{-jknddlZddlmZddlmZddlmZmZmZm Z m Z m Z ddl Z ddlmZddlmZdd lmZdd lmZmZdd lmZd d lmZddlmZmZmZddlmZm Z m!Z!m"Z"ej#edGddeZ$dS)N)deepcopy)sleep)AnyDictListOptionalTupleUnion)logging)pipeline_requires_extra)MarkDownBatchSampler) HPIConfigPaddlePredictorOption) benchmark) BasePipeline)DocumentResult LatexResultMarkdownResult)split_original_textssplit_text_recursivetranslate_code_blocktranslate_html_blocktranscAeZdZdZdgZdddddddddedeedeed eeeefd ee d e d ee eeefe fd e ddffdZ dZdeddfdZdeddfdZdLdZ dMde eeeejeejfdee dee dee dee dee dee dee dee dee eefdee d ee eeeefefd!eed"eed#eed$eed%eed&eed'eed(eed)eed*eed+eed,eed-eed.e d/e d0e d1e d2e d3e def@d4Zd5Zd6Z dNd:eed;eded?ed@edAedBedCedDefdEZdFedefdGZdHedefdIZ dJedefdKZ!xZ"S)OPP_DocTranslation_Pipelinez% PP_ DocTranslation_Pipeline zPP-DocTranslationNF)deviceengine engine_config pp_optionuse_hpip hpi_configinitial_predictorconfigr r!r"r#r$r%r&returnc @tjd||||||d| |d|_||_|dd|_d|_d|_|r*||| |t|_ dS)aInitializes the PP-DocTranslation pipeline. Args: config (Dict): Configuration dictionary containing various settings. device (Optional[str], optional): The device to use for prediction. Defaults to `None`. engine (Optional[str], optional): Inference engine. Defaults to `None`. engine_config (Optional[Dict[str, Any]], optional): Engine-specific config. Defaults to `None`. pp_option (Optional[PaddlePredictorOption], optional): Paddle predictor options. Defaults to `None`. use_hpip (bool, optional): Whether to use HPIP. Defaults to `False`. hpi_config (Optional[Union[Dict[str, Any], HPIConfig]], optional): HPIP configuration. Defaults to `None`. initial_predictor (bool, optional): Whether to initialize the predictor. Defaults to `False`. )r r!r"r#r$r% pipeline_nameuse_layout_parserTN) super__init__r*r'getr+layout_parsing_pipelinechat_botinintial_visual_predictorinintial_chat_predictorrmarkdown_batch_sampler) selfr'r r!r"r#r$r%r&kwargs __class__s v/var/www/html/banglarbhumi/venv/lib/python3.11/site-packages/paddlex/inference/pipelines/pp_doctranslation/pipeline.pyr.z#PP_DocTranslation_Pipeline.__init__.s8  '!      $O4 !',?!F!F'+$  1  * *6 2 2 2  ( ( 0 0 0&:&<&<###cJ|j|jdSdS)N)r0close)r5s r8r;z PP_DocTranslation_Pipeline.closeas.  ' 3  ( . . 0 0 0 0 0 4 3r9c|dd|_|jrF|didddi}|||_dS)a Initializes the visual predictor with the given configuration. Args: config (dict): The configuration dictionary containing the necessary parameters for initializing the predictor. Returns: None r+T SubPipelines LayoutParserpipeline_config_errorz)config error for layout_parsing_pipeline!N)r/r+create_pipeliner0)r5r'layout_parsing_configs r8r2z4PP_DocTranslation_Pipeline.inintial_visual_predictoresu"(,?!F!F  ! W$*JJ~r$B$B$F$F(*UV%% !,0+?+?@U+V+VD (r9c6ddlm}|didddi}|||_ddlm}|didid d d i}|||_d S) a Initializes the chat predictor with the given configuration. Args: config (dict): The configuration dictionary containing the necessary parameters for initializing the predictor. Returns: None rcreate_chat_bot SubModulesLLM_Chatchat_bot_config_errorzconfig error for llm chat bot!)create_prompt_engineeringPromptEngneeringTranslate_CommonTextpe_config_errorz%config error for translate_pe_config!N)rDr/r1rH translate_pe)r5r'rDchat_bot_configrHtranslate_pe_configs r8r3z2PP_DocTranslation_Pipeline.inintial_chat_predictorys '&&&&& **\266::  $&F G  (88 000000 JJ|R ( ( S#R ( ( S&"$KL 656IJJr9c.tjddS)NzPP-Translation Pipeline do not support to call `predict()` directly! Please invoke `visual_predict`, `build_vector`, `chat` sequentially to obtain the result.)r error)r5argsr6s r8predictz"PP_DocTranslation_Pipeline.predicts#  m    