+ i~a^RIt^RIt^RIHtHt^RIHt^RIHtH t ^RI t ^RI H t H t Ht]!R]44t]!R]44t]!R] 44t]]!R R .4,t]P(!R 4t0R kR kRkRkRkRkRkRkRkRkRkRkRkRkRkRkRkRkRkRkR kR!kR"kR#kR$kR%kR&kR'kR(kR)kR*kR+kt]P(!R,4t/tR-t!R.R/4t!R0R1]4t!R2R3]4t!R4R54t!R6R74tR8t R#)9N)BytesIOStringIO)Path) ascii_lettersascii_uppercase)EOF ReparseErrorspace_charactersc#@"TFqP4xK R#5iNencode.0items& [/Users/agent/.openclaw/workspace/venv/lib/python3.14/site-packages/tinyhtml5/inputstream.py r s"N=MT;;===Mc#@"TFqP4xK R#5ir r rs& rrr sH-$ -rc#@"TFqP4xK R#5ir r rs& rrrs!LOD++--Or>I&II FD ! !%f&6&6&8CFCC gff&=&=FKKN63 O O%f777$V6v66c^a]tRt^1toRtRtRtRtRtRt Rt Rt R R lt R t R tVtR #)r zProvides a Unicode stream of characters to the HTMLTokenizer. This class takes care of character encoding and removing or replacing incorrect byte-sequences and also provides column and line tracking. c ^.Vn\R4R3VnVPV4VnVP 4R#)zInitialise the HTMLInputStream. Create a normalized stream from source for use by tinyhtml5. source can be either a file-object, local filename or a string. utf-8certainN) new_lineslookup_encodingencoding open_streamstreamreset)selfr$r%s&&,r__init__HTMLUnicodeInputStream.__init__9s9(19= &&v.  r'chRVn^Vn^Vn.Vn^Vn^VnRVnR#)N)chunk chunk_size chunk_offseterrorsprevious_number_linesprevious_number_columns_buffered_characterr2s&rr1HTMLUnicodeInputStream.resetIs;  &'"'($$( r'c @\VR4'dV#\V4#)zeProduce a file object from source. source can be either a file object, local filename or a string. r)r"rr2r$s&&rr/"HTMLUnicodeInputStream.open_streamWs !00vFhv6FFr'cVPpVPR^V4pVPV,pVPR^V4pVR8XdVPV,pWF3#W^,, pWF3#) )r7countr;rfindr<)r2offsetr7 number_lines position_linelast_line_positionposition_columns&& r _position HTMLUnicodeInputStream._position_s{ {{4F3 22\A "[[q&9  #"::VCO//%Q(>?O//r'c RVPVP4wrV^,V3#)z9Return (line, col) of the current position in the stream.)rMr9)r2linecolumns& rpositionHTMLUnicodeInputStream.positionjs'~~d&7&78 q&!!r'c VPVP8dVP4'g\#VPpVPV,pV^,VnV#)z\Read one character from the stream or queue if available. Return EOF when EOF is reached. )r9r8 read_chunkrr7)r2r9 characters& rrV HTMLUnicodeInputStream.characterosT    /??$$ (( JJ|, (1,r'cVPVP4wVnVnRVn^Vn^VnVP PR4pVP'dVPV,pRVnM V'gR#\V4^8d?