Module dataframe_derive 

Derive macros for bidirectional row ↔ dataframe conversions.

Supports both structs (direct field mapping) and enums (field-name union across variants with Option<T> fill for missing fields).

Modules

enum_expansion 🔒

Enum-specific DataFrame derive expansion.

Structs

AutoExpandVecData 🔒

Data for ResolvedField::AutoExpandVec.

ColumnRegistry 🔒

Accumulates unique columns for an enum-to-dataframe unified schema.

DataFrameAttrs 🔒

Parsed container-level #[dataframe(...)] attributes.

EnumAutoExpandVecData 🔒

Data for EnumResolvedField::AutoExpandVec.

EnumExpandedFixedData 🔒

Data for EnumResolvedField::ExpandedFixed.

EnumExpandedVecData 🔒

Data for EnumResolvedField::ExpandedVec.

EnumMapFieldData 🔒

Data for EnumResolvedField::Map.

EnumSingleFieldData 🔒

Data for EnumResolvedField::Single.

EnumStructFieldData 🔒

Data for EnumResolvedField::Struct.

ExpandedFixedData 🔒

Data for ResolvedField::ExpandedFixed.

ExpandedVecData 🔒

Data for ResolvedField::ExpandedVec.

FieldAttrs 🔒

Parsed field-level #[dataframe(...)] attributes.

MapFieldData 🔒

Data for ResolvedField::Map (struct path, non-String-keyed maps, #919).

ResolvedColumn 🔒

A resolved column in the unified schema across all enum variants.

SingleFieldData 🔒

Data for ResolvedField::Single.

StructFieldData 🔒

Data for ResolvedField::Struct.

VariantInfo 🔒

Parsed and resolved information about a single enum variant for DataFrame codegen.

Enums

EnumResolvedField 🔒

A resolved enum field ready for codegen – either a single column or expanded from an array/Vec into multiple suffixed columns.

FieldTypeKind 🔒

Classification of a field type for DataFrame column expansion.

ResolvedField 🔒

A resolved struct field ready for codegen – determines how this field maps to DataFrame companion struct columns.

VariantShape 🔒

Describes the shape of an enum variant’s fields.

Constants

OPTION_SCALAR_NAMES 🔒

Scalar types accepted inside Option<…> as a struct-path field (#1437): every T here has IntoR for Vec<Option<T>> (typed NA for None) in miniextendr-api::into_r. The reader side is gated separately by is_reader_scalar_ty, so Option<u64> / Option<usize> fields are write-only, like their bare counterparts.

READER_SCALAR_NAMES 🔒

Scalar element types whose Vec<T> (and Vec<Option<T>>) round-trips through TryFromSexp — the supported field types for the parallel from-R reader (try_from_dataframe_par). This is intentionally narrower than FieldTypeKind::Scalar: set/opaque collection types (HashSet<…>, BTreeSet<…>) also classify as Scalar but do NOT implement Vec<_>: TryFromSexp, so they must be excluded from the reader path.

Functions

classify_field_type 🔒

Classify a field type for DataFrame column expansion.

column_sexp 🔒

Convert a companion column without adding R objects to the companion: as_list uses the rooted list builder, as_factor uses UnitEnumFactor, and ordinary columns convert the whole Vec through IntoR.

companion_debug_clone 🔒

Debug and Clone for a companion, each implemented only when every field type has it. #[derive(Debug, Clone)] would demand both of every column’s element type (#2013). fields are the companion’s fields with their types.

companion_lifetime_marker 🔒

Keep lifetimes erased by column expansion or skipped fields in the companion. Columns that still borrow already carry their lifetimes; only missing ones need a marker. This adds no data column or element-type trait requirements.

conditional_where 🔒

where clause: where_clause’s predicates, then for<'__mx_cond> ty: bound for each (ty, bound) pair.

derive_dataframe_row

Derive DataFrameRow: generates a companion DataFrame type with collection fields.

