Providers

Providers

Providers extract schema information from data sources. infermap auto-detects the right provider.

Auto-Detection

Input	Provider
.csv, .parquet, .xlsx	FileProvider
postgresql://, mysql://, sqlite://, duckdb://	DBProvider
.yaml, .yml, .json	SchemaFileProvider
Polars/Pandas DataFrame, list[dict]	InMemoryProvider

Database Support

DB	URI Format	Install
SQLite	sqlite:///path/to/db	built-in
PostgreSQL	postgresql://user:pass@host:5432/db	pip install infermap[postgres]
DuckDB	duckdb:///path/to/db.duckdb	pip install infermap[duckdb]
MySQL	mysql://user:pass@host:3306/db	pip install infermap[mysql]

Schema Definition Files

fields:
  - name: email
    type: string
    aliases: [email_address, e_mail]
    required: true
  - name: phone
    type: string
    aliases: [tel, mobile]

These add extra signals to the scorer pipeline — they don't bypass it.

FieldInfo

Every provider produces FieldInfo objects:

Field	Type	Description
name	str	Column name
dtype	str	Normalized type (string/integer/float/boolean/date/datetime)
sample_values	list[str]	Sampled values (stringified)
null_rate	float	Fraction of nulls (0.0-1.0)
unique_rate	float	Fraction of unique values
value_count	int	Non-null count
metadata	dict	Provider-specific extras