open-index/commoncrawl-urls
收藏资源简介:
Common Crawl URL Seed数据集是Common Crawl快照的完整URL列表,直接从爬虫的列式URL索引中提取,并以分区Parquet格式存储。Common Crawl是一个非营利组织,负责爬取网络并免费发布其存档。每个快照都提供一个列式索引,其中“url”列列出了爬虫捕获的每个页面。该数据集对该列进行去重和分片处理,以便下游爬虫可以从已知的前沿重新播种,而无需再次拉取超过150 GB的索引。当前版本包含爬虫CC-MAIN-2026-25(warc子集),共有2,098,491,742个URL,分256个分片。URL按主机分片,每个分片包含主机哈希到64位主机键空间连续片段的URL集合,确保每个主机的所有URL都在一个分片中,且分片覆盖整个空间无间隙无重叠。数据集模式包括两列:url(爬虫捕获的URL)和host(从URL解析出的主机,作为分片键)。数据集在ODC-By v1.0许可证下发布,使用也需遵守Common Crawl的使用条款。
The Common Crawl URL Seed dataset is a complete list of URLs sourced from Common Crawl snapshots, extracted directly from the crawler's columnar URL index and stored in partitioned Parquet format. Common Crawl is a non-profit organization that crawls the web and freely distributes its archived web content. Each snapshot provides a columnar index, where the "url" column enumerates every page captured by the crawler. This dataset deduplicates and shards this column, allowing downstream crawlers to reseed from a known crawling frontier without having to re-pull the over 150 GB index. The current release includes the crawl CC-MAIN-2026-25 (WARC subset), containing a total of 2,098,491,742 URLs split across 256 shards. URLs are sharded by host: all URLs belonging to the same host are grouped into a single shard by hashing the host into a contiguous segment of the 64-bit host key space, ensuring full coverage of the space with no gaps or overlaps across all shards. The dataset schema consists of two columns: "url" (the URL of the crawled page) and "host" (the host parsed from the URL, used as the sharding key). This dataset is released under the ODC-By v1.0 license, and its usage must also comply with Common Crawl's terms of service.




