j?@ddlmZ ddlZddlmZd dZddZdd ZdS)) annotationsN) html_documentreturntuple[int, int, list, list]c  t|}t}d t|D]\}}d}|D]x}||f|vr |dz }||f|v t|jD]6}t|jD]}|||z||zf 7||jz }t | y fdt|D} fdt|D} t} t|D]\}}d}|D]}||f| vr |dz }||f| v | krn|jt|jnd| ||<|j | ||<t|jD]:}t|jD]#}|s|r| ||z||zf$;||jz }| | | fS)aDerive ``(row_count, col_count, cells, extract)`` from a placement grid. ``placements`` is a ``Table.placements`` row-major grid of ``SpanCell`` colspan/rowspan cells. ``row_count`` is the ```` count; ``col_count`` the column extent after resolving spans (the HTML column count). ``cells`` and ``extract`` are the ``row_count x col_count`` post-span bbox and plain-text matrices, ``None`` where a span covers a slot or the grid has a gap.rrcg|]}dgz SN.0_ col_counts /Users/tedpeters/Documents/Projects/Claude Projects/Misc Tools/notename-sheets/.venv/lib/python3.11/site-packages/pymupdf4llm/helpers/table_html/reconstruct.py z,_placement_grid_matrices..& >>>$)#>>>cg|]}dgz Sr r r s rrz,_placement_grid_matrices..'rrN) lenset enumeraterangerowspancolspanaddmaxbboxlisttext) placements row_countoccupiedrow_idxrowcol_idxcelldrdc bbox_grid text_gridcoveredrs @r_placement_grid_matricesr, smJIuuHI!*-- 0 0  0 0DG$001 G$00DL)) ? ? --??BLL'B," !=>>>>? t| #GIw//II 0?>>>U9-=-=>>>I>>>>U9-=-=>>>IeeG!*-- $ $  $ $DG$//1 G$//)##=AY=R$ty///X\Ig w '*.)Ig w 'DL)) B B --BBBBRB Wr\7R<$@AAAB t| #GG iI 55rct|tj }|rtj|n|} ||}||ddd}dt |ddpgD}|r|n#|r|wwxYwtd|DS)aReconstruct the tables on one PDF page and return them as an HTML string. pdf : PDF file path (str/Path) or an already-open pymupdf.Document. page_index : 0-based page number. returns : concatenated ...
HTML (empty string if no tables). When ``pdf`` is an already-open Document, the target page is derotated in place via ``page.remove_rotation()`` (mutates the caller-owned page). T use_layoutunionrefinec6g|]}|Sr )to_html)r tabs rrzto_html..Js OOOC#++--OOOrtablesNcg|]}d|iS)htmlr )r hs rrzto_html..Ns666!61+666r) isinstancepymupdfDocumentopenremove_rotation find_tablesgetattrcloser)pdf page_indexowns_docdocpagetfr5s rr3r3:sc7#3444H' 0',s   SC:    T$  G GOOGB$,G,G,M2OOO   IIKKK   IIKKKK  66v666 7 77s ABB6rE pymupdf.Page4list[tuple[pymupdf.Rect, str, int, int, list, list]]c  |ddd}g}t|ddpgD]_}t|j\}}}}|t j|j|||||f`|S)aReconstruct one already-open page's tables as payload tuples. Returns one ``(bbox, html, rows, cols, cells, extract)`` tuple per table, in reading order, for the markdown/JSON renderers to drive table emission, reading order and body-text exclusion: * ``bbox`` -- ``tab.bbox`` (a grid-ref table keeps its reported layout box); * ``html`` -- ``tab.to_html()``; * ``rows``/``cols``/``cells``/``extract`` -- the reconstructed grid the ``html`` shows (see :func:`_placement_grid_matrices`): ``extract`` is the per-cell plain-text matrix (``None`` for span-covered slots / grid gaps) and ``cells`` the matching post-span bbox matrix, both ``rows x cols``. The caller must remove page rotation before calling. Not thread-safe on a shared Page: core caches word/vector extraction as attributes on the given ``page``, so concurrent calls must each use their own ``pymupdf.Page``. Tr.r5N) r>r?r,r appendr:Rectrr3)rErFresultr4r!rcellsextracts rpage_html_tablesrOQs$   Td  C CB FHd++1r   /G/W/W, 9eW  SX&&     Mr)rr)r)rErGrrH) __future__rr:corerr,r3rOr rrrRs""""""*6*6*6*6Z8888.      r