…setup so the code is currently broken. WIP.

Convert all partitions, not just the first one.

Fix placeholders of querystring params that need gotten/set for fetching subsequent partitions.

Fix syntax errors to ensure that isn't why connector upload is failing.

Fix more compilation errors.

Wanted a record of the fact that these changes are still returning subsequent partitions in array format, despite literally every response being converted.

Is there maybe some sort of caching of the connector behavior, because it seems like I can't ever get the behavior of my data flow to change at all...

This version of the connector is the most complete example that can successfully be uploaded as a custom connector.

Yet I still can't get the behavior to change no matter what code changes I make.

* Update script.csx

- Fix null detection.

* Fix issue with null handling in Snowflake connector

---------

Co-authored-by: jbrinkman <[email protected]>

- This version of the swaggerjson SHOULD be working, but we are seeing the DataSchema object being flattened out once uploaded to customer connector.

Got the swagger right (was really the code was had checked in before with just a little cleanup.

The custom connector is now failing due to internal server error so we need to find a way to use the test page in powerapps online, despite the fact that it doesn't really handle array data very well. Possibly specifying the raw body data might be a workaround.

- Last few tweaks to get the customer connector to return subsequent partitions in pre-converted format.

- extra body element is required. Caused a whole mess of issues.
- change DataSchema to required and deprecate or delete unused endpoints as needed.
- Remove fetchAllPages feature and separate into its own branch.

- Make log messages more accurate.
- Remove last remnant of fetchAllPartitions.

Add version information into readme documentation

SPC-12/SPC-32 - All partitions should be returned in json class notation.

- The code is acting absolutely insane and returning GetResults method as just a single property "Data" formatted as array. Since this is the 0 partition it should include metadata.
- The interface is also not showing the partition parameter for the execsql method, so something is borked.

- This was a very subtle issue related to the fact that when you call GetResults operation for partition zero you have no request body, so it cannot be parsed as json.
- Change the response of execstmt async to match the schema of the sync version b/c the powerapps ui does not seem to be able to deal with the fact that async/sync have different response formats respectively.

Fix Async always erroring out

* Update script.csx

- Fix async detection based on response code instead of request params, b/c apparently snowflake API can decide to return an async response if a synchronous response takes too long to return.

* Fix typo in script.csx

"BeginFetch" misspelled

---------

Co-authored-by: Joseph Brinkman <[email protected]>

* Update apiDefinition.swagger.json

- change parameter name case to match snowflake docs exactly

* add StatementHandles

Map new response property for multi-statement handling.

* Apply mappings to GetResults

Same statementHandles mapping that was previously added to ExecSql was applied to GetResults to support Async

* Remove async fixes

These changes are already in dev branch, it was just a temporary change for debugging.

Innacurate schema was causing compilation issues in the power apps. Better to leave it as a untyped object since the schema is dynamic.

They were represented as a string before.

* Document limitations per my experience.

* - Tweak readme

* Update language limitations in the Readme documentation.

---------

Co-authored-by: jbrinkman <[email protected]>

I was able to type the untyped objects, but a lot of those openapi spec validation errors are inherent to the fact that the snowflake api routes are technically all partial matches for eachother since the exec stmt path is "/"