j fddlmZmZddlmZmZddlmZmZddl m Z ddl m Z ddl mZmZddlmZerdd lmZeZd d d d ZeeZdZeGddZdedefdZdedefdZdedefdZdeddfdZ deddfdZ!dddedzfdZ"dddedededzfdZ#dededdfdZ$dddedefd Z%deddfd!Z& d=ddded"ee'ezd#e(de(f d$Z)dddeedzeffd%Z*ddded&edeeedzffd'Z+ddded(e(defd)Z,d*edefd+Z-ddd,e(defd-Z.ddd.e(defd/Z/ddd#e(defd0Z0ddd#e(de(dzfd1Z1ddd#e(de(fd2Z2ddd3e(defd4Z3ddd5e(defd6Z4ddded&edeeedzeffd7Z5dddededzfd8Z6ddded&edzdefd9Z7dddefd:Z8ddd;e(defd<Z9dS)>) dataclassfield) TYPE_CHECKINGcast)classify_object_value_commaupdate_inline_container_stack)parse_boolean_or_null)parse_json_llm_block)STRING_DELIMITERSJSONReturnType) ContextValues) JSONParser]}))[{(c(eZdZUdZeed<dZeed<dZeed<dZ eed<dZ eed<dZ eed <dZ eed <e e Zeeed <dZeed <e e Zeeedfeeedzffed<dZeed<dZedzed<dS)StringParseStateFmissing_quotesdoubled_quotes"lstring_delimiterrstring_delimiter string_accunmatched_delimiterpending_inline_container)default_factoryinline_container_stack$object_value_has_no_future_delimiter.Nlookahead_cacher%object_value_unmatched_opening_bracesregex_character_class_start)__name__ __module__ __qualname__rbool__annotations__rrstrrrr r!rlistr#r$dictr%tupleintr&r'FC:\Projects\DSCons\.venv\Lib\site-packages\json_repair\parse_string.pyrrs ND    ND    s    s   J %%%%%*d***(-d(C(C(CDICCC16($666EJU[_E`E`E`OT%S/5cDj+AAB```12)3222.2t22222r3rstatereturnc|jdS)Nrrr5s r4_outer_rstring_delimiterr:$s  "1 %%r3c|jdSNr8r9s r4_active_rstring_delimiterr>(s  "2 &&r3c2t|tkSN)r>LOW_SMART_QUOTE_SENTINELr9s r4_in_low_smart_quote_spanrB,s $U + +/G GGr3Nc0|xjtz c_dSr@)rrAr9s r4_push_low_smart_quote_spanrD0s 77r3c.|jdd|_dSr<r8r9s r4_pop_low_smart_quote_spanrF4s#5crc:Er3selfrc8|dkrdS|jdz}|j}t|tr7|d|}|dkrdS|||}d|vsd|vsd|vrdSnI|}t |}||kr ||}|dkrn|dvrdS|dz }||k ||krdS|||}|dz}t |}||krI|j|r*|dz }||kr|j|*||kr |j|nd}|jj } | tj kr |dkrdSn2| tj kr|d vrdSn| tj kr|d vrdSn|dS|dz|_|S) Nrrr=\  >rJrKrI:>Nr,>NrMr) get_char_atindexjson_str isinstancer-findlenisspacecontextcurrentr OBJECT_KEY OBJECT_VALUEARRAY) rGstartrPendvaluelimitchar next_index next_charcurrent_contexts r4_try_parse_simple_quoted_stringrb8s S  t JNE}H(C  $mmC'' "994s# 5==DEMMTU]]4.;H EkkC=Ds{{)))t 1HC Ekk %<<4s#qJ MME u  z!:!B!B!D!D a  u  z!:!B!B!D!D -7%-?-? j))TIl*O-222   4  M6 6 6 , , ,4 - M/ / / , , ,4 -  tqDJ Lr3 current_charcjt|||xjdz c_|S)Nr)_append_string_contentrOrN)rGr5rcs r4_append_literal_charrfjs5 5,///JJ!OJJ     r3contentc$t|j}|xj|z c_t|D][\}}|dkr|xjdz c_|dkr|jr|xjdzc_:|dkr||zdz|_N|dkrd|_\dS)Nrrrrr)rSr enumerater&r')r5rg start_indexoffsetr^s r4reretse&''K !'**55  3;;  7 71 < 7 7 7 S[[UH[  7 71 < 7 7 7 S[[0;f0Dq0HE - - S[[04E -55r3c|j}|&td|j|dDrdS|dd}||dkS)zOReturn whether the current quote is inside a compact ``[...]