// SQLiteSet
// This file is part of the 'SQLite-based containers for Rust' project (sqlite-containers)
// SPDX-License-Identifier: Unlicense
use crate::{
CharType, Error, SQLiteMap, Size, SizeExt, TryClone, Type,
common::{OPEN_FLAGS, OPEN_FLAGS_NOCREATE, OPEN_FLAGS_RDONLY, SQL_TYPE_TEXT, SQL_TYPES, SqlType, check_constraint_violation, verify_schema},
rusqlite::{Connection, ToSql, Transaction},
};
use cfg_if::cfg_if;
use const_format::concatcp;
use std::{borrow::Borrow, collections::BTreeMap, marker::PhantomData, path::Path, sync::LazyLock};
// ---------------------------------------------------------------------------
// Statements
// ---------------------------------------------------------------------------
const TABLE_NAME: &str = "rusqlite_set_v2";
static SQL_CREATE_TBL: LazyLock<BTreeMap<SqlType, String>> = LazyLock::new(|| {
let mut map = BTreeMap::<SqlType, String>::new();
for sql_type in SQL_TYPES {
cfg_if! {
if #[cfg(feature = "with_rowid")] {
map.insert(sql_type, format!("CREATE TABLE IF NOT EXISTS {TABLE_NAME} (key {sql_type} PRIMARY KEY NOT NULL);"));
} else {
map.insert(sql_type, format!("CREATE TABLE IF NOT EXISTS {TABLE_NAME} (key {sql_type} PRIMARY KEY NOT NULL) WITHOUT ROWID;"));
}
}
}
map
});
cfg_if! {
if #[cfg(feature = "with_rowid")] {
const SQL_CREATE_NOC: &str = concatcp!("CREATE TABLE IF NOT EXISTS ", TABLE_NAME, " (key TEXT PRIMARY KEY NOT NULL COLLATE NOCASE);");
} else {
const SQL_CREATE_NOC: &str = concatcp!("CREATE TABLE IF NOT EXISTS ", TABLE_NAME, " (key TEXT PRIMARY KEY NOT NULL COLLATE NOCASE) WITHOUT ROWID;");
}
}
const SQL_GET_SCHEMA: &str = concatcp!("SELECT sql FROM sqlite_schema WHERE type = 'table' AND name = '", TABLE_NAME, "';");
const SQL_COUNT_KEYS: &str = concatcp!("SELECT COUNT(*) FROM ", TABLE_NAME, ";");
const SQL_EXIST_KEYS: &str = concatcp!("SELECT EXISTS (SELECT 1 FROM ", TABLE_NAME, ");");
const SQL_INSERT_KEY: &str = concatcp!("INSERT INTO ", TABLE_NAME, " (key) VALUES (?1);");
const SQL_EXISTS_KEY: &str = concatcp!("SELECT 1 FROM ", TABLE_NAME, " WHERE key = ?1 LIMIT 1;");
const SQL_EXISTS_NOC: &str = concatcp!("SELECT 1 FROM ", TABLE_NAME, " WHERE key COLLATE NOCASE = ? LIMIT 1;");
const SQL_QUERY_KEYS: &str = concatcp!("SELECT key FROM ", TABLE_NAME, ";");
const SQL_GLOBA_KEYS: &str = concatcp!("SELECT key FROM ", TABLE_NAME, " WHERE key GLOB ?1;");
const SQL_GLOBN_KEYS: &str = concatcp!("SELECT key FROM ", TABLE_NAME, " WHERE key NOT GLOB ?1;");
const SQL_DELETE_KEY: &str = concatcp!("DELETE FROM ", TABLE_NAME, " WHERE key = ?1;");
const SQL_DELETE_NOC: &str = concatcp!("DELETE FROM ", TABLE_NAME, " WHERE key COLLATE NOCASE = ?1;");
const SQL_DELETE_ALL: &str = concatcp!("DELETE FROM ", TABLE_NAME, ";");
// ---------------------------------------------------------------------------
// SQLiteSet
// ---------------------------------------------------------------------------
/// A [set](https://en.wikipedia.org/wiki/Set_(abstract_data_type)) that stores keys of type [`T`](crate::Type), backed by an SQLite in-memory database.
