#sql #macro

macro include-sql

A Yesql inspired macro for using SQL in Rust

9 releases

0.3.1 Mar 11, 2024
0.3.0 Feb 10, 2023
0.2.5 Apr 22, 2022
0.2.3 Feb 22, 2022
0.1.0 Mar 13, 2019

#2636 in Database interfaces

31 downloads per month
Used in 3 crates

MIT license

997 lines

crates.io Documentation MIT

include-sql is a macro for using SQL in Rust.

include-sql was inspired by Yesql. It allows the programmer to write SQL queries in SQL, keep them separate from the Rust code, and easily embed them into Rust programs via the proc-macro that this library provides.

All by itself include-sql actually does very little - it reads and parses SQL file and transforms it into a call to the impl_sql macro. It is expected that impl_sql is provided either by the project that uses include-sql or by an external library. For example, there are several include-sql companion crates, like include-postgres-sql, include-sqlite-sql, and include-oracle-sql, that implement impl_sql. They can simply be used directly if their approaches to embedding SQL are deemed appropriate and convenient. Alternatively, they can be used as a starting point when implementing your own impl_sql.


As include-sql is not intended to be used directly, to illustrate the workflow we'll use include-sqlite-sql.

Add include-sqlite-sql as a dependency:

include-sqlite-sql = "0.2"

Write your SQL and save it in a file. For example, let's say the following is saved as library.sql in the project's sql folder:

-- name: get_loaned_books?
-- Returns the list of books loaned to a patron
-- # Parameters
-- param: user_id: &str - user ID
SELECT book_title
  FROM library
 WHERE loaned_to = :user_id
-- name: loan_books!
-- Updates the book record to reflect loan to a patron
-- # Parameters
-- param: book_titles: &str - book titles
-- param: user_id: &str - user ID
UPDATE library
   SET loaned_to = :user_id
     , loaned_on = current_timestamp
 WHERE book_title IN (:book_titles)

Note that the parameter order is defined by the param declarations.

And then use it in Rust as:

use include_sqlite_sql::{include_sql, impl_sql};
use rusqlite::{Result, Connection};


fn main() -> Result<()> {
    let db = Connection::open("library.db")?;

    db.loan_books(&["Where the Sidewalk Ends", "A Wrinkle in Time", "Dune"], "Penny Teller")?;

    db.get_loaned_books("Leonard Hofstadter", |row| {
        let book_title : &str = row.get_ref("book_title")?.as_str()?;


Note that the path to the SQL file must be specified relative to the project root, i.e. relative to CARGO_MANIFEST_DIR, even if you keep your SQL file alongside rust module that includes it. Because include-sql targets stable Rust this requirement will persist until SourceFile stabilizes.

Under the Hood

After parsing and validating the content of the SQL file include-sql generates the following call:

impl_sql!{ LibrarySql =
    ? get_loaned_books (: user_id (&str))
    " Returns the list of books loaned to a patron\n # Parameters\n * `user_id` - user ID"
    $ "SELECT book_title\n  FROM library\n WHERE loaned_to = " :user_id "\n ORDER BY 1"
    ! loan_books (# book_titles (&str) : user_id (&str))
    " Updates the book records to reflect loan to a patron\n # Parameters\n * `user_id` - user ID\n * `book_titles` - book titles"
    $ "UPDATE library\n   SET loaned_to = " : user_id "\n,     loaned_on = current_timestamp\n WHERE book_title IN (" # book_titles ")"

Which include_sqlite_sql::impl_sql transforms into the following implementation:

trait LibrarySql {
    /// Returns the list of books loaned to a patron
    /// # Parameters
    /// * `user_id` - user ID
    fn get_loaned_books<F>(&self, user_id: &str, row_callback: F) -> rusqlite::Result<()>
    where F: FnMut(&rusqlite::Row) -> rusqlite::Result<()>;

    /// Updates the book records to reflect loan to a patron
    /// # Parameters
    /// * `book_titles` - book titles
    /// * `user_id` - user ID
    fn loan_books(&self, book_ids: &[&str], user_id: &str) -> rusqlite::Result<usize>;

And, of course, it also implements the trait:

impl LibrarySql for rusqlite::Connection {
    /// ...


The included documentation describes the supported SQL file format and provides instructions on writing your own impl_sql macro.

💥 Breaking Changes in Version 0.3

  • Order of the parameters for the generated method is defined by the order of the param descriptors. This is a potentially breaking change as previously parameters of the generated method followed the order of the parameters in the SQL statement. When SQL statement header does not use param descriptors, then the generated generic method parameters will be ordered according to their appearance in the SQL statement.
  • Statements are terminated with the slash / instead of the semicolon ;. This was implemented to allow declaration and use of batches of statements for SQLite and PL/SQL blocks for Oracle. Note that statement terminator is optional when the statement is the last statement in the file or when it is followed by another statement, which header will auto-terminate the preceding statement.


~89K SLoC