\VR ,4pV^ 8XgRTu;8:dR8:dMMVR ,VnVRR p\\\PV444FpVPPR4K VP!RR4pVP!R R4pWn\V4VnR #) r6i(NFiizinvalid-codepointz rD TrE)rMr8r;r<r7r9r0rr=rordrangeinvalid_unicode_refindallr:appendreplace)r2datalast_s& rrU!HTMLUnicodeInputStream.read_chunks'CG>> OOD@"D$@ {{&  # # #++d2D'+D $ t9q=tBx=Dt|v77+/8(CRys-55d;<=A KK  2 3>||FD)||D$' d)r'c  \W3,p.pVP VPVP4pVfVPVP8wdMMOVP4pWpP8wd0VPVPVPV4WpnMAVPVPVPR4VP4'dKRPV4# \dlRPTUu.uFpR\T4R 2NK Muupiup4pT'gRT 2p\P !RT R24pT;p\Y3&EL`i;i)zReturn a string of characters from the stream. String goes up to but does not include any character in 'characters' or EOF. 'characters' must be a container that supports the 'in' method and iteration over its characters. r6z\x02x^[z]+N) characters_until_regexKeyErrorjoinrZrecompilematchr7r9r8endr^rU)r2 charactersoppositerVregexresultrmrns&&& r chars_until"HTMLUnicodeInputStream.chars_untilsO P/0FGJ$$TZZ1B1BCE}$$78iik//)MM$**T->->s"CD(+% MM$**T%6%6%78 9??$$wwvA PGGTIs3y>#"67TUEE7 JJ5'}-EJO OJ/0FG  Ps#C;;E1D. -E1+E10E1cV\JdVP^8Xd3WP,VnV;P^, unR#V;P^,unVPVP,V8XgQhR#R#rN)rr9r7r8)r2chars&&rungetHTMLUnicodeInputStream.ungetsj s?  A% "JJ. 1$!!Q&!zz$"3"34<<< r') r=r7r9r8r.r:r,r<r;r0N)F)__name__ __module__ __qualname____firstlineno____doc__r3r1r/rMrRrVrUrsrx__static_attributes____classdictcell__ __classdict__s@rr r 1s? (G 0" "#J,\ = =r'r c^aa]tRt^toRtR RltV3RltRtRtRt Rt Rt R t Vt V;t#) r#zProvide a binary stream of characters to the HTMLTokenizer. This class takes care of character encoding and removing or replacing incorrect byte-sequences and also provides column and line tracking. c VPV4VnRVnW nW0nW@nWPnW`nVP4Vn VP^,fQhVP4R#)iN) r/ raw_streamnumber_bytes_metaoverride_encodingtransport_encodingsame_origin_parent_encodinglikely_encodingdefault_encodingdetermine_encodingr.r1)r2r$rrrrrr%s&&&&&&&,rr3HTMLBinaryInputStream.__init__sp **62 "&!2"4+F(. 0//1 }}Q+++ r'c<VP^,PPpV!VPR4Vn\ SV`4R#)rr_N)r. codec_info streamreaderrr0superr1)r2r __class__s& rr1HTMLBinaryInputStream.resets8}}Q'22?? "4??I>   r'c\VR4'd;\VR4'dVP4'dV#VP4p\V4#)rseekable)r"rrrrAs&&rr/!HTMLBinaryInputStream.open_stream sB 66 " "vz**v/@/@ [[]Fvr'cPVP4R3pV^,eV#\VP4R3pV^,eV#\VP4R3pV^,eV#VP 4R3pV^,eV#\VP 4R3pV^,e+V^,P PR4'gV#\VP4R3pV^,eV#\VP4R3pV^,eV#\R4R3#)r+ tentativezutf-16 windows-1252) detect_bomr-rrdetect_encoding_metarname startswithrr)r2r.s& rr(HTMLBinaryInputStream.determine_encodings2??$i/ A; "O#4#9#9:IE A; "O#4#:#:;YF A; "O,,. ; A; "O#4#C#CDkQ A; "8A;+;+;+F+Fx+P+PO#4#7#78+E A; "O#4#8#89;F A; "O~. ;;r'cVP^,R8wgQh\V4;pfR#VPR9d\R4pVfQhR#WP^,8XdVP^,R3VnR#VPP ^4VR3VnVP 4\ RVP^, RV 24h)r+Nr*zEncoding changed from z to utf-16beutf-16le)r.r-rrseekr1r )r2 new_encodings&&rchange_encoding%HTMLBinaryInputStream.change_encoding8s}}Q9,,,+L9 9L B     8 8*73L+ ++ ]]1- -!]]1-y9DM OO  #)95DM JJL(q)9(:$|nMO Or'c  \PR\PR\PR\PR\P R/pVP P^4p\V\4'gQhRFGpVPVRV4;p'gK!VP PV4\V4u# VP P^4R#)zAttempt to detect at BOM at the start of the stream. If an encoding can be determined from the BOM return the name of the encoding otherwise return None. r*rrzutf-32lezutf-32beN)) codecsBOM_UTF8 BOM_UTF16_LE BOM_UTF16_BE BOM_UTF32_LE BOM_UTF32_BErrrbytesgetrr-)r2bomsstringrr.s& rr HTMLBinaryInputStream.detect_bomHs OOW              %%a(&%((((D88F5DM22x22$$T*&x00  Qr'c ,VPPVP4p\V\4'gQh\ V4pVPP ^4VP4pVeVPR9d \R4pV#)z1Report the encoding declared by the meta element.r*r) rrrrrEncodingParserr get_encodingrr-)r2bufferparserr.s& rr*HTMLBinaryInputStream.detect_encoding_metaesz%%d&<&<=&%((((' Q&&(  HMM5M$M&w/Hr') rr.rrrrrr0r)NNNNr)rzr{r|r}r~r3r1r/rrrrrr __classcell__)rrs@@rr#r#s42 & s4y  X\**r'cLVP^, ;VnpWV^,#rrrs& rpreviousEncodingBytes.previouss&$(NNQ$66X\**r'chVP\V48d\h\^V4VnR#rv)rMrrmaxrs&&r set_positionEncodingBytes.set_positions& >>SY & Q)r'cVP\V48d\hVP^8d VP#R#rvrr>s&r get_positionEncodingBytes.get_positions3 >>SY &  >>Q >> ! r'c@WPVP^,#r)rRr>s&r current_byteEncodingBytes.current_bytesMM$--!"344r'c VPpV\V48d%WV^,pW19d W nV#V^, pK4W nR#)zSkip past a list of characters.NrRrrMr2rorRrVs&& rskipEncodingBytes.skipsG==T"hl3I*!)  MH!r'cVPpV\V48d%WV^,pW19d W nV#V^, pK4W nR#)rNrrs&& r skip_untilEncodingBytes.skip_untilsG==T"hl3I&!)  MH!r'c VPWP4;p'd#V;P\V4, unV#)zLook for a sequence of bytes at the start of a string. If the bytes are found return True and advance the position to the byte after the match. Otherwise return False and leave the position alone. )rrRr)r2rrrs&& r match_bytesEncodingBytes.match_bytess3__UMM: :6 : MMSZ 'M r'c VPWP4\V4,^, VnR# \d\ hi;i)zLook for the next sequence of bytes matching a given sequence. If a match is found advance the position to the last byte of the match. T)indexrRrrM ValueErrorr)r2rs&&rjump_toEncodingBytes.jump_tosE  !ZZ}}=E JQNDN   s 7;A rN)rzr{r|r}r~rr3rrrrpropertyrRrspace_characters_bytesrrrrrrrs@rrrssf1+ +* "  l3H 555     r'rcZa]tRtRtoRtRtRtRtRtRt Rt R t R t R t R tVtR #)riz@Mini parser for detecting character encoding from meta elements.c4\V4VnRVnR#r )rr`r.r2r`s&&rr3EncodingParser.__init__s!$'  r'c LRVP9dR#RVPRVPRVPRVPRVPRVP /pVPFzpRpVPP R4TP4F0wrETPPT4'gK(T!4pM T'dKoTP# VP# \dTP#i;i \dRpKSi;i) sr`rr>s&rrEncodingParser.handle_commentsyy  ((r'c VPP\9dR#RpRpVP4;pfR#V^,R8Xd%V^,R8HpV'dVe W nR#KGKIV^,R8Xd#V^,p\ V4pVe WPnR#KyV^,R8XgK\ \V^,44pVP4;pfK\ V4pVfKV'd WPnR#TpK)TFNs http-equivs content-typecharsetscontent) r`rr get_attributer.r-ContentAttributeParserrparse)r2 has_pragmapending_encoding attributetentative_encodingcodeccontent_parsers& rrEncodingParser.handle_metas 99 ! !)