///
/// It serves as a drop-in replacement for [`HashSet`](std::collections::HashSet).
///
/// By default, `SQLiteSet<String>` treats its keys as case-sensitive, but a case-insensitive variant is available. Even when using the case-sensitive set variant, for some operations a dedicated "case-insensitive" version is provided.
///
///
///
/// <div class="warning">
///
/// For bulk inserts or updates, it is **strongly recommended** to use an explicit [transaction](Self::transaction). Without one, SQLite executes each insert or update in its own transaction, which can significantly degrade performance.
///
/// </div>
pub struct SQLiteSet<T>
where
T: Type,
{
_type: PhantomData<T>,
connection: Connection,
}
impl<T> SQLiteSet<T>
where
T: Type,
{
/// Creates a new, empty hash set with case-sensitive keys, backed by an *in-memory* SQLite database.
#[inline]
pub fn new() -> Result<Self, Error> {
Self::from_connection(Self::create_connection::<false, &Path>(None, false, false)?)
}
/// Creates a new, empty hash set with case-insensitive keys, backed by an *in-memory* SQLite database.
///
/// Only ASCII characters are case folded. SQLite does **not** attempt to do full UTF case folding!
#[inline]
pub fn with_nocase() -> Result<Self, Error>
where
T: CharType,
{
Self::from_connection(Self::create_connection::<true, &Path>(None, false, false)?)
}
/// Creates a new or re-opens an existing hash set with case-sensitive keys, backed by a *file-based* SQLite database.
///
/// If `readonly` is `true`, the database file is opened in “read-only” mode.
///
/// If `can_create` is `true` (and if `readonly` is `false`), a new empty database file is created, if the file does *not* exist yet.
///
/// Returns [`Error::InvalidSchema`], if the database file already exists but its schema does **not** match the schema expected by this type of set. This means that you **must** re-open a database file with the *same* type of set that was used to create it.
#[inline]
pub fn from_file<P>(path: P, readonly: bool, can_create: bool) -> Result<Self, Error>
where
P: AsRef<Path>,
{
Self::from_connection(Self::create_connection::<false, _>(Some(path), readonly, can_create)?)
}
/// Creates a new or re-opens an existing hash set with case-insensitive keys, backed by a *file-based* SQLite database.
///
/// If `readonly` is `true`, the database file is opened in “read-only” mode.
///
/// If `can_create` is `true` (and if `readonly` is `false`), a new empty database file is created, if the file does *not* exist yet.
///
/// Returns [`Error::InvalidSchema`], if the database file already exists but its schema does **not** match the schema expected by this type of set. This means that you **must** re-open a database file with the *same* type of set that was used to create it.
///
/// Only ASCII characters are case folded. SQLite does **not** attempt to do full UTF case folding!
#[inline]
pub fn from_file_nocase<P>(path: P, readonly: bool, can_create: bool) -> Result<Self, Error>
where
T: CharType,
P: AsRef<Path>,
{
Self::from_connection(Self::create_connection::<true, _>(Some(path), readonly, can_create)?)
}
#[inline(always)]
fn create_connection<const NO_CASE: bool, P>(path_option: Option<P>, readonly: bool, can_create: bool) -> Result<Connection, Error>
where
P: AsRef<Path>,
{
let connection = if let Some(path) = path_option.as_ref() {
Connection::open_with_flags(path, if !readonly { if can_create { *OPEN_FLAGS } else { *OPEN_FLAGS_NOCREATE } } else { *OPEN_FLAGS_RDONLY })?
} else {
Connection::open_in_memory()?
