o 0j- @sddlZddlZddlZddlZddlmZddlmZddlm Z m Z m Z m Z m Z ddlZddlmZddlmZmZmZddlmZmZmZd d Z dad e e eefd e e eefdedefddZde ee e fdede ee e ffddZddZ ddZ!ddZ"dbddZ#dcd d!Z$d"d#Z%d$d%Z&d&d'Z'Gd(d)d)eZ(Gd*d+d+eZ)d,Z*d-Z+d.Z,d/Z-d0Z.d1Z/d2Z0ej1e0d3e0d4ej2d5Z3d6efd7d8Z4d9d:Z5d;e)fdedefd?d@Z7dAefdBdCZ8d6ede edffdDdEZ9 Gddd6edHedIede e eeefdffdJdKZ: L L L MdedNedOedPedHedQedef dRdSZ;dTdUZe1dWej=Z?e1dXZ@e1dXZAdYejBde ee ffdZd[ZCd\ed]ed^ede ee ee fffd_d`ZDdS)fN)Counter)deepcopy)AnyDictListTupleUnion)Image) BaseModelcomputed_fieldmodel_validator)calculate_bbox_areacalculate_overlap_ratio"calculate_projection_overlap_ratiocCs|js|d}|S)Nr)Zis_validbuffer)Zpolyro/var/www/html/Deteccion_Ine/venv/lib/python3.10/site-packages/paddlex/inference/pipelines/paddleocr_vl/uilts.py make_valid"s runionpolygon1polygon2modereturnc Cszddlm}Wn tytdw||}||}t|}t|}||j}||j}|dkr8||S|dkrGt|j|j}||S|dkrVt|j|j} || St d|)a Calculate the overlap ratio between two polygons. Args: polygon1 (List[Tuple[int, int]]): First polygon represented as a list of points. polygon2 (List[Tuple[int, int]]): Second polygon represented as a list of points. mode (str, optional): Overlap calculation mode. Defaults to "union". Returns: float: Overlap ratio value between 0 and 1. r)PolygonzPlease install Shapely library.rsmallZlargezUnknown mode: ) Zshapely.geometryr ImportErrorr intersectionZarearminmax ValueError) rrrrZpoly1Zpoly2rrZ small_areaZ large_arearrrcalculate_polygon_overlap_ratio(s(   r!layout_det_reslayout_shape_modecst|}dd|dD}ttt|D]}||d\}}}}||||} } | dks3| dkr8|t|dt|D]} |vsK| vrLqAt||d|| dd} ||dd ksj|| dd kr| d kr||dd kr{||| dd kr| qA| d kr|d krd ||vrt||d || d d} | d krqAt||d}t|| d}||d|| dh}|hd@rt|dkrd|vs|hdkrqA||kr| qA|qAqfddt|D|d<|S)a. Remove overlapping boxes from layout detection results based on a given overlap ratio. Args: layout_det_res (Dict[str, List[Dict]]): Layout detection result dict containing a 'boxes' list. Returns: Dict[str, List[Dict]]: Filtered dict with overlapping boxes removed. cSsg|] }|ddkr|qS)label referencer).0boxrrr [z(filter_overlap_boxes..boxes coordinaterr$Zinline_formula?gffffff?rectZpolygon_points>imagetablechartZsealr1csg|] \}}|vr|qSrr)r&idxr'Zdropped_indexesrrr(r)) rsetrangelenaddrr!r enumerate)r"r#Zlayout_det_res_filteredr*ix1y1x2y2whjZ overlap_ratioZpoly_overlap_ratioZ box_area_iZ box_area_jlabelsrr4rfilter_overlap_boxesNs\       & rCcCst|tjr|St|S)z Convert the input to a PIL Image. Args: img (PIL.Image or numpy.ndarray): Input image. Returns: PIL.Image: PIL Image object. ) isinstancer Z fromarrayimgrrr to_pil_images  rGcCst|tjr t|S|S)z Convert the input to a numpy array. Args: img (PIL.Image or numpy.ndarray): Input image. Returns: numpy.ndarray: Numpy array image. )rDr nparrayrErrr to_np_arrays rJcCs4dd|D}dd|D}t|}t|}||fS)z Calculate width (max of all) and height (sum) for a vertical merge of images. Args: images (List[PIL.Image or np.ndarray]): List of images. Returns: Tuple[int, int]: (width, height) of merged image. cSg|]}t|jqSr)rGwidthr&rFrrrr(z"calc_merged_wh..cSrKr)rGheightrMrrrr(rN)rsum)imageswidthsZheightsr?r@rrrcalc_merged_whs  rScenterautoc Cs6|sdSt|dkrt|dSt|tr|gt|d}t|t|dkr,tdt|d}tdt|D]]}t||}||d}t|j|j}|j |j }t d||fd} |dkrp||jd} ||jd} n|d kr||j} ||j} nd} } | || df| || |j f| }q9t|S) a+ Merge images vertically with given alignment. Args: images (List[PIL.Image or np.ndarray]): List of images to merge. aligns (str or List[str]): Alignment(s) for each merge step ('center', 'right', 'left'). Returns: np.ndarray: Merged image as numpy array. Nr-rz,The length of aligns must be len(images) - 1RGBrXrXrTr right) r7rJrDstrr rGr6rrLrOr newZpaste) rQalignsr#Zmergedr:Zimg2alignr?r@Znew_imgr;r=rrr merge_imagess4          r^c)sg}i}tD]\}}|dvr|||<q|||fqg}g}g} g} ddfdd} fdd} t|D]\} \}}|sL|g}|g} g} q;|| d\}}|d }|d}|d }|d}t||d }|d ko|d ko||ko|d |d ko|d|dko|d |d t|d |d |d |d dk}|d ko|dvo||ko|d|dkot|d|dt|d|d|d|ddko|d |d |d |d Ao| ||}|rd}n |r| ||}nd}|s|r||| || |q;|| | f|g}|g} g} q;|r(|| | fg}|D]\}}t|t|}}|||||fq,g}td }|tkrQd}|D]\}}}}||kr)tfdd|Dr)d}fdd|D}|r}|ng} t |\}!}"|!d kr|"|!nt d}#|#dkrt|D]!\}$}%|% }|%d|d<d|d<|| |%qn.is_alignedcs0|d|dr dS|d|drdSdS)Nrleftr rYrTr) block_bbox prev_bbox)rbrr get_alignments z#merge_blocks..get_alignmentc s||d}||d}t|d|d}t|d|d}t|d|d}t|d|d}||||g} t|D]\} } | ||fvsJ| dvrKq:| d} t| | dkrYdSq:dS) Nr'rr-r r$TF)rrr9r) block_idxprev_idxblocksrerdr;r<r=r>Zmin_boxr3Z other_blockZ other_bbox)non_merge_labelsrroverlapwith_other_box s     z+merge_blocks..overlapwith_other_boxr-r' horizontalrtextr rgg333333?)rnr.rTNFc3s|]}|vVqdSNrr&r:) used_indicesrr azmerge_blocks..Tcsg|]}|dqSrErrp)rjrrr(cz merge_blocks..infrF merge_alignsZgroup_id)r9appendrrrarr5r7allrSfloatcopyr8r^r6))rjrkr#Zblocks_to_mergeZnon_merge_blocksr3blockZ merged_groupsZ current_groupZcurrent_indicesZcurrent_alignsrfrlr:riZ prev_blockreZ prev_labelrdZ block_labelZiou_hZis_crossZis_updown_alignZ align_modeZ group_rangesZ group_indicesr\startendZ result_blocksZ group_foundZimgsrvr?r@Z aspect_ratiorArhZ merged_imgZ insert_listZn_idxr)rjrbrkrqr merge_blockss      $ $        "           (r~c sddldfdd }dd|D\}}}}||}||} |} j| ||f||fdd d j} d } ||| t|| dd \} }}td t| | }|||d}|| |d}j| |||f| | d|j d| S)aD Fill a rectangular area in the