o ijU @s(dZddlmZddlZddlZddlmZddlmZm Z m Z m Z ddl m Z mZddl mZmZmZmZmZmZdd lmZmZGd d d eZd ZeeZed eZd~ddZdddZdddZidej d fd!ej d"fd#ej dfd$ej!d fd%ej!d"fd&ej!dfd'ej"d fd(ej"d"fd)ej"dfd*ej#d fd+ej#d"fd,ej#dfd-ej$d fd.ej$d"fd/ej$dfd0ej%d fd1ej%d"fid2ej%dfd3ej&d fd4ej&d"fd5ej&dfd6ej'd fd7ej'd"fd8ej'dfd9ej(d fd:ej(d"fd;ej(dfdej)dfd?ej*d fd@ej*d"fdAej*dfdBej+d fidCej+d"fdDej+dfdEej,d fdFej,d"fdGej,dfdHej-d fdIej-d"fdJej-dfdKej.d fdLej/d fdMej0dfdNej1dfdOej2dfdPej3dfdQej4dfdRej5dfdSej6dfej7dfej8dfej9dfej:dfej;dfejdfej?dfej@dfdT ZAddWdXZBddZd[ZCdd]d^ZDdd`daZEddfdgZFddhdiZGGdjdkdkZHGdldmdmeHZIGdndodoeHZJgdpZKddrdsZLddwdxZMdd|d}ZNdS)axParser for assembly programs. Limitations: 1. This is only used to parse programs that we've already assembled and linked successfully, so it's not intended to handle invalid programs gracefully. 2. This is only guaranteed to handle the subset of assembly we use in the book. I've included some support for other common assembly instructions but you shouldn't rely on it. ) annotationsN)Path) GeneratorListOptionalUnion)asmtokenize)Expr ImmediateOpcodeOperandOperatorRegister)TokenTokTypec@eZdZdZdS) ParseErrorzSWe encountered nvalid assembly (or, more likely, valid assembly that don't support)N__name__ __module__ __qualname____doc__rrg/home/gtoal/gtoal.com/languages/algol60/new-unicode-parser/lang/c/tests4/test_framework/parser/parse.pyrrz[_A-Za-z][_A-Za-z0-9]*_lblstrreturnboolcCs"t}tjdkr t}t||duS)zWCould this symbol name be a function or variable name from the original source program?darwinN) C_IDENTIFIERsysplatformMANGLED_C_IDENTIFIERre fullmatch)rregexrrris_valid_c_identifier$s r*toks List[Token]expectedrNonecCs2|d}|j|krtd|d|d|dS)z6Consume next token and fail if it isn't what we expectrz Expected z but found z. Remaining tokens: N)poptok_typer)r+r-next_tokrrr expect_next-s  r2toktuple[Opcode, Optional[int]]cCs|s td||dvrtjdfS|dvrtjdfS|dvr&tjdfS|dr0tjdfS|d rMd}|d d kr@d }n|d d krHd}tj|fS|drjd}|d d kr]d }n|d d kred}tj|fS|dst|drytj dfS|drtj dfS|drtj dfS|drtj d fSgd}dd|D}||}|ddkr|d d|vrtj dfStD];}t|}||r|t|d}d}|d krd }n |d krd}n|dkrd}|tjtjtjfvrd}||fSqtjdfS)zParse an instruction mnemonic; return opcode and inferred size Inferred size is used to normalize immediate values later on; it's None if we can't infer the size or if this is a floating-point instruction zBad mnemonic: )cqocqto)cdqcltd)cltcltqcdqemovsdNmovsbrlmovzcomiucomipxormulset)eggerBlerAbeaaeppopesczcSsg|]}d|qS)nr).0ccrrr sz parse_opcode..rjq)isalnumrr CDQCDQE startswithMOVMOVSMOVZCMPXORIMULSETCCJMPCCrlenPOPPUSHLEAUNKNOWN)r3sizecondition_codesnegated_condition_codesall_condition_codesopcodeopsuffixrrr parse_opcode6sj                        rtraxr7eaxr:alrbxebxblrcxecxclrdxedxdlrdiedidilrsiesisilr8r8dr8br9r9dr9br10r10dr10br11r11dr11br12r12dr12br13r13dr13br14r14dr14br15r15dr15brsprbpripxmm0xmm1xmm2xmm3xmm4xmm5) xmm6xmm7xmm8xmm9xmm10xmm11xmm12xmm13xmm14xmm15 list[Token]tuple[Register, Optional[int]]cCsBt|tjd|dj}zt|WSty td|w)zwParse register and infer its size in bytes ::= "%" ::= "rax" | "eax" | etc. r+r-rz)expected register name after % but found )r2rPERCENTr/tok_str REG_ALIASESKeyErrorr)r+reg_namerrrparse_registers   rr cCst|tjd|d}|j}|tjkrt|jdd}t|S|tj tj fvrQ|d}|jtjkr ::= "$" [ "+" | "-" ] ::= ? decimal or hexadecimal integer ? rrbasezbad immediate value: $zBad immediate value: $) r2rDOLLARr/r0INTintrr PLUS_SIGN MINUS_SIGNr)r+r1r0valnum_tokrrrparse_immediates      rr cCsg} |d}|j}|tjkr||jn=|tjkr'|t|jddn-|tjkr3|t j n!