yekta-sdk

Python SDK for the Yekta data catalog — credential vending, catalog browsing, access control.

Install

pip install yekta-sdk            # Core
pip install yekta-sdk[duckdb]    # + DuckDB integration
pip install yekta-sdk[polars]    # + Polars/Delta integration
pip install yekta-sdk[all]       # Everything

Quick Start

Browse the catalog

from yekta import YektaClient

client = YektaClient.from_env()  # reads YEKTA_URL, YEKTA_TOKEN

# List resources
for r in client.ls("/dev/bronze"):
    print(f"{r.path} ({r.format})")

# Preview data
df = client.preview_df("/dev/bronze/memotech/patents", rows=100)
df.head()

# Schema
schema = client.schema("/dev/bronze/memotech/patents")
for col in schema.columns:
    print(f"  {col.name}: {col.data_type}")

Get temporary credentials (the core feature)

from yekta import CredentialVendor

vendor = CredentialVendor.from_env()  # reads SEPAR_URL, SEPAR_API_KEY

# Vend a short-lived, scoped SAS token
cred = vendor.vend(
    connection_id="019c479e-...",
    resource_path="dev/bronze/memotech/fabric/memotech_imt_patent",
)

print(cred.expires_at)  # 1 hour from now
print(cred.is_expired)  # False

Use with DuckDB

import duckdb

conn = duckdb.connect()
conn.execute(cred.duckdb_secret())

df = conn.sql("""
    SELECT country, COUNT(*) as cnt
    FROM delta_scan('az://dagster-data-pipelines/dev/bronze/.../memotech_imt_patent')
    GROUP BY country ORDER BY cnt DESC
""").df()

Use with Polars

import polars as pl

df = pl.read_delta(
    cred.abfss_uri(),
    storage_options=cred.storage_options(),
)

Use with Spark

for key, value in cred.spark_config().items():
    spark.conf.set(key, value)

df = spark.read.format("delta").load(cred.abfss_uri())

Access control

# List grants
for g in client.grants():
    print(f"{g.path_prefix} → {g.principal_id}: {g.permission}")

# Grant read access
client.grant("/dev/bronze", principal_id="user-uuid", permission="read")

# Revoke
client.revoke(grant_id="...")

Environment Variables

License

Apache-2.0