TAiSrSSKJr SSKrSSKJr SSKrSSKJ r SSKJ r SSK J r SS K J r SS K Jr SS K Jr SS KJr SS KJr SSKJr SSKJr SSKJr SSKJr SSKJr SSKJr SrSr"SS5rSrS$SjrS%SjrS&Sjr S'Sjr!"SS5r""S S!\5r#"S"S#\5r$g)(z'Implementation of network table parser.) annotationsN)Any)LTTextLineHorizontal)LTTextLineVertical)ALL_ALIGNMENTS)HORIZONTAL_ALIGNMENTS)VERTICAL_ALIGNMENTS)TextAlignments) bbox_from_str)bbox_from_textlines)boundaries_to_split_lines)find_columns_boundaries) text_in_bbox)text_in_bbox_per_axis)textlines_overlapping_bbox)TextBaseParserc&\rSrSrSrSrSSjrSrg)TextLine!apA placeholder class to represent a text line with bounding box attributes. Attributes ---------- x0 : float The x-coordinate of the left edge of the text line. x1 : float The x-coordinate of the right edge of the text line. y0 : float The y-coordinate of the bottom edge of the text line. y1 : float The y-coordinate of the top edge of the text line. x0y0x1y1c4XlX lX0lX@lgNr)selfrrrrs S/var/www/html/land-ocr/venv/lib/python3.13/site-packages/camelot/parsers/network.py__init__TextLine.__init__2s)rrrrN)rfloatrr&rr&rr&)__name__ __module__ __qualname____firstlineno____doc__ __slots__r#__static_attributes__r%r"rr!s )Ir%rcSnU[U5:a&X#U:aUS- nU[U5:a X#U:aMUnU[U5:a&X$U:aUS- nU[U5:a X$U:aMXC- $)z=Get the number of columns crossed by a segment [left, right].rrlen)leftright col_anchors index_left index_rights r" column_spreadr79sJ s;' 'K,Cd,Ja  s;' 'K,Cd,JK K( ([-E-Mq  K( ([-E-M  ##r%cSnSnSnSnUupgpUGH8n U RU:aEU RU :dU RU:aM6UbURU R:aU nMWMYXR:aEU RU :dU RU:aMUbURU R:aU nMMU RU:dU RU:aMU RU:a%UbURU R:aU nGMGMXR:dGMUbURU R:dGM6U nGM; UUUUS.$)aSearch for textlines that are closest to the bounding box but outside in all four directions. Parameters ---------- bbox : list of float A list containing the coordinates of the bounding box in the order [left, bottom, right, top]. tls : list of TextLine A list of textline objects to search for the closest lines. Returns ------- dict A dictionary with keys "left", "right", "top", and "bottom", each mapping to the closest textline object in that direction or None if not found. N)r2r3topbottom)rrrr) bboxtlsr2r3r9r: bbox_left bbox_bottom bbox_rightbbox_toptextlines r"find_closest_tlsrBEs4&>BD>BE<@C?CF592YZ ;; "{{X%{)B|tww4 5 ++ %{{X%{)B}8;; 6 !7{{Z'8;;+B{{[(>VYY%<%F&=KK';#&&8;;"6"C16  r%cb/nSnU(avSnUR5HWnURU:dMURU5 URU5 URU:dMIURnSnMY U(aMvUVs/sHoUR UR /PM snU4$s snf)aWExtract zones from the textlines above the body bbox. Parameters ---------- all_above : List[Any] Textlines that are above the bounding box. max_v_gap : float The maximum vertical gap allowed. top : float The current top boundary. Returns ------- Tuple[List[List[float]], float] The extracted zones and the new top boundary. TF)copyrappendremoverrr) all_above max_v_gapr9tls_in_new_row pushed_uprAs r"_extract_zonesrKs&NI  ^^  {{S %%h/  *;;$#++C $I ) 8F F~8[[(++ &~ F KK Fs!B,cURSS9 /nUHGnU(aUSSUS:aURU5 M,[USSUS5USS'MI U$)zMerge overlapping