Struct SeedConfig 

pub struct SeedConfig<'a, D, S> { /* private fields */ }

Builder that generates seed INSERT statements for a schema.

Create one with SeedConfig::sqlite, SeedConfig::postgres, or SeedConfig::mysql (each behind its feature), chain the settings, then call generate / try_generate. Tables and columns passed to the builder must belong to the schema; anything else is a compile error.

Defaults: seed 0, 10 rows per table, no skipped tables. See the crate docs for how values and row counts are chosen.

Implementations

impl<'a, D, S> SeedConfig<'a, D, S>

pub const fn seed(self, seed: u64) -> Self

Sets the RNG seed (default 0). The same seed gives the same rows.

pub const fn default_count(self, count: usize) -> Self

Sets the row count for tables with no explicit count and no seeded parent (default 10).

pub fn max_params(self, limit: usize) -> Self

Caps the bind parameters per INSERT; rows are split across more statements to stay under it.

Without this, a dialect-specific default is used.

Panics

Panics if limit is zero.

impl<D, S> SeedConfig<'_, D, S>

where S: SQLSchemaImpl,

pub fn count<T>(self, table: &T, count: usize) -> Self

where T: SQLTableInfo, S: SchemaHasTable<T>,

Sets the row count for table.

This wins over relation and default_count.

pub fn relation<P, C>(self, parent: &P, child: &C, count: usize) -> Self

where P: SQLTableInfo, C: SQLTableInfo + Relation<P>, S: SchemaHasTable<P> + SchemaHasTable<C>,

Generates count rows of child for each row of parent; each parent row gets count consecutive child rows.

Without this, a child table with no explicit count gets one row per parent row. child must have a foreign key to parent (checked at compile time).

impl<D, S> SeedConfig<'_, D, S>

pub fn skip<T>(self, table: &T) -> Self

where T: SQLTableInfo, S: SchemaHasTable<T>,

Leaves table out: no INSERTs and no reset statements for it.

Foreign keys that point at a skipped table keep their generated (or custom) values, for example to reference rows already in the database.

impl<D, S> SeedConfig<'_, D, S>

where S: SQLSchemaImpl,

Name-based settings, for a Schema built at runtime or names that come from configuration.

They take the same effect as the typed methods. A table name may be qualified as schema.table, and must be when two namespaces have a table with that name. An unknown name is reported as a SeedError by try_generate, try_generate_rows and reset_plan (and makes generate panic).

pub fn count_by_name(self, table: &str, count: usize) -> Self

Sets the row count for the table named table, like count.

pub fn relation_by_name(self, parent: &str, child: &str, count: usize) -> Self

Generates count rows of child for each row of parent, like relation. child must have a foreign key to parent.

pub fn skip_by_name(self, table: &str) -> Self

Leaves the table named table out, like skip.

pub fn kind_by_name( self, table: &str, column: &str, kind: GeneratorKind, ) -> Self

Uses a built-in GeneratorKind for column of table, like kind.

pub fn generator_by_name( self, table: &str, column: &str, generator: impl Generator + 'static, ) -> Self

Uses generator for column of table, like generator.