j9 lUddlZddlZddlmZddlZddlZddlZddlm Z ddl m Z m Z dej zejzejzZe jZdZdZedZe jZe jZe jZe jZe jZ d Z!e"ej#d d d z Z$ee%j&d z Z'da(ej)dze*d<da+e,e*d<dZ-de.de/fdZ0e$fdZ1dZ2d.dZ3d.dZ4de5fdZ6d/de.fdZ7e8dkr8 ddlZddl9Z9ddl:Z:dZ;gZe9j?dZ@eAd e@jBe@D]]ZCe:jDZEe7eCe=!ZFe;e:jDeEz z Z;eFd"r"ee@jBd+zd,ZJejKez+predict_ocr_probability..5s66618A;666r)dtypeNr)rr)rnparrayrfloat32runfloat)rsession input_namexprobass` rpredict_ocr_probabilityr02sm&..GZ 6666 6667rzJJJA [[ A / / 2F   rcJ|}t|D] \}}tj|d}|jrtj|d}|drOtj|d} tj||}tj|d}n #d}YnxYwtjtj|}tj|dd}t |g|d|}t |} t|| }||krd||fcSd||fS) aiStage 2: Separate OCR check for large images. Currently not in use. Args: image_blocks: (list) image blocks in page.get_text("dict") format. prob: (float) OCR probability for rendered full page threshold: (float) probability threshold. If larger recommend OCR. Returns: bool (do OCR), probability, check count imagermaskNbboxTF) enumeratepymupdfPixmapalphacsGRAYrr0max) image_blocksprob threshold best_probiblockpixr3rimg_probs r check_imagesrD:s;Il++&&5nU7^,, 9 ).a((C = >%-00D nS$//nS!,, nW^S11nS#s++$UGU6]C@@*844 8,,  ! !A% % % % " )Q s 1*BB"c|dtkp3|ddkp'|dtzdko|dtzdkS)zIf this is an OCR text span.fontr9r char_flags)TESSERACT_FONT_NAME TEXT_STROKED TEXT_FILLED)spans r is_ocr_spanrLdsW V ++ =A     - 2 6\"[0A5 rFct|d|dt|d|dt|d|dt|d|df}|r|n tj|SSpeedy version using indices.rr)r;minr7Rectr1r2 bbox_onlyr5s rintersect_rectsrXpq 1r!u  s2a5"Q%00#beRU2C2CSAPRSTPUEVEV WD 544',"55rct|d|dt|d|dt|d|dt|d|df}|r|n tj|SrN)rRr;r7rSrTs r join_rectsr[vrYrcJ|d|dkp|d|dkS)rOrrPrrQr")r5s r bbox_is_emptyr]|s' 7d1g  3aDG!33rc |4|dttjd}|j}tj}| }| }d} d} d} d} d} d}d}d}g}g}g}g}|D]}t ||d}|j|jz}|s+|dtkr(|d D]}|d D]}t ||d}|j|jz}|s+|d  }|r| r\| t|z } t|r!|d z }||dtd |D}| |z } t!||}| |z } |r!| |z } ||d||de|dt"krt!||}| |z } |dt$kr$t!||}||z }|ds|d z }||z|z}t'|r2id|ddddddddddddddddddddddddddddddS|d |dz |d!|d z z}|rt|dkr|nd}||rt)||z nd|r| |z nd|rt)||z nd|r| |z nd|rt)||z nd|r||z nd| | |r| |z nd|d|d" } |rt+|t,r|d#dd z|d#<|s i| dddd$S|D]}!||!|tjtjtj%i| d&ddd$S| r(| r&| | z t:kr| | z t:kr i| d&ddd$S|s | s i| dddd$St+|t,r|d'dd z|d'<t=|||}"t+|t,r)|d(rt?j |"tC|"}#|#tDk}$|$r i| |$d)|#d$S|rKtG||#tD\}$}#}%t+|t,r|d*d|%z|d*<i| |$|$rd)nd|#d$S)+aAnalyze the page for the OCR decision. Args: blocks: (dict) output of page.get_text("dict") if already available replace_ocr: (bool) if True, we should make a new OCR text layer stats: (dict) if given fill in execution information (debugging) Returns: A dict with analysis results. The area-related float