4vjZdZddlZddlmZmZGddZdddZd dd Zd dd Zd Z d Z dZ dZ dZ dZdZdZdZdZdZdZdZdZdZdZdZdZdZdZd Zd!Zd"Zd#Z d$Z!dd%d&d'Z"y)(aF Copyright (C) 2023 Artifex Software, Inc. This file is part of PyMuPDF. PyMuPDF is free software: you can redistribute it and/or modify it under the terms of the GNU Affero General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version. PyMuPDF is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Affero General Public License for more details. You should have received a copy of the GNU Affero General Public License along with MuPDF. If not, see Alternative licensing terms are available from the licensor. For commercial licensing, see or contact Artifex Software, Inc., 39 Mesa Street, Suite 108A, San Francisco, CA 94129, USA, for further information. --------------------------------------------------------------------- PyMuPDF table cell-span resolution (opt-in extension). Provides SpanCell and resolve_spans (plus the _span_* helpers), which reconstruct a detected table's merged-cell (colspan / rowspan) structure. Re-exported by pymupdf.table; never runs on the default find_tables() path. Reuses the word-selection helpers of pymupdf._table_refine. N)_refine_is_vertical_or_rotated_refine_page_wordsceZdZdZddZy)SpanCellaOne reconstructed table cell after span resolution (PyMuPDF extension). ``bbox`` is the placement's ``(x0, y0, x1, y1)`` union rect, ``text`` the page text it claims (lines joined by ``\n``), and ``colspan``/``rowspan`` how many grid columns/rows it covers. resolve_spans always sets a real bbox; a caller padding its own grid may construct SpanCells with ``bbox=None``. ``tag`` is the cell's HTML tag (``"td"``/``"th"``), defaulting to ``"td"``. resolve_spans leaves it at the default; find_tables(refine=True) overwrites it from the resolved header region so Table.to_html() can serialize the grid directly, and a caller building its own grid may set it too.cJ||_||_||_||_||_yN)bboxtextcolspanrowspantag)selfr r r r r s G/root/.hermes/venv/lib/python3.12/site-packages/pymupdf/_table_spans.py__init__zSpanCell.__init__7s%    N)td)__name__ __module__ __qualname____doc__rrrrr*s  Drrg@) toleranceczg}t|D]*}|r ||dz |kr|jt|,|S)zGreedy 1-D clustering of edge coordinates into slot boundaries. Keeps a sorted value only when it is more than ``tolerance`` from the last kept boundary, so the first value of each run is the retained boundary.)sortedappendfloat)valuesr boundariesvalues r_span_clustered_boundariesr!