Skip to main content
tiktok_search runs TikTok keyword searches. One result is one video, and every row carries the video’s author fields. A video that matches two keywords is returned once, under the first.

Run it

gofetch-client 0.8.0 does not list tiktok_search yet. client.actor("tiktok_search") works, and Python prints a UserWarning: Unknown actor URL 'tiktok_search', passing through as scraper type you can ignore.

Inputs

Keywords run one after another, in the order you send them. A search also tries close spellings of the keyword, so home workout can return videos for homeworkout.

Output

One row per video. There are no error or placeholder rows: every row is a billed result. A real row from a search for nasa:
This row was captured before song.id became text, so it shows the id as a rounded number. Every id in a row is now text.

What logged-out means here

You get: what TikTok’s search shows a signed-out visitor for each keyword, with the video’s counts and its author’s public fields. Not available: choosing the country or region a search runs for. A location input is accepted and ignored; a value other than US adds a warning to the job’s log. A keyword TikTok did not answer is marked blocked in completeness.keywords, and the job still completes with the other keywords’ rows. The job fails, with error_code BLOCKED, only when it collected nothing and at least one keyword got no answer.

Completeness

scraper_metadata.completeness: