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

// Copyright 2020 The Trax Authors and The HuggingFace Inc. team.
// Copyright (c) 2018, NVIDIA CORPORATION.  All rights reserved.
// Copyright 2020 Guillaume Becquin
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//     http://www.apache.org/licenses/LICENSE-2.0
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.

use crate::error::TokenizerError;
use crate::tokenizer::base_tokenizer::{Token, TokenRef};
use crate::tokenizer::tokenization_utils::{
    bpe, clean_text, decompose_nfkc, fix_mask, is_whitespace, lowercase, split_on_bpe_pairs,
    split_on_special_tokens, whitespace_tokenize, BpeCache,
};
use crate::tokenizer::{MultiThreadedTokenizer, Tokenizer};
use crate::vocab::{BpePairVocab, ReformerVocab, Vocab};
use crate::Mask;
use std::collections::HashMap;
use std::sync::RwLock;

/// # Reformer tokenizer
pub struct ReformerTokenizer {
    vocab: ReformerVocab,
    bpe_ranks: BpePairVocab,
    cache: BpeCache,
    lower_case: bool,
}

impl ReformerTokenizer {
    /// Create a new instance of a `ReformerTokenizer`
    /// Expects a SentencePiece protobuf file as an input.
    ///
    /// # Parameters
    /// - path (`&str`): path to the SentencePiece model file
    /// - lower_case (`bool`): flag indicating if the text should be lower-cased as part of the tokenization
    /// - strip_accents (`bool`): flag indicating if accents should be stripped from the text
    ///
    /// # Example
    ///
    /// ```no_run
    /// use rust_tokenizers::tokenizer::{SentencePieceTokenizer, Tokenizer};
    /// let lower_case = false;
    /// let tokenizer = SentencePieceTokenizer::from_file("path/to/vocab/file", lower_case).unwrap();
    /// ```
    pub fn from_file(path: &str, lower_case: bool) -> Result<ReformerTokenizer, TokenizerError> {
        let vocab = ReformerVocab::from_file(path)?;
        let bpe_ranks = BpePairVocab::from_sentencepiece_file(path)?;
        let cache = RwLock::new(HashMap::new());
        Ok(ReformerTokenizer {
            vocab,
            bpe_ranks,
            cache,
            lower_case,
        })
    }
}

impl Tokenizer<ReformerVocab> for ReformerTokenizer {
    fn vocab(&self) -> &ReformerVocab {
        &self.vocab
    }

    fn tokenize_to_tokens(&self, text: TokenRef) -> Vec<Token> {
        let mut tokens = split_on_special_tokens(text, &self.vocab)
            .into_iter()
            .map(whitespace_tokenize)
            .flatten()
            .map(|token| token.to_owned())
            .collect::<Vec<Token>>();

        let mut sub_tokens = Vec::new();
        for token in tokens.iter_mut() {
            decompose_nfkc(token);
            clean_text(token, true);
            if !token.text.is_empty() {
                token.text = token.text.replace(|c: char| is_whitespace(&c), "\u{2581}");
                if !token.text.starts_with('\u{2581}') {
                    token.text.insert(0, '\u{2581}');
                    token
                        .reference_offsets
                        .insert(0, token.reference_offsets[0]);
                };

                if token.mask != Mask::Special && token.mask != Mask::Unknown {
                    if self.lower_case {
                        lowercase(token);
                    }
                    sub_tokens.extend(split_on_bpe_pairs(
                        token.as_ref(),
                        bpe,
                        &self.bpe_ranks,
                        &self.cache,
                        false,
                    ));
                } else {
                    sub_tokens.push(token.to_owned());
                }
            }

            //     Consolidate consecutive unknown tokens
            let mut prev_is_unk = false;
            let mut indices_to_remove = vec![];
            for (index, sub_token) in sub_tokens.iter_mut().enumerate() {
                if self.vocab.values.get(&sub_token.text).is_none() {
                    sub_token.mask = Mask::Unknown;
                }
                if sub_token.mask == Mask::Unknown {
                    if prev_is_unk {
                        indices_to_remove.push(index);
                    }
                    prev_is_unk = true;
                } else {
                    prev_is_unk = false;
                }
            }
            for index_to_remove in indices_to_remove.into_iter().rev() {
                sub_tokens.remove(index_to_remove);
            }
        }

        fix_mask(&mut sub_tokens);
        sub_tokens
    }

    fn convert_tokens_to_string(&self, tokens: Vec<String>) -> String {
        tokens
            .into_iter()
            .map(|v| v.replace('\u{2581}', " "))
            .collect::<Vec<String>>()
            .join("")
    }
}

impl MultiThreadedTokenizer<ReformerVocab> for ReformerTokenizer {}