@sL J( %*R.0I= %,'( r?cd}tt|dz D]-}||||dzzdz }||z |cxkr ||zks&n)|dz }/td|S)zKHow many [boundaries[i], boundaries[i+1]] slots the span start..end covers.r@)rangelenmax)startendrrcountindexmidpoints r_span_covered_slot_countr.Msn Es:*+u% 519(==D 9  ;C)O ; QJE q%=rc g}tt|dz D]M}||||dzzdz }|j|z |cxkr|j|zks:n=|j |O|syt |t |dzfS)zIThe half-open column-slot range (first, last+1) a rect's x-extent covers.r$r%)rr$)r&r'x0x1rminr()rect x_boundariesrhitsr,r-s r_span_slot_ranger6Ws Ds<(1,- ',uqy*AASH 77Y ( Adgg .A A KK   Is4y1} %%rc2|d|dkxr |d|dkSNrr$rabs r_span_ranges_intersectr<cs% Q4!A$; &1Q4!A$;&rct|j|cxkxrt|jkncxr6t|j|cxkxrt|jkScSr)rr0r1y0y1)xyr3s r_span_point_in_rectrBgsJ >Q 0%. 0 ZU477^q5ZERVRYRYN5ZZ5ZZrc tjtd|Dtd|Dtd|Dtd|DS)Nc34K|]}|jywr)r0.0r3s r z#_span_rect_union..m &DGG &c34K|]}|jywrr>rEs rrGz#_span_rect_union..nrHrIc34K|]}|jywr)r1rEs rrGz#_span_rect_union..orHrIc34K|]}|jywrr?rEs rrGz#_span_rect_union..prHrI)pymupdfRectr2r()rectss r_span_rect_unionrRksK << & && & && & && & &&  rct|j|jt|j|jz Sr)r2r1r(r0r9s r_span_x_overlaprTts) qttQTT?Sqtt_ ,,rcbt||}|dkrytd|j}||z dk\S)aTrue if a text span overlaps a cell by enough x to signal a merge. A few points of overlap is noise (a span drifting across a column line with trailing whitespace or a currency glyph); require >2pt and >=15% of the cell width before treating it as a merged-cell signal.r%Fr"g333333?)rTr(width) span_rect cell_rectoverlap cell_widths r_span_substantial_x_overlapr[xs: i3G#~S)//*J Z 4 ''rc|sgSg}|dx}}|ddD]$}||dzk(r|}|j||f|x}}&|j||f|S)z>Collapse a sorted index list into (start, end) inclusive runs.rr$N)r)indicesrangesr)r*r,s r_span_contiguous_rangesr_sw  F!*EC C!G C  ucl#   MM5#, Mrcg}t|dD]N\}}}||k(r |r ||ddkDr|j|||gf0|d\}}}|t||||gzf|d<P|S)zIMerge overlapping (start, end, text) intervals, accumulating their texts.c|d|dfSr8ritems rz'_span_merge_intervals..sQa?Qrkeyrr$)rrr() intervalsmergedr)r*r prev_startprev_endtextss r_span_merge_intervalsrls F"92QRFsD C< A. MM5#v. / &,Rj# He #h"4etfnEr F Mrc|jdgDcgc]Z}t|jdxsdjr,t|jdxsdj\}}dj|Scc}w)z=Space-joined non-empty span texts of a get_text('dict') line.spansr  )getstrstripjoin)linespanpartss r_span_line_textrxs|HHWb)   txx%2 & , , . DHHV  "#))+ E  88E?  sABc|jd}|r#tj|}|js|S|jdgDcgc];}|jdr(tj|jdxsg=}}|Dcgc]}|jr|}}|syt |Scc}wcc}w)zKBounding rect of a dict line (its bbox, else the union of its span bboxes).r rnN)rqrOrPis_emptyrR)rur r3rvrQs r_span_line_rectr{s 88F D ||D!}}KHHWb)   88F   TXXf%+, E  $ 9d4==T 9E 9  E ""  :s AB7B<#B<ct|dd}||Sg}|jdjdgD]h}|jddvr|jdgD]=}t|st |}|st |}|+|j ||f?j t|d||S#t$rY|SwxYw)zVertical/rotated text lines as (rect, text), cached on the page. Selects non-horizontal lines whose reading order get_text('dict') already preserves._span_vertical_lines_cacheNdictblockstypeNrlines) getattrget_textrqrrxr{rsetattr Exception)pagecachedrblockrur r3s r_span_vertical_text_linesrs T7 >F  Ev&**8R8 ' 99V I - IIgr* 'D1$7"4(D"4(D| LL$ & ' ' 2E: L   L s! B00 B=<B=c\ |syg}t|D]\}}t|jt|jzdz t|jt|j zdz t |sp|j||f|syd}|D]6\}\}} } } }|| zdz | | zdz t fd|Ds2|dz }8|tdt|z dkrytd|Ddkryd jd t|d DS) a"Text of vertical lines centered in rect, when they dominate the rect. Returns the stacked line text only if vertical lines are centered in the rect, cover >=60% of the rect's selected words, and carry >=2 tokens; else None so the caller falls back to horizontal line synthesis.N?rc3>K|]\}}t|ywr)rB)rF line_rect_cxcys rrGz/_span_vertical_text_for_rect..sU,)Q"2r95Usr$g333333?c3TK|] \}}t|j"ywr)r'splitrFrr s rrGz/_span_vertical_text_for_rect..s 7D3tzz|  7s&( c3&K|] \}}| ywrrrs rrGz/_span_vertical_text_for_rect..s At ct|djt|dj|djzdz fS)Nrr)rr0r>r?rbs rrdz._span_vertical_text_for_rect..s;eDGJJ/%ad1gjj9PTW8W2X1XYrre)rrr0r1r>r?rBranyr(r'sumrtr)rr3selected_words candidatesrr vertical_hitsrwx0wy0wx1wy1rrs @@r_span_vertical_text_for_rectrsF J4T:1 4ILL!E),,$773 >ILL!E),,$773 > r2t ,   y$/ 0 1 M&4" "Cc3Ci3 Ci3  U*U U Q M  s1c.122S8 7J 77!; 99 Y  rc |sy|Dcgc]\}}}}td||z }}}}t|t|dz}td|dz}g}t|dD]\}}}} ||zdz } d} |} |D]&} t| t | d z }|| ks#| } |} (| |j | || fgd Y| d j || ft| d }t | d |d z z| z|z | d <g}t|d D]8} |j dj dt| d dD:dj |Scc}}}w)a Join center-point-selected cell words into text, re-synthesizing lines. ``words`` are (y0, x0, y1, text) tuples; they are grouped into lines by an adaptive median-height nearest-line rule, each line's words ordered by x, and the lines joined by newlines.rog?rr%g?c|d|dfSr8rrbs rrdz*_span_words_to_line_text.. sDGT!W;MrreNcenter_y)rwordsrr$ct|dS)Nr)rrbs rrdz*_span_words_to_line_text..suT*5E/Frrpc3&K|] \}}| ywrrrs rrGz+_span_words_to_line_text..s"gGAt4"grc |dSrrrbs rrdz*_span_words_to_line_text..s ^bcd^err)r(rr'absrrrt)rr>rr?heights median_heightline_thresholdrr0r r best_line best_distancerudistancer+ text_liness r_span_words_to_line_textrs 5:;;\RBs3R ;G;7OCLA$56Mmd23N E"5.MNZBD2g_ &  )D2d:&6 778H=( (  )   LLbRJ<@ A '!!2t*-Ig&'!&y'_x1r?r s r_span_word_line_tupler!s1 BCT "IuRy%)SY 77rcg}t|D]X\}}|\}}}}} t| js(||zdz} ||zdz} t| | |sF|j ||fZ|S)z(index, word) pairs whose center lies in rect, index into ``page_words``. The index is what lets resolve_spans claim each page word for exactly one placement (an earlier cell's word is not re-claimed by a later one).r) enumeraterrrsrBr) page_wordsr3selectedr,rrrrrr rrs r_span_select_words_in_rectr's H ,+ t#' S#sD4y Ci3 Ci3  r2t , OOUDM *+ Orc ~t|||}||St|Dcgc]\}}t|c}}Scc}}w)zIText for a rect: vertical-line text if it dominates, else line synthesis.)rrr)rr3r vertical_textrrs r_span_words_text_for_rectr8s@0t^LM  #P^$_WQ%:4%@$_ ``$_s9 ct||Dcgc] \}}||vr||f}}}|D]\}}|j|t|||Scc}}w)zKText of rect's words, skipping words already claimed and claiming the rest.)raddr)rr3r claimed_wordsr,rrrs