Crate neutrino_schema 

Multi-database schema-to-Rust compiler pipeline.

Introspects a live PostgreSQL, MySQL/MariaDB, or SQLite database, normalises column types into a database-agnostic Intermediate Representation (SchemaIR), and generates strongly typed Rust model structs.

Pipeline

1. Introspect — reads schema metadata via DatabaseIntrospector
2. Normalise — raw column types map to DbType; fields, tables, constraints, enums, and indexes collect into SchemaIR
3. Generate — SchemaIR is consumed by generate() which returns a GeneratedOutput containing file contents; the CLI writes them via OutputWriter

Quick start (programmatic)

use neutrino_schema::*;

let tables = vec![
    TableIR {
        name: "users".into(),
        fields: vec![FieldIR {
            name: "email".into(),
            ty: DbType::String,
            nullable: false,
            raw_type: "Varchar".into(),
            default_value: None,
            generated: false,
            comment: None,
        }],
        constraints: vec![],
        comment: None,
        indexes: vec![],
    },
];

let schema = SchemaIR::from_tables(tables, RelationStrategy::NamingHeuristic);

let output = generate(&schema, &GenerateOptions::default());
let generated = output.file("users.rs").map(|f| &f.content[..]).unwrap_or("");
assert!(generated.contains("pub struct Users"));

Quick start (CLI)

# First run — creates neutrino-schema.toml
neutrino-schema generate
# → prompts for database URL, saves it, generates types

# Subsequent runs — just works from config
neutrino-schema generate

# Explicit setup (CI / scripting)
neutrino-schema init --database-url "postgres://localhost/mydb"
neutrino-schema generate

# All flags still work
neutrino-schema generate --database-url "mysql://user:pass@localhost/mydb" --output src/entities

Re-exports

pub use codegen::generate;

pub use codegen::generate_enum_defs;

pub use codegen::generate_files;

pub use codegen::generate_files_with_registry;

pub use codegen::generate_imports;

pub use codegen::generate_struct;

pub use codegen::GenerateOptions;

pub use codegen::GeneratedFile;

pub use codegen::GeneratedOutput;

pub use codegen::OutputWriter;

pub use codegen::RenderMode;

pub use codegen::RustGeneratorConfig;

pub use ir::ConstraintIR;

pub use ir::ConstraintKind;

pub use ir::EnumIR;

pub use ir::EnumVariantIR;

pub use ir::FieldIR;

pub use ir::IndexEntryIR;

pub use ir::IndexIR;

pub use ir::IndexKind;

pub use ir::MatchType;

pub use ir::ReferentialAction;

pub use ir::RelationCardinality;

pub use ir::RelationIR;

pub use ir::RelationInferenceStrategy;

pub use ir::RelationOrigin;

pub use ir::RelationStrategy;

pub use ir::SchemaIR;

pub use ir::SemanticRelationIR;

pub use ir::TableIR;

pub use validation::validate;

pub use validation::ValidationEntry;

pub use validation::ValidationLevel;

pub use validation::ValidationReport;

pub use types::dbtype_to_rust;

pub use types::mysql_to_db_type;

pub use types::sqlite_declared_to_db_type;

pub use types::sqlite_to_db_type;

pub use types::to_db_type;

pub use types::DbType;

pub use types::EnumRef;

pub use types::MysqlType;

pub use types::PgType;

pub use types::RustType;

pub use types::SqliteType;

pub use types::TypeRegistry;

pub use util::naming::deduplicate_identifier;

pub use util::naming::enum_variant_name;

pub use util::naming::sanitize_identifier;

pub use util::naming::to_struct_name;

pub use introspect::Column;

pub use introspect::DatabaseIntrospector;

pub use introspect::PostgresIntrospector;

pub use introspect::SqliteIntrospector;

pub use introspect::MysqlIntrospector;

Modules

cli

CLI entry points for the neutrino-schema binary.

codegen

Rust source code generation from SchemaIR.

config

Configuration for the code generation pipeline.

inference

Relation inference engine.

introspect

Database introspection — reads live schema metadata.

ir

Intermediate Representation (IR) — schema-agnostic model for code generation.

types

Database type representation and normalisation.

util

validation

Semantic validation for SchemaIR documents.