schemarize 0.1.3

A toolkit to infer data schemas from samples

Schemarize

Schemarize is a Python package to infer, inspect, and serialize the schema of data files and objects—whether they’re JSON, CSV, Parquet, or DataFrames. It’s designed for simplicity: just call one method, get a schema, and export as JSON, YAML, or CSV.

GITHUB

Installation

Install from PyPI:

pip install schemarize

---

Quickstart

import pandas as pd
from schemarize import schemarize

df = pd.DataFrame({"a": [1, 2], "b": ["x", "y"]})
schema = schemarize(df)

print(schema.to_json())
print(schema.to_yaml())
print(schema.to_csv())

schema.save("my_schema.json")   # or .yaml, .csv

You can also pass a file path (CSV, JSON, Parquet, etc.):

schema = schemarize("data.csv")
print(schema.to_dict())

---

Main Function

schemarize(source, sample_size=None) -> Schema

• source: File path, DataFrame, Arrow Table, or file-like object.
• sample_size: optional max records to smple.

Returns a Schema object.

---

Schema Methods

The returned Schema object provides:

• to_dict() – Return the schema as a Python dict.
• to_json(pretty=True) – Serialize as JSON string.
• to_yaml() – Serialize as YAML string.
• to_csv() – Serialize as CSV string.
• save(path, format=None) – Save to file, format auto-detected by extension.

Example: Saving a schema

schema.save("my_schema.yaml")
schema.save("my_schema.csv")

---

Supported Inputs

• Pandas DataFrames
• PyArrow Tables
• File paths: CSV, JSON, JSONL, Parquet (compressed files supported)
• File-like objects

---

Supported Outputs

• JSON
• YAML (requires pyyaml)
• CSV (flat field view)

---

License

MIT

---

Links

• Issues