_dir / cache_name(url=url)
parquet_cache_names = cache_dir / "realnewslike_filenames.parquet"
if parquet_cache_names.exists():
news_files = pl.read_parquet(parquet_cache_names)["filename"]
else:
file_names = pl.Series(list_repo_files("allenai/c4", repo_type="dataset"))
# Take all splits of the realnewslike subset (513 files)
news_files = file_names.filter(
file_names.str.starts_with("realnewslike/") & file_names.str.ends_with(".json.gz"),
).str.strip_prefix("realnewslike/")
pl.DataFrame({"filename": news_files}).write_parquet(parquet_cache_names)
c4n_features = {"url": pl.String, "text": pl.String}
aggregator = pl.DataFrame(schema=c4n_features)
domain_capture = r"https?://([^/?]+)"
subpage_capture = r"https?://[^/]+(\/[^/?]+\/)" # Include pre/suffix slashes
domain_match = r"^(news\.bbc\.co\.uk|www\.bbc\.co\.uk|www\.bbc\.com)$"
domain_col = pl.col("url").str.extract(domain_capture)
path_col = pl.col("url").str.extract(subpage_capture)
hf_urls = [
hf_hub_url(
repo_id="allenai/c4",
filename=filename,
subfolder="realnewslike",
repo_type="dataset",
)
for filename in news_files
]
pq_caches = list(map(cache_path, hf_urls))
for json_url, parquet_cache_chunk in tqdm(zip(hf_urls, pq_caches)):
if parquet_cache_chunk.exists():
news_df = pl.read_parquet(parquet_cache_chunk)
else:
print(f"Processing {json_url}")
df = (
pl.read_ndjson(json_url, schema=c4n_features)
.with_columns(pl.col("url").str.extract(r"([^?]+)"))
.filter(
domain_col.str.contains(domain_match),
~pl.col("url").str.contains(r"https?://[^/]+\/\?"), # Path is not `/?`
)
)
# Just match the "/news/" path here
news_df = df.filter(domain_col.str.contains('news').or_(path_col == "/news/"))
news_df.write_parquet(parquet_cache_chunk)
# Reload once all parts completed and upload
aggregator = pl.read_parquet(pq_caches)
news_data = aggregator.to_dict(as_series=False)
news_dataset = Dataset.from_dict(news_data)
news_dataset.push_to_hub("permutans/c4-bbc-news", config_name="realnewslike-bbc-news", private=False)
|