image with a white background and write the given token string. Args: image (np.ndarray): Image to paint on. box (tuple): (x1, y1, x2, y2) coordinates of rectangle. token_str (str): Token string to write. Returns: np.ndarray: Modified image. rN?c szd\}}|}||dkr8||d}j|||dd\\}} } |||kr0| ||kr0|}|}n|}||dks ||| fS)N)g? g{Gz?r r-) thickness)Z getTextSize) rnZfontFaceZ square_size fill_ratiorcrYZ optimal_scalemidr?r@_cv2rrget_optimal_font_scales   z+paint_token..get_optimal_font_scalecSsg|]}t|qSr)int)r&vrrrr(szpaint_token..rW)colorr)rr-r )rrr)ZlineType)r) rrzZ rectangleZFONT_HERSHEY_SIMPLEXrrmathfloorZputTextZLINE_AA)r0r' token_strrr;r<r=r>Zbox_wZbox_hrFZfontZthickness_scale_ratioZ font_scaleZtext_wZtext_hZfont_thicknessZtext_xZtext_yrrr paint_tokens4   rcsdd}ddl}|di}|\}}}} g|t|} || t|D]Q\} } | d\} }}}| |krw||krw||krw|| krw| t|| ||dkrSq&| |||||||g}dt| | d }t|||}| d ||<q&fd d t|D}|||fS) a Replace figures in a table area with tokens, return new image and token map. Args: table_block_img (np.ndarray): Table image. table_box (list): Table bounding box [x_min, y_min, x_max, y_max]. figures (List[Dict]): List of figure dicts (must contain 'coordinate', 'path'). Returns: Tuple[np.ndarray, Dict[str, str], List[str]]: - New table image, - Token-to-img HTML map, - List of figure paths dropped. cSsNhd}g}d}t||kr%tt||@s|||d7}t||ks|S)N>109rr-)r7r5rZrw)numZexclude_digitsseqr:rrrgen_random_maps   z0tokenize_figure_of_table..gen_random_maprNir+[F]pathcs g|] \}}|vr|dqS)rr)r&r:fZ drop_idxesrrr( z,tokenize_figure_of_table..) randomseedr7shuffler9rwrrZr)Ztable_block_imgZ table_boxZfiguresrrZ token_mapZ table_x_minZ table_y_minZ table_x_maxZ table_y_maxZ random_mapZ figure_idZfigureZ figure_x_minZ figure_y_minZ figure_x_maxZ figure_y_maxZdraw_boxrZ drop_figuresrrrtokenize_figure_of_tables8        rcs fdd}d}t|||S)z Replace tokens in a string with their HTML image equivalents. Args: table_res_str (str): Table string with tokens. figure_token_map (dict): Mapping from tokens to HTML img tags. Returns: str: Untokenized string. cs|d}d|d}||d}|d}|dur#|dSg}|d|dddd d|}|jdkrJ|j}|d |d 7}|S) Nr-rrrzImagez-    )groupgetrwformatreplacejoincontent)matchZtoken_idtokenZimg_pathZ img_blockZimg_tagsZ image_infoZ ocr_contentfigure_token_mapimage_path_to_obj_maprrrepls"      z(untokenize_figure_of_table..replz \[F(\d+)\])resub)Z table_res_strrrrpatternrrruntokenize_figure_of_tables rc@seZdZUdZdZeed<dZeed<eed<eed<eed<eed<eed <d Z e ed <d Z e ed <d Z e ed <e ddededefddZdS) TableCella TableCell represents a single cell in a table. Attributes: row_span (int): Number of rows spanned. col_span (int): Number of columns spanned. start_row_offset_idx (int): Start row index. end_row_offset_idx (int): End row index (exclusive). start_col_offset_idx (int): Start column index. end_col_offset_idx (int): End column index (exclusive). text (str): Cell text content. column_header (bool): Whether this cell is a column header. row_header (bool): Whether this cell is a row header. row_section (bool): Whether this cell is a row section. r-row_spancol_spanstart_row_offset_idxend_row_offset_idxstart_col_offset_idxend_col_offset_idxrnF column_header row_header row_sectionbefore)rdatarcCslt|tr4d|vr |S|ddd}t|s0|dd}|r,|D] }||dd7}q!|}||d<|S)z Create TableCell from dict, extracting 'text' property correctly. Args: data (Any): Input data. Returns: Any: TableCell-compatible dict. rnZbboxrrZtext_cell_bboxesNr)rDrrr7popstrip)clsrrnZ text_cellselrrrfrom_dict_formatCs  zTableCell.from_dict_formatN)__name__ __module__ __qualname____doc__rr__annotations__rrZrboolrrr classmethodrrrrrrr's      rc@sXeZdZUdZgZeeed<dZe ed<dZ e ed<e e deeefddZ d S) TableDataz TableData holds a table's cells, row and column counts, and provides a grid property. Attributes: table_cells (List[TableCell]): List of table cells. num_rows (int): Number of rows. num_cols (int): Number of columns. table_cellsrnum_rowsnum_colsrcs|fddtjD}jD],}tt|jjt|jjD]}tt|jjt|jjD]}||||<q1q q|S)z Returns a 2D grid of TableCell objects for the table. Returns: List[List[TableCell]]: Table as 2D grid. cs$g|]fddtjDqS)c s&g|]}tdd||ddqS)rr-)rnrrrr)rr&rAr:rrr(usz-TableData.grid...)r6r)r&selfrrr(ts  z"TableData.grid..) r6rrrrrrrr)r table_datacellr:rArrrgridks        zTableData.gridN)rrrrrrrrrrrr propertyrrrrrr]s   rzzzzzzz+(?:|||||)z.*?(?=z|$))flagssc CsLddttttttgd}t||}t ||}dd|D}||fS)z Extract OTSL tags and text parts from the input string. Args: s (str): OTSL string. Returns: Tuple[List[str], List[str]]: (tokens, text_parts) (|)cSsg|]}|r|qSrr)r&rrrrr(rtz0otsl_extract_tokens_and_text..) rOTSL_NL OTSL_FCEL OTSL_ECEL OTSL_LCEL OTSL_UCEL OTSL_XCELrfindallsplit)rrtokensZ text_partsrrrotsl_extract_tokens_and_texts   rc s~tddt|fddD}g}d}d}|rtdd|D}|D]}t||kr7|tt||ks,q$g}d} |D]N}|D]4} || | t|krv|| | krv| d7} | t|krv|| ttttt t fvrv||| | d7} qB|t| t|kr|| tkr| d7} q>|}d d } d d } t |D]\} }d }|ttfvr!d}d}d}|tkr|| d}d}| |t|kr|| |nd }d }|dt|kr|t||dkr||d|}|tt fvr|| ||d|tt g7}|t t fvr || |||dt t g7}|t | ||||||||d|tttt t fvr/|d7}|tkr:|d7}d}q||fS)a Parse OTSL text and tags into TableCell objects and tag structure. Args: texts (List[str]): List of tokens and text. tokens (List[str]): List of OTSL tags. Returns: Tuple[List[TableCell], List[List[str]]]: (table_cells, split_row_tokens) cSsg|] \}}|st|qSr)list)r&xyrrrr(s z$otsl_parse_texts..cs|kSror)zZ split_wordrrsz"otsl_parse_texts..rcs|]}t|VqdSror7r&rowrrrrrrsz#otsl_parse_texts..r-cSsPd}|}||||vr&|d7}|d7}|t||kr|S||||vs |SNrr-r)rc_idxr_idx which_tokensspanZ c_idx_iterrrr count_rightsz%otsl_parse_texts..count_rightcSsLd}|}||||vr$|d7}|d7}|t|kr|S||||vs |Srr)rrrrrZ r_idx_iterrrr count_downs z$otsl_parse_texts..count_downrr )rnrrrrrr)r itertoolsgroupbyrr7rwrrrrrr9rr)textsrsplit_row_tokensrrrZmax_colsrZ new_textsZtext_idxrrrr:rnZ cell_textrrZ right_offsetZnext_right_cellZnext_bottom_cellrrrotsl_parse_textss            rrc Cs|j}|j}t|jdkrdSd}|j}t|D]f}|d7}t|D]W}|||}|j|j}} |j|j } } | |ks@| |krAq"t |j } |jrNdnd} | }|dkr_|d|d7}| dkrk|d | d7}|d |d | d | d 7}q"|d 7}qd|d}|S)z Export TableData to HTML table. Args: table_data (TableData): TableData object. Returns: str: HTML string. rrzthtdr-z rowspan=""z colspan="<>zzz
)rrr7rrr6rrrrhtmlescapernrr)rZnrowsZncolsbodyrr:rArZrowspanZrowstartZcolspanZcolstartrZcelltagZ opening_tagrrrexport_to_html$s2      rotsl_strcst|tsJ|}t|vr|tS|t}g}|D].}|s!qt|}|s)qt|}d}t|D] \}}| t r@|d}q3| |||dq|sOtS|rZt dd|Dnd} |rgt dd|Dnd} | } t | | } t d} | }t| | dD]tfdd|D}|| kr|} }q}g}|D](}|d }t|}||kr|d |}n tg||}||}| d |qt|tS) z Pad OTSL string to a square (rectangular) format, ensuring each row has equal number of cells. Args: otsl_str (str): OTSL string. Returns: str: Padded OTSL string. rr-) raw_cells total_lenmin_lencs|]}|dVqdS)r Nrrrrrrrirsz%otsl_pad_to_sqr_v2..csr r Nrrrrrrrjrsruc3s |] }t|dVqdSrr`rrLrrrrqsr Nr)rDrZrrrOTSL_FIND_PATTERNrr7r9 startswithrrwrryr6rPrr)r linesZrow_dataliner r r r:Zcell_strZglobal_min_widthZ max_total_lenZ search_startZ search_endZmin_total_costZ optimal_widthZcurrent_total_costZrepaired_linesrcellsZ current_lenZ new_cellspaddingrrrotsl_pad_to_sqr_v2IsZ      r otsl_contentcCsRt|}t|\}}t||\}}tt||r tdd|Dnd|d}t|S)z Convert OTSL-v1.0 string to HTML. Only 6 tags allowed: , , , , , . Args: otsl_content (str): OTSL string. Returns: str: HTML table. csrrorrrrrrrrsz'convert_otsl_to_html..r)rrr)rrrrr7rr)rrZ mixed_textsrrrrrrconvert_otsl_to_htmls rcCsTt|}td|ddD]}||dkr'|d|}||||kr'|Sq dS)z Find the shortest substring that repeats to form the entire string. Args: s (str): Input string. Returns: str or None: Shortest repeating substring, or None if not found. r-r rN)r7r6)rnr:Z substringrrr!find_shortest_repeating_substrings   rr_r  min_repeatscCstt|||ddD]<}|| d}|||rHd}|}||r5|d| }|d7}||s%t|||}|d|||fSq dS)a! Detect if string ends with a repeating phrase. Args: s (str): Input string. min_len (int): Minimum