alyze 0.1.5

High-performance text analysis for full-text search
Documentation 
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
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295

use tpuf_icu_properties_211::{
    CodePointMapData, CodePointMapDataBorrowed, CodePointSetData, CodePointSetDataBorrowed,
    props::{ExtendedPictographic, GeneralCategory, Ideographic, Script, WordBreak},
};

use crate::uax29::break_property_enum;

// Property values for the Word_Break property, as defined in
// [UAX #29](https://www.unicode.org/reports/tr29/#Word_Break_Property).
//
// Each character has an associated Word_Break property value.
break_property_enum! {
    WordBreakProperty {
        Other, ALetter, Format, Katakana, MidLetter, MidNum, Numeric,
        ExtendNumLet, CR, Extend, LF, MidNumLet, Newline,
        RegionalIndicator, HebrewLetter, SingleQuote, DoubleQuote, ZWJ, WSegSpace,
    }
}

/// ASCII characters are very common, so we pre-compute a lookup table for the first 128 code points.
/// A test below verifies that this table is correct against the full dictionary-based lookup.
pub(crate) const ASCII_WORD_BREAK_PROP: [WordBreakProperty; 128] = [
    WordBreakProperty::Other,
    WordBreakProperty::Other,
    WordBreakProperty::Other,
    WordBreakProperty::Other,
    WordBreakProperty::Other,
    WordBreakProperty::Other,
    WordBreakProperty::Other,
    WordBreakProperty::Other,
    WordBreakProperty::Other,
    WordBreakProperty::Other,
    WordBreakProperty::LF,
    WordBreakProperty::Newline,
    WordBreakProperty::Newline,
    WordBreakProperty::CR,
    WordBreakProperty::Other,
    WordBreakProperty::Other,
    WordBreakProperty::Other,
    WordBreakProperty::Other,
    WordBreakProperty::Other,
    WordBreakProperty::Other,
    WordBreakProperty::Other,
    WordBreakProperty::Other,
    WordBreakProperty::Other,
    WordBreakProperty::Other,
    WordBreakProperty::Other,
    WordBreakProperty::Other,
    WordBreakProperty::Other,
    WordBreakProperty::Other,
    WordBreakProperty::Other,
    WordBreakProperty::Other,
    WordBreakProperty::Other,
    WordBreakProperty::Other,
    WordBreakProperty::WSegSpace,
    WordBreakProperty::Other,
    WordBreakProperty::DoubleQuote,
    WordBreakProperty::Other,
    WordBreakProperty::Other,
    WordBreakProperty::Other,
    WordBreakProperty::Other,
    WordBreakProperty::SingleQuote,
    WordBreakProperty::Other,
    WordBreakProperty::Other,
    WordBreakProperty::Other,
    WordBreakProperty::Other,
    WordBreakProperty::MidNum,
    WordBreakProperty::Other,
    WordBreakProperty::MidNumLet,
    WordBreakProperty::Other,
    WordBreakProperty::Numeric,
    WordBreakProperty::Numeric,
    WordBreakProperty::Numeric,
    WordBreakProperty::Numeric,
    WordBreakProperty::Numeric,
    WordBreakProperty::Numeric,
    WordBreakProperty::Numeric,
    WordBreakProperty::Numeric,
    WordBreakProperty::Numeric,
    WordBreakProperty::Numeric,
    WordBreakProperty::MidLetter,
    WordBreakProperty::MidNum,
    WordBreakProperty::Other,
    WordBreakProperty::Other,
    WordBreakProperty::Other,
    WordBreakProperty::Other,
    WordBreakProperty::Other,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::Other,
    WordBreakProperty::Other,
    WordBreakProperty::Other,
    WordBreakProperty::Other,
    WordBreakProperty::ExtendNumLet,
    WordBreakProperty::Other,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::ALetter,
    WordBreakProperty::Other,
    WordBreakProperty::Other,
    WordBreakProperty::Other,
    WordBreakProperty::Other,
    WordBreakProperty::Other,
];

impl WordBreakProperty {
    /// Checks if the property is ExtPictographic.
    /// Notably, a character can be both ExtPictographic and have another Word_Break property value.
    pub fn is_ext_pictographic(c: char) -> bool {
        EXT_PICT.contains(c)
    }
}

/// A dictionary mapping Unicode code points to their corresponding Word_Break property values.
/// This is used to efficiently look up the Word_Break property for any given character.
const WORD_BREAK_PROP: CodePointMapDataBorrowed<'static, WordBreak> =
    CodePointMapData::<WordBreak>::new();
const EXT_PICT: CodePointSetDataBorrowed<'static> = CodePointSetData::new::<ExtendedPictographic>();

/// Lazily-built union of every codepoint whose "strict" word-like check returns true:
/// `ExtendedPictographic ∪ Ideographic ∪ {c | Script(c) ∉ {Common,Inherited,Unknown}} ∪
/// {c | GeneralCategory(c) == OtherNumber}`.
///
/// Stored as a sorted slice of inclusive `(start, end)` ranges so `is_word_like_strict` is a
/// single binary search per char instead of up to four ICU trie lookups. Built on first use.
static WORD_LIKE_STRICT_RANGES: std::sync::OnceLock<Box<[(u32, u32)]>> = std::sync::OnceLock::new();

