phl-ldm/tranco_top_million

Top 1M Tranco Crawl Dataset

A web crawl dataset covering the top 1 million domains from the Tranco list with ID "PYGVJ" (generated July 29, 2026), collected on July 30-31, 2026 using the distcrawl crawler. The crawling process took around 18.5 hours to complete, using ~100 parallel nodes deployed worldwide. The cloud costs for data collection totaled around €20.

Success Rate of the Crawl

Percentage of the 1 million target domains that were successfully crawled (present in site_metadata.parquet):

Successful sites	Success rate
Successful sites	762,110
Success rate	76.21%

Crawl Configuration

Configuration choice	Value
Browser Engine	Chromium
Auto Accept Cookies	Enabled
Execution Mode	Headless
Dwell Time [s]	10
Max Crawl Depth	0
Scroll Sequence [px]	100, 500, -500, -100

• Browser Version: Google Chrome for Testing 148.0.7778.96 (Headless Shell)
• Uses User-Agent: Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) HeadlessChrome/148.0.7778.96 Safari/537.36

Worker Nodes Metadata

Basic worker metadata (country, connection type based on IP address) are included, so filtering per worker country is possible.

File Structure

Each crawl configuration contains a metadata.json file with the exact crawl parameters, and an enriched/ directory with five Parquet tables:

labeled_requests.parquet

Every outgoing HTTP request observed during the crawl, labeled against several filter lists available on easylist.to (based on the latest available filter lists on July 31, 2026).

Column	Type	Description
experiment_id	string	Crawl configuration identifier
request_id	string	Request identifier (unique within a crawl_session_id)
worker_id	string	Worker that performed the crawl
crawl_session_id	string	Session ID for a single domain visit
timestamp	string	Request timestamp
crawled_url	string	Seed URL being crawled
current_page_url	string	Page URL that triggered the request
url	string	Requested URL
frame_url	string	Frame URL that initiated the request
resource_type	string	Resource type (script, image, fetch, document, xhr, etc.)
method	string	HTTP method
headers	string	Request headers (JSON)
crawl_depth	int	Navigation depth from the landing page
blocked_by	string	Filter lists that match this request (JSON array, e.g. ["easylist", "easyprivacy"]; empty [] = not blocked)

responses.parquet

HTTP responses received during the crawl.

Column	Type	Description
experiment_id	string	Crawl configuration identifier
request_id	string	Matching request identifier
crawl_session_id	string	Session ID for a single domain visit
timestamp	string	Response timestamp
crawled_url	string	Seed URL being crawled
url	string	Response URL
status	int	HTTP status code
headers	string	Response headers (JSON)
crawl_depth	int	Navigation depth from the landing page
cookies	string	Cookies set by the response

cookie_warning_consents.parquet

Pages where a cookie consent dialog was detected and clicked.

Column	Type	Description
experiment_id	string	Crawl configuration identifier
crawl_session_id	string	Session ID for a single domain visit
crawled_url	string	Seed URL being crawled
url	string	Page URL where consent was given
timestamp	int	Consent timestamp
crawl_depth	int	Navigation depth from the landing page

site_metadata.parquet

Metadata extracted from each visited site.

Column	Type	Description
crawl_session_id	string	Session ID for a single domain visit
description	string	Site meta description
timestamp	string	Collection timestamp

worker_metadata.parquet

Information about the distributed crawl workers.

Column	Type	Description
worker_id	string	Worker identifier
country_code	string	Worker geographic location
is_residential	bool	Whether the worker uses a residential IP
timestamp	double	Registration timestamp