update python SDK #22
Conversation
nofel-scale
commented
Feb 10, 2021
nofel-scale
commented
- Update README
- add Project API support
- expand batch API support
- update supported task types
nofel-scale
requested review from
IGHARARI,
ivan-scale,
shaun-scale,
mathias-scale and
jonathanfeng-scale
README.md
Outdated
| =================== | ||
| Scale AI | Python SDK | ||
| =================== | ||
|
|
||
| # Installation |
There was a problem hiding this comment.
This file wasn't markdown before - we need to either change the syntax from the .rst file to .md, or save the file again as an .rst file.
The internet does make a reasonably strong case for documenting Python SDKs in .rst
https://eli.thegreenplace.net/2017/restructuredtext-vs-markdown-for-technical-documentation/
https://www.zverovich.net/2016/06/16/rst-vs-markdown.html#:~:text=Both%20are%20lightweight%20markup%20languages,So%20what%20are%20the%20differences%3F
But we're still pretty light-weight and I think 99% of the engineers here just use Markdown.. I'll let you figure out the lift and pros / cons of converting everything.
README.md
Outdated
|
|
||
| client.get_batch( batch_name = "batch_name_01_07_2021" ) | ||
|
|
||
| ## List Batchs |
scaleapi/__init__.py
Outdated
| TASK_TYPES = [ | ||
| 'annotation', | ||
| 'audiotranscription', | ||
| 'categorization', | ||
| 'comparison', | ||
| 'cuboidannotation', | ||
| 'datacollection', | ||
| 'imageannotation', | ||
| 'lineannotation', | ||
| 'namedentityrecognition', | ||
| 'pointannotation', | ||
| 'polygonannotation', | ||
| 'segmentannotation', | ||
| 'transcription', | ||
| 'documenttranscription', | ||
| 'videoannotation', | ||
| 'videoboxannotation', | ||
| 'videocuboidannotation' | ||
| 'videoplaybackannotation', | ||
| 'namedentityrecognition', | ||
| 'textcollection', | ||
| 'documentmodel' | ||
| ] |
There was a problem hiding this comment.
@nofel-scale you're actually missing a big part of why this array exists.
At the bottom of init.py, we have:
def _AddTaskTypeCreator(task_type):
def create_task_wrapper(self, **kwargs):
return self.create_task(task_type, **kwargs)
setattr(ScaleClient, 'create_' + task_type + '_task', create_task_wrapper)
for taskType in TASK_TYPES:
_AddTaskTypeCreator(taskType)
This allows the client to have client.create_imageannotation_task as a method, plus all the other task types. You can still use the raw create_task method for things like Lidar if you wanted, but we should call out in the readme that these other functions do exist. (You removed it for now)
