֦i g 4 d Z ddlZddlmZ ddl ddlmZ eehZ e
e hZ e
ehZe eehz Zeeeefeeeefeee
efiZ ej: fdZ d Z d Z dd
Z!ejD d z Z#de#z dz
Z$d Z%e#e&fd
Z'd Z(d Z)d Z*d Z+d Z,d Z-d Z.d Z/d Z0d Z1d Z2ddZ3y) z Internal support module for sre N )_parser)*) _EXTRA_CASESc * ||z r| | z } | |z | z S N )flags add_flags del_flags
TYPE_FLAGSs %/usr/lib64/python3.12/re/_compiler.py_combine_flagsr s( :
* I ) ++ c
| j }t }t }t }t }t
}d } d }
d } |t z rY|t z sP|t z r't j } t j }
t } n t j } t j }
|D ]d \ } }
| |v r |t z s || ||
&|t z r |t | ||
G | |
s || ||
` |
|
}| s |t | || || vr |t" | || |t$ || } |d | t& u r |t( |f| | z D ] } |t* || |t, || |z
| |< | t. u rt1 |
| |
| \ }}|t z r|t z r
|t2 n*|s
|t. n| s
|t4 n |t$ || } |d t7 ||| || |z
| |< | t8 u r%|t: z r |t< |t8 | |v r|t> z rtA d| tC |
d r` ||| d || } |d ||
d ||
d tE | |
d | |tF || |z
| |< Q ||| d || } |d ||
d ||
d tE | |
d | || |z
| |< ||| d | tH u r^|
\ }}}}|r |tJ ||dz
dz tE | |tM ||| |s |tJ ||dz
dz dz | tN u rE |tN || } |d tE | |
| |tF || |z
| |< f| |v r
|| t| |v r || || } |d |
d dk\ r |d nB|
d jQ \ }}|tR kD r tA d ||k7 r tA d || tE | |
d | |tF || |z
| |< | tT u rp || |tV z rtX j[ |
|
}
|t z rt\ j[ |
|
}
n |t z rt^ j[ |
|
}
||
| t` u r || g }|j }|
d D ]O }
|| } |d tE | |
| |tb | || |d || |z
| |< Q |t, |D ] } || |z
| |< | td u r7 || |t z r
tf |
}
n|t z r th |
}
||
^| tj u rP|t z s || n1|t z r
|tl n| s
|tn n |tp ||
dz
| tr u r || ||
d dz
|| } |d tE | |
d | |
d rM |tb || } |d || |z
dz | |< tE | |
d | || |z
| |< F || |z
dz | |< YtA d| y )Nr z(internal: unsupported template operator r zlooks too much behindz(look-behind requires fixed-width patternz#internal: unsupported operand type ):appendlen_LITERAL_CODES_REPEATING_CODES_SUCCESS_CODES
_ASSERT_CODESSRE_FLAG_IGNORECASESRE_FLAG_LOCALESRE_FLAG_UNICODE_sreunicode_iscasedunicode_tolowerr
ascii_iscased
ascii_tolowerOP_LOCALE_IGNORE OP_IGNOREOP_UNICODE_IGNORE
IN_UNI_IGNORE NOT_LITERALNEGATELITERALFAILUREIN_optimize_charset
IN_LOC_IGNORE IN_IGNORE_compile_charsetANYSRE_FLAG_DOTALLANY_ALLSRE_FLAG_TEMPLATEerror_simple_compileSUCCESS
SUBPATTERNMARKr ATOMIC_GROUPgetwidthMAXCODEATSRE_FLAG_MULTILINE AT_MULTILINEget AT_LOCALE
AT_UNICODEBRANCHJUMPCATEGORY CH_LOCALE
CH_UNICODEGROUPREFGROUPREF_LOC_IGNOREGROUPREF_IGNOREGROUPREF_UNI_IGNOREGROUPREF_EXISTS) codepatternr
emit_len
LITERAL_CODESREPEATING_CODES
SUCCESS_CODES ASSERT_CODESiscasedtolowerfixesopavloskipkcharsethascasedgroupr r phitail
tailappendskipyesskipnos r r4 r4 % s ;;D D"M&O"M LGG E ""5?+B # #**G**G E((G((GB
..RR (%b)*R R[RR R[ 2 'Hu_*2./H ' :DtAw[( V UU2Y . W
Q / M!%dd!2DJ
2X 1"gw N GX**u/F]# RY]#:DtAw WeT 2 dd*DJ
3Y&W
S
?