// See `TokenProperties::is_word_like()` for a detailed explanation of this.
fn word_like_strict_ranges() -> &'static [(u32, u32)] {
    WORD_LIKE_STRICT_RANGES.get_or_init(|| {
        let ideographic = CodePointSetData::new::<Ideographic>();
        let script = CodePointMapData::<Script>::new();
        let gc = CodePointMapData::<GeneralCategory>::new();

        let mut ranges: Vec<(u32, u32)> = Vec::new();
        for r in EXT_PICT.iter_ranges() {
            ranges.push((*r.start(), *r.end()));
        }
        for r in ideographic.iter_ranges() {
            ranges.push((*r.start(), *r.end()));
        }

        // r.value is true for any range where the script is not Common/Inherited/Unknown.
        // E.g. r.value=true when Script is a reasonable writing system (Greek, Thai, Cyrillic, etc.)
        // This operates similar to Rust's .iter().map().filter().
        for r in script.iter_ranges_mapped(|s| {
            !matches!(s, Script::Common | Script::Inherited | Script::Unknown)
        }) {
            if r.value {
                ranges.push((*r.range.start(), *r.range.end()));
            }
        }

        for r in gc.iter_ranges_for_value(GeneralCategory::OtherNumber) {
            ranges.push((*r.start(), *r.end()));
        }

        // Sort + merge adjacent/overlapping ranges so binary search has clean buckets.
        ranges.sort_unstable();
        let mut merged: Vec<(u32, u32)> = Vec::with_capacity(ranges.len());
        for (s, e) in ranges {
            if let Some(last) = merged.last_mut() {
                if s <= last.1.saturating_add(1) {
                    last.1 = last.1.max(e);
                    continue;
                }
            }
            merged.push((s, e));
        }
        merged.into_boxed_slice()
    })
}

/// "Strict" word-like check for chars whose `WordBreakProperty` alone didn't classify them as
/// word-like. Mirrors the residual conditions of turbopuffer's `is_word_like_token_char` /
/// `should_include_token_span` for non-ASCII chars not already covered by `ALetter` /
/// `HebrewLetter` / `Numeric`.
#[inline]
pub(crate) fn is_word_like_strict(c: char) -> bool {
    let cp = c as u32;
    word_like_strict_ranges()
        .binary_search_by(|&(s, e)| {
            if cp < s {
                std::cmp::Ordering::Greater
            } else if cp > e {
                std::cmp::Ordering::Less
            } else {
                std::cmp::Ordering::Equal
            }
        })
        .is_ok()
}

/// Helper function to look up the Word_Break property for a given character using the pre-computed dictionary.
#[inline(never)]
pub(crate) fn lookup_word_break_property_from_dictionary(c: char) -> WordBreakProperty {
    match WORD_BREAK_PROP.get(c) {
        WordBreak::ALetter => WordBreakProperty::ALetter,
        WordBreak::Format => WordBreakProperty::Format,
        WordBreak::Katakana => WordBreakProperty::Katakana,
        WordBreak::MidLetter => WordBreakProperty::MidLetter,
        WordBreak::MidNum => WordBreakProperty::MidNum,
        WordBreak::Numeric => WordBreakProperty::Numeric,
        WordBreak::ExtendNumLet => WordBreakProperty::ExtendNumLet,
        WordBreak::CR => WordBreakProperty::CR,
        WordBreak::Extend => WordBreakProperty::Extend,
        WordBreak::LF => WordBreakProperty::LF,
        WordBreak::MidNumLet => WordBreakProperty::MidNumLet,
        WordBreak::Newline => WordBreakProperty::Newline,
        WordBreak::RegionalIndicator => WordBreakProperty::RegionalIndicator,
        WordBreak::HebrewLetter => WordBreakProperty::HebrewLetter,
        WordBreak::SingleQuote => WordBreakProperty::SingleQuote,
        WordBreak::DoubleQuote => WordBreakProperty::DoubleQuote,
        WordBreak::ZWJ => WordBreakProperty::ZWJ,
        WordBreak::WSegSpace => WordBreakProperty::WSegSpace,
        WordBreak::Other => WordBreakProperty::Other,
        _ => unreachable!(),
    }
}

#[cfg(test)]
mod tests {
    use super::{
        ASCII_WORD_BREAK_PROP, WordBreakProperty, lookup_word_break_property_from_dictionary,
    };

    #[test]
    fn test_ascii_table_correct() {
        let mut expected = [WordBreakProperty::Other; 128];
        for c in 0..=0x7F {
            expected[c as usize] = lookup_word_break_property_from_dictionary(c as u8 as char);
        }

        // If it doesn't match, print the correct table in an easy-to-copy-paste way.
        if ASCII_WORD_BREAK_PROP != expected {
            for i in 0..8 {
                let row = &expected[i * 16..(i + 1) * 16];
                let row_str = row
                    .iter()
                    .map(|p| format!("WordBreakProperty::{:?}", p))
                    .collect::<Vec<_>>()
                    .join(", ");
                println!("    {},", row_str);
            }
            assert!(false);
        }
    }
}