a hA@sddlZddlZddlZddlmZddlmZddlmZm Z m Z ddl Z ddl m Z ddlmZddlmZeeZd d d d d dZdZeddGddde Zeeeefe jdddZeddddZee eefdddZdgZdS)N)Path)copyfile)AnyOptionalUnion)PreTrainedTokenizer)logging)requiresz source.spmz target.spmz vocab.jsonztarget_vocab.jsonztokenizer_config.json) source_spm target_spmvocabtarget_vocab_fileZtokenizer_config_fileu▁) sentencepiece)backendsc s|eZdZdZeZddgZdBeee e fdd fd d Z d dZ e e dddZ ddZe dddZe ee dddZee dddZfddZfdd Zee e d!d"d#ZdCeed$d%d&Zd'd(Zd)d*Zeed$d+d,ZdDe ee ee d-d.d/Zed$d0d1Zd2d3Zd4d5Z ed$d6d7Z!edd8d9d:Z"d;d<Z#d=d>Z$dEeeee%eed?d@dAZ&Z'S)FMarianTokenizeraB 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_maskNF)sp_model_kwargsreturnc  s.| dur in| |_t|s,Jd|| |_t||_t||jvrRtdt| |jvsdJ| rt||_dd|j D|_ g|_ n(dd|j D|_ dd|jD|_ ||_ ||_ ||g|_t||j|_t||j|_|j|_|j|_|tjf||||| | |j|| d | dS) Nzcannot find spm source z token must be in the vocabcSsi|]\}}||qSr.0kvrrj/var/www/html/assistant/venv/lib/python3.9/site-packages/transformers/models/marian/tokenization_marian.py z,MarianTokenizer.__init__..cSsi|]\}}||qSrrrrrrrrcSs$g|]}|dr|dr|qS)>><<) startswithendswith)rrrrr rz,MarianTokenizer.__init__..) source_lang target_lang unk_token eos_token pad_tokenmodel_max_lengthrrseparate_vocabs)rrexistsr+ load_jsonencoderstrKeyErrortarget_encoderitemsdecoderZsupported_language_codesr%r& spm_filesload_spm spm_source spm_target current_spmcurrent_encoder_setup_normalizersuper__init__)selfr r r rr%r&r'r(r)r*rr+kwargs __class__rrr<ksD    zMarianTokenizer.__init__c CsNzddlm}||jj|_Wn*ttfyHtddd|_Yn0dS)Nr)MosesPunctNormalizerz$Recommended: pip install sacremoses.cSs|SNr)xrrrrz3MarianTokenizer._setup_normalizer..) Z sacremosesrAr% normalizepunc_normalizer ImportErrorFileNotFoundErrorwarningswarn)r=rArrrr:s   z!MarianTokenizer._setup_normalizer)rCrcCs|r||SdS)zHCover moses empty string edge case. They return empty list for '' input!)rF)r=rCrrrrEszMarianTokenizer.normalizecCs|j||j|jSrB)r9getr')r=tokenrrr_convert_token_to_idsz$MarianTokenizer._convert_token_to_id)textcCsNg}|drF|d}dkrF||d|d||dd}||fS)z6Remove language codes like >>fr<< before sentencepiecer r!N)r"findappend)r=rOcodeZend_locrrrremove_language_codes z$MarianTokenizer.remove_language_code)rOrcCs&||\}}|jj|td}||S)N)Zout_type)rUr8encoder/)r=rOrTpiecesrrr _tokenizeszMarianTokenizer._tokenize)indexrcCs|j||jS)z?Converts an index (integer) in a token (str) using the decoder.)r3rLr')r=rYrrr_convert_id_to_tokensz$MarianTokenizer._convert_id_to_tokenc stj|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>r?rrr[szMarianTokenizer.batch_decodec stj|fi|S)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>r?rrr]szMarianTokenizer.decode)tokensrcCsv|jr |jn|j}g}d}|D]4}||jvrH||||d7}g}q||q|||7}|td}|S)zQUses source spm if _decode_use_source_tokenizer is True, and target spm otherwiserK ) Z_decode_use_source_tokenizerr6r7Zall_special_tokensZ decode_piecesrSreplaceSPIECE_UNDERLINEstrip)r=r^Zsp_modelZcurrent_sub_tokensZ out_stringrMrrrconvert_tokens_to_strings   z(MarianTokenizer.convert_tokens_to_string)rcCs$|dur||jgS|||jgS)z=Build model inputs from a sequence by appending eos_token_id.N)Z eos_token_id)r= token_ids_0 token_ids_1rrr build_inputs_with_special_tokens s z0MarianTokenizer.build_inputs_with_special_tokenscCs|j|_|j|_dSrB)r6r8r.r9r=rrr_switch_to_input_modesz%MarianTokenizer._switch_to_input_modecCs|j|_|jr|j|_dSrB)r7r8r+r1r9rgrrr_switch_to_target_modesz&MarianTokenizer._switch_to_target_modecCs t|jSrB)lenr.rgrrr vocab_sizeszMarianTokenizer.vocab_size)save_directoryfilename_prefixrc Cstj|s"td|ddSg}|jrtj||r@|dndtd}tj||rb|dndtd}t|j |t|j || || |n8tj||r|dndtd}t|j || |t tdtdg|j |j|jgD]\}}} tj||r|dnd|} tj|tj| kr`tj|r`t|| | | qtj|st| d "} | } | | Wdn1s0Y| | qt|S) NzVocabulary path (z) should be a directory-rKr rr r wb)ospathisdirloggererrorr+joinVOCAB_FILES_NAMES save_jsonr.r1rSzipr4r6r7abspathisfileropenZserialized_model_protowritetuple) r=rlrmZ saved_filesZout_src_vocab_fileZout_tgt_vocab_fileZout_vocab_fileZspm_save_filenameZ spm_orig_pathZ spm_modelZ spm_save_pathfiZcontent_spiece_modelrrrsave_vocabularysN        (    * zMarianTokenizer.save_vocabularycCs|SrB) get_src_vocabrgrrr get_vocabLszMarianTokenizer.get_vocabcCst|jfi|jSrB)dictr.Zadded_tokens_encoderrgrrrrOszMarianTokenizer.get_src_vocabcCst|jfi|jSrB)rr1Zadded_tokens_decoderrgrrr get_tgt_vocabRszMarianTokenizer.get_tgt_vocabcCs"|j}|tgd|S)N)r6r7r8rFr)__dict__copyupdaterfromkeys)r=staterrr __getstate__Us   zMarianTokenizer.__getstate__)drcsF|_tdsi_fddjD\__j_dS)Nrc3s|]}t|jVqdSrB)r5r)rfrgrr crz/MarianTokenizer.__setstate__..)rhasattrrr4r6r7r8r:)r=rrrgr __setstate__\s  zMarianTokenizer.__setstate__cOsdS)zJust EOSr)r=argsr>rrrnum_special_tokens_to_addgsz)MarianTokenizer.num_special_tokens_to_addcs(t|j|jfdd|DS)Ncsg|]}|vrdndqS)rrr)rrCall_special_idsrrr$nrz7MarianTokenizer._special_token_mask..)setrremoveZ unk_token_id)r=seqrrr_special_token_maskks  z#MarianTokenizer._special_token_mask)rdrealready_has_special_tokensrcCs>|r||S|dur&||dgS|||dgSdS)zCGet list where entries are [1] if a token is [eos] or [pad] else 0.Nr)r)r=rdrerrrrget_special_tokens_maskps  z'MarianTokenizer.get_special_tokens_mask) NNNrrrrNF)N)N)NF)(__name__ __module__ __qualname____doc__rvZvocab_files_namesZmodel_input_namesrrr/rr<r:rErNrUlistrXintrZr[r]rcrfrhripropertyrkr}rrrrrrrrboolr __classcell__rrr?rr,sV: >   -  r)rqrrcCstjfi|}|||SrB)rSentencePieceProcessorLoad)rqrZspmrrrr5|s r5)rqrcCs>t|d }tj||ddWdn1s00YdS)NwrQ)indent)r{jsondump)datarqrrrrrws rwcCs8t|d}t|WdS1s*0YdS)Nr)r{rload)rqrrrrr-s r-) rrprIpathlibrshutilrtypingrrrrZtokenization_utilsrutilsr Zutils.import_utilsr Z get_loggerrrsrvrarr/rrr5rwrr-__all__rrrrs2       Q