fix: ignore null items in API responses #240
Conversation
- configure the talk client to use the production API for tests. - test a talk production search that is known to break the API client. - ignore null response data, so that the tests pass.
|
@mcbouslog this fixes a bug that Alisa found earlier this week. |
eatyourgreens
force-pushed
the
ignore-null-comments
branch
from
eee5104
to
4a264d1
Compare
eatyourgreens
force-pushed
the
ignore-null-comments
branch
from
45ab170
to
43bcf0e
Compare
|
Now with mocks, so that the tests aren't running searches against the live API. The mocks are simply JSON snapshots from real Talk searches, including the search that's broken. Includes a real Talk production search query that only returns |
|
zooniverse/Panoptes-Front-End#7071 installs this version of the client in PFE. It needs to be deployed to staging, but here's a link to one of the broken searches: https://pr-7071.pfe-preview.zooniverse.org/talk/search?query=depression&env=production |
There was a problem hiding this comment.
Changes look good, mock talkClient and testing additions look great.
Confirmed using zooniverse/Panoptes-Front-End#7071 locally that existing error message "There was an error with your search. Please try again." is fixed with these changes 👍 .
On page 3 of the results (https://local.zooniverse.org:3735/talk/search?env=production&page=3&query=depression) the pagination shows a page 4 and 5, which are the null response data pages, which is a little confusing, but that's ok and beyond the scope of these changes, just noting for myself.
|
There's definitely something weird going on with that search. After the first 29 results, the remaining 13 are all null. |
nock, using snapshots of actual API responses.