philiprehberger-schema-infer

Infer JSON schemas from sample data.

Installation

pip install philiprehberger-schema-infer

Usage

from philiprehberger_schema_infer import infer

samples = [
    {"name": "Alice", "age": 30, "active": True},
    {"name": "Bob", "age": 25, "email": "bob@test.com"},
]

schema = infer(samples)
# {
#   "type": "object",
#   "properties": {
#     "name": {"type": "string", "minLength": 3, "maxLength": 5},
#     "age": {"type": "integer", "minimum": 25, "maximum": 30},
#     "active": {"type": "boolean"},
#     "email": {"type": "string", "format": "email", ...}
#   },
#   "required": ["age", "name"]
# }

Full JSON Schema output

from philiprehberger_schema_infer import to_json_schema

schema = to_json_schema(samples)
# {
#   "$schema": "https://json-schema.org/draft/2020-12/schema",
#   "type": "object",
#   "properties": { ... },
#   "required": [...]
# }

Single value type inference

from philiprehberger_schema_infer import infer_type

infer_type([1, 2, 3])
# {"type": "array", "items": {"type": "integer"}}

Schema strictness levels

Control how aggressively fields are marked required and constraints are applied:

from philiprehberger_schema_infer import infer

# Loose: no required fields, no numeric/string constraints
schema = infer(samples, strictness="loose")

# Normal (default): fields in all samples are required, constraints included
schema = infer(samples, strictness="normal")

# Strict: all fields required, additionalProperties set to False
schema = infer(samples, strictness="strict")

Custom format detection

Register domain-specific regex patterns for format detection:

from philiprehberger_schema_infer import register_format, infer_type

register_format("phone", r"^\+\d{1,3}-\d{3,14}$")
register_format("credit-card", r"^\d{4}-\d{4}-\d{4}-\d{4}$")

infer_type("+1-5551234567")
# {"type": "string", "format": "phone"}

Merge schemas

Combine multiple inferred schemas with union/intersection logic for required fields:

from philiprehberger_schema_infer import merge_schemas

merged = merge_schemas(schema_a, schema_b, schema_c)

Confidence scores

Analyze how consistently a type was observed across samples for each field:

from philiprehberger_schema_infer import infer_with_confidence

samples = [
    {"name": "Alice", "value": 42},
    {"name": "Bob", "value": "hello"},
    {"name": "Carol", "value": 99},
]

result = infer_with_confidence(samples)
# {
#   "name": {"type": "string", "confidence": 1.0},
#   "value": {"type": ..., "confidence": 0.67}
# }

TypeScript interface output

Generate TypeScript interfaces from sample data:

from philiprehberger_schema_infer import to_typescript

samples = [
    {"name": "Alice", "age": 30, "active": True},
    {"name": "Bob", "age": 25},
]

print(to_typescript(samples, name="User"))
# interface User {
#   active?: boolean;
#   age: number;
#   name: string;
# }

Inference from a .jsonl file

Infer a schema directly from a JSON Lines file without loading it manually:

from philiprehberger_schema_infer import infer_from_jsonl

schema = infer_from_jsonl("events.jsonl")

# Skip lines that aren't valid JSON objects instead of raising
schema = infer_from_jsonl("events.jsonl", skip_invalid=True)

Python dataclass output

Generate Python dataclass definitions from sample data:

from philiprehberger_schema_infer import to_dataclass

samples = [
    {"name": "Alice", "age": 30, "email": "alice@test.com"},
    {"name": "Bob", "age": 25},
]

print(to_dataclass(samples, name="User"))
# @dataclass
# class User:
#     age: int
#     name: str
#     email: str | None = None

API

Development

pip install -e .
python -m pytest tests/ -v

Support

If you find this project useful:

⭐ Star the repo

🐛 Report issues

💡 Suggest features

❤️ Sponsor development

🌐 All Open Source Projects

💻 GitHub Profile

🔗 LinkedIn Profile

License

MIT