r_span_claim_text_in_rectr@so6j$G E4  %  H !q% ! $T4 ::sA ct|dd}||Sg}|jdjdgD]}|jdgD]}|jdgD]}t|jdxsdj }t |d kr>|jd }|sRt j|}|jrt|j||f t|d||S#t$rY|SwxYw) zPage text spans as (rect, text), length>=2, cached on the page. These drive merged-cell detection: a single span whose x-extent crosses a grid column line signals cells the line grid split but text joins._span_text_spans_cacheNr~rrrnr rorr ) rrrqrrrsr'rOrPrzrrr) rrrnrrurvr r r3s r_span_text_spansrLs T3T :F  Ev&**8R8 +IIgr* +D"- +488F+1r288:t9q=xx'||D)== dD\* + + + .6 L   L s- C<< D D c t|dkrgStd|D}td|D}g}|D]\}}|j|jzdz }||dz ks||dzkDr3t |D cgc]\}} t || r|} }} t| dkrkt| D]\} } | | kDs |j| | |f!t|Scc} }w)aNColumn ranges within one grid row that a single text span crosses. ``entries`` are the row's cell rects (in column order). A text span centered in the row band and substantially overlapping >=2 adjacent cells marks those cells for merging; overlapping intervals are merged. Returns (start, end, texts) with end>start.rc34K|]}|jywrrKrFentrys rrGz+_span_crossing_intervals..r/e/rIc34K|]}|jywrrNrs rrGz+_span_crossing_intervals..srrIr%r") r'r2r(r>r?rr[r_rrl) entries text_spansrow_y0row_y1rgrW span_textrpositionr hit_positionsr)r*s r_span_crossing_intervalsris  7|a /w/ /F /w/ /FI * : 9LL9<</36 fsl "h#&= $-W#5 %*9e<   }  ! 1-@ :JE3U{  %i!89 : : ! ++ s:Ccr||k\ry|D],}|ddvr |d|ks|d|k\rt||ds,yy)NFrole>header header_leaf header_grouprowcolsT)r<)row_idxrbase body_startothers r_span_is_leaf_header_rangers\* = I I  <7 "eElj&@  !$f 6  rc jg}t|D]r\}}t|Dcgc]}|| c}D]N\}}tj|}t||} |j |||| | d| dz ||k\rdnddPt|D]&} | ddk7r t | d| d||rd nd | d<(|Scc}w) a Describe every grid cell by row/slot-range/role for the strict-colspan gate. Header cells split into ``header_leaf`` (no header cell below within body) vs ``header_group`` (spans over a lower leaf), which the reject rules compare against the body split.r$rbodyr)rrr3rr rrrrrr)rrOrPr6rr) cellsr4rrrrcellrr3rrcs r_span_build_base_cellsrs D!%(  '#(R$AQ(RS NHd<<%D#D,7D KK" (  #Awa0&-&;F      <8 # *$u+tF|T:V  V   K+)Ss B0 B0 c t|Dchc]}|ddk(s |d|k\s|dc}}|D]}|Dcgc]#}|ddk(r|d|k(rt|dr|%}}|s4t|dk(r |ddk(rNtfd|D}t fd|D}|dks|dk\sy y cc}wcc}w) zDTrue if some body row splits the column range ``cols`` into >1 cell.rrrrr$rc3HK|]}td|ddywrrNr(rFrcrs rrGz._span_body_rows_split_under..s$KdCQfa9K"c3HK|]}td|ddywr$rNr2rs rrGz._span_body_rows_split_under..s$ID#d1gtF|A7IrTF)rr<r'r2r() rrrrc body_rowsrr5 covered_start covered_ends ` r_span_body_rows_split_underrsmV 8NSWX]S^blSlU mnI  F|v%$u+*@E[\`bfgmbnEo     t9>d1gfo5 KdKK IDII DG # tAw(>  n s B>B>B>(Cc |Dcgc]}|ddk(s t|ds|}}|syt|dk(r |ddk(rytfd|D}tfd|D}|dkxr|dk\Scc}w) zGTrue if leaf-header cells split the column range ``cols`` into >1 cell.rrrFr$rc3HK|]}td|ddywrrrs rrGz0_span_header_leafs_split_over..s$G$DGT&\!