Output as generator with lazy-loading pagination #12
Conversation
| @@ -35,7 +35,7 @@ def _request(self, endpoint, data, method='POST'): | |||
| response, data = self.http.request(uri, method=method, headers={'Content-Type': 'application/x-www-form-urlencoded'}) | |||
| else: | |||
| uri = PIPEDRIVE_API_URL + endpoint + '?api_token=' + str(self.api_token) | |||
| response, data = self.http.request(uri, method=method, body=urlencode(data), headers={'Content-Type': 'application/x-www-form-urlencoded'}) | |||
| response, data = self.http.request(uri, method=method, body=json.dumps(data), headers={'Content-Type': 'application/json'}) | |||
There was a problem hiding this comment.
"We recommend using JSON body format when performing API requests" https://developers.pipedrive.com/docs/api/v1/#/
There was a problem hiding this comment.
We can see general issues with Pipedrive API server implementation (wrong specification of API server - they rely on "Content-Type" header instead of "Accept" - look to RFC: https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Accept):
- When using
requestsPython library, I can see HTML instead of JSON the same with curl - Documentation doesn't follow HTTP RFC in case of headers request specification, what is wrong today (maybe 20 years ago it would have been fine, but not in 2024!) -> https://pipedrive.readme.io/docs/core-api-concepts-requests
Sometimes junior developers needs to listen critical comments carefully :-)
This is not comment to you @kevhill - this is for Pipedrive API server developers :-)
|
Looks good to me - but I don't use the library any more so I'll wait and see if there's more feedback on it either way okay? :) |
|
Using it with this PR and it works fine. 👍 |
This is a backward incompatible change to the way you use the pipedrive client, but I found it useful so I'm putting it up here for discussion.
Basically, I ended up writing a bunch of management code to get more than 100 things to handle pipedrive's pagination. Instead I really just wanted to write simple
forloops over the results. So, I decided to change the wrapper into something that returned a python-style generator. The interface is quite nice for basic things, but if you rely on usingresponse['additional_data']then that is now obscured.tldr, you can now do this: