Ng(+LddlmZddlmZddlmZmZddlmZddl m Z m Z ddl Z ddlmZddlmZeGd d Zd%d Zd&dZd'dZeGddZeGddZGddeZGddeZ d(d)d!Zd*d$ZdS)+) annotations)deepcopy) dataclassfield)cached_property)OptionalUnionN)Source) safe_divisionc"eZdZUded<ded<ded<ded<d"dZd"dZd"d Zed#d Zed#d Z ed#d Z ed#dZ d$dZ d$dZ d%d&dZd'dZedZd(dZed)dZd*dZd*dZd+d,d!ZdS)- RectangleUnion[int, float]x1y1x2y2paddingcV|||}|S)zIncreases (or decreases, if padding is negative) the size of the rectangle by extending the boundary outward (resp. inward).)hpadvpadselfr out_objects e/var/www/html/ai-engine/env/lib/python3.11/site-packages/unstructured_inference/inference/elements.pypadz Rectangle.pads)YYw'',,W55 cdt|}|xj|zc_|xj|z c_|S)zIncreases (or decreases, if padding is negative) the size of the rectangle by extending the left and right sides of the boundary outward (resp. inward).)rrrrs rrzRectangle.hpad5d^^     rcdt|}|xj|zc_|xj|z c_|S)zIncreases (or decreases, if padding is negative) the size of the rectangle by extending the top and bottom of the boundary outward (resp. inward).)rrrrs rrzRectangle.vpad#rrreturnc |j|jz S)zWidth of rectanglerrrs rwidthzRectangle.width+w  rc |j|jz S)zHeight of rectanglerrr#s rheightzRectangle.height0r%rc&|j|jzdz S)z,Finds the horizontal midpoint of the object.r"r#s r x_midpointzRectangle.x_midpoint5$'!Q&&rc&|j|jzdz S)z*Finds the vertical midpoint of the object.r*r'r#s r y_midpointzRectangle.y_midpoint:r,rotherboolc.|| S)zAChecks whether this rectangle is disjoint from another rectangle.) intersectsrr/s r is_disjointzRectangle.is_disjoint?s??5))))rc|s|rdSt||dS)z;Checks whether this rectangle intersects another rectangle.F)r) _has_none intersectionsr3s rr2zRectangle.intersectsCs? >>   u00 5T5))$//rN error_marginOptional[Union[int, float]]c|||n|}t|j|jk|j|jk|j|jk|j|jkgS)zDChecks whether this rectangle is contained within another rectangle.)rallrrrr)rr/r9 padded_others ris_inzRectangle.is_inIse2>2Juyy...PU LO+LO+LO+LO+     rc`t|jdu|jdu|jdu|jdufS)z)return false when one of the coord is nanN)anyrrrrr#s rr7zRectangle._has_noneUs2DGtOTW_dgotwRVWXXXrcn|j|jf|j|jf|j|jf|j|jffS)z!Gets coordinates of the rectangle)rrrrr#s r coordinateszRectangle.coordinatesYs9$'"TWdg$6$'8JTWVZV]L^__rOptional[Rectangle]cf|s|rdSt|j|j}t|j|j}t|j|j}t|j|j}||ks||krdSt||||S)zoGives the rectangle that is the intersection of two rectangles, or None if the rectangles are disjoint.N)r7maxrminrrrr )rr/rrrrs r intersectionzRectangle.intersection^s >>   u00 4 %( # # %( # # %( # # %( # # 77b2gg4RR(((rfloatc |j|jzS)z Gives the area of the rectangle.)r$r(r#s rareazRectangle.areakszDK''rc||}|dn|j}|j|jz|z }t||S)zGives the intersection-over-union of two rectangles. This tends to be a good metric of how similar the regions are. Returns 0 for disjoint rectangles, 1 for two identical rectangles -- area of intersection / area of union.NrGrJr )rr/rGintersection_area union_areas rintersection_over_unionz!Rectangle.intersection_over_unionpsN((// #/#7CC\=NY+.?? . ;;;rc||}|dn|j}t|j|j}t||S)zGives the area-of-intersection over the minimum of the areas of the rectangles. Useful for identifying when one rectangle is almost-a-subset of the other. Returns 0 for disjoint rectangles, 1 when either is a subset of the other.NrL)rGrJrFr )rr/rGrNmin_areas rintersection_over_minimumz#Rectangle.intersection_over_minimumysM((// #/#7CC\=Nty%*--.999r?subregion_thresholdc||}|dn|j}|t||jko|j|jkS)zReturns whether this region is almost a subregion of other. This is determined by comparing the intersection area over self area to some threshold, and checking whether self is the smaller rectangle.NrLrM)rr/rUrGrNs ris_almost_subregion_ofz Rectangle.is_almost_subregion_ofsT((// #/#7CC\=N#m4Ety&Q&QQ I # r)rr)r r)r/r r r0N)r/r r9r:r r0)r r0)r/r r rC)r rH)r/r r rHrT)r/r rUrHr r0)__name__ __module__ __qualname____annotations__rrrpropertyr$r(r+r.r4r2r>r7rBrGrJrPrSrWrrr r s !!!X!!!!X!'''X''''X'****0000      YYYY``X` ) ) ) )(((X(<<<<::::       rr regionsr ctd|D}td|D}td|D}td|D}t||||S)zHReturns the smallest rectangular region that contains all regions passedc3$K|] }|jV dSrX)r.0regions r z,minimal_containing_region..