"((B PQQr!u~_R(+,Dz47RU RU r!ue,W
!$Z$ .T
_R(+,Dz47RU RU r!ue, !$Z$ .T
_R(+,
:
-/ *E9iT
eAgq[! T1nUIy I JT
eAgq[]#
<
:DtAw T2u % M dd*DJ
=
H
<
H:DtAw!uzQA )B< 7888 JKKR T2a5% ( M dd*DJ
2X H))!%%b"-& ]]2r*)) ^^B+ H
6\ HD JeDz47r5)T
4:&Q !$Z$ .T
M !$Z$ .T
8^ H& r])) ^ H
8^..R (()_%() AJ
?
" H AqM4jG$q' T2a5% (!uT
dT!W $T
W 4q 8W
r!ue, #DzF 2V $T
W 4q 8W
2GH H} r c |j }| D ] \ }} || |t u r|t u r || (|t u s|t u r ||d ||d O|t
u r|j
| i|t u r|j
| |t u r;|t z r |t | |t z r |t | || t d| |t y )Nr r z#internal: unsupported set operator )r r& r' RANGERANGE_UNI_IGNORECHARSETextend
BIGCHARSETrC r rD r rE r2 r( )r[ r
rK rM rV rW s r r- r- s ;;DB R
<
7] H
5[B"2 2 AK AK
7] KKO
:
KKO
8^&Yr]#))Z^$R2GH H+ , Mr c ^ g }g }t d }d}| D ] \ }} |t u r7|r/ || } d|| < |r| |v r|| D ] }
d||
< |s || rd}nd|| < n|t u rt | d | d dz } |r[|r+t || D ] } d|| < | |v s
|| D ] }
d||
< nt || D ] } d|| < |sRt t || }n<| D ] } d|| < n/|t u r|j || f n|j || f g }
d} |j d| }|dk rnYt |
dk\ rd }
nH|j d| }|dk r |
j |t | f n|
j ||f q|
l|
D ]A \ }}||z
dk( r|j t |f &|j t ||dz
ff C ||z
}|st | t | k r||fS | |fS t | dk( r+t | }|j t |f ||z
}||fS t | }i }t d }d}t }t dd d D ]1 } || | dz }||v r || || dz < |x|| dz < ||< |dz
}||z
}3 t | }|gt | z |dd |j t |f ||z
}||fS # t $ rQ t | dk( r |ddz z
}Y |r |t u r |rt }d}n |s
|| rd}|j || f Y w xY w)
N FTr r i r i ) bytearrayr' re rangemapanyr& r
IndexErrorr rf find
_mk_bitmaprg bytes_bytes_to_codesri )r[ rS fixuprU outr` charmapr\ rV rW rZ rirunsqr^ datacompsmappingblockchunks r r* r* s
C
DnGHB;
&= "2Y&' R5[%*2Y-.
&/ 'GBK'+H&' 5[ beRU1W-A %(]-.
#$:-21X56
.6 &3 &)]-.
&3 ''*3w?';H!"A)*GAJ "#6\JJBx(KKR)8
} B D A
LLA q5 t9>D LLA q5 KKCL) * QF DAq1uz
GQ<(
EAq1u:./
t s3x#g,.= 7|s'"
GT?# t H} , GnG EnG
E ;D
1eS
!1s7# E> %e GAH /4 4GAH e QJE EMD " d D/'22D1IJJ
D!"4KC =M
&w<3&uv~-G U{ !1B#' (GBK'+H RH%5
&s A8KA |S )aj
Generate an overlap table for the following prefix.
