/* * SPDX-FileCopyrightText: 2020 Stalwart Labs LLC * * SPDX-License-Identifier: AGPL-3.0-only OR LicenseRef-SEL */ pub mod chinese; pub mod japanese; pub mod space; pub mod stream; pub mod types; pub mod word; use std::borrow::Cow; #[derive(Debug, Clone, PartialEq, Eq)] pub struct Token { pub word: T, pub from: usize, pub to: usize, } pub trait InnerToken<'x>: Sized { fn new_alphabetic(value: impl Into>) -> Self; fn unwrap_alphabetic(self) -> Cow<'x, str>; fn is_alphabetic(&self) -> bool; fn is_alphabetic_8bit(&self) -> bool; } impl<'x> InnerToken<'x> for Cow<'x, str> { fn new_alphabetic(value: impl Into>) -> Self { value.into() } fn is_alphabetic(&self) -> bool { true } fn is_alphabetic_8bit(&self) -> bool { !self.is_ascii() } fn unwrap_alphabetic(self) -> Cow<'x, str> { self } } impl Token { pub fn new(offset: usize, len: usize, word: T) -> Token { debug_assert!(offset <= u32::MAX as usize); debug_assert!(len <= u8::MAX as usize); Token { from: offset, to: offset + len, word, } } }