Pages with role="text"

Related Usage of ARIA attributes

question from James Craig

Thanks for this. Is there any want to trace back the 421 instances of
role=“text” to specific domains? e.g. Can I get more information about
where those were used?

  SELECT page, url, REGEXP_EXTRACT(LOWER(body), r'(<[a-z-]+\s(?:[^>]+\s)?role\s*=\s*["\']?text(?:[^a-z0-9][^>]*>|>))') AS tag
  FROM [httparchive:har.2016_06_15_chrome_requests_bodies]
) WHERE tag != "null"

This query for some reason found 600 matches in the same dataset. I’m not sure why; removing duplicate url entries doesn’t bring it down to 421. Anyway, results at