U md: @szzddlZejZWn(eefk r:ddlmZdZYnXddlmZmZm Z ddl m Z m Z zddlZejZWn(eefk rddlmZdZYnXee e fZ ee e fZee ZeeZe edfZeeZee ZdZejejejejejejejejejejejd ee ee edd d Zeeed d d ZeeeedddZejejejejejejejejejejdeeeee e e dddZejejejejejejejejedd)eee edddZ ejejejejejejeejdeddfeee e dddZ!e"edddZ#eeed d!d"Z$d*eee edd$d%Z%d+eeee ed&d'd(Z&dS),N)cythonF)SequenceTupleUnion)IntegralReal) jnx1x2d1d2scalexd)coordsrc1rd1rc2rd2cCsddg}dD]}g||<}||||||||f\}} } } || kr|t|} | | krj|| g| q |dg| q || kr| |}} | | } } | | | |} |D]B}||}||kr| }n|| kr| }n| ||| }||qq t|S)zGiven two reference coordinates `rc1` & `rc2` and their respective delta vectors `rd1` & `rd2`, returns interpolated deltas for the set of coordinates `coords`.Nrr)lenextendappendzip)rrrrrZ out_arraysr outr r r rr rpairrrrM/home/sam/Atlas/atlas_env/lib/python3.8/site-packages/fontTools/varLib/iup.py iup_segment)s. $  r!)deltasrreturnc Cst|t|kstd|kr |St|}ddt|D}|sHdg|Sg}t|}t|}|dkrd|||df\}}} } |t||||| || || || ||||D]j} | |dkr|d| || f\}}} } |t||||| || || || ||| | }q||dkr|d|||df\}}} } |t||||| || || || t|t|kstt|t|f|S)zFor the contour given in `coords`, interpolate any missing delta values in delta vector `deltas`. Returns fully filled-out delta vector.NcSsg|]\}}|dk r|qSNr).0ivrrr msziup_contour..)rrrr)rAssertionError enumerateiternextrr!r) r"rr indicesritstarti1i2Zri1Zri2endrrr iup_contourasb    &r4)r"rendsr#cCst||kr,t||r"|ddnddks0tt|}||d|d|d|dg}g}d}|D]4}|d7}t||||||}|||}qd|S)zFor the outline given in `coords`, with contour endpoints given in sorted increasing order in `ends`, interpolate any missing delta values in delta vector `deltas`. Returns fully filled-out delta vector.r)rr)sortedrr*r4r)r"rr5r rr0r3contourrrr iup_deltas 0  r;)r&r tolerancerypq)r"rr&r r<csh||dkstt||d|||||||||}||d|}tfddt||DS)zReturn true if the deltas for points at `i` and `j` (`i < j`) can be successfully used to interpolate deltas for points in between them within provided error tolerance.r8rc3s4|],\\}}\}}tt||||kVqdSr$abscomplex)r%rr=r>r?r<rr sz%can_iup_in_between..)r*r!allr)r"rr&r r<ZinterprrCr can_iup_in_betweens , rF)cjdjlcjldjncjndjforceforced)r"rr<r#cCst|t|kstt|}t}tt|dddD]}||d||d}}||||}} |||d|||d} } dD]} | | } || }|| }|| }| | }| | }||kr||}}||}}n||}}||}}d}||kr*t|||krt||krd}n|| krB|krznn4t||||krrt|||ksnd}n||kr| |krt||krt|||kr|||k||kkrd}n:t||krt|||kr|||k||kkrd}|r||q6qq6|S)aThe forced set is a conservative set of points on the contour that must be encoded explicitly (ie. cannot be interpolated). Calculating this set allows for significantly speeding up the dynamic-programming, as well as resolve circularity in DP. The set is precise; that is, if an index is in the returned set, then there is no way that IUP can generate delta for that point, given `coords` and `deltas`. rr)rFT)rr*setrangerAminmaxadd)r"rr<r rNr&ldlcrcndncr rGrHrIrJrKrLc1c2r rrMrrr _iup_contour_bound_forced_sets\"       .     r[)r&r best_costZbest_jcostrNr<)r"rr<lookbackc Cst|}|dkr|}t|t}ddi}ddi}td|D]}||dd} | ||<|d||<|d|krnq8t|dt||ddD]H} || d} | | krt||| ||r| ||<} | ||<| |krq8qq8||fS)aStraightforward Dynamic-Programming. For each index i, find least-costly encoding of points 0 to i where i is explicitly encoded. We find this by considering all previous explicit points j and check whether interpolation can fill points between j and i. Note that solution always encodes last point explicitly. Higher-level is responsible for removing that restriction. As major speedup, we stop looking further whenever we see a "forced" point.Nr)rrr8)rrQ MAX_LOOKBACKrPrRrF) r"rrNr<r^r costschainr&r\r r]rrr _iup_contour_optimize_dp+s(     rc)lkcCs8t|}||;}|s|S|||d|d||S)z{Rotate list by k items forward. Ie. item at position 0 will be at position k in returned list. Negative k is allowed.N)r)rdrer rrr _rot_list`s rfsrer cs$;s|Sfdd|DS)Ncsh|]}|qSrr)r%r'rer rr nsz_rot_set..rrgrrir _rot_setjsrkc st}tfddDr(dg|S|dkr4SdtfddDrfgdg|dSt|}|r6|dt|}|dkstt|t||}t|||}t||\}}t|d}|dk r |||}qԈ d|kst|ffdd t |Dt| nt||||\}}d|d} t |dt|dD]n} t| }|| |kr ||||}q|| |krv|| || |} | | krv| } qv|kst|ffd d t |DS) zFor contour with coordinates `coords`, optimize a set of delta values `deltas` within error `tolerance`. Returns delta vector that has most number of None items instead of the input delta. c3s|]}tt|kVqdSr$r@)r%r>rCrr rDsz'iup_contour_optimize..Nrrc3s|]}|kVqdSr$r)r%r)d0rr rDsr)cs g|]}|kr|ndqSr$rr%r&)r"solutionrr r(sz(iup_contour_optimize..cs g|]}|kr|ndqSr$rrn)best_solr"rr r(s) rrEr[rRr*rfrkrcrOrSremoverP) r"rr<r rNrerbrar&r\r0r]r)rprmr"ror<r iup_contour_optimizeqs\            rr)r"rr5r<r#c Cst||kr,t||r"|ddnddks0tt|}||d|d|d|dg}g}d}|D]R}t|||d|||d|}t|||dkst|||d}qd|S)aFor the outline given in `coords`, with contour endpoints given in sorted increasing order in `ends`, optimize a set of delta values `deltas` within error `tolerance`. Returns delta vector that has most number of None items instead of the input delta. r)rrr6r7r8)r9rr*rrr) r"rr5r<r rr0r3r:rrr iup_delta_optimizes 0   rs)r)rl)rl)'rcompiledZCOMPILEDAttributeError ImportErrorZfontTools.misctypingrrrnumbersrrZ_PointZ_DeltaZ _PointSegmentZ _DeltaSegmentZ _DeltaOrNoneZ_DeltaOrNoneSegmentZ _Endpointsr`Zcfunclocalsintdoubler!r4r;inlineZreturnsrFrOr[rclistrfrkrrrsrrrr s          ,5   Q ,  i