U
_m8 @ s< d dl Z d dlZd dlmZ d dlmZ G dd deZdS ) N)lex)TOKENc @ sJ e Zd ZdZdd Zdd Zdd Zdd Zd
d Zd d
Z dd Z
dd Z dZ i Z
e D ]4Zedkrnee
d< qXedkree
d< qXee
e < qXe d ZdZdZdZdZd Zd Zd e d e d Zd!e Zee e Zee e Zd"Zd#Z d$Z d%Z d&Z d'e d( e d( e d) Z d*e d Z!d+e! d+ Z"d,e" Z#d-e! d. e! d/ Z$d-e! d0 e d1 Z%d2e d Z&d3e& d4 Z'd,e' Z(d3e& d5 e e& d4 Z)d6Z*d7Z+d8e+ d e* d9 e* d: Z,d;Z-d<e d= e d> e d? Z.d@e d@ e d( e. d e- dA Z/dBZ0dCdD Z1e2e'dEdF Z3e2edGdH Z4dIdJ Z5dKdL Z6dMZ7dNdO Z8dPdQ Z9dRdS Z:dTZ;e2e'dUdV Z<e2edWdX Z=dYdZ Z>dMZ?d[d\ Z@d]ZAd^ZBd_ZCd`ZDdaZEdbZFdcZGddZHdeZIdfZJdgZKdhZLdiZMdjZNdkZOdlZPdmZQdnZRdoZSdpZTdqZUdrZVdsZWdtZXduZYdvZZdwZ[dxZ\dyZ]dzZ^d{Z_d|Z`d}Zad~ZbdZcdZddZedZfdZgdZhdZidZjdZkdZle2ddd Zme2ddd Zne'Zoe2e,dd Zpe2e/dd Zqe2edd Zre2edd Zse2edd Zte2edd Zue2edd Zve2e"dd Zwe2e#dd Zxe2e$dd Zye2e%dd Zze2e(dd Z{e2e)dd Z|e2edd Z}dd Z~dS )CLexera A lexer for the C language. After building it, set the
input text with input(), and call token() to get new
tokens.
The public attribute filename can be set to an initial
filaneme, but the lexer will update it upon #line
directives.
c C s@ || _ || _|| _|| _d| _d| _td| _td| _ dS )ab Create a new Lexer.
error_func:
An error function. Will be called with an error
message, line and column as arguments, in case of
an error during lexing.
on_lbrace_func, on_rbrace_func:
Called when an LBRACE or RBRACE is encountered
(likely to push/pop type_lookup_func's scope)
type_lookup_func:
A type lookup function. Given a string, it must
return True IFF this string is a name of a type
that was defined with a typedef earlier.
Nz([ ]*line\W)|([ ]*\d+)z
[ ]*pragma\W)
error_funcon_lbrace_funcon_rbrace_functype_lookup_funcfilename
last_tokenrecompile line_patternpragma_pattern)selfr r r r r /c_lexer.py__init__ s zCLexer.__init__c K s t j f d| i|| _dS )z Builds the lexer from the specification. Must be
called after the lexer object is created.
This method exists separately, because the PLY
manual warns against calling lex.lex inside
__init__
objectN)r lexer)r kwargsr r r build: s z CLexer.buildc C s d| j _dS )z? Resets the internal line number counter of the lexer.
N)r linenor r r r reset_linenoD s zCLexer.reset_linenoc C s | j | d S N)r input)r textr r r r I s z CLexer.inputc C s | j | _| jS r )r tokenr r r r r r L s z CLexer.tokenc C s | j jdd|j}|j| S )z3 Find the column of the token in its line.
