Skip to content

Row Constraints API

NotNull

ydata.constraints.rows.constraint.NotNull

Bases: RowConstraint

Check that values in the specified columns are not null.

Parameters:

Name Type Description Default
columns str | list[str]

Column(s) that must not contain null / NaN values.

required
name str | None

Human-readable label shown in engine summaries.

None

validate(dataset)

Validate the constraint against a dataset.

Parameters:

Name Type Description Default
dataset Dataset

Dataset to check.

required

Returns:

Name Type Description
pdDataFrame DataFrame

Boolean mask with one row per dataset row and one column per constrained column, True where the row satisfies the constraint.

Unique

ydata.constraints.rows.constraint.Unique

Bases: RowConstraint

Check that values in the specified columns contain no duplicates.

A row is flagged as violating when its value appears more than once in that column. This is the standard uniqueness / primary-key check.

Parameters:

Name Type Description Default
columns str | list[str]

Column(s) whose values must be unique.

required
name str | None

Human-readable label shown in engine summaries.

None

validate(dataset)

Validate the constraint against a dataset.

Parameters:

Name Type Description Default
dataset Dataset

Dataset to check.

required

Returns:

Name Type Description
pdDataFrame DataFrame

Boolean mask with one row per dataset row and one column per constrained column, True where the row satisfies the constraint.

GreaterThan

ydata.constraints.rows.constraint.GreaterThan

Bases: RowConstraint

Check that values in one or more columns are greater than a value or another column.

Against a number the comparison is strict (>). Against another column it is inclusive (>=), and only one column can be constrained.

Parameters:

Name Type Description Default
columns str | list[str]

Column(s) to check.

required
value float | str

A number, or the name of the column to compare against.

required
Example

GreaterThan(columns="end_date", value="start_date")

validate(dataset)

Validate the constraint against a dataset.

Args: dataset (Dataset): Dataset to test

Returns:

Name Type Description
pdDataFrame DataFrame

constraint mask

LowerThan

ydata.constraints.rows.constraint.LowerThan

Bases: RowConstraint

Check that values in one or more columns are lower than a value or another column.

Against a number the comparison is strict (<). Against another column it is inclusive (<=), and only one column can be constrained.

Parameters:

Name Type Description Default
columns str | list[str]

Column(s) to check.

required
value float | str

A number, or the name of the column to compare against.

required
Example

LowerThan(columns="discount", value=100)

validate(dataset)

Validate the constraint against a dataset.

Args: dataset (Dataset): Dataset to test

Returns:

Name Type Description
pdDataFrame DataFrame

constraint mask

Between

ydata.constraints.rows.constraint.Between

Bases: RowConstraint

Check that values in one or more columns lie in [lower_bound, upper_bound].

Both bounds are inclusive.

Parameters:

Name Type Description Default
columns str | list[str]

Column(s) to check.

required
lower_bound int | float

Minimum allowed value.

required
upper_bound int | float

Maximum allowed value.

required
Example

Between(columns="age", lower_bound=18, upper_bound=99)

validate(dataset)

Validate the constraint against a dataset.

Args: dataset (Dataset): Dataset to test

Returns:

Name Type Description
pdDataFrame DataFrame

constraint mask

Positive

ydata.constraints.rows.constraint.Positive

Bases: GreaterThan

Check that values in one or more columns are strictly greater than 0.

Parameters:

Name Type Description Default
columns str | list[str]

Column(s) to check.

required

validate(dataset)

Validate the constraint against a dataset.

Parameters:

Name Type Description Default
dataset Dataset

Dataset to test

required

Returns:

Name Type Description
pdDataFrame DataFrame

constraint mask

NotIncludedIn

ydata.constraints.rows.constraint.NotIncludedIn

Bases: RowConstraint

Check that values in a column are NOT in a specified blocklist.

This is the complement of :class:IncludedIn — rows containing any of the forbidden values are flagged as violations.

Parameters:

Name Type Description Default
column str

Column on which the constraint is applied.

required
values list[Any] | Any

Single value or list of values that are not allowed.

required
name str | None

Human-readable label shown in engine summaries.

None

validate(dataset)

Validate the constraint against a dataset.

Parameters:

Name Type Description Default
dataset Dataset

Dataset to check.

required

Returns:

Name Type Description
pdDataFrame DataFrame

Boolean mask with one row per dataset row and one column per constrained column, True where the row satisfies the constraint.

IncludedIn

ydata.constraints.rows.constraint.IncludedIn

Bases: RowConstraint

Check that values in a column belong to an allowed set.

Null values are violations.

Parameters:

Name Type Description Default
column str

Column to check.

required
values list[Any] | Any

Allowed value, or list of allowed values.

required
Example

IncludedIn(column="status", values=["open", "closed"])

validate(dataset)

Validate the constraint against a dataset.

Args: dataset (Dataset): Dataset to test

Returns:

Name Type Description
pdDataFrame DataFrame

constraint mask

StringLength

ydata.constraints.rows.constraint.StringLength

Bases: RowConstraint

Check that string values in the specified columns have an acceptable length.

Length is measured in characters after coercing to str. Both bounds are inclusive. Omit max_length to only enforce a minimum.

Parameters:

Name Type Description Default
columns str | list[str]

Column(s) to check.

required
min_length int

Minimum allowed string length (inclusive). Defaults to 0.

0
max_length int | None

Maximum allowed string length (inclusive). None means no upper bound.

None
name str | None

Human-readable label shown in engine summaries.

None

validate(dataset)

Validate the constraint against a dataset.

