TAiNXSrSSKJr SSKrSSKrSSKrSSKrSSKJr SSK J r SSK J r J r SSKJr SSKJr SS KJrJrJr SS KJrJrJrJrJrJrJrJrJrJ r SS K!J"r# SS K!J$r$ \RJ"\&5r'S r(\RRr)\"SS55r*"SS\+5r,"SS5r-g)zhOCR transform implementation.) annotationsN) dataclass)pairwise)atanpi)Path) ElementTree)MatrixName Rectangle) BLACKBLUECYAN DARKGREENGREENMAGENTAREDCanvasText TextDirection) EncodableFont) GlyphlessFontgR@cn\rSrSr%SrSrS\S'SrS\S'SrS\S'Sr S\S'Sr S\S 'Sr S\S 'S r g ) DebugRenderOptions-z'A class for managing rendering options.Fboolrender_paragraph_bboxrender_baselinerender_trianglerender_line_bboxrender_word_bboxrender_space_bboxN) __name__ __module__ __qualname____firstlineno____doc__r__annotations__rrr r!r"__static_attributes__r#X/var/www/html/land-ocr/venv/lib/python3.13/site-packages/ocrmypdf/hocrtransform/_hocr.pyrr-sE1"'4'!OT!!OT!"d""d"#t#r+rc\rSrSrSrSrg)HocrTransformError9z$Error while applying hOCR transform.r#N)r$r%r&r'r(r*r#r+r,r.r.9s.r+r.c|\rSrSrSr\R "S\R5r\R "S\R5r \R "S\R5r S\ "S5\ "5SS .S"S jjr S#S jr\S$S j5r\S%S j5r\S&Sj5rS'S(Sjjr\S)Sj5rSSS.S*SjjrSrSr\S5rS+SjrS,Sjr\4S-Sjjr\4S.Sjjr\S4S/Sjjr \!S4S/Sjjr"\#S4S/Sjjr$\%S4S.S jjr&S!r'g)0 HocrTransform=zA class for converting documents from the hOCR format. For details of the hOCR format, see: http://kba.github.io/hocr-spec/1.2/. z bbox \s+ (\d+) \s+ # left: uint (\d+) \s+ # top: uint (\d+) \s+ # right: uint (\d+) # bottom: uint z| baseline \s+ ([\-\+]?\d*\.?\d*) \s+ # +/- decimal float ([\-\+]?\d+) # +/- int zO textangle \s+ ([\-\+]?\d*\.?\d*) # +/- decimal float Fz/f-0-0N)debugfontnamefontdebug_render_optionsc LU(a.[RS[5 [UUSUSSS9UlOU=(d [5UlX l[ R"[R"U55Ul X@l XPl [R"SURR5R 5nSUlU(aUR%S5UlURR'UR)SS55HnUR+U5n U (d [-S 5eU R.U R0- UR [2- - UlU R6U R8- UR [2- - Ul g g ) z$Initialize the HocrTransform object.z Use debug_render_options insteadF)rrr r!rr"z ({.*})htmldivocr_pagez$hocr file is missing page dimensionsN)logwarningDeprecationWarningrrender_optionsdpir parseosfspathhocr _fontname_fontrematchgetroottagxmlnsgroupfindall _child_xpathelement_coordinatesr.urxllxINCHwidthuryllyheight) self hocr_filenamer@r3r4r5r6matchesr:coordss r,__init__HocrTransform.__init__^s=  KK:? ! ((=$))*;*;*=*A*AB  q)DJ99$$T%6%6uj%IJC--c2F()OPP **vzz1dhhoFDJ!:: 2txx$GDK Kr+cURb UROSnUHnX RU5- nM X!RbUR- nU$S- nU$)z;Return the textual content of the element and its children.r8)text_get_element_texttail)rWelementr^childs r,r_HocrTransform._get_element_texts]&||7w||RE **51 1D 8 @ ?A@ r+c XURRURRSS55nU(dg[ [ UR S55[ UR S55[ UR S55[ UR S555$)z6Get coordinates of the bounding box around an element.titler8Nr9) box_patternsearchattribgetr floatrLclsrarYs r,rO!HocrTransform.element_coordinatess//((););GR)HI '--" # '--" # '--" # '--" #   r+cURRURRSS55nU(dg[ UR S55[ UR S554$)z#Get baseline's slope and intercept.rer8)rrr9rf)baseline_patternrjrkrlrmrLintrns r,baselineHocrTransform.baselinesX&&--gnn.@.@".MNW]]1%&GMM!,<(===r+cURRURRSS55nU(dg[ UR S55$)zGet text angle of an element.rer8rrr9)textangle_patternrjrkrlrmrLrns r, textangleHocrTransform.textanglesE''..w~~/A/A'2/NOW]]1%&&r+cHSURU3nU(a USUS3- nU$)Nz.