derive_struct_dataframe 🔒

Generate DataFrameRow expansion for struct types.

expanded_elem_idents 🔒

The variables fixed_array_elems / width_elems bind for the n columns of the field key names. Leading underscores are dropped from key so the variables stay snake case (_0 gives __mxe1_0).

field_is_borrowed_ref 🔒

True if ty is a borrowed reference (&[T], &[T; N], &str, …). Such expansion fields can’t be reconstructed by value in the R→Rust reader.

field_reader_capable 🔒

True if a resolved struct field can be read back out of an R data.frame.

fixed_array_elems 🔒

Bind each element of fixed-array field value access to its own variable in elems: moved out of an owned array (indexing would need Copy), cloned out of a borrowed one, which is the only way to own an element behind a reference.

from_rows_par_override 🔒

The from_rows_par override of a companion’s ColumnarFrame impl: the companion’s ParallelColumns impl where it applies, from_rows otherwise (miniextendr_api::dataframe::FromRowsParProbe).

generic_type_arg_count 🔒

Number of type arguments on ty’s last path segment (0 when not angle-bracketed). Rejects HashMap<K, V, S> custom-hasher maps from the reader path — the Vec<HashMap<String, V>>: TryFromSexp impl only covers the default hasher.

is_bare_reader_scalar_ty 🔒

True if ty names a known scalar, qualified or bare (no Option or generics).

is_known_scalar_ty 🔒

Match the final scalar name, as field classification does. Absolute and relative paths have the same conversion support; associated types and generic payloads still need their own classification.

is_option_scalar_inner 🔒

True if ty names a type from OPTION_SCALAR_NAMES (possibly qualified, without generic args), a valid payload for a single NA-able column.

is_raw_scalar_ty 🔒

True if ty is u8, bare or qualified. R stores it in a raw vector, which has no NA, so a u8 column cannot pad or omit cells (Vec<Option<u8>> has no IntoR).

is_reader_scalar_ty 🔒

True if ty is a known scalar path, or Option<known-scalar>.

is_string_ty 🔒

True if ty’s last path segment is the bare ident String (path-identity check, same convention as classify_field_type).

list_column_fn 🔒

into_list_column::<#ty> (or into_option_list_column::<#ty>, for a column with absent rows), spanned at the field type: a type without IntoList is then reported at the field instead of at the derive.

option_payload 🔒

The payload of Option<T> (possibly qualified), if ty is one.

owned_elems_iter 🔒

The elements of an expanded Vec<T> / Box<[T]> / &[T] field value access as an iterator of owned T: moved out of an owned container, cloned out of a borrowed slice. IntoIterator::into_iter is called by path so a Box<[T]> iterates by value in every edition.

parallel_columns_impl 🔒

The parallel column fill of a companion: ParallelColumns with body as from_rows_parallel(rows), implemented only where bounds hold (see miniextendr_api::dataframe::ParallelColumns). Emitted only when miniextendr-api has rayon; pair it with from_rows_par_override.

parse_dataframe_attrs 🔒

Parse container-level #[dataframe(...)] attributes from the derive input.

parse_field_attrs 🔒

Parse field-level #[dataframe(...)] attributes from a syn::Field.

raw_na_column_error 🔒

A u8 in a column whose cells can be NA. why names the source of the NA cells and fix the alternative, so the error points at the field instead of the missing IntoR impl inside the generated code (#2025).

resolve_struct_field 🔒

Resolve a struct field into a ResolvedField, applying field attributes.

spanned_clone_fn 🔒

<ty as Clone>::clone, spanned at ty: a type without Clone is then reported at the user’s field instead of at the derive.

validate_factor_field_type 🔒

The same as_factor shapes are accepted on struct and enum fields. The generated conversion checks UnitEnumFactor on the selected type.

width_elems 🔒

Bind the first elems.len() elements of a width = N field value access to Option<T> variables (None past its end; elements past N are dropped), through iterator variable iter.