tokenization_binaryllm.py

#!/usr/bin/env python3
# tokenization_binaryllm.py
# ============================================================
# BinaryLLMTokenizer (AutoTokenizer compatible) EXACTEMENT comme
# llmTalk (mode base=65536) + infer_tagged12.py:
#
# - Base vocab: 0..65535 (radix)
# - BOS id = 65536
# - EOS id = 65537
# - UNK alias = EOS (pas de nouvel id)
# - Encodage MANUEL: UTF-8 bytes -> digits base65536 BIG-ENDIAN (chunks 2 bytes)
#   (si byte impair: dernier chunk = 1 byte => id 0..255)
# - Décodage: digits -> bytes BIG-ENDIAN -> UTF-8 (errors=replace)
# - build_inputs_with_special_tokens:
#     single: [BOS] + ids + [EOS]
#     pair  : [BOS] + ids0 + [EOS] + ids1 + [EOS]
#
# IMPORTANT:
# - Ce tokenizer NE génère PAS ton pattern "...[EOS][BOS]" tout seul,
#   parce que HuggingFace standard = BOS ... EOS.
#   Pour llmTalk / infer_tagged*, c’est ton script qui ajoute le BOS final.
#
# Usage:
# - Mets ce fichier dans le repo HF (avec __init__.py si tu veux) et
#   ajoute dans tokenizer_config.json:
#     {"tokenizer_class": "BinaryLLMTokenizer", "auto_map": {...}} si besoin.
# - Puis:
#   tok = AutoTokenizer.from_pretrained(repo, trust_remote_code=True)
# ============================================================

from __future__ import annotations

import json
import os
import re
from typing import Dict, List, Optional, Tuple, Any

from transformers import PreTrainedTokenizer


class BinaryLLMTokenizer(PreTrainedTokenizer):
    model_input_names = ["input_ids", "attention_mask"]

    TOKEN_RE = re.compile(r"^<U([0-9A-Fa-f]{4})>$")

    def __init__(
        self,
        bos_token: str = "<BOS>",
        eos_token: str = "<EOS>",
        unk_token: str = "<UNK>",
        pad_token: Optional[str] = None,
        **kwargs: Any,
    ):
        # base ids 0..65535 reserved for radix tokens (strict)
        self._base_vocab_size = 65536

        # reserve ids
        self._bos_id = 65536
        self._eos_id = 65537

        # UNK is an alias to EOS to preserve radix purity (no new base id)
        self._unk_id = self._eos_id

        # special token strings
        self._bos_str = bos_token
        self._eos_str = eos_token
        self._unk_str = unk_token
        self._pad_str = pad_token

        super().__init__(
            bos_token=bos_token,
            eos_token=eos_token,
            unk_token=unk_token,
            pad_token=pad_token,
            **kwargs,
        )

    @property
    def vocab_size(self) -> int:
        return 65538

    def get_vocab(self) -> Dict[str, int]:
        # IMPORTANT: never call self.unk_token_id here (it triggers recursion)
        v = {
            self._bos_str: self._bos_id,
            self._eos_str: self._eos_id,
            self._unk_str: self._unk_id,
        }
        if self.pad_token is not None:
            # if pad_token == "<EOS>", it will map to eos id via _convert_token_to_id()
            v[self.pad_token] = self._convert_token_to_id(self.pad_token)
        return v

    # -----------------------------
    # Core manual base65536 codec
    # -----------------------------

    def _encode_base65536_be(self, text: str) -> List[int]:
        b = bytearray(text.encode("utf-8", errors="strict"))
        if len(b) == 0:
            return [0]

        out: List[int] = []
        i = 0
        n = len(b)

        # chunks of 2 bytes, BIG-ENDIAN
        while i + 1 < n:
            out.append((b[i] << 8) | b[i + 1])
            i += 2

        # last odd byte => 0..255
        if i < n:
            out.append(int(b[i]))

        return out

    def _decode_base65536_be(self, ids: List[int]) -> str:
        bb = bytearray()
        for x in ids:
            xi = int(x) & 0xFFFFFFFF
            if 0 <= xi <= 255:
                bb.append(xi)
            else:
                bb.append((xi >> 8) & 0xFF)
                bb.append(xi & 0xFF)
        return bytes(bb).decode("utf-8", errors="replace")

    # -----------------------------
    # HF required overrides
    # -----------------------------

    def _tokenize(self, text: str) -> List[str]:
        ids = self._encode_base65536_be(text)
        return [self._id_to_token_base(i) for i in ids]

    def _convert_token_to_id(self, token: str) -> int:
        if token == self._bos_str:
            return self._bos_id
        if token == self._eos_str:
            return self._eos_id
        if token == self._unk_str:
            return self._unk_id

        if self.pad_token is not None and token == self.pad_token:
            # common case: pad_token is "<EOS>"
            if self.pad_token == self._eos_str:
                return self._eos_id
            # otherwise: no dedicated PAD id in this vocab, alias to EOS
            return self._eos_id

        m = self.TOKEN_RE.match(token)
        if m:
            return int(m.group(1), 16)