_5Grc3HK|]}td|ddywrrrs rrGz0_span_header_leafs_split_over..s$Ec$q'4<?3Er)r<r'r2r()rrrcr5rrs` r_span_header_leafs_split_overrs! rTT&\]%BG]^bdhiodpGqD rD r  4yA~$q'&/T1G$GGMEEEK DG # > tAw(>> ss BBBc|d|dz dkry||krt||||syt|||ryyt||ryy)zJWhether to reject a candidate merge over ``cols`` as a (reason,) mismatch.r$r)Fro)T.header_leaf_colspan_changed_against_body_split)T.body_colspan_changed_against_header_leaf_split)rrrrrrrs r#_span_reject_colspan_mismatch_mergersU AwaA)'4zJ &tT: >I$T40E rctg}|||dzD]*}t||}|jt|||,|S)Nr$)rrr)rrr)r*rrkrrs r_span_cell_texts_for_entriesrsL Eq)D*:u= .tUEBCD Lrc|dk7ryt|||||Dcgc]}|j}}t| Scc}w)zAllow an otherwise-rejected header merge when a covered part is empty. A header leaf that would be rejected against a body split is still merged if one of the merged parts has no text (an empty header slot the body fills).rF)rrsall)reasonrrr)r*rr part_textss r0_span_allow_header_colspan_merge_with_empty_partrsL AA+GgW\^acm+no4$**,oJo: ps=F)header_row_countstrict_colspanc 2t|}g}g}|D]o}|D]h}|tj|} |j| j| j g|j| j | jgjqt|} t|} |r#tdtt|xsd|nd} |r t|| | ng} t|}t|}t}g}t!|D]\}}g}|Dcgc]}|tj|}}t#||}|Dcic] \}}}|||f }}}}d}|t|kr|j%|}|d|\}}t'|||dz} |rFt)| | }t+||| | \}} |rt-| |||||rd}|r ||} |dz }n|dz}n|dz}n ||} |dz }| j.rt1| j| j | }!t1| j | j| }"|j3t5t7| t9|| |||!|"|t|kr|j3||Scc}wcc}}}w)aResolve a detected table's merged-cell (colspan/rowspan) structure. ``cells`` is a row-major grid: a list of rows, each a list of ``[x0, y0, x1, y1]`` cell rectangles (``None`` for a gap) -- the same grid shape :func:`refine_grid` accepts. Returns a row-major grid of :class:`SpanCell` placements, ragged where cells span: each placement carries its union ``bbox``, the page ``text`` it claims, and its ``colspan``/``rowspan`` (how many clustered column/row slots it covers). Every page word is claimed by at most one placement. ``header_row_count`` is the number of leading header rows (default a conservative 1); it only matters together with ``strict_colspan``. ``strict_colspan`` (default False), when set, refuses a merge whose colspan would contradict the header/body column split. This is a PyMuPDF extension; it reads page text/graphics but does not mutate the page. r$rr)rrrr)r*rF)r r r r )r'rOrPextendr0r1r>r?r!r(r2intrrrsetrrrqrRr6rrrzr.rrtupler)#rrrrrowsx_edgesy_edgesrrr3r4 y_boundariesr base_cellsrrr placementsr placement_rowrrgr)r*rkintervals_by_startrinterval end_position_crossing_textscandidate_colsrejectrr r s# r resolve_spansrs" u:DGG/ /D|<<%D NNDGGTWW- . NNDGGTWW- .  //.g6L-g6LBFQC 0 5A6=>AJLZ'|ZH`bJ!$'J#D)JEMJ!%(3)  25J$9I7<<%JJ,WjA KTUU6GeS%ec5\1UUW%)--h7H#08- o'rsB ,58 CF7: &'[- (  #$: F!B8 "a ;:,B @( ? 48[r