r9Tinputuse_doc_orientation_classifyuse_doc_unwarpinguse_textline_orientationuse_seal_recognitionuse_table_recognitionuse_formula_recognitionuse_chart_recognitionuse_region_detectionlayout_threshold layout_nmslayout_unclip_ratiolayout_merge_bboxes_modetext_det_limit_side_lentext_det_limit_typetext_det_threshtext_det_box_threshtext_det_unclip_ratiotext_rec_score_threshseal_det_limit_side_lenseal_det_limit_typeseal_det_threshseal_det_box_threshseal_det_unclip_ratioseal_rec_score_thresh#use_wired_table_cells_trans_to_html&use_wireless_table_cells_trans_to_htmluse_table_orientation_classify use_ocr_results_with_table_cellsuse_e2e_wired_table_rec_model use_e2e_wireless_table_rec_modelc +K|jdkrtjdddiV|j.tjd||j|jj|fid|d|d|d |d |d |d |d | d| d| d| d| d|d|d|d|d|d|d|d|d|d|d|d|d|d|d |d!|d"|d#|D] }!d$|!i}"|"V dS)%a This function takes an input image or a list of images and performs various visual prediction tasks such as document orientation classification, document unwarping, general OCR, seal recognition, and table recognition based on the provided flags. Args: input (Union[str, list[str], np.ndarray, list[np.ndarray]]): Input image path, list of image paths, numpy array of an image, or list of numpy arrays. use_doc_orientation_classify (Optional[bool]): Whether to use document orientation classification. use_doc_unwarping (Optional[bool]): Whether to use document unwarping. use_textline_orientation (Optional[bool]): Whether to use textline orientation prediction. use_seal_recognition (Optional[bool]): Whether to use seal recognition. use_table_recognition (Optional[bool]): Whether to use table recognition. use_formula_recognition (Optional[bool]): Whether to use formula recognition. use_region_detection (Optional[bool]): Whether to use region detection. layout_threshold (Optional[float]): The threshold value to filter out low-confidence predictions. Default is None. layout_nms (Optional[bool], optional): Whether to use layout-aware NMS. Defaults to `False`. layout_unclip_ratio (Optional[Union[float, Tuple[float, float]]], optional): The ratio of unclipping the bounding box. Defaults to `None`. If it's a single number, then both width and height are used. If it's a tuple of two numbers, then they are used separately for width and height respectively. If it's None, then no unclipping will be performed. layout_merge_bboxes_mode (Optional[str], optional): The mode for merging bounding boxes. Defaults to `None`. text_det_limit_side_len (Optional[int]): Maximum side length for text detection. text_det_limit_type (Optional[str]): Type of limit to apply for text detection. text_det_thresh (Optional[float]): Threshold for text detection. text_det_box_thresh (Optional[float]): Threshold for text detection boxes. text_det_unclip_ratio (Optional[float]): Ratio for unclipping text detection boxes. text_rec_score_thresh (Optional[float]): Score threshold for text recognition. seal_det_limit_side_len (Optional[int]): Maximum side length for seal detection. seal_det_limit_type (Optional[str]): Type of limit to apply for seal detection. seal_det_thresh (Optional[float]): Threshold for seal detection. seal_det_box_thresh (Optional[float]): Threshold for seal detection boxes. seal_det_unclip_ratio (Optional[float]): Ratio for unclipping seal detection boxes. seal_rec_score_thresh (Optional[float]): Score threshold for seal recognition. use_wired_table_cells_trans_to_html (bool): Whether to use wired table cells trans to HTML. use_wireless_table_cells_trans_to_html (bool): Whether to use wireless table cells trans to HTML. use_table_orientation_classify (bool): Whether to use table orientation classification. use_ocr_results_with_table_cells (bool): Whether to use OCR results processed