1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174

#[cfg(feature = "indexer")]
use std::collections::HashSet;

use infisearch_common::dictionary::Dictionary;
use infisearch_common::utils::split_incl::SplitIncl;
#[cfg(feature = "indexer")]
use regex::Regex;

#[cfg(feature = "indexer")]
use crate::ascii_folding_filter;
use crate::spelling::BestTermCorrector;
use crate::stop_words::get_stop_words;
use crate::utils;
use infisearch_common::language::InfiLanguageConfig;
#[cfg(feature = "indexer")]
use infisearch_common::tokenize::{IndexerTokenizer, TermIter};
use infisearch_common::tokenize::{self, SearchTokenizeResult, SearchTokenizer, SearchTokenizeTerm};

#[cfg(feature = "indexer")]
lazy_static! {
    pub static ref SENTENCE_SPLITTER: Regex = Regex::new(r#"[.,;?!]\s+"#).unwrap();
}

pub struct Tokenizer {
    // Remove HashSet from the search binary, where speed benefits are minimal
    #[cfg(feature = "indexer")]
    pub stop_words: HashSet<String>,
    #[cfg(not(feature = "indexer"))]
    pub stop_words: Vec<String>,

    ignore_stop_words: bool,

    // Just needs to be filtered during indexing
    #[cfg(feature = "indexer")]
    max_term_len: usize,

    best_term_corrector: BestTermCorrector,
}

pub fn new_with_options(lang_config: &InfiLanguageConfig) -> Tokenizer {
    let stop_words = get_stop_words(lang_config, &[
        // Same list from tantivy
        "a", "an", "and", "are", "as", "at", "be", "but", "by", "for", "if", "in", "into", "is", "it", "no",
        "not", "of", "on", "or", "such", "that", "the", "their", "then", "there", "these", "they", "this",
        "to", "was", "will", "with"
    ]);

    #[cfg(feature = "indexer")]
    let max_term_len = lang_config.options.max_term_len.unwrap_or(80).min(250);

    Tokenizer {
        stop_words,
        ignore_stop_words: lang_config.options.ignore_stop_words.unwrap_or(false),
        #[cfg(feature = "indexer")]
        max_term_len,
        best_term_corrector: BestTermCorrector::new(),
    }
}

#[cfg(feature = "indexer")]
impl IndexerTokenizer for Tokenizer {
    fn tokenize<'a>(&'a self, text: &'a mut str) -> TermIter<'a> {
        text.make_ascii_lowercase();
        let it = SENTENCE_SPLITTER.split(text)
            .flat_map(move |sent_slice| {
                sent_slice.split(utils::split_terms)
                    .filter(|&s| !s.is_empty())
                    .map(|term_slice| utils::term_filter(ascii_folding_filter::to_ascii(term_slice)))
                    .filter(move |term| {
                        let term_byte_len = term.len();
                        term_byte_len > 0
                            && term_byte_len <= self.max_term_len
                            && !(self.ignore_stop_words && self.stop_words.contains(term.as_ref()))
                    })
                    .map(Some).chain(std::iter::once(None))
            });

        Box::new(it)
    }
}

impl SearchTokenizer for Tokenizer {
    fn search_tokenize(
        &mut self,
        query_chars: &[char],
        query_chars_offset: usize,
        query_chars_offset_end: usize,
        escape_indices: &[usize],
        dict: &Dictionary,
    ) -> SearchTokenizeResult {
        let mut text: String = unsafe { query_chars.get_unchecked(query_chars_offset..query_chars_offset_end) }.iter().collect();
        text.make_ascii_lowercase();

        let should_expand = !text.ends_with(' ');

        let mut terms = Vec::new();
        let split: Vec<_> = SplitIncl::split(
            &text,
            utils::split_terms,
        ).collect();

        for (idx, (char_idx, s)) in split.iter().enumerate() {
            if s.is_empty() {
                continue;
            }

            let suffix_wildcard = (idx + 1 != split.len()) && unsafe { split.get_unchecked(idx + 1) }.1 == "*";
            let prefix_ops = tokenize::get_prefix_ops(
                *char_idx + query_chars_offset, 1, query_chars_offset, query_chars, escape_indices, self,
            );

            let mut term_inflections = Vec::new();

            let preprocessed = utils::ascii_and_nonword_filter(&mut term_inflections, s, utils::term_filter);
            if preprocessed.is_empty() {
                continue;
            }

            let original_term = preprocessed.clone().into_owned();
            let mut is_corrected = false;

            // This comes before spelling correction,
            // as ignore_stop_words removes from the index (won't be present in the dictionary)
            if self.ignore_stop_words && self.is_stop_word(&preprocessed) {
                terms.push(SearchTokenizeTerm {
                    term: None,
                    term_inflections,
                    original_term,
                    suffix_wildcard,
                    is_corrected,
                    prefix_ops,
                });
                continue;
            }

            let term = if dict.get_term_info(&preprocessed).is_none() {
                if suffix_wildcard {
                    None
                } else if let Some(corrected_term) = self.best_term_corrector.get_best_corrected_term(dict, &preprocessed) {
                    term_inflections.push(corrected_term.clone());
                    is_corrected = true;
                    Some(corrected_term)
                } else {
                    None
                }
            } else {
                Some(preprocessed.into_owned())
            };

            terms.push(SearchTokenizeTerm {
                term,
                term_inflections,
                original_term,
                suffix_wildcard,
                is_corrected,
                prefix_ops,
            })
        }

        SearchTokenizeResult {
            terms,
            auto_suffix_wildcard: should_expand,
        }
    }

    #[inline(never)]
    fn is_stop_word(&self, term: &str) -> bool {
        self.stop_words.iter().any(|t| t == term)
    }

    fn is_valid_prefix_op_terminator(&self, c: char) -> bool {
        c.is_ascii_whitespace()
    }
}