r )r lexdatarfindlexpos)r r Zlast_crr r r find_tok_columnP s zCLexer.find_tok_columnc C s0 | |}| ||d |d | jd d S )Nr r )_make_tok_locationr r skip)r msgr locationr r r _error[ s
z
CLexer._errorc C s |j | |fS r )r r$ )r r r r r r% ` s zCLexer._make_tok_location)%_BOOL_COMPLEXZAUTOZBREAKZCASEZCHARZCONSTZCONTINUEZDEFAULTZDOZDOUBLEELSEZENUMZEXTERNZFLOATZFORZGOTOZIFZINLINEZINTZLONGZREGISTERZOFFSETOFZRESTRICTZRETURNZSHORTZSIGNEDZSIZEOFZSTATICZSTRUCTZSWITCHZTYPEDEFZUNIONZUNSIGNEDZVOIDZVOLATILEZWHILEr* Z_Boolr+ Z_Complex);IDTYPEIDZ
INT_CONST_DECZ
INT_CONST_OCTZ
INT_CONST_HEXZ
INT_CONST_BINZ FLOAT_CONSTZHEX_FLOAT_CONSTZ
CHAR_CONSTZ WCHAR_CONSTZSTRING_LITERALZWSTRING_LITERALPLUSMINUSZTIMESZDIVIDEZMODORZANDZNOTZXORZLSHIFTZRSHIFTZLORZLANDZLNOTZLTZLEZGTZGEZEQZNEZEQUALSZ
TIMESEQUALZDIVEQUALZMODEQUAL PLUSEQUALZ
MINUSEQUALZ LSHIFTEQUALZ RSHIFTEQUALZANDEQUALZXOREQUALZOREQUALZPLUSPLUSZ
MINUSMINUSZARROWZCONDOPZLPARENZRPARENZLBRACKETZRBRACKETLBRACERBRACECOMMAZPERIODSEMICOLONELLIPSISPPHASHz[a-zA-Z_$][0-9a-zA-Z_$]*z0[xX]z [0-9a-fA-F]+z0[bB]z[01]+zD(([uU]ll)|([uU]LL)|(ll[uU]?)|(LL[uU]?)|([uU][lL])|([lL][uU]?)|[uU])?z(0z)|([1-9][0-9]*)z0[0-7]*z 0[0-7]*[89]z([a-zA-Z._~!=&\^\-\\?'"])z(\d+)z(x[0-9a-fA-F]+)z#([\\][^a-zA-Z._~^!=&\^\-\\?'"x0-7])z(\\(|z))z
([^'\\\n]|'Lz('z*\n)|('z*$)z[^'
]+')|('')|('z [^'\n]*')z
([^"\\\n]|"z*"*z([eE][-+]?[0-9]+)z([0-9]*\.[0-9]+)|([0-9]+\.)z((((z
?)|([0-9]+z
))[FfLl]?)z([pP][+-]?[0-9]+)z(((z)?\.z)|(z\.))(z[FfLl]?)))ppline exclusive)pppragmarB c C sf | j j|jj|jjdr2|jd d | _| _n0| jj|jj|jjdrX|jd n
d|_ |S dS )z[ \t]*\#)posrA NrC r9 )
r matchr r! r# beginpp_line pp_filenamer typer tr r r t_PPHASH s zCLexer.t_PPHASHc C s0 | j d kr| d| n|jdd| _d S )Nz$filename before line number in #liner> )rG r) valuelstriprstriprH rJ r r r t_ppline_FILENAME s
zCLexer.t_ppline_FILENAMEc C s | j d kr|j| _ n d S r )rG rM rJ r r r t_ppline_LINE_NUMBER
s
zCLexer.t_ppline_LINE_NUMBERc C sH | j dkr| d| n t| j | j_| jdk r8| j| _|jd dS )\nNz line number missing in #lineINITIAL)rG r) intr r rH r
rF rJ r r r t_ppline_NEWLINE s
zCLexer.t_ppline_NEWLINEc C s dS )lineNr rJ r r r t_ppline_PPLINE s zCLexer.t_ppline_PPLINEz c C s | d| d S )Nzinvalid #line directiver) rJ r r r t_ppline_error& s zCLexer.t_ppline_errorc C s |j jd7 _|j d dS )rR r rS N)r r rF rJ r r r t_pppragma_NEWLINE, s zCLexer.t_pppragma_NEWLINEc C s dS )ZpragmaNr rJ r r r t_pppragma_PPPRAGMA1 s zCLexer.t_pppragma_PPPRAGMAz$ <>.-{}();=+-*/$%@&^~!?:,0123456789c C s d S r r rJ r r r t_pppragma_STR7 s zCLexer.t_pppragma_STRc C s d S r r rJ r r r