Parameters:

Name Type Description Default
dataset Dataset

Dataset to check.

required

Returns:

Name Type Description
pdDataFrame DataFrame

Boolean mask with one row per dataset row and one column per constrained column, True where the row satisfies the constraint.

Monotonic

ydata.constraints.rows.constraint.Monotonic

Bases: RowConstraint

Check that values in a column are monotonically ordered.

Row i is flagged as violating when it breaks the ordering relative to row i-1. The first row always passes (no predecessor to compare to).

Works with numeric and datetime columns. For strictly increasing/decreasing sequences, equal consecutive values are treated as violations.

Parameters:

Name Type Description Default
columns str | list[str]

Column(s) to check.

required
increasing bool

If True (default), values must be non-decreasing. If False, values must be non-increasing.

True
strict bool

If True, equal consecutive values are violations (strictly monotonic). Defaults to False.

False
name str | None

Human-readable label shown in engine summaries.

None

validate(dataset)

Validate the constraint against a dataset.

Parameters:

Name Type Description Default
dataset Dataset

Dataset to check.

required

Returns:

Name Type Description
pdDataFrame DataFrame

Boolean mask with one row per dataset row and one column per constrained column, True where the row satisfies the constraint.

Regex

ydata.constraints.rows.constraint.Regex

Bases: RowConstraint

Check that values in a column fully match a regular expression.

Values are converted to str before matching; the whole value must match (re.fullmatch).

Parameters:

Name Type Description Default
column str

Column to check.

required
regex str

Regular expression pattern.

required
Example

Regex(column="zip_code", regex=r"\d{5}")

validate(dataset)

Validate the constraint against a dataset.

Args: dataset (Dataset): Dataset to test

Returns:

Name Type Description
pdDataFrame DataFrame

constraint mask

BetweenDates

ydata.constraints.rows.constraint.BetweenDates

Bases: RowConstraint

Check that a date column is within a window of days relative to another date column.

A row passes when reference + lower_bound days <= constrained <= reference + upper_bound days (both bounds inclusive).

Parameters:

Name Type Description Default
constrained_column str

Date column being checked.

required
reference_column str

Date column the window is anchored to.

required
lower_bound int

Start of the window, in days from the reference.

required
upper_bound int

End of the window, in days from the reference.

required
Example

BetweenDates(constrained_column="ship_date", reference_column="order_date", ... lower_bound=0, upper_bound=7)

validate(dataset)

Validate the constraint against a dataset.

Args: dataset (Dataset): Dataset to test

Returns:

Name Type Description
pdDataFrame DataFrame

constraint mask

RelationConstraint

ydata.constraints.rows.constraint.RelationConstraint

Bases: RowConstraint

Check that each row's key and values match an allowed combination in a reference table.

A row passes when its key (key_columns) exists in the reference and each value column holds one of the values allowed for that key. In the reference, a value cell can be a scalar or a list of allowed values. Rows with a null key or value are violations.

Parameters:

Name Type Description Default
reference DataFrame

DataFrame with the allowed combinations.

required
key_columns list[str]

Columns that form the lookup key. Must hold scalars in the reference.

required
value_columns list[str]

Columns whose allowed values depend on the key.

required
Example

ref = pd.DataFrame({"country": ["PT", "US"], "currency": ["EUR", "USD"]}) RelationConstraint(reference=ref, key_columns=["country"], value_columns=["currency"])

validate(dataset)

Validate the constraint against a dataset.

Parameters:

Name Type Description Default
dataset Dataset

Dataset to test

required

Returns:

Name Type Description
pdDataFrame DataFrame

constraint mask (True = satisfied, False = violated)

CombineConstraints

ydata.constraints.rows.constraint.CombineConstraints

Bases: RowConstraint

Combine several row constraints into one.

Each constraint is first reduced to one boolean per row (a row passes when all its constrained columns pass). The results are then combined:

  • Operation.MERGE: the row passes when every constraint passes (AND).
  • Operation.XNOR (default): the row passes when either all constraints pass or none do. Operation.CHAIN is an alias kept for backwards compatibility.

Parameters:

Name Type Description Default
constraints list[RowConstraint]

Row constraints to combine.

required
operation Operation

How to combine them. Defaults to Operation.XNOR.

XNOR

validate(dataset)

Validate the constraint against a dataset.

Args: dataset (Dataset): Dataset to test

Returns:

Name Type Description
pdDataFrame DataFrame

constraint mask

RowCustomConstraint

ydata.constraints.rows.constraint.RowCustomConstraint

Bases: RowConstraint

Check a user-defined condition row by row.

With axis="column" (default), check receives the dataset as a pandas.DataFrame (or each entity group) and returns a boolean DataFrame or Series. With axis="rows", check is applied to each row and returns a boolean.

Parameters:

Name Type Description Default
check Callable

Callable returning the boolean mask (or a boolean per row).

required
columns str | list[str] | None

Column(s) the result is reported on. When omitted, every column returned by check is kept.

None
name str | None

Human-readable label shown in engine summaries.

None
axis Axis

"column" (default) for vectorized checks, "rows" to apply check one row at a time.

'column'
available_columns list[str] | None

Extra columns the check reads but which are not reported on.

None
entity str | None

Column that identifies an entity (e.g. a customer id). When set, check is applied within each entity separately.

None
Example

RowCustomConstraint(check=lambda df: df["end"] >= df["start"], columns="end")

validate(dataset)

Validate the constraint against a dataset.

Parameters:

Name Type Description Default
dataset Dataset

Dataset to test

required

Returns:

Name Type Description
pdDataFrame DataFrame

constraint mask