};
let connection = Self::initialize_connection(connection)?;
#[cfg(feature = "checked")]
if path_option.is_some() {
crate::common::integrity_check(&connection)?;
}
let sql_create_table = if !NO_CASE {
SQL_CREATE_TBL.get(&T::SQL_TYPE).unwrap()
} else {
debug_assert_eq!(T::SQL_TYPE, SQL_TYPE_TEXT);
SQL_CREATE_NOC
};
connection.execute(sql_create_table, [])?;
if path_option.is_some() {
let sql_schema_table: String = connection.query_one(SQL_GET_SCHEMA, [], |row| row.get(0))?;
if !verify_schema(sql_create_table, &sql_schema_table) {
return Err(Error::InvalidSchema);
}
}
Ok(connection)
}
#[inline(always)]
fn initialize_connection(connection: Connection) -> Result<Connection, Error> {
connection.pragma_update(None, "synchronous", "OFF")?;
connection.pragma_update(None, "journal_mode", "OFF")?;
Ok(connection)
}
#[inline(always)]
fn from_connection(connection: Connection) -> Result<Self, Error> {
Ok(Self { _type: PhantomData, connection })
}
/// Starts a new SQLite transaction for this set.
///
/// Please note that using an explicit SQLite transaction allows for much more efficient bulk inserts 🚀
///
/// Returns the new [`SQLiteSetTransaction`] instance.
#[inline]
pub fn transaction(&mut self) -> Result<SQLiteSetTransaction<'_, T>, Error> {
SQLiteSetTransaction::from(&mut self.connection)
}
/// Inserts the given key into the set.
///
/// Returns `true`, if the set did not already contain the key; otherwise returns `false`.
#[inline]
pub fn insert<Q>(&mut self, key: &Q) -> Result<bool, Error>
where
T: Borrow<Q>,
Q: ?Sized + ToSql,
{
let mut statement = self.connection.prepare_cached(SQL_INSERT_KEY)?;
match statement.execute([key]) {
Ok(_) => Ok(true),
Err(error) => check_constraint_violation(error),
}
}
/// Inserts all keys from another set into this set, ignoring existing keys.
///
/// Returns the number of new keys that have been inserted.
#[inline]
pub fn union(&mut self, other: &Self) -> Result<Size, Error> {
let trx = self.transaction()?;
let mut statement_insert = trx.transaction.prepare_cached(SQL_INSERT_KEY)?;
let mut statement_query = other.connection.prepare_cached(SQL_QUERY_KEYS)?;
let mut counter = Size::default();
for key in statement_query.query_map([], |row| row.get::<_, T>(0))? {
if match statement_insert.execute([key?]) {
Ok(_) => true,
Err(error) => check_constraint_violation(error)?,
} {
counter += 1;
}
}
Ok(counter)
}
/// Checks whether the set contains the specified key.
///
/// For case-sensitive sets, the check is case-sensitive; for case-insensitive sets, the check is case-insensitive.
///
/// Returns `true`, if the set contains the key; otherwise returns `false`.
#[inline]
pub fn contains<Q>(&self, key: &Q) -> Result<bool, Error>
where
T: Borrow<Q>,
Q: ?Sized + ToSql,
{
let mut statement = self.connection.prepare_cached(SQL_EXISTS_KEY)?;
Ok(statement.exists([key])?)
}
/// This is the "case-insensitive" version of the [`contains()`](Self::contains) function.
///
/// The check is *always* performed case-insensitive.
///
/// Returns `true`, if the set contains the key; otherwise returns `false`.
#[inline]
pub fn contains_nocase<Q>(&self, key: &Q) -> Result<bool, Error>
where
T: Borrow<Q> + CharType,
Q: ?Sized + ToSql,
{
let mut statement = self.connection.prepare_cached(SQL_EXISTS_NOC)?;
Ok(statement.exists([key])?)
}
/// Removes the specified key from the set, if present.
///
/// For case-sensitive sets, the key is treated as case-sensitive; for case-insensitive sets, it is treated as case-insensitive.
///
/// Returns `true`, if the set contained the key; otherwise returns `false`.