        return self._unk_id

    def _convert_id_to_token(self, index: int) -> str:
        if index == self._bos_id:
            return self._bos_str
        if index == self._eos_id:
            return self._eos_str
        if index == self._unk_id:
            return self._unk_str
        if self.pad_token is not None and index == self.pad_token_id:
            return self.pad_token

        if 0 <= index < self._base_vocab_size:
            return self._id_to_token_base(index)

        return self._unk_str

    def convert_tokens_to_string(self, tokens: List[str]) -> str:
        ids: List[int] = []
        for t in tokens:
            if t in (self._bos_str, self._eos_str, self._unk_str):
                continue
            if self.pad_token is not None and t == self.pad_token:
                continue
            m = self.TOKEN_RE.match(t)
            if m:
                ids.append(int(m.group(1), 16))
        return self._decode_base65536_be(ids)

    def build_inputs_with_special_tokens(
        self,
        token_ids_0: List[int],
        token_ids_1: Optional[List[int]] = None,
    ) -> List[int]:
        if token_ids_1 is None:
            return [self._bos_id] + token_ids_0 + [self._eos_id]
        return [self._bos_id] + token_ids_0 + [self._eos_id] + token_ids_1 + [self._eos_id]

    def get_special_tokens_mask(
        self,
        token_ids_0: List[int],
        token_ids_1: Optional[List[int]] = None,
        already_has_special_tokens: bool = False,
    ) -> List[int]:
        pad_id = self.pad_token_id if self.pad_token is not None else -1

        if already_has_special_tokens:
            return [
                1 if t in (self._bos_id, self._eos_id, self._unk_id, pad_id) else 0
                for t in token_ids_0
            ]

        if token_ids_1 is None:
            return [1] + [0] * len(token_ids_0) + [1]
        return [1] + [0] * len(token_ids_0) + [1] + [0] * len(token_ids_1) + [1]

    def create_token_type_ids_from_sequences(
        self,
        token_ids_0: List[int],
        token_ids_1: Optional[List[int]] = None,
    ) -> List[int]:
        if token_ids_1 is None:
            return [0] * (len(token_ids_0) + 2)
        return [0] * (len(token_ids_0) + len(token_ids_1) + 3)

    def save_vocabulary(self, save_directory: str, filename_prefix: Optional[str] = None) -> Tuple[str]:
        if not os.path.isdir(save_directory):
            os.makedirs(save_directory, exist_ok=True)

        name = (filename_prefix + "-" if filename_prefix else "") + "binaryllm_vocab.json"
        path = os.path.join(save_directory, name)

        data = {
            "base_vocab_size": 65536,
            "vocab_size": 65538,
            "bos_token": self._bos_str,
            "bos_token_id": self._bos_id,
            "eos_token": self._eos_str,
            "eos_token_id": self._eos_id,
            "unk_token": self._unk_str,
            "unk_token_id": self._unk_id,
            "pad_token": self.pad_token,
            "pad_token_id": self.pad_token_id,
            "encoding": "utf-8",
            "packing": "base65536_big_endian_2bytes",
        }
        with open(path, "w", encoding="utf-8") as f:
            json.dump(data, f, ensure_ascii=False, indent=2)

        return (path,)

    # -----------------------------
    # Utilities
    # -----------------------------

    def _id_to_token_base(self, i: int) -> str:
        return f"<U{i:04X}>"

    # -----------------------------
    # Make HF "fast path" consistent
    # -----------------------------

    def _decode(
        self,
        token_ids: List[int],
        skip_special_tokens: bool = False,
        clean_up_tokenization_spaces: bool = False,
        **kwargs: Any,
    ) -> str:
        ids: List[int] = []
        for t in token_ids:
            ti = int(t)
            if skip_special_tokens and ti in (self._bos_id, self._eos_id, self._unk_id):
                continue
            if skip_special_tokens and self.pad_token is not None and ti == self.pad_token_id:
                continue
            if 0 <= ti < self._base_vocab_size:
                ids.append(ti)
            else:
                if not skip_special_tokens:
                    # ignore out-of-range unknowns in decode body; specials are handled above
                    pass
        return self._decode_base65536_be(ids)

    def __call__(
        self,
        text: str,
        add_special_tokens: bool = True,
        return_attention_mask: bool = True,
        return_tensors: Optional[str] = None,
        **kwargs: Any,
    ) -> Dict[str, Any]:
        ids = self._encode_base65536_be(text)

        if add_special_tokens:
            ids = self.build_inputs_with_special_tokens(ids)

        attn = [1] * len(ids) if return_attention_mask else None

        if return_tensors is None:
            out: Dict[str, Any] = {"input_ids": ids}
            if return_attention_mask:
                out["attention_mask"] = attn
            return out

        rt = str(return_tensors).lower().strip()
        if rt != "pt":
            raise ValueError("Only return_tensors='pt' is supported in this tokenizer.")

        input_ids_t = torch.tensor([ids], dtype=torch.long)
        out_t: Dict[str, Any] = {"input_ids": input_ids_t}
        if return_attention_mask:
            out_t["attention_mask"] = torch.tensor([attn], dtype=torch.long)
        return out_t