by table cells. use_e2e_wired_table_rec_model (bool): Whether to use end-to-end wired table recognition model. use_e2e_wireless_table_rec_model (bool): Whether to use end-to-end wireless table recognition model. **kwargs (Any): Additional settings to extend functionality. Returns: dict: A dictionary containing the layout parsing result. Fz1The models for layout parser are not initialized.rQNzGThe layout parsing pipeline is not initialized, will initialize it now.rUrVrWrXrYrZr[r\r]r^r_r`rarbrcrdrerfrjrgrhrirkrlrmrnrorprqrrlayout_parsing_result)r+r rQr0warningr2r'rS)#r5rTrUrVrWrXrYrZr[r\r]r^r_r`rarbrcrdrerfrgrhrirjrkrlrmrnrorprqrrr6rtvisual_predict_ress# r8visual_predictz)PP_DocTranslation_Pipeline.visual_predictsPb  !U * * MM N N NOP P P P  ' / OY     * *4; 7 7 7%IT%A%I  & & & )E)E & 0/ & &>%= & "6!5 & #8"7 & %<$; & #8"7 & "6!5 & .- & "z & !4 3 & &>%= & %<$; & !4 3 & ,O! & "!4 3# & $#8"7% & &#8"7' & (!4 3) & *%<$;+ & ,!4 3- & .,O/ & 0#8"71 & 2#8"73 & 41T0S5 & 64Z3Y7 & 8,J+I9 & :.N-M; & <+H*G= & >.N-M? & % %% % !F()>" % $ $ $ $K% %% %r9cg}|j|D]E}|jd}|jd}|d|dd}|t |F|S)NrTT) input_path page_indexmarkdown_textspage_continuation_flags)r4sample instances input_pathsappendr)r5rTmarkdown_info_listmarkdown_samplemarkdown_contentrz markdown_infos r8load_from_markdownz-PP_DocTranslation_Pipeline.load_from_markdown s#:AA%HH E EO.8; (4Q7J(""2+7 M  % %n]&C&C D D D D!!r9c g}d}tjdt|dtjdt|D]\}}|\}} |dkr|r2|||d}tjd|dzdt|d t | |||t| |kr|d krt|t| z|kr |d | zz }|r[tjd|dzdt|d |||| }Mtjd|dzdt|d |r2|||d}|d kr&|t| |||d ks|d krt| ||| td||r0|||d |S)a Chunks the given markdown blocks into smaller chunks of size `chunk_size` and translates them using the given translate function. Args: md_blocks (list): A list of tuples representing each block of markdown content. Each tuple consists of a string indicating the block type ('text', 'code') and the actual content of the block. chunk_size (int): The maximum size of each chunk. translate_func (callable): A callable that accepts a string argument and returns the translated version of that string. Returns: str: A string containing all the translated chunks concatenated together with newlines between them. rLzSplit the original text into z blockszStarting translation...codezTranslating block r/z...text text_with_htmlhtmlzUnknown block type: ) r infolen enumeratestriprrrr ValueErrorjoin) r5 md_blocks chunk_sizetranslate_functranslation_resultschunkidxblock block_type block_contents r8chunk_translatez*PP_DocTranslation_Pipeline.chunk_translate/s! LS^^LLLMMM .///#I..! J! JJC(- %J V##;;=='..~~ekkmm/L/LMMME M#a%MM#i..MMMNNN$!:~?R]##j00Z65I5Iu::M 2 22Z??Vm33EE{{}}R %U#a%%U%U#i..%U%U%UVVV+22>>%++--3P3PQQQ)EE M#a%MM#i..MMMNNN;;=='..~~ekkmm/L/LMMME'''..,]JWW #333zV7K7K(%z>CV%%HJ%H%HIII ;;== F  & &~~ekkmm'D'D E E E{{.///r9zh ori_md_info_listtarget_languagertask_description output_format rules_strfew_shot_demo_text_contentfew_shot_demo_key_value_listglossaryllm_request_intervalrNc + Kj.tjdj| ddlm} | | njt|tr1|d d |g}t tst  t| ts | Jdd}| O| D]:\}}t|trd |}||d |d z };|dkr|n d z |z  f d }jd }t#|}||kr||z }nt%d|d|d|D]Y}|d}t'|}|||}t+|d|d|d|dVZdS)a Translate the given original text into the specified target language using the configured translation model. Args: ori_md_info_list (List[Dict]): A list of dictionaries containing information about the original markdown text to be translated. target_language (str, optional): The desired target language code. Defaults to "zh". chunk_size (int, optional): The maximum number of characters allowed per chunk when splitting long texts. Defaults to 5000. task_description (str, optional): A description of the task being performed by the translation