#[inline]
pub fn remove<Q>(&mut self, key: &Q) -> Result<bool, Error>
where
T: Borrow<Q>,
Q: ?Sized + ToSql,
{
let mut statement = self.connection.prepare_cached(SQL_DELETE_KEY)?;
Ok(statement.execute([key])? != 0)
}
/// This is the "case-insensitive" version of the [`remove()`](Self::remove) function.
///
/// The key is *always* treated as case-insensitive.
///
/// Returns `true`, if the set contained the key; otherwise returns `false`.
#[inline]
pub fn remove_nocase<Q>(&mut self, key: &Q) -> Result<bool, Error>
where
T: Borrow<Q> + CharType,
Q: ?Sized + ToSql,
{
let mut statement = self.connection.prepare_cached(SQL_DELETE_NOC)?;
Ok(statement.execute([key])? != 0)
}
/// Invokes the given `callback` function for each key that is currently contained in the set.
///
/// This function does **not** guarantee a specific iteration order.
#[inline]
pub fn for_each<F>(&self, mut callback: F) -> Result<(), Error>
where
F: FnMut(T),
{
let mut statement = self.connection.prepare_cached(SQL_QUERY_KEYS)?;
for key in statement.query_map([], |row| row.get(0))? {
callback(key?);
}
Ok(())
}
/// Searches the set for the first key that satisfies the given `predicate`.
///
/// Returns the first key that satisfies the given predicate, or `None` if **no** key satisfies the predicate or the set is empty.
///
/// This function does **not** guarantee a specific iteration order.
///
/// Also, the predicate is **not** always tested on *all* keys, because the function returns at the first match.
#[inline]
pub fn find<P, Q>(&self, predicate: P) -> Result<Option<T>, Error>
where
P: Fn(&Q) -> bool,
T: Borrow<Q>,
Q: ?Sized,
{
let mut statement = self.connection.prepare_cached(SQL_QUERY_KEYS)?;
for key in statement.query_map([], |row| row.get(0))? {
let key: T = key?;
if predicate(key.borrow()) {
return Ok(Some(key));
}
}
Ok(None)
}
/// Invokes the given `callback` function for each key that matches the given pattern.
///
/// The pattern string is interpreted according to the [UNIX *globbing*](https://man7.org/linux/man-pages/man7/glob.7.html) rules. Specifically, a `?` character (not between brackets) matches any single character, and a `*` character (not between brackets) matches any string, including the empty string. Additionally, an expression `[…]` where the first character after the leading `[` is *not* an `!` matches a single character that can be any of the characters enclosed by the brackets; and an expression `[!…]` matches a single character that is *not* matched by the expression.
///
/// This function is case-sensitive and does **not** guarantee a specific iteration order.
///
/// If `invert` is `true`, returns all keys that do **not** match the pattern; otherwise the matching keys are returned.
#[inline]
pub fn glob<F, Q>(&self, invert: bool, pattern: &Q, mut callback: F) -> Result<(), Error>
where
F: FnMut(T),
T: Borrow<Q> + CharType,
Q: ?Sized + ToSql,
{
let mut statement = self.connection.prepare_cached(if invert { SQL_GLOBN_KEYS } else { SQL_GLOBA_KEYS })?;
for key in statement.query_map([pattern], |row| row.get(0))? {
callback(key?);
}
Ok(())
}
/// Returns the number of unique keys in the set.
#[inline]
pub fn len(&self) -> Result<Size, Error> {
let mut statement = self.connection.prepare_cached(SQL_COUNT_KEYS)?;
let count: i64 = statement.query_one([], |row| row.get(0))?;
Ok(Size::from_i64(count))
}
/// Returns `true` if the set contains **no** keys; otherwise returns `false`.
#[inline]
pub fn is_empty(&self) -> Result<bool, Error> {
let mut statement = self.connection.prepare_cached(SQL_EXIST_KEYS)?;
let any_elements: bool = statement.query_one([], |row| row.get(0))?;
Ok(!any_elements)
}
/// Removes *all* keys from the set.