`` character class.Nc3>K|]}|VdSr@)rT).0r^s r4 z:_quote_belongs_to_regex_character_class..s*PPtDLLNNPPPPPPr3Frridx)r'anyrskip_to_characterrN)rGr5rZclosing_bracket_idxs r4'_quote_belongs_to_regex_character_classrusp  -E }PPu7G7OPPPPP}u00!0<<   / 0 0C 77r3cd|_d|_t|jD]X\}}|dkr|xjdz c_|dkr|jr|xjdzc_:|dkr |dz|_K|dkrd|_YdS)Nrrrrrr)r&r'rir)r5rOr^s r4!_rebuild_unmatched_opening_bracesrws23E/(,E% !12255 t 3;;  7 71 < 7 7 7 S[[UH[  7 71 < 7 7 7 S[[05 E - - S[[04E -55r3 characterrqct|tr|fnt|}|j|z}|j|}|E|\}}|"||krt |j|jz S|||cxkr|kr nn ||jz S|||} | | } | s|df|j|<| S|j| z} | dks|j| dz dkr || f|j|<| S)NrrrI) rQr-r0rOr%getrSrPrsrN) rGr5rxrqtargetsrjcached cached_start cached_match match_offsetmatch match_indexs r4_cached_skip_to_characterrs? )C88NylleI>N>NG*s"K  " & &w / /F %+" l  K<$?$?t}%% 2 2  # (S(S(S(S|(S(S(S(S(S$*, ,)))S99L   \ * *E *5t)<g&*|+Ka4=q9TAA*5{)Cg& r3c|}|dvrd|fS|r`|tvrW|sC|xjdz c_|}|r|tvr|C|sdSt |}|d|fSt }|dkrdx|_|_n|dkrd|_d|_nx|rd| dvr2|j j tj krt|\}}|r||fS|dd |_|js|xjdz c_|d kr,t#|}|d ur||fS|d ||jkr|j j tj kr|dd ksb|j j tjkr|ddvs1|j j tjkr+|ddvr|xjdz c_|dfS|d|jkr/|d|jrt+d|dfS||jd}||dz|jkr-|dd |_|xjdz c_n|d}||}|gtddvr?|d|jrt+d|xjdz c_|dfS|dvr%|d|xjdz c_|t2fS)N)#/r)Nr'“”)tfnz=While parsing a string, we found a literal instead of a quoteT`FzoWhile parsing a string, we found code fences but they did not enclose valid JSON, continuing parsing the stringrLrMr)rMrrzTWhile parsing a string, we found a doubled quote and then a quote again, ignoring itz/Found doubled quotes followed by another quote.rxrqz?While parsing a string, we found a valid starting doubled quoterprrz_While parsing a string, we found a doubled quote but also another quote afterwards, ignoring itzFFound doubled quotes followed by another quote while parsing a string.rMrrzYWhile parsing a string, we found a doubled quote but it was a mistake, removing one quote)rN parse_commentr isalnumrOrbrrrlowerrUrVrrWr logrr rXrYstrict ValueErrorrsrscroll_whitespacesNO_DIRECT_RESULT) rGr^fast_path_valuer5 found_literalr\ret_valinext_cs r4_prepare_string_entryrsR     D zT'')))) "40000 a !! "40000 x5d;;O"_$$   E s{{#> M5 $e|#  K    $   a  S  &t,, %  '> !  }    U444 \ !]%= = =$BRBRSTBUBUY\B\B\ $ (BBBtGWGWXYGZGZ^hGhGh $ (;;;@P@PQR@S@SWa@a@a JJ!OJJ"9    A  %"9 9 9 HHf   { T !RSSS"9   " "U-D! " L L   AE " "e&= = = HHQ   $(E JJ!OJJJ''A'..A%%a((F7,7c73777u;$` a by _,,o a " ""r3r^c|dt|}t|ra|dkr[|jdd|z|_t |t ||xjdz c_d|fS|dkr|jdz }|jdz}|t|j kr?|j |dkr.|dz }|t|j kr|j |dk.||z }|||jz }|dzdkrO||krI|jddd|dzzz|_t |||_d|fS||d d d d dfvr|jdd|_d dddd}|xj| ||z c_t ||xjdz c_|}|r|jr|jddkrp||dfvrj|jdd|z|_t ||xjdz c_|}|r|jr|jddkr||dfvjd|fS|dvr|dkrdnd} |j |jdz|jdz| z} t| | krtd| Dr|d|jddtt| dz|_t ||xjd| zz c_d|fSnv|dks|tvrg||kra|d|jdd|z|_t ||xjdz c_d|fSd|fS)Nz-Found a stray escape sequence, normalizing itrr=rTrIrrrrb rJrK)rrrr)uxrc3K|]}|dvV dS)0123456789abcdefABCDEFNr2)rncs r4roz-_normalize_escape_sequence..