An overlap table is a table of the same size as the prefix which
informs about the potential self-overlap for each index in the prefix:
- if overlap[i] == 0, prefix[i:] can't overlap prefix[0:...]
- if overlap[i] == k with 0 < k <= i, prefix[i-k+1:i+1] overlaps with
prefix[0:k]
r r )r rn )prefixtablerz idxs r _generate_overlap_tabler s C#f+ E
1c&k
"AElQi6#;&ax aa.C Qi6#;& QwE!H # Lr c h | t z sy | t z rt j S t j S r )r r r r r )r
s r _get_iscasedr s/ & & ! !###!!!r c g }|j }d }t | }| j D ] \ }}|t u r|r
|| r n || #|t u rn|\ }} }
} t || |
} | t z r | t z r nNt | | \ }
}}| | t | }n|t | |z }|j |
|r n n ||dfS ||dfS )NTF) r r r} r' r6 r r r _get_literal_prefixr rh )rL r
r prefixappend prefix_skiprS rV rW r] r r r^ flags1prefix1 prefix_skip1got_alls r r r s
F==LK5!G,,B
=72;
:
-/ *E9i#E9i@F++ 0H-@F-K *G\7"$"%f+K!-"%f+ "? # " r c |j \ }}|t kD rt }|dk( r| j t dd||g y g }d}g }|t z r |t
z s t
|| \ }}}|s t || }| j } | t t | }
| d d} |rt } |r| t z } n |r | t z } | | |t k r | | n | t |d t } | | |rR | t | | t | } | | | j | | j t | n |rt | \ }} t ||| t | |
z
| |
< y )Nr )r9 r: rh INFOr r r r r r SRE_INFO_PREFIXSRE_INFO_LITERALSRE_INFO_CHARSETr r* r- )
rK rL r
rX r_ r r r[ r rM rY maskr\ s
r
_compile_infor sv
FB G|
Qw T1aR()
FKG ' 'EO,C':7E'J$ W)'59G ;;DJ t9Dd1g D
7**D &&J G| R W
!H
S[ v;K [ F +F34 -g 6%.TT!DJr c . t | t t f S r )
isinstancestrrt )objs r isstringr ? s cC< ((r c | j j |z }g }t || | t || j | |j t |S r )stater
r r4 r} r r5 )r^ r
rK s r _coder B sJ
GGMME !E
D $5!
T1665!KK Kr c 8 ddj d | D z S )N[%s], c 3 R K | ] }d t j dz dz |fz ! yw)z%#0*xr N)r CODESIZE.0xs r z _hex_code..R s( M1g q ):A(> >s %')join)rK s r _hex_coder Q s DII M MM MMr c dd l t dt t t dz
fd dt y )Nr r c l
d d fd
}fd}dz
}||k r| | }|dz
}t | }|t t t t t
t t fv r
|| n|t t t t t t t t fv r#| }|dz
} ||d|t! | fz ns|t" u r*| }|dz
}t% t& | } |||dd nA|t( u r*| }|dz
}t% t* | } |||dd n|t, t. t0 t2 fv r)| } |||||z |dz ||z ||z
}n|t4 t6 fv r4||dz \ }} |dz
} ||d || t! | t! | fz n|t8 u r- ||t; ||d
t< z z |d
t< z z
}nW|t> u r| }|dz
}tA d jC fd ||d
tD jF z z D }
||||
|d
tD jF z z
}dz
tI | D ], } |t; ||d
t< z z |d
t< z z
}. dz n|tJ tL tN tP tR fv r| }|dz
} ||| ns|tT u r| } |||||z |dz
}nQ|tV u rW| } |||||z |r; |dz ||z ||z
}| | }|r |d
|||z n |t |r;|dz