|tj kr?|t j n|tj krK|t j n |d| |Sq)zParse an expression (used as displacement in memory operand) NOTE: we don't normalize these, so +10(%rbp) and 10(%rbp) will NOT compare equal ::= { | | "+" | "-" }+ Trr)r/r0rSYMBOLappendrrrrrPLUSrMINUSATinsert)r+exprr1tok_typrrr parse_expr s$       rtuple[Operand, Optional[int]]cCs,d}d}d}d}|djtjkrt|}t|tjd|djtjkr)t|\}}|d}|jtjkr=t j ||ddfS|jtj krKt dt ||dj}|tjkr`t|djdd}n%t|\}}|djtj kr|d|djtjkrt|djdd}t|tjdt ||||dfS)a Parse memory operand ::= [ ] "(" ")" ::= [ "," ] // base, with or without other stuff | "," // no base ::= | [ "," [ ]] Nrrr)disprz8Unexpected token after base register in memory operand: r)r0r OPEN_PARENrr2rrr/ CLOSE_PARENr MemoryCOMMArrrrr)r+rridxscalerr1 next_tok_typerrrparse_memory_operand's4          rrrrrm Optional[int]cCsRt|tr'|dur td|dkr|S|j|ddd}tj|ddd}t|S|S) z3Normalize immediate values to signed representationNz:Can't interpret immediate b/c instruction size is ambigousrlittleF)length byteordersignedT)rr) isinstancer rto_bytesr from_bytes)rrrmas_bytesrrrr fix_immediate`s rcCs|dj}|tjkrt|S|tjkrt|dfSt|dkr.|tjkr.|dj }|dfS|tj kr<|dt|St|dkrl|tjkrl|djtj krl|djtjkrl|dj d|dj }| |dfSt |S)zlParse the next operand in list of tokens ::= | | ["@" ] rNr@)r0rrrrrrhrr/rSTARrclearr)r+start_tok_typetargetrrr parse_operandrs&         rc@r) Directivez Any directiveNrrrrrrrrc@r)EnterTextSectionz5Directive that makes text section the current sectionNrrrrrrrrc@r)LeaveTextSectionzLDirective that makes any section other than text section the current sectionNrrrrrrrr)z.bssz.data.cstringz.rodataz.literal4.literal8z .literal16rtok_listcCs|djdkr tS|djtvrtS|djdvr%t|djd|djdkr[|djtjkr>td|dj|dj}|dkrJtS|dkrX|d jd krXtStStS) abParse a directive and figure out whether it enters the text section, exits it, or neither NOTE: unlike earlier parse_* statements, we don't need to consume these tokens. tok_list represents a single line of assembly and once we've identified the kind of directive we can just discard the rest of the line ::= | | ::= ".text" | ".section" "__TEXT" "," "__text" | ".section" ".text" ::= ".section" | ".bss" | ".data" | etc. ::= { }+ ::= ? anything other than .text or __TEXT,__text ? ::= ? any symbol starting with "." other than ".section", ".text", ".bss", etc ? rz.text)z .popsectionz .pushsectionz .previousz not supportedz.sectionrz5Expected section name after section directive, found __TEXTr__text) rrNON_TEXT_SECTIONSrrr0rrr)r section_namerrrparse_directives$ rtokensGenerator[Token, None, None]Union[asm.AsmItem, Directive]c s^t|}|jtjtjfvrt|}|jtjtjfvs |jtjkr(td|jttjd}t||}|jtj kr?t |jS|g}|jtjtjfvr^| |t||}|jtjtjfvsK|dj drjt|S|d}t|j\}g}|rt|\}} | || rs| |rt|tjd|std|szdurfdd |D}t ||S) a]Parse the next instruction, label or directive Grammar: ::=