Semogram Docs
PluginsTransforms reference

Transforms reference

Parameters, bindings and lineage for all bundled transformations.

All 20 capabilities belong to @craven/transforms. Install the bundle once, then choose the required capability in a pipeline.

CapabilityPurposeLineage
select-projectKeep, reorder, and optionally rename fields from upstream data.row_preserving
filterKeep only rows that satisfy all configured deterministic predicates.explicit
lookupEnrich the primary input with fields from a reference input using key-based matching.explicit
drop-duplicatesRemove duplicate rows using selected key fields.explicit
rename-fieldsRename selected fields without changing row counts.row_preserving
cast-fieldsConvert fields to deterministic scalar target types.row_preserving
windowAdd partitioned row-wise analytics such as row numbers and running totals.row_preserving
unionAppend rows from two compatible inputs into one stream.explicit
flattenPromote nested object fields into top-level columns using a delimiter.row_preserving
fill-missing-valuesFill null or empty values with deterministic defaults.row_preserving
standardize-valuesNormalize values into a canonical deterministic representation.row_preserving
joinJoin two inputs on matching keys and append mapped fields from the right input.explicit
aggregateAggregate upstream records into grouped outputs.explicit
pivotTurn row values into columns within grouped outputs.explicit
unpivotTurn selected columns into key/value rows.explicit
classificationClassify rows into a bounded set of business labels using model reasoning.row_preserving
extract-structured-dataExtract schema-bounded structured fields from unstructured text columns.row_preserving
semantic-mapMap messy source records into a target semantic schema using bounded model output.row_preserving
data-quality-repairRepair invalid or inconsistent values inside declared field boundaries.row_preserving
entity-resolutionUse AI to propose provisional entity clusters; reconcile source identities through the persistent registry before treating them as canonical.row_preserving