lance_tokenizer/
whitespace_tokenizer.rs

1// SPDX-License-Identifier: Apache-2.0
2// SPDX-FileCopyrightText: Copyright The Lance Authors
3// SPDX-License-Identifier: MIT
4// Adapted from Tantivy v0.24.2 whitespace tokenizer.
5// Copyright (c) 2017-present Tantivy contributors.
6
7use std::str::CharIndices;
8
9use crate::{Token, TokenStream, Tokenizer};
10
11#[derive(Clone, Default)]
12pub struct WhitespaceTokenizer {
13    token: Token,
14}
15
16pub struct WhitespaceTokenStream<'a> {
17    text: &'a str,
18    chars: CharIndices<'a>,
19    token: &'a mut Token,
20}
21
22impl Tokenizer for WhitespaceTokenizer {
23    type TokenStream<'a> = WhitespaceTokenStream<'a>;
24
25    fn token_stream<'a>(&'a mut self, text: &'a str) -> Self::TokenStream<'a> {
26        self.token.reset();
27        WhitespaceTokenStream {
28            text,
29            chars: text.char_indices(),
30            token: &mut self.token,
31        }
32    }
33}
34
35impl WhitespaceTokenStream<'_> {
36    fn search_token_end(&mut self) -> usize {
37        (&mut self.chars)
38            .filter(|(_, c)| c.is_ascii_whitespace())
39            .map(|(offset, _)| offset)
40            .next()
41            .unwrap_or(self.text.len())
42    }
43}
44
45impl TokenStream for WhitespaceTokenStream<'_> {
46    fn advance(&mut self) -> bool {
47        self.token.text.clear();
48        self.token.position = self.token.position.wrapping_add(1);
49        while let Some((offset_from, c)) = self.chars.next() {
50            if !c.is_ascii_whitespace() {
51                let offset_to = self.search_token_end();
52                self.token.offset_from = offset_from;
53                self.token.offset_to = offset_to;
54                self.token.text.push_str(&self.text[offset_from..offset_to]);
55                return true;
56            }
57        }
58        false
59    }
60
61    fn token(&self) -> &Token {
62        self.token
63    }
64
65    fn token_mut(&mut self) -> &mut Token {
66        self.token
67    }
68}
lance_tokenizer/whitespace_tokenizer.rs

lance_tokenizer/
whitespace_tokenizer.rs
