Ng  ZddlmZmZmZmZmZddlmZmZddl Z ddl m Z ddl mZddlmZmZ ddeed eeegefd eeegefd ed eef d Z ddeededed eefdZ ddeeded eeeeeeffdZdS))ListUnionAnyCallableIterable)partialreduceN) csr_matrix)connected_components)BaseLayoutElement TextBlocksequence scoring_funcaggregation_funcdefault_score_valuereturnc|d}t}tj||f|z}t|D]9}t|dz|D]#}||||||<$:t |}t |dd\} } g} t| D]J} tj| | kd} | |fd| DK| S) aPerform connected componenet analysis for any 1D sequence based on the scoring function and the aggregation function. It will generate the adjacency_matrix for the 1D sequence object using the provided `scoring_func` and find the connected componenets. The `aggregation_func` will be used to aggregate all elements within identified components (when not set, it will be the identity function). Args: sequence (List[Any]): The provided 1D sequence of objects. scoring_func (Callable[[Any, Any], int]): The scoring function used to construct the adjacency_matrix. It should take two objects in the sequence and produe a integer. aggregation_func (Callable[[List[Any]], Any], optional): The function used to aggregate the elements within an identified component. Defaults to the identify function: `lambda x: x`. default_score_value (int, optional): Used to set the default (background) score values that should be not considered when running connected component analysis. Defaults to 0. Returns: List[Any]: A list of length n - the number of the detected componenets. Nc|SN)xs _/var/www/html/ai-engine/env/lib/python3.11/site-packages/layoutparser/tools/shape_operations.pyz=generalized_connected_component_analysis_1d..:sQFT)csgraphdirected return_labelsrc g|] }| Srr).0irs r z?generalized_connected_component_analysis_1d..Ks1S1S1S!(1+1S1S1Sr)lennponesranger r whereappend)rrrrseq_lenadjacency_matrixr"jgraph n_componentslabelsgrouped_sequencecomp_idx element_idxs` r+generalized_connected_component_analysis_1dr3sE@&;(mmGw1225HH 7^^LLq1ug&& L LA%1\(1+x{%K%K Q  " " L ' ( (E/TL&,''VVhv12215  0 01S1S1S1S{1S1S1S T TUUUU r layout x_tolerance y_tolerancecXd}tt|||fd}|S)aFPerform line detection based on connected component analysis. The is_line_wise_close is the scoring function, which returns True if the y-difference is smaller than the y_tolerance AND the x-difference (the horizontal gap between two boxes) is also smaller than the x_tolerance, and False otherwise. All the detected components will then be passed into aggregation_func, which returns the overall union box of all the elements, or the line box. Args: layout (Iterable): A list (or Layout) of BaseLayoutElement x_tolerance (int, optional): The value used for specifying the maximum allowed y-difference when considered whether two tokens are from the same line. Defaults to 10. y_tolerance (int, optional): The value used for specifying the maximum allowed horizontal gap when considered whether two tokens are from the same line. Defaults to 10. Returns: List[BaseLayoutElement]: A list of BaseLayoutElement, denoting the line boxes. c>|jjd}|jjd}|jjddd\}}|jjddd\}} t||z |ko3t t|| z t||z |kS)Nrrr )blockcenter coordinatesabsmin) token_atoken_br6r7y_ay_ba_lefta_rightb_leftb_rights ris_line_wise_closez1simple_line_detection..is_line_wise_closensm"1%m"1%!-3ADqD9!-3ADqD9 c NNk ) QC())3w/?+@+@AA[P r)r7r6cDtdjj|SNrr __class__unionseqr5s rrz'simple_line_detection..sVF1I,?,Es%K%Krrr)r3r)r5r6r7rGdetected_liness` rsimple_line_detectionrQPs\<    A K[   LKKK N rT union_groupcB|rfd}nd}td|}|S)a0Group textblocks based on their category (block.type). Args: layout (Iterable): A list (or Layout) of BaseLayoutElement union_group (bool): Whether to union the boxes within each group. Defaults to True. Returns: List[TextBlock]: When `union_group=True`, it produces a list of TextBlocks, denoting the boundaries of each texblock group. List[List[TextBlock]]: When `union_group=False`, it preserves the elements within each group for further processing. cDtdjj|SrIrJrMs rrz4group_textblocks_based_on_category..svfQi.A.G'M'MrNc"|j|jkSr)type)abs rrz4group_textblocks_based_on_category..s!&AF"2rrO)r3)r5rRrdetected_group_boxess` r"group_textblocks_based_on_categoryrZsN& MMMMF22) rrI)r4r4)T)typingrrrrr functoolsrr numpyr% scipy.sparser scipy.sparse.csgraphr elementsr rintr3rQboolrZrrrrcs87777777777777%%%%%%%%######55555533333333 48 443i4C:s?+4S {C/04 4  #Y 4444pTV66 & '6696MP6 6666t6:  Y  .2  4 ?Di1 12      r