@s)/b/bRS5M0M/b/b/b/b/b/br3z/Found a unicode escape sequence, normalizing it„zPFound a delimiter that was escaped but shouldn't be escaped, removing the escapeF)rr>rBrrwrFrOrNrSrPrzallchrr1r ) rGr5r^active_rstring_delimiter run_startrun_end run_lengthr` escape_seqs num_chars next_charss r4_normalize_escape_sequencers\  HH <===8??&&(43;; +CRC047)%000!%((( a T%%'''' t||JN *q.DM****t}W/E/M/M qLGDM****t}W/E/M/My( $$Wtz%9:: >Q  90H#H#H$/4 a8PQE  -e 4 4 4 DJ))+++ + (#sCdCCC +CRC0 t$TBB  KOOD$777)%000 a $$&&  +  + $,,6===$/4y@E  -e 4 4 4 JJ!OJJ((**I  +  + $,,6=== Y zAA! ]4:>DJNY4N#NO z??i ' 'C/b/bWa/b/b/b,b,b ' HHF G G G$/4s3z2;N;N7O7OOE  -e 4 4 4 JJ!i- 'JJ))+++ + $"333@X8X8X cddd +CRC047)%000 a T%%'''' $;r3 fence_idxc|dz}t||}d}||tvr(t||}|t ||rdSd}|}t |}|||}|||krp||dz} || } | dvr |rd|_dS|||dz}|||kpdS)NFTrrrp)rMrrN) %_scroll_comment_prefixed_member_startrNINLINE_CONTAINER_OPENERS_skip_inline_container(_post_fence_container_starts_next_memberr:rsrr!) rGr5rquote_search_idxnext_content_idxkeep_post_fence_containercontainer_end_idxouter_rstring_delimiter quote_idxafter_quote_idx after_quotes r4*_brace_before_code_fence_belongs_to_stringrOsN !1}OPP u(, %0 6u==&&1HN^&__I   9 % %)@ @ @11i!m1DD&&77 / / /( 615.4**5LR[^_R_*``    9 % %)@ @ @ 5r3 delimiterc|dkrdn|S)Nrrr2)rs r4_matching_string_delimiterrms&&55I5r3key_idxc4||}|r|s|dksdS ||}|r|s|dvsn|dz }6||}||dkS)N_FT)r-rrprL)rNrr)rGrkey_chars r4_bare_key_is_followed_by_colonrqs((H H,,..(c//u##G,,  0 0 2 2 h*6L6L 1   %%'%22G   G $ $ ++r3rc||}||}|dvrdS|dkrdSt||dz}||dvpt||S)NrprNTrMFrrrNr_object_member_starts_at)rGrafter_container_idxafter_containernext_member_idxs r4rrs116G1HH&&':;;O+%%t#u;DBUXYBYZZO   O , , ; n?WX\^m?n?nnr3c||}|dz }|dkr||}|dS|s|tvrdS|dvrdS||dz}||}|dvr4|dd gtd td d d vp|duo|S|dkrdS|dgtvrdS|dkot ||S|dz}|dkdS)NrrT)rMrLFrp)rrrrrrrrrrrL)rNrTrrr isdigitr)rGrqopening_delimiterprev_idx prev_charnext_idxr`s r4_starts_nested_inline_containerrsj((--QwH a--$$X..  4  "" W444t **u..37.;;H((22I J.. S#$q0A$q3$qIa$qcf$qhk$qmp$qqT)Ai.?.?.A.A!C''uS5#4555t#V(FtX(V(V VA + a--, 4r3c||}|tvr|St|g}|dz}|r||}|sdS|tvrEt|}|||dz}|||krdSna|tvr1t ||r!|t|n'||dkr||s|dzS|dz }|dS)Nrrr=)rN#INLINE_CONTAINER_CLOSING_DELIMITERSr rrsrappendpop)rGrqrstackrr^ end_delimiters r4rrs@((-- CCC 01B C DE aA "" 4 $ $ $6t<|xjdzc_|}|dd|dfSd|dfS|j rd|_ t|||}d|dfSd}||}d}|r|||j fvr|r|rd}tj|jjvr|dvs\tj |jjvr|d ks>tj|jjvr|d vs"|r!