#[inline]
pub fn clear(&mut self) -> Result<(), Error> {
let mut statement = self.connection.prepare_cached(SQL_DELETE_ALL)?;
statement.execute([])?;
Ok(())
}
#[cfg(feature = "backup")]
#[cfg_attr(docsrs, doc(cfg(feature = "backup")))]
#[inline]
/// Stores all contents of this set in the specified database file.
///
/// The file will be replaced, if it already exists!
pub fn to_file<P>(&self, path: P) -> Result<(), Error>
where
P: AsRef<Path>,
{
let mut connection_out = Self::initialize_connection(Connection::open_with_flags(path, *OPEN_FLAGS)?)?;
crate::common::backup_db(&self.connection, &mut connection_out)?;
Ok(())
}
}
cfg_if! {
if #[cfg(feature = "backup")] {
impl<T> TryClone for SQLiteSet<T>
where
T: Type,
{
type Error = Error;
/// Creates an exact, independent *copy* of the given set instance.
///
/// This function fails with [`Error::Unsupported`](crate::Error::Unsupported), if the optional [`backup`](crate#optional-features) feature is **not** enabled 😥
#[cfg_attr(docsrs, doc(cfg(feature = "backup")))]
#[inline]
fn try_clone(&self) -> Result<Self, Self::Error> {
let mut connection_cloned = Self::initialize_connection(Connection::open_in_memory()?)?;
crate::common::backup_db(&self.connection, &mut connection_cloned)?;
Self::from_connection(connection_cloned)
}
}
} else {
impl<T> TryClone for SQLiteSet<T>
where
T: Type,
{
type Error = Error;
/// Creates an exact, independent *copy* of the given set instance.
///
/// This function fails with [`Error::Unsupported`](crate::Error::Unsupported), if the optional [`backup`](crate#optional-features) feature is **not** enabled 😥
#[inline]
fn try_clone(&self) -> Result<Self, Self::Error> {
Err(Error::Unsupported)
}
}
}
}
// ---------------------------------------------------------------------------
// SQLiteSet Transaction
// ---------------------------------------------------------------------------
/// Represents an active SQLite transaction for a [`SQLiteSet`].
///
/// Most functions provided by this struct mirror the corresponding functions of the `SQLiteSet` struct.
///
/// The transaction is committed when the `SQLiteSetTransaction` is dropped.
pub struct SQLiteSetTransaction<'a, T>
where
T: Type,
{
_type: PhantomData<T>,
transaction: Transaction<'a>,
}
impl<'a, T> SQLiteSetTransaction<'a, T>
where
T: Type,
{
#[inline]
fn from(connection: &'a mut Connection) -> Result<Self, Error> {
let mut transaction = connection.transaction()?;
transaction.set_drop_behavior(rusqlite::DropBehavior::Commit);
Ok(Self { _type: PhantomData, transaction })
}
/// Drops the `SQLiteSetTransaction`, thereby committing the SQLite transaction.
#[inline]
pub fn commit(self) {}
/// This function is equivalent to [`SQLiteSet::insert()`].
#[inline]
pub fn insert<Q>(&mut self, key: &Q) -> Result<bool, Error>
where
T: Borrow<Q>,
Q: ?Sized + ToSql,
{
let mut statement = self.transaction.prepare_cached(SQL_INSERT_KEY)?;
match statement.execute([key]) {
Ok(_) => Ok(true),
Err(error) => check_constraint_violation(error),
}
}
/// Insert all keys from the given [`SQLiteMap`] into this set.
#[inline]
pub(crate) fn insert_mapped_keys<Q>(&mut self, map: &mut SQLiteMap<T, Q>) -> Result<(), Error>
where
Q: Type,
{
let mut statement = self.transaction.prepare_cached(SQL_INSERT_KEY)?;
let trx = map.transaction()?;
trx.try_for_each_key(|key| match statement.execute([key]) {
Ok(_) => Ok(()),
Err(error) => check_constraint_violation(error).map(|_| ()),
})
}
/// This function is equivalent to [`SQLiteSet::contains()`].
