Accessing Tumblr Data

matthewb:

Mostly as a reference for myself, I’ve summarised the types of data we extract from Tumblr, and highlighted the few areas where support is inconsistent across the methods provided.

I’ve noted previously, for example, that while we can get content by post type using the Tumblr API, the same can’t be achieved using a formatted URL. Conversely, there is now a URL format for displaying content by tag, while the API does not allow this directly.

The chart below depicts the five key ways in which posts are filtered (including the new native search), along with other commonly accessed data: tags, feeds and avatar images. Tumblr has announced changes to its theme engine which may in turn affect the API and URL schemes as well, so I’ll keep this chart updated where possible.

API Theme URL Posts Yes Yes Yes — by Type Yes Soon No — by Tag No ¹ Soon Yes — by Keyword No Yes Yes — by Date No No Yes — by Quantity Yes Soon No Tags (per post) Yes Yes Yes Tags (all) No ¹ No N/A Feed Yes ² Yes Yes Avatar No Yes No

Yes Supported No Not Supported Soon Announced N/A Not Applicable

¹ It’s technically possible to request every post in a Tumblr site using multiple API calls, and then filter or collate by tag, but this isn’t ideal.

² The feed URL is not specifically included in the API response, however it’s trivial to construct it given the username.

Is there data for reblogged posts? It would seem like a goldmine of interesting stuff, someone could make a Summize/Twitturly for Tumblr…