Struct TabularData 

pub struct TabularData {
    pub headers: Vec<String>,
    pub rows: Vec<Vec<Value>>,
    pub has_header: bool,
    pub key: Vec<String>,
    pub column_types: Vec<Option<String>>,
    pub column_metadata: Vec<Value>,
    pub table_metadata: Value,
}

Format-neutral tabular data: an ordered list of records with a shared column schema. Produced by CSV, JSON record arrays, JSONL, Excel, Parquet, DB, and other tabular parsers; consumed by tabular writers, compaction rules, and extractors.

This is the codec type for the tabular_v1 artifact format. Serialize with serde_json::to_vec, deserialize with serde_json::from_slice.

The shape spectrum (rectangular?, named columns?, typed cells?) is derived from the data via TabularData::is_rectangular, TabularData::has_named_columns, and the cell Value variants — rules gate their behavior on those facts rather than on artifact subtypes.

Fields

headers: Vec<String>

Column names. For headerless sources these are synthesized positional labels and has_header is false.

rows: Vec<Vec<Value>>

has_header: bool

Whether the source supplied real column names (CSV header, object keys).

key: Vec<String>

Declared identity column names, in order. Empty when the source declares no key; drives keyed row alignment when present.

column_types: Vec<Option<String>>

Optional source-declared type per column (parallel to headers), when the source format carries one (DB, Parquet, Stata). Empty means “none”.

column_metadata: Vec<Value>

Optional per-column metadata bag (parallel to headers), when the source format carries column-scoped facts a generic tabular consumer would not otherwise see — labels, display formats, value-label set names, units. Each entry is an open object (or Null for a column with no metadata). Empty means “none”. This is tier 1 of the tiered-metadata design (see the tiered-artifact-metadata ADR): facts keyed to a column.

table_metadata: Value

Optional table-scoped metadata bag — facts about this table as a whole that are not per-column and not per-file (a single-table file folds its source-format facts here; a table inside a multi-table container carries only its own facts, e.g. dataset name/label). Null means “none”. This is tier 2 of the tiered-metadata design.

Implementations

impl TabularData

pub fn from_string_rows(headers: Vec<String>, rows: Vec<Vec<String>>) -> Self

Construct from all-string cells (CSV and other untyped sources). Cells are wrapped in Value::String; the result is byte-identical in behavior to the legacy all-string tabular model.

pub fn new(headers: Vec<String>, rows: Vec<Vec<Value>>) -> Self

Construct from typed rows with a named header.

pub fn with_column_metadata(self, column_metadata: Vec<Value>) -> Self

Attach tier-1 per-column metadata (parallel to headers). Builder-style so producers can enrich a table without restating every field.

pub fn with_table_metadata(self, table_metadata: Value) -> Self

Attach tier-2 table-scoped metadata.

pub fn column_index(&self, name: &str) -> Option<usize>

pub fn column_values(&self, name: &str) -> Option<Vec<&Value>>

pub fn is_rectangular(&self) -> bool

Every row has arity equal to the column count.

pub fn has_named_columns(&self) -> bool

The source supplied real, usable column names.

pub fn stable_columns(&self) -> bool

Columns can be identified across rows and snapshots — the precondition for cell-grain and column-grain edits. Otherwise the writer degrades to row-grain output.

pub fn to_csv(&self) -> String

Trait Implementations

impl Clone for TabularData

fn clone(&self) -> TabularData

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for TabularData

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for TabularData

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl PartialEq for TabularData

fn eq(&self, other: &TabularData) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl Serialize for TabularData

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.

impl Eq for TabularData

impl StructuralPartialEq for TabularData