model. Defaults to `None`. output_format (str, optional): The desired output format of the translation result. Defaults to `None`. rules_str (str, optional): Rules or guidelines for the translation model to follow. Defaults to `None`. few_shot_demo_text_content (str, optional): Demo text content for the translation model. Defaults to `None`. few_shot_demo_key_value_list (str, optional): Demo text key-value list for the translation model. Defaults to `None`. glossary (Dict, optional): A dictionary containing terms and their corresponding definitions. Defaults to `None`. llm_request_interval (float, optional): The interval in seconds between each request to the LLM. Defaults to 0.0. chat_bot_config (Dict, optional): Configuration for the chat bot used in the translation process. Defaults to `None`. **kwargs: Additional keyword arguments passed to the translation model. Yields: MarkdownResult: A dictionary containing the translation result in the target language. Nz;The LLM chat bot is not initialized,will initialize it now.rrCrr{zglossary must be a dictrLu或z:  c T t j| }|dd}d|vrt d|t d|dd}|S) z Translate the given text using the configured translation model. Args: text (str): The text to be translated. Returns: str: The translated text in the target language.  original_textlanguagerrrrr)promptcontentrLz<>zThe translation did not reach the end. This may happen if your chunk_size is too large. Please reduce chunk_size and try again.Nz#The call to the large model failed.)rrMgenerate_promptgenerate_chat_resultsr/ Exceptionreplacerstrip) rr translater1rrrrrr5rrs r8rz.translate_funcs & ' ' '&66"(!1+#+E-I7F!66f6EEII)UWXXI ))o  EFFF!)))R88??AAI r9rz:Chunk size should be greater than the base prompt length (z ), but got .r|rzr})rrzr{r}r|)r1r rur3r'rLrD isinstancelistr/concatenate_markdown_pagesfloatdictitemsrrMrrrrrr)r5rrrrrrrrrrrNr6rD glossary_strkvrbase_prompt_contentbase_prompt_lengthori_mdoriginal_textsrtarget_language_textsr1s` ` ````` ` @r8rz$PP_DocTranslation_Pipeline.translatehsF = OM     ( ( 5 5 5  & * * * * * *&77HH}H ' . . S #'' 55A!% ? ?@P Q QR .66 ?#()=#>#> (D))XX-=-=?X-=-==   (( . .1a&&& 1 A1  - 2  +3/;,,,4,, <,             @#/??$-''A)E@  !!455 * * *#&88JJyM_yylvyyy '  F#$45N,^<>Q  ?@@ @ %.y%9%9 & & !Hk$]3 1 1%e__ +3 <($$Z0000&x0 & & W%%%% &,'l<8     r9latex_info_listc^t|dkrtdg}g}t|D]Y\}}|dD]+}t|}||d<||,|dD]}||Zt |||dddS)a  Concatenate LaTeX content from multiple pages into a single document. Args: latex_info_list (list): A list containing LaTeX data for each page. Returns: tuple: A tuple containing the processed LaTeX document. rz&The length of latex_info_list is zero. latex_blocksr{rrz)rrrz)rrrrrr) r5rr merged_imagesrrrrrs r8concatenate_latex_pagesz2PP_DocTranslation_Pipeline.concatenate_latex_pagesWs   1 $ $EFF F  %.%?%? . . !Hk$^4 1 1%e__ +3 <($$Z0000&x0 . .$$W---- . -'-a0>     r9)r(N)NNNNNNNNNNNNNNNNNNNNNNNNFFTTFT) rrNNNNNNrN)#__name__ __module__ __qualname____doc__entitiesrrstrrrboolr rr.r;rr2r3rSrnpndarrayrr intrwrrrrtuplerrr __classcell__)r7s@r8rr%s$$H !% $2659AE"'1=1=1=1= 1=  1=  S#X/ 1=121=1=U4S>9#<=>1= 1= 1=1=1=1=1=1=f111(dt>8<,037/3042604/39=%)QU2615-1+//3151515-1+//31515497 @% #+4. @% 'tn @% (~@%"*$@% (~@%'tn@%#5#56@%TN@%&eE53F,L&MN@%#+3-@%"*#@% &c]!@%""%#@%$&e_%@%& ('@%( ()@%*"*#+@%,&c]-@%."%/@%0&e_1@%2 (3@%4 (5@%6.27@%8159@%:)-;@%<+/=@%>(,?@%@+/A@%D E@%@%@%@%D " " "707070x $ $!*.,0&) $LLt*LL L  L  LL%(L'*LL$LLLLL\?2?2?2?2?2?2B      @ t         r9r)%rcopyrtimertypingrrrrr r numpyrutilsr utils.depsr common.batch_samplerrmodelsrrutils.benchmarkrbaserresultrrrrrrr time_methodsrr,r9r8rs ::::::::::::::::22222288888866666666((((((?????????? !!O  O  O  O  O  O  O  "!O  O  O  r9