length of unit. min_repeats (int): Minimum repeat count. Returns: Tuple[str, str, int] or None: (prefix, unit, count) if found, else None. r-rNr)r6r7endswith)rr rr:unitcountZtemp_s start_indexrrrfind_repeating_suffixs  r!r rline_thresholdchar_threshold min_countcCst||kr|S|}|s|Sd|vr6t|dkr6t|ddd}|r6|\}}} t|| t|dkr6|Sd|vrTt||krTt|}|rTt|t|} | |krT|Sdd|dD} | sb|St| } | |krl|St| } | d d \} } | |kr| | d kr| S|S) a Detect and truncate character-level, phrase-level, or line-level repetition in content. Args: content (str): Input text. line_threshold (int): Min lines for line-level truncation. char_threshold (int): Min repeats for char-level truncation. min_len (int): Min length for char-level check. Returns: Union[str, str]: (truncated_content, info_string) rdrr_)r rr.cSsg|] }|r|qSrr)r&rrrrr(sz/truncate_repetitive_content..r-rg?)r7rr!rrr most_common)rr#r$r r%Zstripped_contentZ suffix_matchprefixZrepeating_unitrrZ total_linesZ line_countsZmost_common_linerrrtruncate_repetitive_contents8  r)cCsddl}t|jdkr|||j}n|}|jtjkr#| tj}| }| }||kr1|S||||d}| tj}| |dd|j \}}||}|durW|S||\} } } } || | | | | | f} | S)NrrgrX)rr7shapeZcvtColorZCOLOR_BGR2GRAYrzZdtyperHZuint8Zastyperr thresholdZTHRESH_BINARY_INVZ findNonZeroZ boundingRect)rFrgrayZmax_valZmin_valrrbinaryZcoordsrrr?r@Zcroppedrrr crop_margins&    r/z#<\|TEXT_START\|>(.*?)<\|TEXT_END\|>z!<\|LOC_BEGIN\|>(.*?)<\|LOC_END\|>z<\|LOC_(\d+)\|>r0cCs|jdd\}}|dkr?|dkr?t|}|d|d}}ztjj}Wn ty0tj}Ynw|||f|}t|}|S|}|S)zG Post-process the input image to extract location information. Nr i)r+rGr Z ResamplingZLANCZOSAttributeErrorresizerJ)r0r@r?Z process_wZ process_hZresample_filterZ inference_imgrrrpre_process_for_spotting%s   r2 input_strr?r@cst|tsJt|}t|}g}g}tt|t|}t|D]B}||} t ||} t| dkr8q"t t t | ddfddtdddD} fdd| D} | | | | q"|ri|st t|} d} d}|dt| kr| ||d}d d|Dfd dtdddD} fd d| D} || |d}|} | | | | |d } |d7}|dt| ks|d |}||d}||fS)zL Post-process the input string to extract text and location blocks. rNc g|] }||dfqSr-rrvalsrrr(Rrz-post_process_for_spotting..rr c,g|]}|dd|ddfqSrg@@r-rr&pr@r?rrr(S,cSsg|] }t|dqSr5)rr)r&mrrrr(^scr4r5rrr6rrr(_rcr8r9rr:r<rrr(`r=rr) rec_polys rec_texts)rDrZ ANNOT_TEXT_REr LOC_BLOCK_RErr7r6r LOC_ITEM_RErmaprrw LOC_TOKEN_REfinditerr|r}r)r3r?r@rZ loc_blocksr@rArr:txtZ loc_itemsZptsmatchesZlast_endrZ text_spanZ result_strZ spotting_resr)r@r7r?rpost_process_for_spotting:sF           rJ)r)rTrU)rU)rr_)rrrr")Errrr collectionsrrzrtypingrrrrrnumpyrHZPILr Zpydanticr r r Zlayout_parsing.utilsrrrrrrZryr!rCrGrJrSr^r~rrrrrrrrrrrZNON_CAPTURING_TAG_GROUPcompileDOTALLrrrrrrrr!r)r/SrBrCrDrFZndarrayr2rJrrrrs     & C  )!><$62s%:  9