Python API

`scan_project`

from pandera_ui import scan_project

schemas: list[SchemaMetadata] = scan_project(project_path)

Walks project_path recursively and returns a list of SchemaMetadata objects — one per discovered Pandera schema.

Parameters

Parameter	Type	Description
`project_path`	`str \\| Path`	Root directory to scan

Returns list[SchemaMetadata]

Skipped directories: .venv, venv, __pycache__, .git, node_modules, dist, build.

Models

All models are Pydantic BaseModel subclasses and can be serialized with .model_dump().

`SchemaMetadata`

Top-level schema descriptor.

Field	Type	Description
`name`	`str`	Schema name (from `name=` kwarg or class name)
`var_name`	`str`	Python variable or class name in source
`file_path`	`str`	Path relative to the scanned root
`source_class`	`str \\| None`	Class name if defined via `DataFrameModel`
`title`	`str \\| None`	Short display title
`description`	`str \\| None`	Schema documentation
`coerce`	`bool`	Whether dtypes are coerced automatically
`columns`	`list[ColumnMetadata]`	Column descriptors
`index`	`IndexMetadata \\| None`	Index descriptor
`metadata_source`	`"runtime" \\| "ast"`	How the schema was extracted

`ColumnMetadata`

Field	Type	Description
`name`	`str`	Column name
`dtype`	`str \\| None`	Pandas dtype string, e.g. `"int64"`
`nullable`	`bool`	Whether `NaN`/`None` values are allowed
`required`	`bool`	Whether the column must be present
`checks`	`list[CheckMetadata]`	Validation checks
`title`	`str \\| None`	Short display label
`description`	`str \\| None`	Column documentation

`CheckMetadata`

Field	Type	Description
`name`	`str`	Check function name, e.g. `"greater_than"`
`statistics`	`dict[str, Any]`	Check parameters, e.g. `{"min_value": 0}`
`description`	`str \\| None`	Human-readable description
`error`	`str \\| None`	Custom error message

`IndexMetadata`

Field	Type	Description
`name`	`str \\| None`	Index name
`dtype`	`str \\| None`	Index dtype
`nullable`	`bool`	—
`checks`	`list[CheckMetadata]`	—

`CoverageReport`

Returned by GET /api/coverage and produced by the --coverage flag.

Field	Type	Description
`schema_count`	`int`	Total schemas
`schemas_with_title`	`int`	Schemas with a `title`
`schemas_with_description`	`int`	Schemas with a `description`
`column_count`	`int`	Total columns across all schemas
`columns_with_title`	`int`	Columns with a `title`
`columns_with_description`	`int`	Columns with a `description`
`schema_title_pct`	`float`	% of schemas with a title
`schema_description_pct`	`float`	% of schemas with a description
`column_title_pct`	`float`	% of columns with a title
`column_description_pct`	`float`	% of columns with a description

Examples

Iterate columns and checks

from pandera_ui import scan_project

for schema in scan_project("./myproject"):
    print(f"=== {schema.name} ({schema.file_path}) ===")
    for col in schema.columns:
        checks = ", ".join(c.name for c in col.checks) or "none"
        print(f"  {col.name}: {col.dtype}, nullable={col.nullable}, checks=[{checks}]")

Export to JSON

import json
from pandera_ui import scan_project

schemas = scan_project("./myproject")
with open("schemas.json", "w") as f:
    json.dump([s.model_dump() for s in schemas], f, indent=2, default=str)

Compute coverage programmatically

from pandera_ui import scan_project
from pandera_ui._coverage import compute_coverage, format_coverage

schemas = scan_project("./myproject")
report = compute_coverage(schemas)
print(format_coverage(report))

Export to Markdown

from pandera_ui import scan_project
from pandera_ui._export import to_markdown

schemas = scan_project("./myproject")
print(to_markdown(schemas))