_j]4SrSSKJr SSKJr SSKJr SSKJrJ r J r SSK J r SSK JrJrJrJrJrJrJrJrJrJrJrJrJrJr \(aSS KJr S r"S S \5r\"S S9SSj5r \"S S9SSj5r!\"S S9SSj5r"\"S S9SSj5r#\"S S9SSj5r$"SS\ 5r%\"S S9SSj5r&S Sjr'S!S"Sjjr(S#Sjr)S$Sjr*S!S"Sjjr+g)%z Grapheme cluster segmentation following Unicode Standard Annex #29. This module provides pure-Python implementation of the grapheme cluster boundary algorithm as defined in UAX #29: Unicode Text Segmentation. https://www.unicode.org/reports/tr29/ ) annotations)IntEnum) lru_cache) TYPE_CHECKINGOptional NamedTuple)bisearch) GRAPHEME_L GRAPHEME_T GRAPHEME_V GRAPHEME_LV INCB_EXTEND INCB_LINKER GRAPHEME_LVTINCB_CONSONANTGRAPHEME_EXTENDGRAPHEME_CONTROLGRAPHEME_PREPENDGRAPHEME_SPACINGMARKEXTENDED_PICTOGRAPHICGRAPHEME_REGIONAL_INDICATOR)Iterator cP\rSrSrSrSrSrSrSrSr Sr S r S r S r S rS rSrSrSrSrg)GCB,z'Grapheme Cluster Break property values.rr  N)__name__ __module__ __qualname____firstlineno____doc__OTHERCRLFCONTROLEXTENDZWJREGIONAL_INDICATORPREPEND SPACING_MARKLVTLVLVT__static_attributes__r*E/home/rurouni/.local/lib/python3.13/site-packages/wcwidth/grapheme.pyrr,sL1 E B BG F CGL A A A B Cr?ri)maxsizecUS:Xa[R$US:Xa[R$US:Xa[R$[ U[ 5(a[R $[ U[5(a[R$[ U[5(a[R$[ U[5(a[R$[ U[5(a[R$[ U[5(a[R $[ U["5(a[R$$[ U[&5(a[R($[ U[*5(a[R,$[ U[.5(a[R0$[R2$)z;Return the Grapheme_Cluster_Break property for a codepoint.r)r&i )rr1r2r5 _bisearchrr3rr4rr6rr7rr8r r9r r:r r;rr<rr=r0ucss r@_grapheme_cluster_breakrFBs'  f}vv  f}vv  f}ww&''{{o&&zz122%%%&''{{*++j!!uu j!!uu j!!uu k""vv l##ww 99r?c4[[U[55$)z6Check if codepoint has Extended_Pictographic property.)boolrCrrDs r@_is_extended_pictographicrIes  #45 66r?c4[[U[55$)z,Check if codepoint has InCB=Linker property.)rHrCrrDs r@_is_incb_linkerrKk  #{+ ,,r?c4[[U[55$)z/Check if codepoint has InCB=Consonant property.)rHrCrrDs r@_is_incb_consonantrNqs  #~. //r?c4[[U[55$)z,Check if codepoint has InCB=Extend property.)rHrCrrDs r@_is_incb_extendrPwrLr?c.\rSrSr%SrS\S'S\S'Srg) BreakResult}z*Result of grapheme cluster break decision.rH should_breakintri_countr*N)r+r,r-r.r/__annotations__r>r*r?r@rRrR}s4Mr?rRcU[R:XaU[R:Xa [SSS9$U[R[R[R4;a [SSS9$U[R[R[R4;a [SSS9$U[R :XaLU[R [R [R[R4;a [SSS9$U[R[R 4;a.U[R [R4;a [SSS9$U[R[R4;aU[R:Xa [SSS9$U[R:Xa [SSS9$U[R:Xa [SSS9$U[R:Xa [SSS9$g)z Check simple GCB-pair-based break rules (cacheable). Returns BreakResult for rules that can be determined from GCB properties alone, or None if complex lookback rules (GB9c, GB11) need to be checked. FrrTrVTN) rr1r2rRr3r9r:r<r=r;r4r8r7)prev_gcbcurr_gcbs r@_simple_break_checkr\st366h#&&0::CKK00q99CKK00q99355X#%%)HH::CFFCEE?"xCEE355>'A::CGGSUU##CEE(9::3::::3###::3;;:: r?cZ[X5nUbU$U[R:Xa [SSS9$[ X#5n[ U5(asSnUS- nUS:af[ X(5n [ U 5(aSnUS-nO9[U 5(aUS-nO#[ U 5(aU(a [SSS9$O OUS:aMfU[R:Xap[U5(a`US- nUS:aU[ X(5n [U 5n U [R:XaUS-nO[U 5(a [SSS9$OUS:aMUU[R:Xa4U[R:Xa US-S:Xa [SUS-S9$[SSS9$U[R:XaSOSn[SUS9$)z Determine if there should be a grapheme cluster break between prev and curr. Implements UAX #29 grapheme cluster boundary rules. FrrYr Tr) r\rr5rRordrNrKrPrIrFr4r6) rZr[textcurr_idxrVresultcurr_ucs has_linkeriprev_ucs prev_props r@ _should_breakrgs! 