o ZhB@sddlZddlZddlZddlZddlmZddlmZddlm Z m Z m Z m Z m Z mZddlZddlmZddlmZddlmZeeZd d d d d dZdZeddGdddeZdede ee fdejfddZdeddfddZdedee e ffddZ dgZ!dS)N)Path)copyfile)AnyDictListOptionalTupleUnion)PreTrainedTokenizer)logging)requiresz source.spmz target.spmz vocab.jsonztarget_vocab.jsonztokenizer_config.json) source_spm target_spmvocabtarget_vocab_fileZtokenizer_config_fileu▁) sentencepiece)backendsc seZdZdZeZddgZedZ        dEd e e e e fd dffd d ZddZde d e fddZddZde fddZde d ee fddZded e fddZfddZfd d!Zd"ee d e fd#d$ZdFd eefd%d&Zd'd(Zd)d*Zed efd+d,ZdFd-e d.e e d ee fd/d0Z d e fd1d2Z!d3d4Z"d5d6Z#d e fd7d8Z$d9e d dfd:d;Z%dd?Z' dGd@edAe edBe(d eefdCdDZ)Z*S)HMarianTokenizeraB Construct a Marian tokenizer. Based on [SentencePiece](https://github.com/google/sentencepiece). This tokenizer inherits from [`PreTrainedTokenizer`] which contains most of the main methods. Users should refer to this superclass for more information regarding those methods. Args: source_spm (`str`): [SentencePiece](https://github.com/google/sentencepiece) file (generally has a .spm extension) that contains the vocabulary for the source language. target_spm (`str`): [SentencePiece](https://github.com/google/sentencepiece) file (generally has a .spm extension) that contains the vocabulary for the target language. source_lang (`str`, *optional*): A string representing the source language. target_lang (`str`, *optional*): A string representing the target language. unk_token (`str`, *optional*, defaults to `""`): The unknown token. A token that is not in the vocabulary cannot be converted to an ID and is set to be this token instead. eos_token (`str`, *optional*, defaults to `""`): The end of sequence token. pad_token (`str`, *optional*, defaults to `""`): The token used for padding, for example when batching sequences of different lengths. model_max_length (`int`, *optional*, defaults to 512): The maximum sentence length the model accepts. additional_special_tokens (`List[str]`, *optional*, defaults to `["", ""]`): Additional special tokens used by the tokenizer. sp_model_kwargs (`dict`, *optional*): Will be passed to the `SentencePieceProcessor.__init__()` method. The [Python wrapper for SentencePiece](https://github.com/google/sentencepiece/tree/master/python) can be used, among other things, to set: - `enable_sampling`: Enable subword regularization. - `nbest_size`: Sampling parameters for unigram. Invalid for BPE-Dropout. - `nbest_size = {0,1}`: No sampling is performed. - `nbest_size > 1`: samples from the nbest_size results. - `nbest_size < 0`: assuming that nbest_size is infinite and samples from the all hypothesis (lattice) using forward-filtering-and-backward-sampling algorithm. - `alpha`: Smoothing parameter for unigram sampling, and dropout probability of merge operations for BPE-dropout. Examples: ```python >>> from transformers import MarianForCausalLM, MarianTokenizer >>> model = MarianForCausalLM.from_pretrained("Helsinki-NLP/opus-mt-en-de") >>> tokenizer = MarianTokenizer.from_pretrained("Helsinki-NLP/opus-mt-en-de") >>> src_texts = ["I am a small frog.", "Tom asked his teacher for advice."] >>> tgt_texts = ["Ich bin ein kleiner Frosch.", "Tom bat seinen Lehrer um Rat."] # optional >>> inputs = tokenizer(src_texts, text_target=tgt_texts, return_tensors="pt", padding=True) >>> outputs = model(**inputs) # should work ```Z input_idsZattention_maskz>>.+<Fsp_model_kwargsreturnc  s.