//z [@class='z'])rK)rWhtml_tag html_classxpaths r,rNHocrTransform._child_xpaths1djj\(,  y B/ /E r+c0[R"SU5$)z;Normalize the given text using the NFKC normalization form.NFKC) unicodedata normalize)ross r,normalize_textHocrTransform.normalize_texts$$VQ//r+T)image_filenameinvisible_textc [URUR4S9nURURUR 5 [ 5RSUR5RSS5R[UR- [UR- 5n[RU5 URRUS9 URU5 SnUR R#UR%SS55HknS UR#UR%S 555H=nS nUR'U5n UR)U5n UR+UUS UU U 5 M? Mm U(dRUR R-UR%S S55n UR'U 5n UR+UU S UU S 5 SSS5 Ub1URR/USSURURS9 UR15R3U5 g!,(df  Nb=f)apCreates a PDF file with an image superimposed on top of the text. Text is positioned according to the bounding box of the lines in the hOCR file. The image need not be identical to the image used to create the hOCR file. It can have a lower resolution, different color mode, etc. Arguments: out_filename: Path of PDF to write. image_filename: Image to use for this file. If omitted, the OCR text is shown. invisible_text: If True, text is rendered invisible so that is selectable but never drawn. If False, text is visible and may be seen if the image is skipped or deleted in Acrobat. ) page_sizerr9cmFpocr_parc3t# UH.nSUR;dMURSS;dM*Uv M0 g7f)class>ocr_line ocr_header ocr_caption ocr_textfloatN)rk).0ras r, 'HocrTransform.to_pdf..sA#J'..0 w/QRG#Js 88 8spanT ocrx_wordr:r;N)rSrV)rrSrVadd_fontrErFr translatedscaledrRr@r<r3do save_state_debug_draw_paragraph_boxesrDiterfindrN_get_text_direction_get_inject_word_breaks_do_linefind draw_imageto_pdfsave) rW out_filenamerrcanvas page_matrix found_linesparline directioninject_word_breaksroots r,rHocrTransform.to_pdfs24::t{{";< 3 H Z4;; ' VAr] VD488OTDHH_ 5  + YY ! ![ ! 1  , ,V 4Kyy))$*;*;C*KL#&<<0A0A&0I#JD#'K $ 8 8 =I)-)E)Ec)J&MM#&!* M(yy~~d&7&7z&JK 44T:  " 72H  % II 1DJJt{{ !   \*U2 1s DH== I cUc[R$URRSS5S:Xa[R$[R$)zGet the text direction of the paragraph. Arabic, Hebrew, Persian, are right-to-left languages. When the paragraph element is None, defaults to left-to-right. dirltrrtl)rLTRrkrlRTL)rWrs r,r!HocrTransform._get_text_directionsN ; $$ $zz~~eU+u4    "" r+ctURRSS5n[RU5 US;agg)aDetermine whether word breaks should be injected. In Chinese, Japanese, and Korean, word breaks are not injected, because words are usually one or two characters and separators are usually explicit. In all other languages, we inject word breaks to help word segmentation. langr8>jpnkorchi_simchi_traFT)rkrlr<r3)rWrrs r,r%HocrTransform._get_inject_word_breakss2zz~~fb) $ 7 7r+cX!S-US-$)z/Calculate the value of a polynomial at a point.rr9r#)ropolyxs r,polyvalHocrTransform.polyval$s7{T!W$$r+c UcgURU5nU(dgURUR::a'[R SUUR U55 gUR X5 URUR4n[5R"U6RURU5*5n U R5RU5n URU5up[U 5S:aSn [!U 5n U RSU R"5RSU 5RU [$- S-5nUR&R)US9 [+US9nU R"U -nUR-UR.U5 UR1U(aS OS5 UR3XR5RU5S5 UR&R5[65 UR9UR;S U55n[=US/-5HunnUR?UUUUUUUU5 M UR&RAU5 SSS5 g!,(df  g=f) zRender the text for a given line. The canvas's coordinate system must be configured so that hOCR pixel coordinates are mapped to PDF coordinates. Nz:line box is invalid so we cannot render it: box=%s text=%sg{Gzt?rrrr)rrgr)!rOrTrUr<errorr__debug_draw_line_bboxrQr rrotatedryinverse transformruabsrrVrrrrr5rE render_mode_debug_draw_baseline fill_colorr rMrNr _do_line_word draw_text)rWrr elemclassrtext_directionr line_min_aabbtop_left_cornerline_size_aabb_matrixline_size_aabbslope intercept slope_anglebaseline_matrixr^fontsizeelementselem next_elems r,rHocrTransform._do_line)sF < 006      1 1 1 IIL&&t,   ""69),,m.?.?@ H Z ( *WdnnT** +  /668BB=Q==. u: E5k " Z>00 1 Z9 % W[2%+ , YY ! !