toklen 0.2.0

A single-threaded, lightweight, and fast token counter.
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

#[cfg(feature = "split")]
mod split;

#[cfg(feature = "byte_level")]
mod byte_level;

#[cfg(feature = "whitespace")]
mod whitespace;

#[cfg(feature = "whitespace_split")]
mod whitespace_split;

#[cfg(feature = "punctuation")]
mod punctuation;

#[cfg(feature = "metaspace")]
mod metaspace;

#[cfg(feature = "digits")]
mod digits;

#[cfg(feature = "char_delimiter_split")]
mod char_delimiter_split;

// ── PreTokenizerConfig ──────────────────────────────────────────────────────

use serde::Deserialize;

/// Pre-tokenizer pipeline step from `tokenizer.json`.
#[derive(Clone, Debug, Deserialize)]
#[serde(tag = "type")]
pub enum PreTokenizerConfig {
    Sequence {
        #[serde(default)]
        pretokenizers: Vec<PreTokenizerConfig>,
    },
    #[cfg(feature = "byte_level")]
    ByteLevel(byte_level::ByteLevel),
    #[cfg(feature = "split")]
    Split(split::Split),
    #[cfg(feature = "whitespace")]
    Whitespace,
    #[cfg(feature = "whitespace_split")]
    WhitespaceSplit,
    #[cfg(feature = "digits")]
    Digits(digits::Digits),
    #[cfg(feature = "metaspace")]
    Metaspace(metaspace::Metaspace),
    #[cfg(feature = "punctuation")]
    Punctuation(punctuation::Punctuation),
    #[cfg(feature = "char_delimiter_split")]
    CharDelimiterSplit(char_delimiter_split::CharDelimiterSplit),

    // ------------------------------------------------------------------------
    // Unsupported settings.
    #[serde(untagged)]
    Other(serde_json::Value),
    // BertPreTokenizer,
    // UnicodeScripts,
}

// ── PreTokenizer ────────────────────────────────────────────────────────────

use crate::pre_tokenized::PreTokenizedString;

/// Errors from constructing or running a pre-tokenizer.
#[derive(Debug, thiserror::Error)]
pub enum Error {
    #[error("invalid config value: {0}")]
    Json(#[from] serde_json::Error),

    #[error("regex error: {0}")]
    Regex(#[from] fancy_regex::Error),

    #[error("unsupported pre-tokenizer type: {0}")]
    Unsupported(String),
}

/// A compiled pre-tokenizer ready for use.
#[derive(Clone, Debug)]
pub enum PreTokenizer {
    Sequence(Vec<PreTokenizer>),

    #[cfg(feature = "split")]
    Split(split::Split),
    #[cfg(feature = "byte_level")]
    ByteLevel(byte_level::ByteLevel),
    #[cfg(feature = "whitespace")]
    Whitespace,
    #[cfg(feature = "whitespace_split")]
    WhitespaceSplit,
    #[cfg(feature = "digits")]
    Digits(digits::Digits),
    #[cfg(feature = "metaspace")]
    Metaspace(metaspace::Metaspace),
    #[cfg(feature = "punctuation")]
    Punctuation(punctuation::Punctuation),
    #[cfg(feature = "char_delimiter_split")]
    CharDelimiterSplit(char_delimiter_split::CharDelimiterSplit),
}

impl PreTokenizer {
    /// Build a pre-tokenizer from its JSON configuration.
    pub fn from_config(config: PreTokenizerConfig) -> Result<Self, Error> {
        match config {
            #[cfg(feature = "split")]
            PreTokenizerConfig::Split(s) => Ok(Self::Split(s)),
            #[cfg(feature = "byte_level")]
            PreTokenizerConfig::ByteLevel(bl) => Ok(Self::ByteLevel(bl)),
            #[cfg(feature = "whitespace")]
            PreTokenizerConfig::Whitespace => Ok(Self::Whitespace),
            #[cfg(feature = "whitespace_split")]
            PreTokenizerConfig::WhitespaceSplit => Ok(Self::WhitespaceSplit),
            #[cfg(feature = "digits")]
            PreTokenizerConfig::Digits(d) => Ok(Self::Digits(d)),
            #[cfg(feature = "metaspace")]
            PreTokenizerConfig::Metaspace(m) => Ok(Self::Metaspace(m)),
            #[cfg(feature = "punctuation")]
            PreTokenizerConfig::Punctuation(p) => Ok(Self::Punctuation(p)),
            #[cfg(feature = "char_delimiter_split")]
            PreTokenizerConfig::CharDelimiterSplit(c) => Ok(Self::CharDelimiterSplit(c)),
            PreTokenizerConfig::Sequence { pretokenizers } => {
                let mut steps: Vec<Self> = Vec::with_capacity(pretokenizers.len());
                for pretokenizer in pretokenizers {
                    steps.push(Self::from_config(pretokenizer)?);
                }
                Ok(Self::Sequence(steps))
            }
            PreTokenizerConfig::Other(v) => Err(Error::Unsupported(v.to_string())),
        }
    }

    /// Refine the splits of `pts` in place.
    pub fn pre_tokenize(&self, pts: &mut PreTokenizedString) -> Result<(), Error> {
        match self {
            #[cfg(feature = "split")]
            Self::Split(s) => s.pre_tokenize(pts),
            #[cfg(feature = "byte_level")]
            Self::ByteLevel(bl) => bl.pre_tokenize(pts),
            #[cfg(feature = "whitespace")]
            Self::Whitespace => whitespace::pre_tokenize(pts),
            #[cfg(feature = "whitespace_split")]
            Self::WhitespaceSplit => whitespace_split::pre_tokenize(pts),
            #[cfg(feature = "digits")]
            Self::Digits(d) => d.pre_tokenize(pts),
            #[cfg(feature = "metaspace")]
            Self::Metaspace(m) => m.pre_tokenize(pts),
            #[cfg(feature = "punctuation")]
            Self::Punctuation(p) => p.pre_tokenize(pts),
            #[cfg(feature = "char_delimiter_split")]
            Self::CharDelimiterSplit(c) => c.pre_tokenize(pts),
            Self::Sequence(steps) => {
                for step in steps {
                    step.pre_tokenize(pts)?;
                }
                Ok(())
            }
        }
    }
}