Upload fetch_cc_from_bucket.py with huggingface_hub

fetch_cc_from_bucket.py

@@ -0,0 +1,68 @@
+# /// script
+# requires-python = ">=3.8"
+# dependencies = [
+#   "warcio",
+#   "requests",
+# ]
+# ///
+"""Fetch a page's content from a NFS-mounted Common Crawl Hugging Face Bucket."""
+
+import argparse
+import io
+import json
+from pathlib import Path
+from urllib.parse import quote_plus
+
+import requests
+from warcio.archiveiterator import ArchiveIterator
+
+SERVER = "http://index.commoncrawl.org/"
+INDEX_NAME = "CC-MAIN-2026-17"
+USER_AGENT = "cc-hf-jobs/1.0 (Buckets Example)"
+
+def search_cc_index(url):
+    encoded_url = quote_plus(url)
+    index_url = f"{SERVER}{INDEX_NAME}-index?url={encoded_url}&output=json"
+    response = requests.get(index_url, headers={"user-agent": USER_AGENT})
+    response.raise_for_status()
+    return [json.loads(line) for line in response.text.strip().split("\n")]
+
+def fetch_page_from_bucket(records, bucket_root):
+    for record in records:
+        offset = int(record["offset"])
+        length = int(record["length"])
+        warc_path = bucket_root / record["filename"]
+
+        if not warc_path.exists():
+            print(f"Not in bucket: {warc_path}")
+            continue
+
+        with warc_path.open("rb") as f:
+            f.seek(offset)
+            chunk = f.read(length)
+
+        # The byte range covers exactly one gzipped WARC record, so we can
+        # hand it to ArchiveIterator as an in-memory stream.
+        for warc_record in ArchiveIterator(io.BytesIO(chunk)):
+            if warc_record.rec_type == "response":
+                return warc_record.content_stream().read()
+
+    print("No response record found")
+    return None
+
+def main():
+    parser = argparse.ArgumentParser(description=__doc__.splitlines()[0])
+    parser.add_argument("url", help="URL to look up in the CC index (e.g. huggingface.co/blog)")
+    parser.add_argument("bucket_path", type=Path, help="Path to the mounted Common Crawl bucket")
+    args = parser.parse_args()
+
+    records = search_cc_index(args.url)
+    print(f"Found {len(records)} record(s) for {args.url}")
+
+    content = fetch_page_from_bucket(records, args.bucket_path)
+    if content is not None:
+        print(f"Fetched {len(content)} bytes")
+        print(content[:500].decode("utf-8", errors="replace"))
+
+if __name__ == "__main__":
+    main()