ygg 0.4.8

Type-friendly utilities for moving data between Python objects, Arrow, Polars, Pandas, Spark, and Databricks

Yggdrasil (Python)

Schema-aware utilities for moving data between Python objects, Arrow, Polars, pandas, Spark, and Databricks. Define types once — cast everywhere.

Install

pip install ygg                           # core (Arrow, requests, pyutils)
pip install "ygg[polars]"                # + Polars
pip install "ygg[pandas]"                # + pandas
pip install "ygg[spark]"                 # + PySpark
pip install "ygg[databricks]"            # + Databricks SDK

From source (dev):

cd python/
uv venv .venv && source .venv/bin/activate
uv pip install -e ".[dev]"

Quickstart

Infer Arrow schema from a dataclass

from dataclasses import dataclass
from yggdrasil.dataclasses import dataclass_to_arrow_field

@dataclass
class Order:
    id: int
    amount: float
    country: str | None = None

field = dataclass_to_arrow_field(Order)
print(field.type)           # struct<id: int64, amount: double, country: string>
schema = field.type.to_schema()

Cast any table to an Arrow schema

import pyarrow as pa
from yggdrasil.arrow.cast import cast_arrow_tabular
from yggdrasil.data.cast import CastOptions

target = pa.schema([
    pa.field("id", pa.int64()),
    pa.field("amount", pa.float64()),
])
raw = pa.table({"id": ["1", "2"], "amount": ["10.5", "20.0"]})
out = cast_arrow_tabular(raw, CastOptions(target_field=target))

Retry + parallel

from yggdrasil.pyutils import retry, parallelize

@retry(tries=5, delay=0.5, backoff=2.0)
def fetch(url: str) -> bytes: ...

@parallelize(max_workers=8)
def process(item: str) -> dict:
    return {"result": item.upper()}

results = list(process(["a", "b", "c"]))

Databricks — SQL with typed results

from yggdrasil.databricks.workspaces import Workspace
from yggdrasil.databricks.sql import SQLEngine

ws = Workspace(host="https://<workspace>", token="<pat>").connect()
engine = SQLEngine(catalog_name="main", schema_name="analytics", workspace=ws)

result = engine.execute("SELECT id, amount FROM transactions LIMIT 100")
df = result.to_pandas()
arrow_table = result.to_arrow_table()

Module map

Module	Key exports
yggdrasil.arrow	arrow_field_from_hint
yggdrasil.arrow.cast	cast_arrow_tabular, cast_arrow_array
yggdrasil.data.cast	CastOptions, convert, register_converter
yggdrasil.dataclasses	dataclass_to_arrow_field
yggdrasil.pandas.cast	cast_pandas_dataframe
yggdrasil.polars.cast	cast_polars_dataframe, cast_polars_lazyframe
yggdrasil.spark.cast	cast_spark_dataframe
yggdrasil.pyutils	retry, parallelize
yggdrasil.concurrent	JobPoolExecutor, Job
yggdrasil.requests	YGGSession
yggdrasil.io	BytesIO, Codec, MediaType
yggdrasil.deltalake	DeltaTable
yggdrasil.databricks	Workspace, SQLEngine, Cluster, NotebookConfig

Docs

Module reference →

Test

cd python/
pytest
ruff check .
mypy