zones into consolidated zones. Parameters ---------- zones : List[List[float]] A list of zones defined by their x-coordinates. Returns ------- List[List[float]] A list of merged zones. c US$Nrr.)zs r"_merge_zones..sQqTr%keyrr)sortrEmax)zones merged_zoneszones r" _merge_zonesrZss JJ>J"&(L|B/2T!W<    %"%l2&6q&947"CL Q   r%c ^UnUupVpx/n Sn U (Ga\Sn Sn /n UHn SU RU R--nU RU:dM1X^s=:aU:dM?O MCU RU 5 Sn UVs/sHDnURU:dMUSURUR--s=:aU:dM>O MBUPMF n nU (dM[ U SS9n M U (aU RX-:ar[ XU R 5up[U 5n[U4SjU55nU[ [[R"[T5S- 55::aXVXx4nSn U (aGM\U$s snf) aExpand a bounding box (bbox) vertically by looking for plausible headers. The core algorithm is based on fairly strict alignment of text. It works for the table body but might fail on table headers since they tend to be in a different font, alignment (e.g., vertical), etc. This method evaluates the area above the table body's bbox for characteristics of a table header: close to the top of the body, with cells that fit within the horizontal bounds identified. Parameters ---------- body_bbox : Tuple[float, float, float, float] The bounding box of the body in the format (left, bottom, right, top). textlines : List[Any] A list of textline objects, each with properties x0, x1, y0, and y1. col_anchors : List[float] A list of x-coordinates representing column anchors. max_v_gap : float The maximum vertical gap allowed to consider a header plausible. Returns ------- Tuple[float, float, float, float] The expanded bounding box in the format (left, bottom, right, top). TFNg?cUR$r r)tls r"rP.search_header_from_body_bbox..s"%%r%rRc3J># UHn[USUST5v M g7f)rrN)r7).0rYr4s r" /search_header_from_body_bbox..s(IU d1gtAw <<s #r) rrrrEminrKrrZrVMAX_COL_SPREAD_IN_HEADERmathceilr1) body_bbox textlinesr4rHnew_bboxr2r:r3r9rWkeep_searching closest_aboverGrAtextline_centerr^rX max_spreads ` r"search_header_from_body_bboxros<H!*T5!EN   !H!X[[8;;%>?O{{S T%Ce%C%C  * $ ('uus{'+cRUURUU].C'Ke'K'K' 9$' 7G$HM" ]--?( m>N>NOJE'.LIUJS($))C 4Dq4H*I!%5"&g .h OQs7E3%E38E3<E3cR\rSrSrSrSrSrSrSSjrSr S r S r S r S r S rg)AlignmentCounteriz For a given textline, represent all other textlines aligned with it. A textline can be vertically aligned with others if their bbox match on left, right, or middle coord, and horizontally aligned if they match top, bottom, or center coord. cJ0Ul[Hn/URU'M gr )alignment_to_occurrencesr)r! alignments r"r#AlignmentCounter.__init__'s$(*%'I79D ) )) 4(r%c URU$)z/Get the value of a property to the given value.rs)r!rSs r" __getitem__AlignmentCounter.__getitem__,s,,S11r%c"X RU'U$)z/Set the value of a property to the given value.rw)r!rSvalues r" __setitem__AlignmentCounter.__setitem__0s-2%%c* r%Nc^U=(d TRR5n[U4SjU5n[USS9$)z=Get the alignment dimension with the max number of textlines.c&>UTRU4$r rw) alignment_idr!s r"rP1AlignmentCounter.max_alignments..9s--l;"r%c[US5$Nrr0)items r"rPr?s Sa\r%rR)rskeysmaprV)r! alignment_idsalignment_itemss` r"max_alignmentsAlignmentCounter.max_alignments5s@%M)F)F)K)K)M     ?