|jjtj kr|d krn'|dz }||}|r |||j fv|d kr|jjtj kr|dz }||| }||}|dz }| |}||}|d vr+|dt|||}d|dfSn||kr||dz dkrut%||r/|jjtj krt'||sd|dfS|jjtj krt'||r+|dt|||}d|dfS|||dz }|dz }||}|rO|dkrI|dvs"||kr||dz dkrn"|dz }||}|r|dkI|dkr8|d|j |_ t|||}d|dfSn|jjtjkr||k}||krq||dg|dz }||}||krd}n7||dg|dz }||}||kq|r8|d|j |_ t|||}d|dfSd|dfS|jjtjkr+|dt|||}d|dfSd|dfS)Nrz=While parsing a string, we found a doubled quote, ignoring itTFrprLzIn a string with missing quotes and object value context, I found a delimeter but it turns out it was the beginning on the next key. Stopping here.)rLrrrrMrMr)rrMzWhile parsing a string, we found a misplaced quote that would have closed the string but has a different meaning here, ignoring itrIrrzWhile parsing a string in Array context, we detected a quoted section that would have closed the string but has a different meaning here, ignoring itzWhile parsing a string in Object Key context, we detected a quoted section that would have closed the string but has a different meaning here, ignoring it)r:rrNrrOrrUrVrrXrrr rfisalpharWrYrs_only_whitespace_untilr) rGr5r^rrrr`check_comma_in_object_valueeven_delimiterss r4!_handle_right_delimiter_candidaters 7u== ! 0 0 3 37N N N PQQQ a T5   " 4 8R R R !!!$$ ) #  #(    FA%%a((F  ) #  #(    . FA''A'..A""c)) a  ,,.. ji--dE!!  &$)!(ud;; Y%% A   a F"& %V $ ' 06>>+;+; 0*/ '  %)= = =&JBVBV*dl.BBBvQT}}#t|';;;*@T@T+AU04 0D Hb0b0bgmqtgtgt  Q!!!$$ %V $}}-1KKK Q  " "-D! " L L!!!$$ Q  # # # * *!!!$$ Z   HHU   -T5$??IE) ) * * *t/?/?A/F/F$/N/N !$ * * % L M$> > >C`aeghCiCi >$$ $ < =#= = =,T155 .Y1udCC Y--&&1HaRSe&TTA FA%%a((F -Vs]]_,,;R1R1RW[WgWghilmhmWnWnrvWvWvQ))!,,  -Vs]] }}Y160I,I)0udCC Y-- \ !]%8 8 8$(??O333**6Ms5SYZ]^Y^*__))!,,444&+O**6Ms5SYZ]^Y^*__))!,,333 .l160I,I)0udCC Y--$$ $ \ !]%= = = HHm   -T5$??IE) ) $ r3cht}d*dtttzdtdtffd }}|r||kst rÉjrjjtj kr1|dks| r dnqjjtj kr|dvr d n<|d kr9jrjd d kr!tt!|}t r(|d kr"t#t!|}js]jjtjkr|dkrd d krt)dr|t,vrjrjd d krt/d}| dd_jt5jjj|zxj|z c_}2jsjjtjkr|dkrjs}jsvjrdnt?|}|dkr dn?|dkrd_|dk_ dt!|}tA|jj\_}|rt!|}jsىjjtjkr|dkrjrjd |krj!rt!|}rd}"dd krd}tG|d} | } | r| dz } | } | } | r| dvrd}n׉$j%| } | } | sd}n| dz} | } | r| dkrd}nh$dd} | } | rn7d} $d| } | | z dkrd}|r dnjsb|dkr\tj jjvrDjrjd |kr,$|} | sn}jjtjkr|dkr҉d} | } | d kr| dzd krl| d!zd krPtM| r( d"t!|}! d#n| s d$n|Jt5|xjdz c_}|Djr;jr4jd d kr#jdd _tOn jr)jd d krtQ|\} }| r|dkrjsjjtj kr̉$j%d} | } | r| dz } $|| } | } | rO| dz } | } | } | dvr d%| d&nn d'nt r(|d(kr"t#t!|})||krVjjtjkr.cached_skip_to_characters(uiEEEr3rLzdWhile parsing a string missing the left delimiter in object key context, we found a :, stopping hererzdWhile parsing a string missing the left delimiter in array context, we found a ] or ,, stopping hererr=rIrrrrpz{While parsing a string in object value context, we found a balanced inline container that belongs to the string, keeping itFrMstringmemberzWhile parsing a string missing the right delimiter in object value context, we found a comma that starts the next object member. Stopping herestring_no_future_delimiterT containerzgWhile parsing a string in object value context, we found a comma that belongs to the string, keeping