ij3dZddlmZddlmZddlmZddlmZm Z m Z ddgZ dd l m Z dd lmZmZmZmZmZmZmZmZmZmZmZmZmZmZerdd lmZd ZGd deZ edddZ!edd dZ"edd dZ#edd dZ$edd dZ%Gdde Z&edd!dZ' d"dZ( d# d$dZ)d%dZ*d&dZ+ d# d'dZ,e)Z-y)(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 NamedTuplezwcwidth.bisearchzwcwidth.table_grapheme)bisearch) GRAPHEME_L GRAPHEME_T GRAPHEME_V GRAPHEME_LV INCB_EXTEND INCB_LINKER GRAPHEME_LVTINCB_CONSONANTGRAPHEME_EXTENDGRAPHEME_CONTROLGRAPHEME_PREPENDGRAPHEME_SPACINGMARKEXTENDED_PICTOGRAPHICGRAPHEME_REGIONAL_INDICATOR)Iterator cHeZdZdZdZdZdZdZdZdZ dZ d Z d Z d Z d Zd ZdZdZy)GCBz'Grapheme Cluster Break property values.rr  N)__name__ __module__ __qualname____doc__OTHERCRLFCONTROLEXTENDZWJREGIONAL_INDICATORPREPEND SPACING_MARKLVTLVLVTY/home/ubuntu/.cache/pipx/cdb5059130eadca/lib/python3.12/site-packages/wcwidth/grapheme.pyrr0sL1 E B BG F CGL A A A B Cr<ri)maxsizec |dk(rtjS|dk(rtjS|dk(rtjSt |t rtj St |trtjSt |trtjSt |trtjSt |trtjSt |trtj St |t"rtj$St |t&rtj(St |t*rtj,St |t.rtj0Stj2S)z;Return the Grapheme_Cluster_Break property for a codepoint.r(r%i )rr.r/r2 _bisearchrr0rr1rr3rr4rr5r r6r r7r r8rr9rr:r-ucss r=_grapheme_cluster_breakrCFs   f}vv  f}vv  f}ww&'{{o&zz12%%%&'{{*+j!uu j!uu j!uu k"vv l#ww 99r<c4tt|tS)z6Check if codepoint has Extended_Pictographic property.)boolr@rrAs r=_is_extended_pictographicrFis  #45 66r<c4tt|tS)z,Check if codepoint has InCB=Linker property.)rEr@rrAs r=_is_incb_linkerrHo  #{+ ,,r<c4tt|tS)z/Check if codepoint has InCB=Consonant property.)rEr@rrAs r=_is_incb_consonantrKus  #~. //r<c4tt|tS)z,Check if codepoint has InCB=Extend property.)rEr@rrAs r=_is_incb_extendrM{rIr<c&eZdZUdZded<ded<y) BreakResultz*Result of grapheme cluster break decision.rE should_breakintri_countN)r)r*r+r,__annotations__r;r<r=rOrOs4Mr<rOc|tjk(r |tjk(r tddS|tjtjtjfvr tddS|tjtjtjfvr tddS|tj k(rM|tj tj tjtjfvr tddS|tjtj fvr/|tj tjfvr tddS|tjtjfvr |tjk(r tddS|tjk(r tddS|tjk(r tddS|tjk(r tddSy)z Check simple GCB-pair-based break rules (cacheable). Returns BreakResult for rules that can be determined from GCB properties alone, or None if complex lookback rules (GB9c, GB11) need to be checked. FrrPrRTN) rr.r/rOr0r6r7r9r:r8r1r5r4)prev_gcbcurr_gcbs r=_simple_break_checkrXst366h#&&0::CKK00q99CKK00q99355X#%%)HH::CFFCEE?"xCEE355>'A::CGGSUU##CEE(9::3::::3###::3;;:: r<c<t||}||S|tjk(r tddSt ||}t |rl|dz }|dk\r9t t ||r"|dz}|dk\rt t ||r"|dk\r$tt ||r tddS|tjk(rft|r[|dz }|dk\rQt ||}t|} | tjk(r|dz}nt|r tddSn|dk\rQ|tjk(r8|tjk(r%|dzdk(rtd|dzStddS|tjk(rdnd}td|S)z Determine if there should be a grapheme cluster break between prev and curr. Implements UAX #29 grapheme cluster boundary rules. FrrUr