(ABBr%c,UR[5$)z8Tuple (alignment_id, textlines) of largest vertical row.)rr r!s r"max_vAlignmentCounter.max_vAs ""#899r%c,UR[5$)z:Tuple (alignment_id, textlines) of largest horizontal col.)rr rs r"max_hAlignmentCounter.max_hHs""#677r%c:[UR5S5$)zuMaximum vertical count. Return the maximum number of alignments along one of the vertical axis (left/right/middle). r)r1rrs r" max_v_countAlignmentCounter.max_v_countL 4::<?##r%c:[UR5S5$)zyMaximum horizontal count. Return the maximum number of alignments along one of the horizontal axis (bottom/top/center). r)r1rrs r" max_h_countAlignmentCounter.max_h_countTrr%cPUR5S- UR5S- -$)zReturn the alignment score. We define the alignment score of a textline as the product of the number of aligned elements - 1. The -1 is to avoid favoring singletons on a long line. r)rrrs r"alignment_score AlignmentCounter.alignment_score\s+  "Q&4+;+;+=+ABBr%rwr )r'r(r)r*r+r#rxr|rrrrrrr-r.r%r"rqrqs6: 2 C:8$$Cr%rqc^\rSrSrSrU4SjrSrSrSrSr Sr S r SS jr SS jr SS jrSS jrSSjrSSjrSSjrSrU=r$) TextNetworksifzText elements connected by vertical AND horizontal alignments. The alignment dict has six keys based on the hor/vert alignments, and each key's value is a list of camelot.core.TextAlignment objects. c:>[TU][5 0Ulgr )superr#r_textline_to_alignments)r! __class__s r"r#TextNetworks.__init__ms ((*$r%c&URX25 gr )register_aligned_textline)r!rtcoordrAs r"_update_alignmentTextNetworks._update_alignmentss++H alignment object.N)_text_alignmentsitemsrirgetrq)r!align_id textedgestextedgerA alignmentss r"_compute_alignment_counts&TextNetworks._compute_alignment_counts}s{#'#8#8#>#>#@ H% ( 2 2H!%!=!=!A!A(D!QJ!)%5%7 AK44X>+3+=+=J( !3&$Ar%cSnU(aSnURR5HnUHn/n[[UR55H\nURUnUR UnUR 5S::dUR5S::dMKURU5 M^ [U5HnURU SnM M M 0UlUR5 U(aMgg)zRemove elements which are only connected on one dimension. Elements should be connected to others both vertically and horizontally. TFrN) rvaluesranger1rirrrrEreversedr) r!removed_singletonstext_alignmentstext_alignment to_removeirArindexs r"remove_unconnected_edges%TextNetworks.remove_unconnected_edgess" !& #'#8#8#?#?#A'6N "I"3~'?'?#@A#1#;#;A#>%)%A%A(%K '2249)5571<%,,Q/B"*)!4*44U;-1*"5!'6$B0,.D (  * * ,9! r%cR^[TRR5U4SjSS9$)z-Retrieve the textline that is most connected.cp>TRUR5UR*UR*4$r )rrrr)rAr!s r"rP6TextNetworks.most_connected_textline..s1,,X6FFH  "r%N)rSdefault)rVrrrs`r"most_connected_textline$TextNetworks.most_connected_textlines.   ( ( - - /   r%c <UR5nUcgURRU5nUcgUR5up4UR 5up5[ U5S::d[ U5S::ag[ USS9n[ USS9n[R"[S[ U55Vs/sH#nXhRXhS- R- PM% sn5n [R"[S[ U55Vs/sH#nXxRXxS- R- PM% sn5n U RS:XdU RS:XagS[R"U S5-S[R"U S5-4n U $s snfs snf) zEvaluate plausible gaps between cells. Both horizontally and vertically based on the textlines aligned with the most connected textline. Returns ------- gaps_hv : tuple (horizontal_gap, vertical_gap) in pdf coordinate space. NrcUR$r )rrAs r"rP5TextNetworks.compute_plausible_gaps..8;;r%rRcUR$r r]rs r"rPrrr%rg@K)rrrrrr1sortednparrayrrrsize percentile) r!most_aligned_tlbest_alignment__ref_h_textlinesref_v_textlines h_textlines v_textlinesrh_gapsv_gapsgaps_hvs r"compute_plausible_gaps#TextNetworks.compute_plausible_gapss668  "5599/J  !