}n|tX tZ t\ t^ t` fv r=||dz \ }} }
|
tb k( rd}
|||| |
||z |dz ||z ||z
}n|td u r!||dz \ }} ||||||z |dz
}no|tf th fv r0||dz \ }} ||||||z |dz ||z ||z
}n1|tj u r)| } |||||z |dz ||z ||z
}n |tl u r||dz \ }}} }
|
tb k( rd}
|||to | | |
||z |dz |tp z r|dz |dz \ }} |d| |dz |z } |dddjC d |D z ddjC ts t | z |z
|d |z |z
|tt z r dz
|d ||z dz ||z
}n tw | ||k rdz y )N)toc | j | |d| fz fz
}t dv rdndfz ddz
z t | y )Nz(to %d)z%*d%s :.z r end)addprint)r argslabelslevel offset_widthstarts r print_z!dis..dis_..print_\ sW ~
2bU*,, (lE%6/3s SSE!Gn
& 4Lr c @ t ddz z z t | y )N r r )r )r r r s r print_2z"dis..dis_..print_2d s c.dis_.. s- (R1PA )*
4==#--(P1Ps 47branch MAXREPEATr z
prefix_skipz prefixr r c 3 & K | ] }d |z yw)z%#02xNr r s r r z$dis..dis_.. s .KFqw{Fs z(%r) z overlapin)<OPCODESr5 r( r. r0 MAX_UNTIL MIN_UNTILr& r' r% LITERAL_IGNORENOT_LITERAL_IGNORELITERAL_UNI_IGNORENOT_LITERAL_UNI_IGNORELITERAL_LOC_IGNORENOT_LITERAL_LOC_IGNOREchrr; r ATCODESrC CHCODESr) r, r$ r+ re rf rg r r ri listr r r rn r7 rF rH rI rG rB rA REPEAT
REPEAT_ONEMIN_REPEAT_ONEPOSSESSIVE_REPEATPOSSESSIVE_REPEAT_ONEr rJ ASSERT
ASSERT_NOTr8 r binr ro r
ValueError)r r r r rz rV argrY rX r_ r jminmaxr
prefix_lenr r rK dis_r r r r s ` r r zdis..dis_[ s !
#gEaB
FABgwW F4 4r
&(:*,B*,BD D 1gQr<3C/9:r1gQ'#,'r3qr7#x 1gQ'#,'r3qr7#I}mDDAwr4AdF+QqS!D&!T /00 a1 BQr0BCGSW3MMNw r9T!Qi -?%@ABS)^#z!1gQ sxx (R15aS$--=O9O1P(R R Sr3(S$--''
sAId1a#y..@&A BCi 'A $
h9L+- -1gQr3tAwr4AdF+Qv Awr4AdF+1af%IA E 7D x!D&9 w Q
N)+@B B!%a1 c3)#%Cr4caf5QqS!D&!T & AaCL Tr340Q
++ AaCL cr440QqS!D&!T |#Awr4AdF+QqS!D&!T t(,Q! %eS#)#%Cr4US#!D&A!?*.21Q3!n+J O[9 EE !%z)9 :FJ "TYY.KF.K%K K "RWWSf-=%> >@ Z'EKeU:5E)FGZ'E++QJEDM$'QJET n$o #gr
r )r setr r )rK r r r r r s `@@@@@r disr T sF
UF
Es3t9q=)*LI IV CIr c
t | r| }t j | | } nd }t | | }|t z rt t
| | j j }d g| j j z }|j D ]
\ }}|||< t j ||| j j z || j j dz
|t | S )Nr )r r parser SRE_FLAG_DEBUGr r r groupdictgroupsitemsr compiler
tuple)r^ r
rL rK
groupindex
indexgrouprZ rz s r r r s { MM!U # E?D ~
D ""J!''..(J "1
1
# << & qE*%
r )NNN)r )4__doc__r r r
_constants_casefixr r' r% r r5 r( r r r r r. r) r
MIN_REPEATr r r
MAX_REPEATr r r r r r
r r4 r- r* r r r: r intrs ru r3 r r r r r r r r r r r r r r s & " ;'7#$
Ry( N3J/)74IJ & 0 0, qIf6Ob
MMA >Q
(s 3
*" &>*X9"v)
NRj
r