$ - -6VY - - - - - -rc3$K|] }|jV dSrX)rrcs rrfz,minimal_containing_region..rgrc3$K|] }|jV dSrX)rrcs rrfz,minimal_containing_region..rgrc3$K|] }|jV dSrX)rrcs rrfz,minimal_containing_region..rgr)rFrEr )r`rrrrs rminimal_containing_regionrks - -W - - - - -B - -W - - - - -B - -W - - - - -B - -W - - - - -B RR $ $$rrectsc\tjd|D}t|S)zReturns a square boolean matrix of intersections of an arbitrary number of rectangles, i.e. the ijth entry of the matrix is True if and only if the ith Rectangle and jth Rectangle intersect.cBg|]}|j|j|j|jgSr_)rrrr)rdrs r z!intersections..s)???AadAD!$/???r)nparraycoords_intersections)rlcoordss rr8r8s0 X????? @ @F  ' ''rrt np.ndarrayc|dddf|dddf|dddf|dddff\}}}}|d|dk|d|dkz|d|dkjz|d|dkjz}|S)zReturns a square boolean matrix of intersections of given stack of coords, i.e. the ijth entry of the matrix is True if and only if the ith coords and jth coords intersect.Nrr6r*).N)T)rtx1sy1sx2sy2sr8s rrsrss 1vaaad|VAAAqD\6!!!Q$<OCc3 TS^ # t9s9~ % ' t9s9~ % ( ) t9s9~ % ( )M rc\eZdZUded<dZded<dZded<dd Ze dddZdS) TextRegionr bboxN Optional[str]textOptional[Source]sourcer strc*t|jSrX)rrr#s r__str__zTextRegion.__str__s49~~rrrrrrc >t||||}|d|||d|S)z%Constructs a region from coordinates.)rrrr_)r ) clsrrrrrrkwargsrs r from_coordszTextRegion.from_coordss6RR((sAV$AA&AAAr)r r)NN)rrrrrrrrrrrrr r~) rZr[r\r]rrr classmethodrr_rrr~r~sOOOD#F######' B B B B[ B B Brr~ceZdZUded<edZded<dZded<d Zdd Zd Z e ddZ dZ e dZe dZe dZe dZeddZdS) TextRegionsruelement_coordsc*tjgSrX)rqrrr_rrzTextRegions.sbhrllr)default_factorytextsNz Source | Nonerc|jjdkr?|jjdkr1tjdg|jjdz|_dSdSdSNr)rsizerrqrrshaper#s r __post_init__zTextRegions.__post_init__sV :?a  D$7$+DQ+G"GHHDJJJ $@$@rr c\t|j||j||jS)z#slice text regions based on indicesrrr)rrrr)rindicess rslicezTextRegions.slices3.w7*W%;    rcjfdjDSfdtjjDS)z9return a list of TextRegion objects representing the dataNc bg|]+\}}}}t||||dj,SrXr~rr)rdrrrrrs rrpz'TextRegions.as_list..sK$RR&&r2r2tT[IIrc hg|].\\}}}}}t|||||j/Sr_r)rdrrrrrrs rrpz'TextRegions.as_list..sP   & RR$  " "2r2r4 E E   r)rrzipr#s`ras_listzTextRegions.as_listss : (,(;     *-d.A4:*N*N    rr`listcDgg}}|D]]}||jj|jj|jj|jjf||j^|tj|tj||dj S)zccreate TextRegions from a list of TextRegion objects; the objects must have the same sourcerr) appendrrrrrrrqrrr)rr`rtrres r from_listzTextRegions.from_listsB & &F MM6;>6;>6;>6;>Z [ [ [ LL % % % %s"(6"2"2"(5//RYZ[R\Rcddddrc&|jjdSr)rrr#s r__len__zTextRegions.__len__s"(++rc$|jdddfS)zleft coordinateNrrr#s rrzTextRegions.x1"111a4((rc$|jdddfS)ztop coordinateNr6rr#s rrzTextRegions.y1rrc$|jdddfS)zright coordinateNr*rr#s rrzTextRegions.x2rrc$|jdddfS)zbottom coordinateNrwrr#s rrzTextRegions.y2 rrc@|j|jz |j|jz zS)z7areas of each region; only compute it when it is needed)rrrrr#s rareaszTextRegions.areass!$'!dg&788r)r r)r`r)r ru)rZr[r\r]rrrrrrrrrr^rrrrrrr_rrrrsI.B.BCCCECCCC F    III        eee[e,,,))X)))X)))X)))X)999_999rrceZdZdS)EmbeddedTextRegionNrZr[r\r_rrrrDrrceZdZdS)ImageTextRegionNrr_rrrrrrrrTregion1region2same_region_thresholdrHr0c4|||kS)zReturns whether bounding boxes are almost the same. This is determined by checking if the intersection over union is above some threshold.)rP)rrrs r)region_bounding_boxes_are_almost_the_samers  * *7 3 36K KKrregion_to_growregion_to_matchct||j\\}}}\}}}||||f\|_|_|_|_dS)zEGrows a region to the minimum size necessary to contain both regions.N)rkrBrrrr)rrnew_x1new_y1_new_x2new_y2s rgrow_region_to_match_regionr(sb/H00-VVa)&&1  RNN~(.*;^=N=N=Nr)r`r r r )rlr )rtrur rurY)rr rr rrHr r0)rr rr ) __future__rcopyr dataclassesrr functoolsrtypingrr numpyrq unstructured_inference.constantsr unstructured_inference.mathr r rkr8rsr~rrrrrr_rrrs"""""""((((((((%%%%%%""""""""333333555555 { { { { { { {  { |%%%%((((. BBBBBBB B2 A9A9A9A9A9A9A9 A9H             j   $(LLLLL      r