_ ! 5.1D%,,y8H IIdnnh /   .Qa 8  % %//1;;MJA  II  '||D$5$5fi$HIH#+Hv,=#>i""#"& $? II   %/6 5 5s *D I== J c <UcgURURU5R55n U S:XagURU5n U cgUR 5R U 5n UR RX5n URX5 URX5 U[R:Xa[RSU 5 U S:aU[R:Xa*UR[!SSSSU R"S55 OJU[R:Xa6UR[!SSSSU R"U R$-S55 UR'SU R$-U - 5 UR)UR R+U 55 UbURU5OSn U cgU(dgUR 5R U 5nU[R:Xa7[-U R.U R0UR"UR25nOJU[R:Xa6[-UR.U R0U R"UR25nUR5UW5 UR RSU5nUS:aUR$S:aU[R:Xa*UR[!SSSSUR"S55 OJU[R:Xa6UR[!SSSSUR"UR$-S55 UR'SUR$-U- 5 UR)UR R+S55 ggg) z"Render the text for a single word.Nr8zRTL: %srr9rd )rr_striprOrrrF text_width_debug_draw_word_triangle_debug_draw_word_bboxrrr<infortext_transformr rQrS horiz_scaleshow text_encoder rPrUrT_debug_draw_space_bbox)rWr line_matrixr^rrrrrelemtxthocr_boxbox font_width hocr_next_boxnext_box space_box space_widths r,rHocrTransform._do_line_words < %%d&<&!2!22##F1aB$CD=#4#44##F2q!R3999La$PQ   S399_z9 : IIdjj,,W5 64=3HD $ $Y /d    " &&(22=A ].. .!#''377HLL(,,OI }00 0!(,,(,,OI ##FI6jj++C: ?y2!2!22##F1aB q$IJ=#4#44##2q!R)H!L   S9??2[@ A IIdjj,,S1 2 3?r+c rURR(dgURR5 URR U5R S5 UR RURSS55HnURU5R5n[U5S:XaM3URU5nUcMIURRURURUR UR"SS9 M SSS5 g!,(df  g=f)z-Draw boxes around paragraphs in the document.N皙?rrrFfill)r?rrr stroke_color line_widthrDrrNr_rlenrOrectrQrUrSrV)rWrcolorrrrs r,r)HocrTransform._debug_draw_paragraph_boxess""88  YY ! ! # II " "5 ) 4 4S 9 **4+<+w<1$2248? KKgmmW^^RWN$ # #s C(D(( D6c jURR(dgURR5 URR U5R S5R URURURURSS9 SSS5 g!,(df  g=f)z'Render the bounding box of a text line.Ng333333?Fr) r?r rrrrrrQrUrSrV)rWrline_boxrs r,r#HocrTransform._debug_draw_line_bboxsz""33  YY ! ! # II " "5 ) 4 4T : ? ? hllHNNHOORW @ $ # # A$B$$ B2rcTURR(dgURR5 URR U5R U5R URURURUR5R URURURUR5R URURURUR5 SSS5 g!,(df  g=f)zARender a triangle that conveys word height and drawing direction.N) r?rrrrrrrQrUrPrTrWrrrrs r,r'HocrTransform._debug_draw_word_triangles""22  YY ! ! # II " "5 ) 4 4Z @ E E#''377 d377CGGSWWcgg6tt#''3778$ # #s CD D'c jURR(dgURR5 URR U5R U5R URURURURSS9 SSS5 g!,(df  g=f)z Render a box depicting the word.NFr) r?r!rrrrrrQrUrSrVrs r,r#HocrTransform._debug_draw_word_bboxs{""33  YY ! ! # II " "5 ) 4 4Z @ E E#))SZZe F $ # #rc jURR(dgURR5 URR U5R U5R URURURURSS9 SSS5 g!,(df  g=f)z3Render a box depicting the space between two words.NTr) r?r"rrrrrrQrUrSrVrs r,r$HocrTransform._debug_draw_space_bboxs{""44  YY ! ! # II  ' 2 2: > C C#))SZZd D $ # #rg?cDURR(dgURR5 URR U5R U5R URUURU5 SSS5 g!,(df  g=f)zRender the text baseline.N) r?rrrrrrrQrP)rWrr baseline_llyrrs r,r"HocrTransform._debug_draw_baseline sn""22  YY ! ! # II " "5 ) 4 4Z @ E E    $ # #s AB B)rFrEr@rVrDr?rSrK) rXz str | Pathr@rmr3rr4r r5Fontr6zDebugRenderOptions | None)raElementreturnstr)rarrzRectangle | None)rarrztuple[float, float])rarrrm)N)r|rr}z str | Nonerr)rrrr)rrrz Path | NonerrrNone) rrrElement | Nonerrrrrrrr)rrrr r^rrrmrrrrrrrr)rr)rrrr )rrrr )(r$r%r&r'r(rGcompileVERBOSErirsrxr rr[r_ classmethodrOruryrNrrrrrrrrrrrrrrrrrrrr*r#r+r,r1r1=s **   Kzz    h"_:>*"* *  *  **8*X    >>'' 00'+# L+L+$ L+  L+  L+\   %%^&^&^& ^&  ^& & ^&!^&@B3B3B3 B3  B3  B3"B3&B3!B3HAE$PT58C  #, 5:c  #, 5>#  #,   r+r1).r( __future__rloggingrBrGr dataclassesr itertoolsrmathrrpathlibr xml.etreer pikepdfr r r pikepdf.canvasr rrrrrrrrrocrmypdf.hocrtransform._fontrrr getLoggerr$r<rRrr Exceptionr.r1r#r+r,r&s%" !!++   ?6 !     $$ $//__r+