values are computed as fractions of the total covered area. "covered": Rect, page area covered by content "img_joins": float, fraction of area of the joined images "img_area": float, fraction of sum of image area sizes "txt_joins": float, fraction of area of the joined text spans "txt_area": float, fraction of sum of text span bbox area sizes "vec_joins": float, fraction of area of the joined vector characters "vec_area": float, fraction of sum of vectors "vec_norects": int, count of vectors with isrect=False (not in a rectangle) "chars_total": int, count of visible characters "chars_bad": int, count of Replacement Unicode characters "bad_areas": float, fraction of text areas having bad characters "ocr_spans": int, count: text spans with ignored text (render mode 3) "pixmap": Pixmap of page with all "good" text removed - for OCR use "needs_ocr": bool, final decision "reason": str | None, reason for the OCR decision "probability": float | None, OCR probability Ndict)flagsclipblocksrgr5typelinesspanstextrc30K|]}|tk dVdS)rN)REPLACEMENT_CHARACTER)r#cs r zanalyze_page..s-#R#R!q~ &((      I!##HyHyHKIIHHHKINNOL 66y!F)44zDK'   V9 " "wZ : :7::A(qy99B h2G"! V9??,,D!4<<>>! 3t99,K"1~~!!Q &--ai888 !$#R#Rt#R#R#R R RI*I)(B77H'H :!W, &--ai8888'..qy99995:6  V9 # #!(D11H  H  V9 $ $!(D11H  HX; !q  %!H,GW  w              1  1        d    d 4!  &!*wqz)gaj71:.EFJ    A % %   5?Hc(mmj00S-7@Hz))S5?Hc(mmj00S-7@Hz))S5?Hc(mmj00S-7@Hz))S"/9BY++s"H$  eT " " ;$yyA66:E)  YXhXUdSWXXX X  % %A  ! !! $ $ $ $ 05/      !            #&8 8 8  #5 5 5  !       x        %?$yy::Q>m 488H%599_#=#= ( "8 , ,D++I   "      K'3 $ 3( ( $ 4 eT " " K"'))L!"<"<{"JE,     )3**t    r__main__rTzOCR-Analysis for doc.name=)rrzr{zNEW: duration: z.2fz s, per page: z.4fzs OCR: z pages (z%)z.txtw)rstream)F)NFr^N)Mosrpathlibrnumpyr' onnxruntimerr7rcompute_ocr_featuresrrTEXT_ACCURATE_BBOXESTEXT_PRESERVE_IMAGESTEXT_COLLECT_VECTORSrfz_device_grayGRAYTYPE3_FONT_NAMErHchrriFZ_STEXT_STROKEDrIFZ_STEXT_FILLEDrJFZ_STEXT_BLOCK_TEXTrFZ_STEXT_BLOCK_IMAGErFZ_STEXT_BLOCK_VECTORrrintgetenvr__file__parentrr r__annotations__rrrr`r+r0rDrLrXr[boolr]r__name__systimeduration ocr_pages STATISTICSopenargvdocprintrr perf_countert0rrnumberroutrcloser"rrrs9 AAAAAAAA "# "#"# u%F % #  & ( * c)")$A2FFGG#Md8nn#&?? (,#  %,,, S!!!du0C''''T   6666 6666 444444 yySWyyyyx z!MMMJJJKKKHI!4(J ',sx{ # #C E )ch ) )***@@ T  <J777%D%''",, K  @   dk8H+=> ? ? ? E O( O O OXcc#hh5F O O OSVSVW`SaSa O Oknknoxkykyz}z}B{C{ClCDGlG O O O O $sx& # & &C FIis3////IIKKK FIj/r