t_pppragma_ID: s zCLexer.t_pppragma_IDc C s | d| d S )Nzinvalid #pragma directiverX rJ r r r t_pppragma_error= s zCLexer.t_pppragma_errorc C s |j j|jd7 _dS )z\n+r N)r r rM countrJ r r r t_NEWLINEF s zCLexer.t_NEWLINEz\+-z\*/%z\|&~z\^z<<z>>z\|\|z&&!<>z<=z>=z==z!==z\*=z/=z%=z\+=z-=z<<=z>>=z&=z\|=z\^=z\+\+z--z->z\?z\(z\)z\[z\],z\.;:z\.\.\.z\{c C s | |S r )r rJ r r r t_LBRACE s zCLexer.t_LBRACEz\}c C s | |S r )r rJ r r r t_RBRACE s zCLexer.t_RBRACEc C s |S r r rJ r r r
t_FLOAT_CONST s zCLexer.t_FLOAT_CONSTc C s |S r r rJ r r r t_HEX_FLOAT_CONST s zCLexer.t_HEX_FLOAT_CONSTc C s |S r r rJ r r r t_INT_CONST_HEX s zCLexer.t_INT_CONST_HEXc C s |S r r rJ r r r t_INT_CONST_BIN s zCLexer.t_INT_CONST_BINc C s d}| || d S )NzInvalid octal constantrX r rK r' r r r t_BAD_CONST_OCT s zCLexer.t_BAD_CONST_OCTc C s |S r r rJ r r r t_INT_CONST_OCT s zCLexer.t_INT_CONST_OCTc C s |S r r rJ r r r t_INT_CONST_DEC s zCLexer.t_INT_CONST_DECc C s |S r r rJ r r r t_CHAR_CONST s zCLexer.t_CHAR_CONSTc C s |S r r rJ r r r
t_WCHAR_CONST s zCLexer.t_WCHAR_CONSTc C s d}| || d S )Nz Unmatched 'rX rs r r r t_UNMATCHED_QUOTE s zCLexer.t_UNMATCHED_QUOTEc C s d|j }| || d S )NzInvalid char constant %s)rM r) rs r r r t_BAD_CHAR_CONST s
zCLexer.t_BAD_CHAR_CONSTc C s |S r r rJ r r r t_WSTRING_LITERAL s zCLexer.t_WSTRING_LITERALc C s d}| || d S )Nz#String contains invalid escape coderX rs r r r t_BAD_STRING_LITERAL s zCLexer.t_BAD_STRING_LITERALc C s2 | j |jd|_|jdkr.| |jr.d|_|S )Nr- r. ) keyword_mapgetrM rI r rJ r r r t_ID s z CLexer.t_IDc C s" dt |jd }| || d S )NzIllegal character %sr )reprrM r) rs r r r t_error s zCLexer.t_errorN)__name__
__module__ __qualname____doc__r r r r r r$ r) r% keywordsr} keywordlowertokensZ
identifierZ
hex_prefixZ
hex_digitsZ
bin_prefixZ
bin_digitsZinteger_suffix_optZdecimal_constantZoctal_constantZ hex_constantZ bin_constantZbad_octal_constantZ
simple_escapeZdecimal_escapeZ
hex_escapeZ
bad_escapeZescape_sequenceZ cconst_charZ
char_constZ wchar_constZunmatched_quoteZbad_char_constZ string_charZstring_literalZwstring_literalZbad_string_literalZ
exponent_partZfractional_constantZfloating_constantZbinary_exponent_partZhex_fractional_constantZhex_floating_constantZstatesrL r rP rQ rU rW Zt_ppline_ignorerY rZ r[ Zt_pppragma_ignorer\ r] r^ Zt_ignorer` Zt_PLUSZt_MINUSZt_TIMESZt_DIVIDEZt_MODZt_ORZt_ANDZt_NOTZt_XORZt_LSHIFTZt_RSHIFTZt_LORZt_LANDZt_LNOTZt_LTZt_GTZt_LEZt_GEZt_EQZt_NEZt_EQUALSZ t_TIMESEQUALZ
t_DIVEQUALZ
t_MODEQUALZ t_PLUSEQUALZ t_MINUSEQUALZ
t_LSHIFTEQUALZ
t_RSHIFTEQUALZ
t_ANDEQUALZ t_OREQUALZ
t_XOREQUALZ
t_PLUSPLUSZ t_MINUSMINUSZt_ARROWZt_CONDOPZt_LPARENZt_RPARENZ
t_LBRACKETZ
t_RBRACKETZt_COMMAZt_PERIODZt_SEMIZt_COLONZ
t_ELLIPSISrm rn Zt_STRING_LITERALro rp rq rr rt ru rv rw rx ry rz r{ r| r r r r r r r s& !
; $
r )r sysZplyr Zply.lexr r r r r r r <module> s