kh  ddlmZmZmZmZmZddlmZmZddl Z ddl m Z ddl mZddlmZmZ ddeed eeegefd eeegefd ed eef d Z ddeededed eefdZ ddeeded eeeeeeffdZy))ListUnionAnyCallableIterable)partialreduceN) csr_matrix)connected_components)BaseLayoutElement TextBlocksequence scoring_funcaggregation_funcdefault_score_valuereturnc |d}t|}tj||f|z}t|D]+}t|dz|D]}||||||||<-t |}t |dd\} } g} t| D]E} tj | | k(d} | j|| Dcgc]}|| c}G| Scc}w)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. c|SN)xs _/var/www/teggl/fontify/venv/lib/python3.12/site-packages/layoutparser/tools/shape_operations.pyz=generalized_connected_component_analysis_1d..:sQFT)csgraphdirected return_labelsr)lennponesranger r whereappend)rrrrseq_lenadjacency_matrixijgraph n_componentslabelsgrouped_sequencecomp_idx element_idxs r+generalized_connected_component_analysis_1dr0s@&(mGww125HH 7^Lq1ug& LA%1(1+x{%K Q  " LL ' (E/TL&,'Vhhv1215  0{1S!(1+1S TUV 2Ts< Clayout x_tolerance y_tolerancecHd}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. cV|jjd}|jjd}|jjddd\}}|jjddd\}} t||z |kxr't t|| z t||z |kS)Nrrr )blockcenter coordinatesabsmin) token_atoken_br2r3y_ay_ba_lefta_rightb_leftb_rights ris_line_wise_closez1simple_line_detection..is_line_wise_closensmm""1%mm""1%!--33ADqD9!--33ADqD9 c Nk ) QC()3w/?+@A[P r)r3r2cJtdjj|SNrr __class__unionseqr1s rrz'simple_line_detection..sVF1I,?,?,E,Es%Krrr)r0r)r1r2r3rCdetected_liness` rsimple_line_detectionrMPs5<  A K[ L N r union_groupc:|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. cJtdjj|SrErFrIs rrz4group_textblocks_based_on_category..svfQi.A.A.G.G'MrNc4|j|jk(Sr)type)abs rrz4group_textblocks_based_on_category..s!&&AFF"2rrK)r0)r1rNrdetected_group_boxess` r"group_textblocks_based_on_categoryrVs0&MF2) rrE) rW)T)typingrrrrr functoolsrr numpyr! scipy.sparser scipy.sparse.csgraphr elementsr rintr0rMboolrVrrrr`s87%#53 48 43i4C:s?+4S {C/04 4  #Y 4pTV6 & '6696MP6 6t6: Y  .2  4 ?Di1 12 r