Expand description

Diesel

Diesel is an ORM and query builder designed to reduce the boilerplate for database interactions. If this is your first time reading this documentation, we recommend you start with the getting started guide. We also have many other long form guides.

Where to find things

Declaring your schema

For Diesel to validate your queries at compile time it requires you to specify your schema in your code, which you can do with the table! macro. diesel print-schema can be used to automatically generate these macro calls (by connecting to your database and querying its schema).

Getting started

Queries usually start from either a table, or a function like update. Those functions can be found here.

Diesel provides a prelude module, which exports most of the typically used traits and types. We are conservative about what goes in this module, and avoid anything which has a generic name. Files which use Diesel are expected to have use diesel::prelude::*;.

Constructing a query

The tools the query builder gives you can be put into these three categories:

  • “Query builder methods” are things that map to portions of a whole query (such as ORDER and WHERE). These methods usually have the same name as the SQL they map to, except for WHERE which is called filter in Diesel (To not conflict with the Rust keyword). These methods live in the query_dsl module.
  • “Expression methods” are things you would call on columns or other individual values. These methods live in the expression_methods module You can often find these by thinking “what would this be called” if it were a method and typing that into the search bar (e.g. LIKE is called like in Diesel). Most operators are named based on the Rust function which maps to that operator in std::ops (For example == is called .eq, and != is called .ne).
  • “Bare functions” are normal SQL functions such as sum. They live in the dsl module. Diesel only supports a very small number of these functions. You can declare additional functions you want to use with the sql_function! macro.

Serializing and Deserializing

Types which represent the result of a SQL query implement a trait called Queryable.

Diesel maps “Rust types” (e.g. i32) to and from “SQL types” (e.g. diesel::sql_types::Integer). You can find all the types supported by Diesel in the sql_types module. These types are only used to represent a SQL type. You should never put them on your Queryable structs.

To find all the Rust types which can be used with a given SQL type, see the documentation for that SQL type.

To find all the SQL types which can be used with a Rust type, go to the docs for either ToSql or FromSql, go to the “Implementors” section, and find the Rust type you want to use.

Getting help

If you run into problems, Diesel has an active community. Either open a new discussion thread at diesel github repository or use the active Gitter room at gitter.im/diesel-rs/diesel

Crate feature flags

The following feature flags are considered to be part of diesels public API. Any feature flag that is not listed here is not considered to be part of the public API and can disappear at any point in time:

  • sqlite: This feature enables the diesel sqlite backend. Enabling this feature requires a compatible copy of libsqlite3 for your target architecture.
  • postgres: This feature enables the diesel postgres backend. Enabling this feature requires a compatible copy of libpq for your target architecture. This features implies postgres_backend
  • mysql: This feature enables the idesel mysql backend. Enabling this feature requires a compatible copy of libmysqlclient for your target architecture. This feature implies mysql_backend
  • postgres_backend: This feature enables those parts of diesels postgres backend, that are not dependend on libpq. Diesel does not provide any connection implementation with only this feature enabled. This feature can be used to implement a custom implementation of diesels Connection trait for the postgres backend outside of diesel itself, while reusing the existing query dsl extensions for the postgres backend
  • mysql_backend: This feature enables those parts of diesels mysql backend, that are not dependend on libmysqlclient. Diesel does not provide any connection implementation with only this feature enabled. This feature can be used to implement a custom implementation of diesels Connection trait for the mysql backend outside of diesel itself, while reusing the existing query dsl extensions for the mysql backend
  • returning_clauses_for_sqlite_3_35: This feature enables support for RETURNING clauses in the sqlite backend. Enabling this feature requires sqlite 3.35.0 or newer.
  • 32-column-tables: This feature enables support for tables with up to 32 columns. This feature is enabled by default. Consider disabling this feature if you write a library crate providing general extensions for diesel or if you do not need to support tables with more than 16 columns and you want to minimize your compile times.
  • 64-column-tables: This feature enables support for tables with up to 64 columns. It implies the 32-column-tables feature. Enabling this feature will increase your compile times.
  • 128-column-tables: This feature enables support for tables with up to 128 columns. It implies the 64-column-tables feature. Enabling this feature will increase your compile times significantly.
  • i-implement-a-third-party-backend-and-opt-into-breaking-changes: This feature opens up some otherwise private API, that can be useful to implement a third party Backend or write a custom Connection implementation. Do not use this feature for any other usecase. By enabling this feature you explicitly opt out diesel stability guarantees. We explicitly reserve us the right to break API’s exported under this feature flag in any upcoming minor version release. If you publish a crate depending on this feature flag consider to restrict the supported diesel version to the currently released minor version.
  • serde_json: This feature flag enables support for (de)serializing json values from the database using types provided by serde_json.
  • chrono: This feature flags enables support for (de)serializing date/time values from the database using types provided by chrono
  • uuid: This feature flag enables support for (de)serializing uuid values from the database using types provided by uuid
  • network-address: This feature flag enables support for (de)serializing IP values from the database using types provided by ipnetwork.
  • ipnet-address: This feature flag enables support for (de)serializing IP values from the database using types provided by ipnet.
  • numeric: This feature flag enables support for (de)serializing numeric values from the database using types provided by bigdecimal
  • r2d2: This feature flag enables support for the r2d2 connection pool implementation.
  • extras: This feature enables the feature flaged support for any third party crate. This implies the following feature flags: serde_json, chrono, uuid, network-address, numeric, r2d2
  • with-deprecated: This feature enables items marked as #[deprecated]. It is enabled by default. disabling this feature explicitly opts out diesels stability guarantee.
  • without-deprecated: This feature disables any item marked as #[deprecated]. Enabling this feature explicitly opts out the stability guarantee given by diesel. This feature overrides the with-deprecated. Note that this may also remove items that are not shown as #[deprecated] in our documentation, due to various bugs in rustdoc. It can be used to check if you depend on any such hidden #[deprecated] item.
  • nightly-error-messages: This feature enables the generation of improved compiler error messages for common mistakes using diesel. This feature requires a nightly rust compiler and is considered to be unstable. We might remove it in future diesel versions without replacement or deprecation.