-224,224  1 $O(<(A_2NO _2NO q#k"23 3A!!KA$6$9$993   q#k"23 3A!!KA$6$9$993   ;;! v{{a/ "--+ + "--+ +  -   s 4*F *Fc UR5nUupESnUbUU/S.nURU5 URURURUR /n[ URR55nURU5 U/n Sn URUR4/n X:waDUbUSRU5 Un [Xx5n URUU UUU U 5up{pX:waMD[U 5[:aU$g)aBuild a candidate bounding box for the body of a table using network algorithm. Parameters ---------- gaps_hv : tuple of float The maximum distance allowed to consider surrounding lines/columns as part of the same table. parse_details : list Optional parameter list, in which to store extra information to help later visualization of the table creation. Returns ------- list of float or None The bounding box of the table body as a list of four floats [x0, y0, x1, y1] or None if not enough textlines are found. N) max_h_gaprH iterationsr)rrErrrrlistrrrFrB expand_bboxr1MINIMUM_TEXTLINES_IN_TABLE) r!r parse_detailsrrrHparse_details_searchr;tls_search_space tls_in_bbox last_bboxlast_cols_bounds closest_tlss r"search_table_bodyTextNetworks.search_table_bodys<,668& 6:  $&& $  !5 6                < < A A CD0&'  ,//1C1CDE#/$\299$?I*4BKDHDTDT   E ADK  { 9 9Kr%cUR5nUR5HupU bURXyXH5(dM!URXyU5n [ X5n X-n UR XU5(dM[[ [U 55=p[U 5nURU 5 URX;5 M XXc4$)aLExpand the bounding box based on closest textlines. Parameters ---------- bbox : list of float The current bounding box. closest_tls : dict The closest textlines found. tls_search_space : list The list of textlines available for searching. gaps_hv : tuple of float The maximum allowed horizontal and vertical gaps. last_cols_bounds : list of tuple The boundaries of the last found columns. tls_in_bbox : list The textlines currently in the bounding box. Returns ------- tuple The updated bounding box, column boundaries, textlines in bbox, and search space. ) rDrcan_expand_bboxget_expanded_bboxris_valid_expansionrr rextendupdate_search_space) r!r;rrrrr cand_bbox directionrAexpanded_cand_bboxnew_tlstls_in_new_boxs r"rTextNetworks.expand_bbox=s>IIK #.#4#4#6 It';';W((!%!7!7 Y!W "#5HG$2N**9FVWW#$7$GH HD6~F    w '  $ $%5 ?!$7${DDr%cUS:XaUSUR- US:*$US:XaURUS- US:*$US:XaUSUR- US:*$US:XaURUS- US:*$g ) aCheck if the bounding box can be expanded in the given direction. Parameters ---------- cand_bbox : list of float The candidate bounding box. textline : Any The textline to check against. gaps_hv : tuple of float The maximum allowed horizontal and vertical gaps. direction : str The direction to check for expansion. Returns ------- bool True if the bounding box can be expanded, otherwise False. r2rr3rr:rr9rF)rrrr)r!rrArrs r"rTextNetworks.can_expand_bboxrs6  Q<(++-; ; ' !;;1-; ; ( "Q<(++-; ; % ;;1-; ;r%cUR5nUS:XaURUS'U$US:XaURUS'U$US:XaURUS'U$US:XaURUS'U$) aMGet