4F  377:: 4>"H(## qL1f47|Hx((! Q **Q#H--&EAFF1f3778BB qL1f47|H/9ICJJ&Q*844"BB1f3)))h#:P:P.P a<1 EHqLI Iq99 6 66qAH D8 <>> list(iter_graphemes('cafe\u0301')) ['c', 'a', 'f', 'e\u0301'] >>> list(iter_graphemes('ok\U0001F468\u200D\U0001F469\u200D\U0001F467')) ['o', 'k', '\U0001F468\u200D\U0001F469\u200D\U0001F467'] >>> list(iter_graphemes('ok\U0001F1FA\U0001F1F8')) ['o', 'k', '\U0001F1FA\U0001F1F8'] .. versionadded:: 0.3.0 Nrr ) lenminrFr^rr6rangergrVrT) unistrstartendlength cluster_startrVrZidxr[ras r@iter_graphemesrrs8  [F { |u c CMH's6='9:H3)))UQY$*3v{+;<x6I??   s+ +M% s ##sCCc[XS- 5nUS:XaUS:aXS- S:XaUS- $US:aRUS:aGUS:aA[XS- 5nUS:a+[U5[R:Xa[ XS- 5$US- $US- nUS:aaX- [ :aU[X5nSUs=::aS:aO OO7[U5[R :XaOUS-nUS:aX- [ :aMUUn[[X55nU[R:XaSOSn[US-U5HGn [[X 55n [XzX U5n U RnU R(aU nU nMI U$)ac Find the start of the grapheme cluster containing the character before pos. Scans backwards from pos to find a safe starting point, then iterates forward using standard break rules to find the actual cluster boundary. :param text: The Unicode string. :param pos: Position to search before (exclusive). :returns: Start position of the grapheme cluster. r r&r rr) r^rFrr7_find_cluster_startMAX_GRAPHEME_SCANr3r6rkrgrVrT) r_pos target_cpprev_cp safe_startcprpleft_gcbrVrd right_gcbras r@rvrv<sDqM"IDSAX$Qw-4*?Qw4 !8 T)$Qw-(G$#:7#Cs{{#R*4q99QwqJ q.c.2CC ! " 2    "2 &#++ 5 a q.c.2CCM&s4+;'<=H 6 66qAH :>3 '+CL9 xDXF??   M ( r?c NUS::ag[U[U[U555$)a Find the grapheme cluster boundary immediately before a position. :param unistr: The Unicode string to search. :param pos: Position in the string (0 < pos <= len(unistr)). :returns: Start index of the grapheme cluster containing the character at pos-1. Example:: >>> grapheme_boundary_before('Hello \U0001F44B\U0001F3FB', 8) 6 >>> grapheme_boundary_before('a\r\nb', 3) 1 .. versionadded:: 0.3.6 r)rvrjri)rlrxs r@grapheme_boundary_beforerps&" ax vs3F '< ==r?c## U(dg[U5nUcUO [X#5n[US5nX:dX:agUnXA:a![X5nXQ:agXUv UnXA:aM gg7f)ay Iterate over grapheme clusters in reverse order (last to first). :param unistr: The Unicode string to segment. :param start: Starting index (default 0). :param end: Ending index (default len(unistr)). :yields: Grapheme cluster substrings in reverse order. Example:: >>> list(iter_graphemes_reverse('cafe\u0301')) ['e\u0301', 'f', 'a', 'c'] .. versionadded:: 0.3.6 Nr)rirjmaxrv)rlrmrnrorxrps r@iter_graphemes_reversersq(  [FK&S%5C qME |u C ++F8  3'' +s A A&$A&)rErUreturnr)rErUrrH)rZrr[rrzOptional[BreakResult]) rZrr[rr_strr`rUrVrUrrR)rN)rlrrmrUrnz Optional[int]rz Iterator[str])r_rrxrUrrU)rlrrxrUrrU),r/ __future__renumr functoolsrtypingrrrr rCtable_graphemer r r rrrrrrrrrrrcollections.abcrrwrrFrIrKrNrPrRr\rgrrrvrrr*r?r@rs#66, : : : :( ' , 4D 477  4--  400  4-- * 4--`@=@=@= @= @=  @=  @=JA$ A$ A$ A$ A$H1h>0& & & & &r?