By default the following features are enabled:

  • with-deprecated
  • 32-column-tables

Re-exports

pub use crate::result::Error::NotFound;
pub use crate::prelude::*;

Modules

Traits related to relationships between multiple tables.
Types which represent various database backends
Types related to database connections
Structs to represent the primitive equivalent of SQL types where there is no existing Rust primitive, or where using it would be confusing (such as date and time types). This module will re-export all backend specific data structures when compiled against that backend.
Types and traits related to deserializing values from the database
Includes various helper types and bare functions which are named too generically to be included in prelude, but are often used when using Diesel.
AST types representing various typed SQL expressions.
Adds various methods to construct new expressions. These traits are exported by default, and implemented automatically.
Provide helper types for concisely writing the return type of functions. As with iterators, it is unfortunately difficult to return a partially constructed query without exposing the exact implementation of the function. Without higher kinded types, these various DSLs can’t be combined into a single trait for boxing purposes.
Representation of migrations
Re-exports important traits and types. Meant to be glob imported when using Diesel.
Contains traits responsible for the actual construction of SQL statements
Traits that construct SELECT statements
Types related to describing schema, and interactions between tables.
Connection pooling via r2d2.
Errors, type aliases, and functions related to working with Result.
Contains the Row trait
Types and traits related to serializing values for the database
Types which represent a SQL data type.
Types and functions related to PG’s and Sqlite’s ON CONFLICT clause

Macros

Declare a new alias for a table
Allow two or more columns which are otherwise unrelated to be used together in a group by clause.
Allow two or more tables which are otherwise unrelated to be used together in a query.
Useful for libraries adding support for new SQL types. Apps should never need to call this.
Allow two tables to be referenced in a join query without providing an explicit ON clause.
Declare a 0 argument SQL function for use in your code. This will generate a unit struct, which is an expression representing calling this function. See now for example output. now was generated using:
Indicates that an expression allows all numeric operators. If you create new SQL functions that return a numeric type, you should invoke this macro that type. Unfortunately, Rust disallows us from automatically implementing Add for types which implement Expression, under its orphan rules.
Implements the Rust operator for a given type. If you create a new SQL function, which returns a type that you’d like to use an operator on, you should invoke this macro. Unfortunately, Rust disallows us from automatically implementing Add and other traits from std::ops, under its orphan rules.
Useful for libraries adding support for new SQL types. Apps should never need to call this.
Useful for libraries adding support for new SQL types. Apps should never need to call this.
Specifies that a table exists, and what columns it has. This will create a new public module, with the same name, as the name of the table. In this module, you’ll find a unit struct named table, and a unit struct with the names of each of the columns.

Functions

Takes a query QueryFragment expression as an argument and returns a type that implements fmt::Display and fmt::Debug to show the query.
Creates a DELETE statement.
Creates an INSERT statement for the target table.
Creates an INSERT [OR] IGNORE statement.
Creates a REPLACE statement.
Creates a bare select statement, with no from clause. Primarily used for testing diesel itself, but likely useful for third party crates as well. The given expressions must be selectable from anywhere.
Construct a full SQL query using raw SQL.
Creates an UPDATE statement.