rT) rXrr2rOordrKrMrHrFrCr1r3) rVrWtextcurr_idxrRresultcurr_ucsiprev_ucs prev_props r= _should_breakrbs!8 4F  377::4>"H(# qL1fT!W6 FA1fT!W6 6oc$q'l3EA> >3778B qL1f47|H/9ICJJ&Q*84"BB1f3)))h#:P:P.P a<1 EHqLI Iq99 6 66qAH D8 <>> list(iter_graphemes('cafe\u0301')) ['c', 'a', 'f', 'é'] >>> list(iter_graphemes('ok\U0001F468\u200D\U0001F469\u200D\U0001F467')) ['o', 'k', '👨\u200d👩\u200d👧'] >>> list(iter_graphemes('ok\U0001F1FA\U0001F1F8')) ['o', 'k', '🇺🇸'] .. versionadded:: 0.3.0 Nrr ) lenminrCrZrr3rangerbrRrP) unistrstartendlength cluster_startrRrVidxrWr]s r=_iter_graphemes_pythonrms8  [F { |u c6 CMH's6%='9:H3)))UQY$ *3vc{+;<x63I??   s+ +M  s ##sB?Cct||dz }|dk(r|dk\r||dz dk(r|dz S|dkrP|dk\rF|dk\rAt||dz }|dk\r+t|tjk(rt ||dz S|dz S|dz }|dkDr]||z t krQt||}d|cxkrdkrnnn4t|tj k(rn|dz}|dkDr ||z t krQ|}tt||}|tjk(rdnd}t|dz|D]D} tt|| } t|| || |} | j}| jr| }| }F|S)a 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) rZrCrr4_find_cluster_startMAX_GRAPHEME_SCANr0r3rfrbrRrP) r[pos target_cpprev_cp safe_startcprkleft_gcbrRr_ right_gcbr]s r=rqrq5sDqM"IDSAX$sQw-4*?Qw4 !8 T)$sQw-(G$#:7#Cs{{#R*4q99QwqJ q.cJ.2CC j! " 2    "2 &#++ 5 a q.cJ.2CCM&s4 +;'<=H 6 66qAH :>3 '+CQL9 xD!XF??   M  r<c L|dkryt|t|t|S)a Find the grapheme cluster boundary immediately before a position. :param unistr: The Unicode string to search. :param pos: Position in the string (0 < pos <= len(unistr)). :returns: Start index of the grapheme cluster containing the character at pos-1. Example:: >>> grapheme_boundary_before('Hello \U0001F44B\U0001F3FB', 8) 6 >>> grapheme_boundary_before('a\r\nb', 3) 1 .. versionadded:: 0.3.6 r)rqrerd)rgrss r=grapheme_boundary_beforer{is&" ax vs3F '< ==r<c#K|syt|}||n t||}t|d}||k\s||k\ry|}||kDr"t||}||kry||||}||kDr!yyw)a Iterate over grapheme clusters in reverse order (last to first). :param unistr: The Unicode string to segment. :param start: Starting index (default 0). :param end: Ending index (default len(unistr)). :yields: Grapheme cluster substrings in reverse order. Example:: >>> list(iter_graphemes_reverse('cafe\u0301')) ['e\u0301', 'f', 'a', 'c'] .. versionadded:: 0.3.6 Nr)rdremaxrq)rgrhrirjrsrks r=iter_graphemes_reverser~s(  [FK&Sf%5C qME |u C ++FC8 5 ]3'' +s AA" A")rBrQreturnr)rBrQrrE)rVrrWrrzOptional[BreakResult]) rVrrWrr[strr\rQrRrQrrO)rN)rgrrhrQriz int | Noner Iterator[str])r[rrsrQrrQ)rgrrsrQrrQ)rgrrhrQriz Optional[int]rr).r, __future__renumr functoolsrtypingrrr__lazy_modules__r r@table_graphemer r r rrrrrrrrrrrcollections.abcrrrrrCrFrHrKrMrOrXrbrmrqr{r~iter_graphemesr;r<r=rs#66 , : : : :( ' , 4D 477  4--  400  4-- * 4--`5=5=5= 5= 5=  5=  5=tA$ A$ A$ A$ A$H1h>0& & & & &T(r<