kh"jdZddlmZmZmZmZddlmZddlmZm Z ddl m Z ddl m Z Gdd e Z y ) zA group of ``Shape`` instances.)ShapeStrokeFill Hyperlink)RectType) CollectionElementCollection)share) constantsceZdZdZdefdZdefdZedZ edZ edZ ed Z ed Z ed Zd ed efdZdefdZdZedZdZy)Shapesz= A collection of ``Shape`` instances: ``Stroke`` or ``Fill``.rawsc|j|D]>}d|vr t|}nd|vr t|}n t|}|j |@|S)z;Clean current instances and restore them from source dicts.starturi)resetrrrappend)selfrrawshapes Q/var/www/teggl/fontify/venv/lib/python3.12/site-packages/pdf2docx/shape/Shapes.pyrestorezShapes.restore sV  C#~s #!#S  KK   ecy)z override. Do nothing.N)rrs r _update_bboxzShapes._update_bboxsrcXttd|j}t|S)zJ Stroke Shapes, including table border, text underline and strike-through.c"t|tSN) isinstancerrs rz Shapes.strokes..%s *UF3rlistfilter _instancesrr instancess rstrokeszShapes.strokes!s+ 3T__FG i  rcXttd|j}t|S)z3 Fill Shapes, including cell shading and highlight.cjt|txr"|jtjdk7S)N)rrr)r"rcolorr rgb_valuer#s rr$z!Shapes.fillings...s**UD18 uw77rr%r)s rfillingszShapes.fillings)s0 89=JK i  rcXttd|j}t|S)z Hyperlink Shapes.c"t|tSr!)r"rr#s rr$z#Shapes.hyperlinks..7s *UI6rr%r)s r hyperlinkszShapes.hyperlinks3s+ 6IJ i  rcXttd|j}t|S)zPotential table borders.c@|jtjSr!)has_potential_typerBORDERr#s rr$z&Shapes.table_strokes..?s%228??Crr&r'r(r r)s r table_strokeszShapes.table_strokes;s+ CT__VW  ++rcXttd|j}t|S)zPotential table shadings.c@|jtjSr!)r6rSHADINGr#s rr$z'Shapes.table_fillings..Gs%2283C3CDrr8r)s rtable_fillingszShapes.table_fillingsCs+ DdooWX  ++rc\d}tt||j}t|S)zcPotential text style based shapes, e.g. underline, strike-through, highlight and hyperlink.c|jtjxsa|jtjxs@|jtjxs|jtj Sr!)r6r HIGHLIGHT UNDERLINESTRIKE HYPERLINKr#s rfz#Shapes.text_style_shapes..fOsj++H,>,>?A,,X-?-?@A,,X__=A,,X-?-?@ Ar)setr'r(r )rrDr*s rtext_style_shapeszShapes.text_style_shapesKs* A q$//23  ++rmax_border_widthshape_min_dimensioncb|jsy|jj}g}|D]}t|jj|jj |kr;|jj |}|jrc|j|j||j|}g}|D]J} t| tr'| j|} |j| r| n| :|j| L|j||jy)aClean rectangles. * Delete shapes out of page. * Delete small shapes (either width or height). * Merge shapes with same filling color. * Detect semantic type. Args: max_border_width (float): The max border width. shape_min_dimension (float): Ignore shape if both width and height is lower than this value. N)r(parentbboxmaxwidthheight intersectis_emptyr update_bbox _merge_shapesr"r to_stroker_parse_semantic_type) rrGrH page_bboxcleaned_shapess bbox_in_page merged_shapesshapesrstrokes rclean_upzShapes.clean_upXsKK$$  ?A166<</0CCX66++I6L$$h  ! !!-- "= >  ?**>: " %E%&)9: fE: e$  % 6 !!#rtablescp|sy|Dcgc]}g}}g}|jD]}|jtjs|jtjr|j |St ||D]c\}}|jj|jr|j ||jj|jrde|j |t ||D]\}}|s |j||j|ycc}w)zAdd Shape to associated cells of given tables. Args: tables (list): A list of TableBlock instances. N) r( equal_to_typerr7r<rziprKcontains intersects assign_shapesr)rr]_shapes_in_tablesrZrtableshapes_in_tables rassign_to_tableszShapes.assign_to_tabless v)//1B//__ %E""8??3u7J7J8K[K[7\ e$*-f6F*G %&::&&uzz2#**51..uzz: % e$% %*'*&2B&C 1 "E?"H    0 1 690s D3cd}|jD]}|j||d}|jD]}|j||d}|jD]}|j||y)aZPlot shapes for debug purpose. Different colors are used to display the shapes in detected semantic types, e.g. yellow for text based shape (stroke, underline and highlight). Due to overlaps between Stroke and Fill related groups, some shapes are plot twice. Args: page (fitz.Page): pdf page. )?g?rj)rkrk)rrrkN)r=plotr9rF)rpager.rs rrlz Shapes.plotsq,((AE%**T5*AA''@ED%)@@++DEUZZe-DDrcttd|}d}t|j|}g}|D]}|jj }t d|D}||z tjk\r.|j|dj|js|j|td|}|j||S)znMerge shapes if same filling color. Note the merged bbox must match source shapes as more as possible.c|j Sr!) is_determinedr#s rr$z&Shapes._merge_shapes..se111rc|j|jk(xr8|jj|jtj Sr!)r.rKrbget_expand_bboxr TINY_DIST)abs rrDzShapes._merge_shapes..fs:77AGG#a(9(9!:K:KIL_L_:`(a arc3PK|]}|jj ywr!)rKget_area).0rs r z'Shapes._merge_shapes..sDU5::..0Ds$&rkc@|jtjSr!)r_rrCr#s rr$z&Shapes._merge_shapes..s%*=*=h>P>P*Qr) r&r'r grouprKrwsumr FACTOR_ALMOSTrrQextend) rZ normal_shapesrDgroupsrYr{ merged_areasum_arear3s rrRzShapes._merge_shapess V 16;<  bM*003  ,E**--/KDeDDH #y'>'>>$$U1X%9%9%**%EF$$U+  ,QSYZ Z(rc|jj}|j|jD]}|j |y)a Detect shape type based on the position to text blocks. .. note:: Stroke shapes are grouped on connectivity to each other, but in some cases, the gap between borders and underlines/strikes are very close, which leads to an incorrect table structure. So, it's required to distinguish them in advance, though we needn't to ensure 100% accuracy. They are finally determined when parsing table structure and text format. N)_parentblockssort_in_reading_orderr(parse_semantic_type)rrrs rrTzShapes._parse_semantic_typesB$$$$&__ .E  % %f - .rN)__name__ __module__ __qualname____doc__r&rrrpropertyr+r0r3r9r=rFfloatr\rhrl staticmethodrRrTrrrrr sG 4  %U%!!!!!!,,,, , ,'$'$5'$T%d%PE66.rrN)rrrrr common.sharercommon.Collectionr r commonr r rrrrrs)%11#=g. g.r