? ? !//11 :|},&q\_< "2">$4M #?:1+%.q\"'(:;$$)M %1+!7 iPQl8S!T*8*>*>*@@&M+,>?E(%,1M#(+0(r'c&VPRR7#)Fend_tag)handle_possible_tagr>s&rr(EncodingParser.handle_possible_start_tags'''66r'cP\VP4VPRR7#)Tr)nextr`rr>s&rr&EncodingParser.handle_possible_end_tags! TYY'''55r'c$VPpVP\9d+V'd!VP4VP 4R#VP \ 4pVR8XdVP4R#VP4eKR#)Tr)r`rascii_letters_bytesrrrspaces_angle_bracketsr)r2rr`rVs&& rr"EncodingParser.handle_possible_tagsxyy   $7 7 !!#OO$9:   MMO %%'/r'c8VPPR4#)rrr>s&rrEncodingParser.handle_other5syy  &&r'c $VPpVP\\R.4,4pVe\ V4^8XgQhVR9dR#.p.pVR8Xd V'dMV\9dVP4pMiVR9dRP V4R3#V\ 9d!VPVP44MVfR#VPV4\V4pKVR8wd$VP4RP V4R3#\V4VP4pT;pR9d\V4pW%8Xd.\V4RP V4RP V43#V\ 9d"VPVP44KkVPV4K~VR8XdRP V4R3#V\ 9d!VPVP44MVfR#VPV4\V4pV\9d#RP V4RP V43#V\ 9d"VPVP44KeVfR#VPV4K~) zkReturn a (name, value) pair for the next attribute in the stream. If no attribute is found, return None. /Nr=r')rN)rr)'") r`rr frozensetrrjascii_uppercase_bytesr^rrrr)r2r`rVattribute_nameattribute_valuequotes& rrEncodingParser.get_attribute8s5 yyII4y$7HHI  C Na$777  $D ^44 IIK l*xx/4433%%ioo&78"%%i0T I   MMO88N+S0 0 T IIK  E< / J %J88N3SXXo5NNN"77#**9??+<=$**95 $ 88N+S0 0 / /  " "9??#4 5    " "9 -T I11xx//1JJJ33&&y'89"&&y1r')r`r.N)rzr{r|r}r~r3rrrrrrrrrrrs@rrrs=J>)1B76.'L2L2r'rc,a]tRtRtoRtRtRtVtR#)ricB\V\4'gQhWnR#r )rrr`rs&&rr3ContentAttributeParser.__init__s$&&&& r'cVPPR4VP;P^, unVPP4VPPR8XgR#VP;P^, unVPP4VPPR9dVPPpVP;P^, unVPPpVPPV4'd#VPW PP#R#VPPpVPP \ 4VPW PP# \dTPTRu#i;i \dR#i;i)rrN)r r)r`rrRrrrrr)r2r% old_positions& rrContentAttributeParser.parses_  II  j ) II  ! #  IINN 99))T1 II  ! #  IINN yy%%5 .. ""a'"#yy11 99$$U++99\))2D2DEE $yy11 4II(()?@99\))2D2DEE$499\]334  sDA3G7CG?!G#G:AF;;GGGG G*)G*)r`N)rzr{r|r}r3rrrrs@rrrsr'rc\V\4'dVPR4pVe\P !V4#R# \dR#i;i \ dR#i;i)zReturn the Python codec name corresponding to an encoding. Return None if the string doesn't correspond to a valid encoding. asciiN)rrdecodeUnicodeDecodeError webencodingslookupAttributeError)r.s&rr-r-sn (E"" w/H &&x0 0"     s"AA AA A'&A')!rrkiorrpathlibrrrrr1 constantsrr r r!rrr"rrlr\non_bmp_invalid_codepointsascii_punctuation_rerhr&r r#rrrrr-r'rr:s5 1::#"N=M"NNH-HH!!LO!LL.D$<1HHZZ=>N N NN&N(/N18N:ANCJN N NN&N(/N18N:ANCJN NNN 'N)0N29N;BNDKN N N N ' N)0 N29 N;C NEM N zzOQ7o=o=dM2M`VEVrx2x2v##Lr'