#[inline]
pub fn contains<Q>(&self, key: &Q) -> Result<bool, Error>
where
T: Borrow<Q>,
Q: ?Sized + ToSql,
{
let mut statement = self.transaction.prepare_cached(SQL_EXISTS_KEY)?;
Ok(statement.exists([key])?)
}
/// This function is equivalent to [`SQLiteSet::contains_nocase()`].
#[inline]
pub fn contains_nocase<Q>(&self, key: &Q) -> Result<bool, Error>
where
T: Borrow<Q> + CharType,
Q: ?Sized + ToSql,
{
let mut statement = self.transaction.prepare_cached(SQL_EXISTS_NOC)?;
Ok(statement.exists([key])?)
}
/// This function is equivalent to [`SQLiteSet::remove()`].
#[inline]
pub fn remove<Q>(&mut self, key: &Q) -> Result<bool, Error>
where
T: Borrow<Q>,
Q: ?Sized + ToSql,
{
let mut statement = self.transaction.prepare_cached(SQL_DELETE_KEY)?;
Ok(statement.execute([key])? != 0)
}
/// This function is equivalent to [`SQLiteSet::remove_nocase()`].
#[inline]
pub fn remove_nocase<Q>(&mut self, key: &Q) -> Result<bool, Error>
where
T: Borrow<Q> + CharType,
Q: ?Sized + ToSql,
{
let mut statement = self.transaction.prepare_cached(SQL_DELETE_NOC)?;
Ok(statement.execute([key])? != 0)
}
/// This function is equivalent to [`SQLiteSet::for_each()`].
#[inline]
pub fn for_each<F>(&self, mut callback: F) -> Result<(), Error>
where
F: FnMut(T),
{
let mut statement = self.transaction.prepare_cached(SQL_QUERY_KEYS)?;
for key in statement.query_map([], |row| row.get(0))? {
callback(key?);
}
Ok(())
}
/// This function is equivalent to [`SQLiteSet::find()`].
#[inline]
pub fn find<P, Q>(&self, predicate: P) -> Result<Option<T>, Error>
where
P: Fn(&Q) -> bool,
T: Borrow<Q>,
Q: ?Sized,
{
let mut statement = self.transaction.prepare_cached(SQL_QUERY_KEYS)?;
for key in statement.query_map([], |row| row.get(0))? {
let key: T = key?;
if predicate(key.borrow()) {
return Ok(Some(key));
}
}
Ok(None)
}
/// This function is equivalent to [`SQLiteSet::glob()`].
#[inline]
pub fn glob<F, Q>(&self, invert: bool, pattern: &Q, mut callback: F) -> Result<(), Error>
where
F: FnMut(T),
T: Borrow<Q> + CharType,
Q: ?Sized + ToSql,
{
let mut statement = self.transaction.prepare_cached(if invert { SQL_GLOBN_KEYS } else { SQL_GLOBA_KEYS })?;
for key in statement.query_map([pattern], |row| row.get(0))? {
callback(key?);
}
Ok(())
}
/// This function is equivalent to [`SQLiteSet::len()`].
#[inline]
pub fn len(&self) -> Result<Size, Error> {
let mut statement = self.transaction.prepare_cached(SQL_COUNT_KEYS)?;
let count: i64 = statement.query_one([], |row| row.get(0))?;
Ok(Size::from_i64(count))
}
/// This function is equivalent to [`SQLiteSet::is_empty()`].
#[inline]
pub fn is_empty(&self) -> Result<bool, Error> {
let mut statement = self.transaction.prepare_cached(SQL_EXIST_KEYS)?;
let any_elements: bool = statement.query_one([], |row| row.get(0))?;
Ok(!any_elements)
}
/// This function is equivalent to [`SQLiteSet::clear()`].
#[inline]
pub fn clear(&mut self) -> Result<(), Error> {
let mut statement = self.transaction.prepare_cached(SQL_DELETE_ALL)?;
statement.execute([])?;
Ok(())
}
}