the expanded bounding box based on the textline in the specified direction. Parameters ---------- cand_bbox : list of float The candidate bounding box. textline : Any The textline to expand the bounding box with. direction : str The direction to expand. Returns ------- list of float The expanded bounding box. r2rr3rr:rr9r)rDrrrr)r!rrArrs r"rTextNetworks.get_expanded_bboxs&'^^-  $,KK q !"! ' !$,KK q ! "! ( "$,KK q !"!% $,KK q !!!r%ch[U5nUS;=(a [U5[U5:(+$)aGCheck if the new expansion is valid. Parameters ---------- direction : str The direction of expansion. tls_in_new_box : list The textlines in the new bounding box. last_cols_bounds : list of tuple The boundaries of the last found columns. Returns ------- bool True if the expansion is valid, otherwise False. )r:r9)rr1)r!rrr cols_boundss r"rTextNetworks.is_valid_expansions6,.n= * * Ws;/?#FVBW/W  r%c^[[U5S- SS5HnXnXB;dMX M g)zUpdate the search space by removing textlines in the new bounding box. Parameters ---------- tls_search_space : list The current search space of textlines. new_tls : list The new textlines added to the bounding box. rrTN)rr1)r!rrrrAs r"r TextNetworks.update_search_spaces6s+,q0"b9A'*H"$':r%cFURU5 UR5 g)zGenerate the text edge dictionaries based on the input textlines. Parameters ---------- textlines : list List of textline objects to be processed. N)rr)r!ris r"generateTextNetworks.generates %%i0 &&(r%)r)rtuple[float, float]rzlist[Any] | Nonereturnzlist[float] | None)r; list[float]rzdict[str, Any]r list[Any]rr rr rr r z3tuple[list[float], list[Any], list[Any], list[Any]])rr rArrr rstr)rr rArrr r r )rr rr rr r bool)rr rr r None)rir r r)r'r(r)r*r+r#rrrrrrrrrrrrrr- __classcell__rs@r"rrfsH * =2 >%-N :x@$@(@  @D3E3E$3E$ 3E % 3E $ 3E3E =3Ej###% #  #J"$"03"@C" "<  " $    6( )(4=( (" ) )r%rch^\rSrSrSrS U4SjjrSrSrSrSr Sr S r S r U=r $) NetworkiaNetwork method looks for spaces between text to parse the table. If you want to specify columns when specifying multiple table areas, make sure that the length of both lists are equal. Parameters ---------- table_regions : list, optional (default: None) List of page regions that may contain tables of the form x1,y1,x2,y2 where (x1, y1) -> left-top and (x2, y2) -> right-bottom in PDF coordinate space. table_areas : list, optional (default: None) List of table area strings of the form x1,y1,x2,y2 where (x1, y1) -> left-top and (x2, y2) -> right-bottom in PDF coordinate space. columns : list, optional (default: None) List of column x-coordinates strings where the coordinates are comma-separated. split_text : bool, optional (default: False) Split text that spans across multiple cells. flag_size : bool, optional (default: False) Flag text based on font size. Useful to detect super/subscripts. Adds around flagged text. strip_text : str, optional (default: '') Characters that should be stripped from a string