| durin| |_t|sJd|| |_t||_t||jvr)tdt| |jvs2J| rHt||_dd|j D|_ g|_ ndd|j D|_ dd|jD|_ ||_ ||_ ||g|_t||j|_t||j|_|j|_|j|_|tjd ||||| | |j|| d | dS) Nzcannot find spm source z token must be in the vocabcSi|]\}}||qSr.0kvrr]/var/www/auris/lib/python3.10/site-packages/transformers/models/marian/tokenization_marian.py z,MarianTokenizer.__init__..cSrrrrrrr!r"r#cSs$g|]}|dr|dr|qS)z>>z<<) startswithendswith)rrrrr! s$z,MarianTokenizer.__init__..) source_lang target_lang unk_token eos_token pad_tokenmodel_max_lengthrrseparate_vocabsr)rrexistsr- load_jsonencoderstrKeyErrortarget_encoderitemsdecoderZsupported_language_codesr'r( spm_filesload_spm spm_source spm_target current_spmcurrent_encoder_setup_normalizersuper__init__)selfrrrrr'r(r)r*r+r,rr-kwargs __class__rr!r>msD    zMarianTokenizer.__init__c CsNzddlm}||jj|_WdSttfy&tddd|_YdSw)Nr)MosesPunctNormalizerz$Recommended: pip install sacremoses.cSs|SNr)xrrr!sz3MarianTokenizer._setup_normalizer..) Z sacremosesrCr' normalizepunc_normalizer ImportErrorFileNotFoundErrorwarningswarn)r?rCrrr!r<s  z!MarianTokenizer._setup_normalizerrEcCs|r||SdS)zHCover moses empty string edge case. They return empty list for '' input!)rH)r?rErrr!rGszMarianTokenizer.normalizecCs|j||j|jSrD)r;getr))r?tokenrrr!_convert_token_to_idsz$MarianTokenizer._convert_token_to_idtextcCs2|j|}|r|dgng}||jd|fS)z6Remove language codes like >>fr<< before sentencepiecerrM)language_code_rematchgroupsub)r?rQrScoderrr!remove_language_codes z$MarianTokenizer.remove_language_codecCs&||\}}|jj|td}||S)N)Zout_type)rWr:encoder1)r?rQrVpiecesrrr! _tokenizeszMarianTokenizer._tokenizeindexcCs|j||jS)z?Converts an index (integer) in a token (str) using the decoder.)r5rNr))r?r[rrr!_convert_id_to_tokensz$MarianTokenizer._convert_id_to_tokenc tj|fi|S)ao Convert a list of lists of token ids into a list of strings by calling decode. Args: sequences (`Union[List[int], List[List[int]], np.ndarray, torch.Tensor, tf.Tensor]`): List of tokenized input ids. Can be obtained using the `__call__` method. skip_special_tokens (`bool`, *optional*, defaults to `False`): Whether or not to remove special tokens in the decoding. clean_up_tokenization_spaces (`bool`, *optional*): Whether or not to clean up the tokenization spaces. If `None`, will default to `self.clean_up_tokenization_spaces` (available in the `tokenizer_config`). use_source_tokenizer (`bool`, *optional*, defaults to `False`): Whether or not to use the source tokenizer to decode sequences (only applicable in sequence-to-sequence problems). kwargs (additional keyword arguments, *optional*): Will be passed to the underlying model specific decode method. Returns: `List[str]`: The list of decoded sentences. )r= batch_decode)r? sequencesr@rArr!r^szMarianTokenizer.batch_decodec r])a Converts a sequence of ids in a string, using the tokenizer and vocabulary with options to remove special tokens and clean up tokenization spaces. Similar to doing `self.convert_tokens_to_string(self.convert_ids_to_tokens(token_ids))`. Args: token_ids (`Union[int, List[int], np.ndarray, torch.Tensor, tf.Tensor]`): List of tokenized input ids. Can be obtained using the `__call__` method. skip_special_tokens (`bool`, *optional*, defaults to `False`): Whether or not to remove special tokens in the decoding. clean_up_tokenization_spaces (`bool`, *optional*): Whether or not to clean up the tokenization spaces. If `None`, will default to `self.clean_up_tokenization_spaces` (available in the `tokenizer_config`). use_source_tokenizer (`bool`, *optional*, defaults to `False`): Whether or not to use the source tokenizer to decode sequences (only applicable in sequence-to-sequence problems). kwargs (additional keyword arguments, *optional*): Will be passed to the underlying model specific decode method. Returns: `str`: The decoded sentence. )r=decode)r?Z token_idsr@rArr!r`szMarianTokenizer.decodetokenscCsv|jr|jn|j}g}d}|D]}||jvr$||||d7}g}q||q|||7}|td}|S)zQUses source spm if _decode_use_source_tokenizer is True, and target spm otherwiserM ) Z_decode_use_source_tokenizerr8r9Zall_special_tokensZ decode_piecesappendreplaceSPIECE_UNDERLINEstrip)r?raZsp_modelZcurrent_sub_tokensZ out_stringrOrrr!convert_tokens_to_strings   z(MarianTokenizer.convert_tokens_to_stringcCs$|dur ||jgS|||jgS)z=Build model inputs from a sequence by appending eos_token_id.N)Z eos_token_id)r? token_ids_0 token_ids_1rrr! build_inputs_with_special_tokens s z0MarianTokenizer.build_inputs_with_special_tokenscCs|j|_|j|_dSrD)r8r:r0r;r?rrr!_switch_to_input_modes z%MarianTokenizer._switch_to_input_modecCs|j|_|jr |j|_dSdSrD)r9r:r-r3r;rkrrr!_switch_to_target_modes z&MarianTokenizer._switch_to_target_modecCs t|jSrD)lenr0rkrrr! vocab_sizes zMarianTokenizer.vocab_sizesave_directoryfilename_prefixc Cstj|std|ddSg}|jrOtj||r |dndtd}tj||r1|dndtd}t|j |t|j || || |ntj||rY|dndtd}t|j || |t tdtdg|j |j|jgD]Z\}}} tj||r|dnd|} tj|tj| krtj|rt|| | | q|tj|st| d } | } | | Wdn1swY| | q|t|S) NzVocabulary path (z) should be a directory-rMrrrrwb)ospathisdirloggererrorr-joinVOCAB_FILES_NAMES save_jsonr0r3rczipr6r8r9abspathisfileropenZserialized_model_protowritetuple) r?rprqZ saved_filesZout_src_vocab_fileZout_tgt_vocab_fileZout_vocab_fileZspm_save_filenameZ spm_orig_pathZ spm_modelZ spm_save_pathfiZcontent_spiece_modelrrr!save_vocabularysR        $      zMarianTokenizer.save_vocabularycCs|SrD) get_src_vocabrkrrr! get_vocabLszMarianTokenizer.get_vocabcCt|jfi|jSrD)dictr0Zadded_tokens_encoderrkrrr!rOzMarianTokenizer.get_src_vocabcCrrD)rr3Zadded_tokens_decoderrkrrr! get_tgt_vocabRrzMarianTokenizer.get_tgt_vocabcCs"|j}|tgd|S)N)r8r9r:rHr)__dict__copyupdaterfromkeys)r?staterrr! __getstate__Us  zMarianTokenizer.__getstate__dcsF|_tds i_fddjD\__j_dS)Nrc3s|] }t|jVqdSrD)r7r)rfrkrr! csz/MarianTokenizer.__setstate__..)rhasattrrr6r8r9r:r<)r?rrrkr! __setstate__\s   zMarianTokenizer.__setstate__cOsdS)zJust EOSr)r?argsr@rrr!num_special_tokens_to_addgsz)MarianTokenizer.num_special_tokens_to_addcs(t|j|jfdd|DS)Ncsg|] }|vr dndqS)rrr)rrEall_special_idsrr!r&nsz7MarianTokenizer._special_token_mask..)setrremoveZ unk_token_id)r?seqrrr!_special_token_maskks  z#MarianTokenizer._special_token_maskrhrialready_has_special_tokenscCs:|r||S|dur||dgS|||dgS)zCGet list where entries are [1] if a token is [eos] or [pad] else 0.Nr)r)r?rhrirrrr!get_special_tokens_maskps  z'MarianTokenizer.get_special_tokens_mask) NNNrrrrNFrD)NF)+__name__ __module__ __qualname____doc__rzZvocab_files_namesZmodel_input_namesrecompilerRrrr1rr>r<rGrPrWrrZintr\r^r`rgrjrlrmpropertyrorrrrrrrrrboolr __classcell__rrrAr!r-sf:  >    - rrurrcCstjdi|}|||S)Nr)rSentencePieceProcessorLoad)rurZspmrrr!r7|s r7cCs@t|d}tj||ddWddS1swYdS)Nw)indent)rjsondump)datarurrrr!r{s "r{cCs8t|d }t|WdS1swYdS)Nr)rrload)rurrrr!r/s $r/)"rrtrrKpathlibrshutilrtypingrrrrrr rZtokenization_utilsr utilsr Zutils.import_utilsr Z get_loggerrrwrzrerr1rr7r{r/__all__rrrr!s6          P