API Reference

recx.Rec(columns, check_all=True, check_missing_indices=True, check_extra_indices=True, align_date_col=None)

Configure and run reconciliation between two DataFrames.

Steps performed when :meth:run is invoked:

1. Optionally clip both frames to their last common date (align_date_col).
2. Collect index presence checks (extra indices on each side).
3. Execute explicit per-column checks provided in columns.
4. Apply a default_check to any remaining columns not explicitly covered (unless default_check is None).

Parameters:

Name	Type	Description	Default
columns	dict[str, ColumnCheck | None]	Mapping of column names (or regex patterns if the associated check has regex=True) to checks. A value of None skips that column.	required
check_all	bool	All unspecified columns will be checked with :class:EqualCheck if True.	True
align_date_col	str	Optional date/datetime column (or index level) name used to clip both frames to their last common date before comparison.	None

run(baseline, candidate, raise_on_failure=False)

Execute all configured checks.

Parameters:

Name	Type	Description	Default
baseline	DataFrame	Baseline frame.	required
candidate	DataFrame	Candidate frame to reconcile against the baseline.	required
raise_on_failure	bool	If True raise :class:RecFailedException when any check fails.	False

Returns:

Type	Description
RecResult	The full list of check results (passing + failing). When raise_on_failure is True and failures occur this method raises an exception.

recx.ColumnCheck(regex=False, **kwargs)

Bases: ABC

check(baseline, candidate) abstractmethod

Evaluate the check for a single column.

Parameters:

Name	Type	Description	Default
baseline	Series	Baseline series.	required
candidate	Series	Candidate series (index-aligned with baseline).	required

Returns:

Type	Description
DataFrame	Frame of failing rows (may include additional diagnostic columns) or an empty frame if the check passed entirely.

run(baseline, candidate, column)

Execute the check for one (possibly regex) column pattern.

Parameters:

Name	Type	Description	Default
baseline	DataFrame	Baseline frame containing the columns.	required
candidate	DataFrame	Candidate frame.	required
column	str	Exact column name or regex pattern (if regex) selecting columns to test.	required

Returns:

Type	Description
list[CheckResult]	One result per concrete column matched.

recx.EqualCheck(regex=False, **kwargs)

Bases: ColumnCheck

Check that baseline and candidate values are exactly equal.

NaNs (nulls) in the same position are treated as equal.

recx.AbsTolCheck(tol, sort=None, regex=False)

Bases: ColumnCheck

Check absolute difference within tolerance.

The absolute error abs(baseline - candidate) must be <= tol or the row is flagged. Matching nulls are ignored.

Parameters:

Name	Type	Description	Default
tol	float	Maximum permitted absolute difference.	required
sort	(asc, desc)	Sort order for failing rows by absolute error.	'asc'
regex	bool	Treat the provided column spec as a regex pattern.	False

recx.RelTolCheck(tol, sort=None, regex=False)

Bases: ColumnCheck

Check relative difference within tolerance.

Relative error is computed as abs(baseline - candidate) / |candidate|. A small stabiliser 1e-10 is used to avoid division by zero.

Parameters:

Name	Type	Description	Default
tol	float	Maximum permitted relative error.	required
sort	(asc, desc)	Sort order for failing rows by relative error.	'asc'
regex	bool	Treat the provided column spec as a regex pattern.	False

recx.CheckResult(failed_rows, check_name, total_rows, column=None, check_args=None, min_dots=5, disp_rows=20)

Result of an individual column (or index) check.

Parameters:

Name	Type	Description	Default
failed_rows	DataFrame	Subset of rows (and associated columns) that failed the check. A zero-length frame indicates success.	required
check_name	str	Name of the check (usually the class name of the checker).	required
total_rows	int	Total number of rows evaluated by the check (denominator for failure ratios).	required
column	str	Column name the check was applied to (None for index-wide checks).	None
check_args	dict	Mapping of argument names to values used to parameterise the check.	None
min_dots	int	Minimum number of dots when formatting one-line summaries.	5

outcome()

Return formatted outcome string.

"PASSED" if there are no failing rows, otherwise a summary of the form "[<count>/<total> (<pct>%)] FAILED".

one_liner(width=None)

Return the result formatted as a single line.

Parameters:

Name	Type	Description	Default
width	int	Target total width. If provided the line is padded with dots; if the width is insufficient a ValueError is raised.	None

Returns:

Type	Description
str	One-line summary "<signature> ... <outcome>".

Raises:

Type	Description
ValueError	If width is smaller than the minimum space required.

recx.RecResult(results, baseline, candidate)

Represents the result of a diff between two DataFrames.

Parameters:

Name	Type	Description	Default
results	list[CheckResult]	All individual check results (passing and failing) in execution order.	required

summary(log=False)

Print a summary.