Versioned recipes
Save strict JSON or YAML rules for explicit strings, numbers, dates, booleans, categories, mappings and fill values.
Released 19 October 2026
EazyDataFix v2.0.0Reusable, reviewable preparation.
Define explicit rules once, preview the transformation, approve the evidence and replay the same recipe on future datasets—without changing the familiar v1 convenience APIs.
terminal
bash
pip install --upgrade eazydatafix==2.0.0Review before apply
Preview selected stages, inspect evidence and apply only when the source, recipe and proposed output still match.
Bounded failure policies
Preserve, reject or quarantine invalid conversions with source values and row positions recorded as evidence.
One pipeline interface
Reuse the same recipe model across Analysis Ready, leakage-safe ML Ready and multi-table Power BI Ready workflows.
A reviewed pipeline in Python
Recipes are portable; review plans are deliberately process-local and bound to the exact source, recipe, selected operations and previewed output.
v2_pipeline.py
python
1import eazydatafix as edf23recipe = edf.PipelineRecipe(4 name="monthly_sales",5 columns={6 "customer_id": edf.ColumnRule(dtype="string"),7 "amount": edf.ColumnRule(dtype="number", currency_symbols=("₹",)),8 "order_date": edf.ColumnRule(dtype="datetime", date_format="%d/%m/%Y"),9 },10)1112pipeline = edf.Pipeline(recipe)13plan = pipeline.preview("sales.csv")14result = pipeline.apply("sales.csv", plan)1516if result.is_ready:17 result.save("sales_output")Compatibility and boundaries
Existing fix(), analysis_ready(), ml_ready(), powerbi_ready(), EDA and reporting APIs remain supported.
CSV and Excel loading protects textual identifiers before explicit type conversion.
Exports require a ready, applied and unchanged result; failed previews remain inspectable.
Streaming, SQL connectors, persistent cross-process approvals and browser Studio execution are not included yet.