before assigning it to a cell. edge_tol : int, optional (default: 50) Tolerance parameter for extending textedges vertically. row_tol : int, optional (default: 2) Tolerance parameter used to combine text vertically, to generate rows. column_tol : int, optional (default: 0) Tolerance parameter used to combine text horizontally, to generate columns. c  4>[T U]SUUUUUUUUU U S9 g)Nnetwork) table_regions table_areascolumns flag_size split_text strip_textedge_tolrow_tol column_toldebug)rr#) r!rrrrrrrrrrkwargsrs r"r#Network.__init__s8  '#!!!  r%cUR5n[UR55nU(a [U5O/n[5n0UlUR b-/UR S'/UR S'/UR S'U(aUR X5upVUcg[XS5nU(dg[U5n[U5n URUUUUU U5n [U [5(a [U 5n UUU U S.n XRU 'UR bUR SRU 5 URU5 UV s/sH oU;dM U PM nn U(dgU(aMggs sn f)Nnetwork_searches bbox_searches col_searches) bbox_bodycols_boundaries cols_anchors bbox_full)_get_user_provided_bboxesr_get_filtered_textlinessettable_bbox_parsesr_get_bbox_bodyrrr_get_full_bbox isinstancetuplerEupdate) r!user_provided_bboxesfiltered_textlinesritextlines_processedr&rrr'r(r) table_parser^s r"_generate_table_bboxNetwork._generate_table_bbox/s#==?!  ( ( * 1CD+,  E "$    )57D  1 224D   /13D  ~ .!%!4!45I!U I 4YJK5kBO4_EL++$ I)T**!), '#2 ,& K1< " "9 -!!-"">299+F & &{ 3&/Qi=P3PiIQYiPRs  F(FcUb![U5S:aUR5S4$[5nURU5 UR 5 UR 5nUcgUSUR cUSO UR 4nURUUR(aURSOSS9nURbURSRU5 Xd4$)Nr)NNrr$)rr#) r1poprrrrrrrrE)r!r3ri text_networkr edge_tol_hvr&s r"r.Network._get_bbox_bodyns  +'(1,(,,. $~ i(--/557 ? AJ--/GAJT]] !22 7;7I7I""?3t3     )   1 2 9 9, G!!r%cnUb[U5S:aU$U$[U5nUb[XtXVS5$U$)Nrr)r1r ro)r!r3r&rrir(rbbox_body_from_tlss r"r/Network._get_full_bboxs\ +'(1,  - "5[!A !-3&<r%cURUR-Vs/sH2n[UR5R 55S:dM0UPM4 nnUR U5$s snfrN)horizontal_text vertical_textr1rr_apply_regions_filter)r!t all_textliness r"r+Network._get_filtered_textlinessi))D,>,>> >1::<%%'(1, >  ))-88  s /A(A(ctURb&URVs/sHn[U5PM sn$gs snfr )rr )r!area_strs r"r*!Network._get_user_provided_bboxess9    '<@.s S):):)<)B)B)D%E%Ir% horizontalverticalc4UR*UR4$r )rrrs r"rPrMsx{{lHKK%@r%rR)rrrr(c>TUTUS-/$rr.)idxr4s r"rPrMs[%5{377K$Lr%zWarning: Bounding box z not found in table_bbox_parses.)rrBrCt_bboxrrfilterr _group_rowsr _join_rowsrr1r-rprint)r!r; user_colsall_tls text_x_min text_y_min text_x_max text_y_max rows_groupedrowscolsrrr4s @r"_generate_columns_and_rows"Network._generate_columns_and_rowssn+ &&(:(:   IKK - J0GGA   :MW9U6 '''F |D  <)+zl:D49!SY]4KL4Kq!Wdq5k*4KDLD--- $ 6 6t < +N; La[!1A!56.tf4TUV2r2~%4%%'MsE)rSr-) NNNFFNrrF)r'r(r)r*r+r#r7r.r/r+r*rar-rrs@r"rrsS#N 8=~"@(9 -&-&r%r)r;r r<z/list[LTTextLineHorizontal | LTTextLineVertical]r z;dict[str, LTTextLineHorizontal | LTTextLineVertical | None])rGr rHr&r9r&r ztuple[list[list[float]], float])rWlist[list[float]]r rd) rh!tuple[float, float, float, float]rir r4r rHr&r re)%r+ __future__rrftypingrnumpyrpdfminer.layoutrrcorerr r r utilsr r rrrrrbaserrerrr7rBrKrZrorqrrr.r%r"rms-" 0.!(&!!'-+ ). 0 $9 9K9@9x&L&L%*&L16&L$&LR2W0WWW W ' WtFCFCRD)>D)N o&no&r%