itrrrzWhile parsing a string missing the left delimiter in object value context, we found a , or } and we couldn't determine that a right delimiter was present. Stopping hererrrzsWhile parsing a string in object value context, we found a literal fenced snippet after }, keeping it in the stringzuWhile parsing a string in object value context, we found a } that closes the object before code fences, stopping herezbWhile parsing a string in object value context, we found a } that closes the object, stopping herezUWhile parsing a string missing the right delimiter in object key context, we found a z stopping herezeWhile parsing a string missing the right delimiter in object key context, we found a :, stopping hererzXWhile parsing a string, we found a bare quote inside a regex character class, keeping itr)+r:r-r.r1rNrBrrUrVrrWrTrrYrrDrfrFr!rXrrrrr#clearrerPrO stream_stabler$rr r&skip_whitespacesrrsrrrwrrur)rGr5rrr^rcomma_classificationkeep_inline_container_charrstring_delimiter_missingrrjhandled_escapechhandled_delimiter should_breaks`` r4_scan_string_bodyrs 7u==FFC$s)OF#FcFFFFFFF     D PD3337OPU7V7V3   |#}'???TS[[TXT`T`TbTb[z|#}':::tz?Q?Qz 5==%"2=e6Fr6Jd6R6R &u - - -'eT::D  #E * * tu}} %e , , ,'eT::D . L(M,FFF ((,,446tT=T=TYZ=T=[=[\\5000%1).)9")=)E)E 6tQ ? ?  ,R27.,22444&udmDJVgIg>(, %  ! ! # # #""d**,1))$7NTUVVVA%%a((F 6Q+++22))!,, >:!5!505--..9PVW.XXA!--a00F!>4911 33A3>>!%!1!1!!4!4!>fmm8=5**Sa*@@))!,,+++22**Sa*@@q519905-( "  #t|';;;%<).)9")=AX)X)X&&'>??A##A&&  < =#= = =$#++''A'..A%%a((F}}!1!1!a%!8!8C!?!?DDTDTUVYZUZD[D[_bDbDb=dE1MMHHN0eTBBDL xud+++ a !! <! 9e&6 95;KB;OSW;W;W#(#3CRC#8 1%888     0 4 < <#=dE4#P#P ND  3;;u3; 8LP]Ph8h8h&&1Ha&PPA%%a((F Q**5LRS*TT))!,,FA//A/66A))!,,BZ''GtvGGG{ #E * * ts{{ %e , , ,'eT::D  + + + $ (BBB7eDDC HHj   ###'eT::D  * * *u/? *EDTUWDX\`D`D`###4UVZ\acg4h4h 1 t\   a PD3337OPU7V7V3b Kr3cHt|}|rv|jro|jjtjkrU|rA|d|| dvrdS||kr;|j s3|d|j |_ n|xj dz c_ |j s=|js|j r/|j ddkr|j |_ |j S)NzWhile parsing a string, handling an extreme corner case in which the LLM added a comment instead of valid string, invalidate the string and return an empty value)rLrMrz=While parsing a string, we missed the closing quote, ignoringrr=rJ)r:rrUrVrrWrTrrrNrrrstriprO)rGr5r^rs r4_finalize_string_resultr qsA 7u== $)=AY)Y)Y^b^j^j^l^l)Y  p         Z / /2 &&&! 9 HHO    %/6688E  a  55#75Er+rBrDrFrbrfrerurwr.r1rrrrrrrrrrrrrrr r rr2r3r4rsK((((((((&&&&&&&&qqqqqqqqMMMMMMKKKKKK>>>>>>>>------(''''''688,/c&D&D# 5!DEE  3 3 3 3 3 3 3  3&$4&&&&&'%5'#''''H$4HHHHH8&6848888;%5;$;;;;/,/3:////d   4Z  5"2 5S 5T 5 5 5 5 8  8  8  8 8 8 8 5-= 5$ 5 5 5 5$   T#Y   8V# V# d "F *+V#V#V#V#r: : : : 4t  ::::z   <6#6#6666, , , ,,,,$ o  o o  o o o o   >   4Z<# # # ####L ;  ; ;  ; ; ; ;  ,y y y y 4tT !" yyyyx[ [ [ 4Z[[[[|   * :6|66666CDr3