flipsidecrypto / sdk

Flipside SDK (formerly known as ShroomDK)

Programmatic access to the most reliable & comprehensive blockchain data in Web3.

You've found yourself at the FlipsideCrypto SDK repository, the official SDK to programmatically query all of Flipside Crypto's data.

🧩 The Data

Flipside Crypto's Analytics Team has curated dozens of blockchain data sets with more being added each week. All tables available to query in Flipside's Data Studio can be queried programmatically via our API and library of SDKs.

📖 Official Docs

https://docs.flipsidecrypto.com/flipside-api/get-started

🗝 Want access? Genrate an API Key for Free

Get your free API key here

SDKs

Language	Version	Status
✅ Python	2.0.8
✅ JS/TypeScript	2.0.1
✅ R	0.2.2	Available on CRAN

sdk's Issues

Query that returns 0 results throws an error in SDK

Query that returns 0 results throws an error in SDK

I would expect the behavior to be that the resulting records are and empty list and run_stats.record_count would be 0.
In the UI when a query returns no results the query still executes successfully.

Query error time is in seconds

I received this error:
https://github.com/FlipsideCrypto/sdk/edit/main/python/shroomdk/errors/query_run_errors.py#L22
although the time I got was 1400+ minutes. I feel it was more like 1400+ seconds. Just a minor issue though, not breaking anything.

Query Run Timeout Error returning incorrect time unit, minutes for all

Unclear Error Message

Description

When running a query using the Python SDK with 'TIMEOUT MINUTES' set, if the query times out the error message text returned expresses the timeout in minutes when it should either be seconds or convert seconds to minutes.

For example my query was set to run with a 2 min timeout, and ran for ~136 seconds, the error message returned was as follows:

raise QueryRunTimeoutError(elapsed_seconds)
flipside.errors.query_run_errors.QueryRunTimeoutError: QUERY_RUN_TIMEOUT_ERROR: your query has timed out after 136.0 minutes.

It seems that the timeout is being passed elapsed_seconds and not converting to minutes.

https://github.com/FlipsideCrypto/sdk/blob/73e52bf82507dc53e4026554802e13339d3e3505/python/src/errors/query_run_errors.py#LL24C8-L26C39

To Reproduce

1. Create a query that will reliably run for longer than 2 minutes.
2. Use the python SDK and set TIMEOUT_MINUTES to 2
3. Issue the query and review the error message raised. The error message will show that the query has timed out after x number of 'minutes', where x is the number of seconds instead of minutes.

API ERROR

Hello we started noticing this error all of a sudden from the sdk.

Our sql query and API key still the same as before

QueryResultSetBuilder {
    error: ApiError: UnknownAPIError: message=, code=-1
        at getExceptionByErrorCode (/Users/charles/Desktop/pear-protocol/pear-backend/node_modules/@flipsidecrypto/sdk/src/errors/api-error.ts:37:12)
        at QueryIntegration.<anonymous> (/Users/charles/Desktop/pear-protocol/pear-backend/node_modules/@flipsidecrypto/sdk/src/integrations/query-integration/index.ts:62:39)
        at Generator.next (<anonymous>)
        at fulfilled (/Users/charles/Desktop/pear-protocol/pear-backend/node_modules/@flipsidecrypto/sdk/dist/src/integrations/query-integration/index.js:5:58)
        at processTicksAndRejections (node:internal/process/task_queues:95:5),
    queryId: null,
    status: null,
    columns: null,
    columnTypes: null,
    rows: null,
    runStats: null,
    records: null,
    page: null
  }

We are not sure what could be the cause

requirements.txt version constraints too strict

The version constraints on requirements.txt in the /python directory are unnecessarily strict. This breaks by environment since some packages require older versions of requests (super common package) or newer versions of pydantic (fairly popular pacakge with version 2.x out now)

I propose you lower the requests requirement to >=2.20.0 or similar

Also, pydantic is on version 2.0.2 now so something like pydantic>=1.10.9 (granted 2.0.2 passes tests) would be preferable

DEFAULT_PAGE_SIZE = 1000 instead of 100,000 as stated in docs

DEFAULT_PAGE_SIZE = 1000, this used to be 100,000 differs from documentation stating that default page size is 100,000.
This requires users of the SDK that were regularly returning more than 1000 records to update every query they run when adopting the new version of the SDK.

sdk/python/src/flipside.py

Line 19 in aafde1b

DEFAULT_PAGE_SIZE = 1000

Docs

https://docs.flipsidecrypto.com/flipside-api/get-started/python

QUERY_RUN_EXECUTION_ERROR: SQL compilation error: Missing column specification

Python SDK: 2.0.7

Query works great on xyz studio app and has been working fine up until today:

  SELECT RECORDED_HOUR, MAX(CLOSE)
    FROM solana.core.fact_token_prices_hourly
    WHERE symbol = 'ATLAS'
    AND RECORDED_HOUR > '{timestamp}'::timestamp without time zone
    GROUP BY RECORDED_HOUR
    ORDER BY RECORDED_HOUR

We'd been using the FS SDK, 2.0.3, for last week or two and had no problems up till yesterday, everything broke this morning.

Column names don't match what is seen in Data Studio

columnNames returned by the Python SDK are all lowercase. In the Data Studio, they are uppercase, or case can be made explicit using double quotes. The behavior of shroomDK v1.0.2 matched the data studio. This change effects downstream processing, which is case sensitive for column names.

Is it possible to re-enable the old behavior?

Example:

select 
  block_timestamp::date as "Date",
  count(distinct tx_id) as tx_count,
  program_id
from
  solana.core.fact_events
where
  block_timestamp::date = '2023-05-21'
group by
  program_id,
  "Date"

Expected results

Column names:

• Date
• TX_COUNT
• PROGRAM_ID

Actual results

Column names:

• date
• tx_count
• program_id
• __row_index

Note, an additional __row_index column is included even though it wasn't included in the query

Question about the pydantic version

Hi flipside team,

Wondering if there's any planned update to the pydantic version pinned in the python sdk?
Its currently at v1 which is incompatible with every other python library because they depend on pydantic v2

Thank you!

JS/TS SDK Doesn't Work with Cloudflare Workers

When used inside a Cloudflare Worker, this is the error message that appears:

Error: UNEXPECTED_SDK_ERROR: adapter is not a function

Which according to this StackOverflow post, is probably due to Axios not recognizing the environment its in and using the wrong code to try and execute the request.

Until this is fixed I'll be using the RPC API directly with fetch.

Error when imported @flipsidecrypto/sdk

Hi team, I got this error while import the @flipsidecrypto/sdk

Please let me know how can I fix it.
Thanks

Pydantic pinned to buggy version, breaking compatibility with another package

A bug (pydantic/pydantic#5821) in pydantic==1.10.7 has been fixed in pydantic==1.10.8

The flipside python API is pinned to pydantic==1.10.7. Another package I'm using is pinning around pydantic==1.10.7 to avoid the referenced bug.

Suggest relaxing the pydantic to maintain compatibility, thanks.