Question

我正在尝试学习BNF并尝试组装一些Z80 ASM代码。由于我是两个领域的新手，我的问题是，我是否在正确的轨道上？我正在尝试将Z80 ASM的格式编写为EBNF，以便我可以从那里找出从源头创建机器代码的位置。目前我有以下内容：

Assignment = Identifier, ":" ;

Instruction = Opcode, [ Operand ], [ Operand ] ;

Operand = Identifier | Something* ;

Something* = "(" , Identifier, ")" ;

Identifier = Alpha, { Numeric | Alpha } ;

Opcode = Alpha, Alpha ;

Int = [ "-" ], Numeric, { Numeric } ;

Alpha = "A" | "B" | "C" | "D" | "E" | "F" | 
        "G" | "H" | "I" | "J" | "K" | "L" | 
        "M" | "N" | "O" | "P" | "Q" | "R" | 
        "S" | "T" | "U" | "V" | "W" | "X" | 
        "Y" | "Z" ;

Numeric = "0" | "1" | "2" | "3"| "4" | 
          "5" | "6" | "7" | "8" | "9" ;

如果我出错，任何方向反馈都会非常好。

Answer 1

旧式汇编程序通常使用汇编程序进行手工编码，并使用adhoc解析技术处理汇编源代码行以生成实际的汇编程序代码。当汇编语法很简单时（例如总是OPCODE REG，OPERAND），这种方法运行得很好。

现代机器具有杂乱，讨厌的指令集，其中包含许多指令变体和操作数，可以用复杂的语法表达，允许多个索引寄存器参与操作数表达式。允许使用具有固定和可重定位常量的复杂的汇编时表达式以及各种类型的加法运算符使这变得复杂。允许条件编译，宏，结构化数据声明等的复杂汇编程序都增加了对语法的新要求。通过ad hoc方法处理所有这些语法非常困难，这也是解析器生成器被发明的原因。

使用BNF和解析器生成器是构建现代汇编器的非常合理的方法，即使对于诸如Z80的传统处理器也是如此。我为摩托罗拉8位机器（如6800/6809）构建了这样的汇编程序，并准备为现代x86做同样的事情。我认为你正走在正确的道路上。

**********编辑**************** OP询问了lexer和解析器定义。我在这里提供了两个。

这些是6809 asssembler的真实规格的摘录。完整的定义是这里样本大小的2-3倍。

为了减少空间，我已经编辑了大部分暗角复杂性这是这些定义的重点。有人可能会因为显而易见的复杂性而感到沮丧;该关键是，有了这样的定义，你试图描述语言的形状，而不是在程序上编码。如果你，你会付出更高的复杂性以特别的方式对所有这些进行编码，这将是远远不够的不太可维护。

了解这些定义也会有所帮助用于高端程序分析系统有lexing /解析工具作为子系统，称为 The DMS Software Reengineering Toolkit。 DMS将自动从中构建AST 解析器specfication中的语法规则，这使它成为一个很容易建立解析工具。最后，解析器规范包含所谓的“prettyprinter” 声明，允许DMS从AST重新生成源文本。（语法的真正目的是允许我们构建代表汇编程序的AST 说明，然后将它们吐出来送给真正的汇编程序！）

有一点值得注意：如何陈述词汇和语法规则（metasyntxax！）在不同的词法分析器/解析器生成器系统之间有所不同。该基于DMS的规范的语法也不例外。 DMS相对复杂它自己的语法规则，在这里可用的空间中解释真的不切实际。您将不得不忍受其他系统使用类似符号的想法 EBNF用于规则和lexemes的正则表达式变体。

鉴于OP的兴趣，他可以实现类似的词法分析器/解析器使用任何词法分析器/解析器生成器工具，例如FLEX / YACC， JAVACC，ANTLR，...

********** LEXER **************

-- M6809.lex: Lexical Description for M6809
-- Copyright (C) 1989,1999-2002 Ira D. Baxter

%%
#mainmode Label

#macro digit "[0-9]"
#macro hexadecimaldigit "<digit>|[a-fA-F]"

#macro comment_body_character "[\u0009 \u0020-\u007E]" -- does not include NEWLINE

#macro blank "[\u0000 \ \u0009]"

#macro hblanks "<blank>+"

#macro newline "\u000d \u000a? \u000c? | \u000a \u000c?" -- form feed allowed only after newline

#macro bare_semicolon_comment "\; <comment_body_character>* "

#macro bare_asterisk_comment "\* <comment_body_character>* "

...[snip]

#macro hexadecimal_digit "<digit> | [a-fA-F]"

#macro binary_digit "[01]"

#macro squoted_character "\' [\u0021-\u007E]"

#macro string_character "[\u0009 \u0020-\u007E]"

%%Label -- (First mode) processes left hand side of line: labels, opcodes, etc.

#skip "(<blank>*<newline>)+"
#skip "(<blank>*<newline>)*<blank>+"
  << (GotoOpcodeField ?) >>

#precomment "<comment_line><newline>"

#preskip "(<blank>*<newline>)+"
#preskip "(<blank>*<newline>)*<blank>+"
  << (GotoOpcodeField ?) >>

-- Note that an apparant register name is accepted as a label in this mode
#token LABEL [STRING] "<identifier>"
  <<  (local (;; (= [TokenScan natural] 1) ; process all string characters
         (= [TokenLength natural] ?:TokenCharacterCount)=
         (= [TokenString (reference TokenBodyT)] (. ?:TokenCharacters))
         (= [Result (reference string)] (. ?:Lexeme:Literal:String:Value))
         [ThisCharacterCode natural]
         (define Ordinala #61)
         (define Ordinalf #66)
         (define OrdinalA #41)
         (define OrdinalF #46)
     );;
     (;; (= (@ Result) `') ; start with empty string
     (while (<= TokenScan TokenLength)
      (;;   (= ThisCharacterCode (coerce natural TokenString:TokenScan))  
        (+= TokenScan) ; bump past character
        (ifthen (>= ThisCharacterCode Ordinala)
           (-= ThisCharacterCode #20) ; fold to upper case
        )ifthen
        (= (@ Result) (append (@ Result) (coerce character ThisCharacterCode)))=

        );;
     )while
     );;
  )local
  (= ?:Lexeme:Literal:String:Format (LiteralFormat:MakeCompactStringLiteralFormat 0))  ; nothing interesting in string
  (GotoLabelList ?)
  >>

%%OpcodeField

#skip "<hblanks>"
  << (GotoEOLComment ?) >>
#ifnotoken
  << (GotoEOLComment ?) >>

-- Opcode field tokens
#token 'ABA'       "[aA][bB][aA]"
   << (GotoEOLComment ?) >>
#token 'ABX'       "[aA][bB][xX]"
   << (GotoEOLComment ?) >>
#token 'ADC'       "[aA][dD][cC]"
   << (GotoABregister ?) >>
#token 'ADCA'      "[aA][dD][cC][aA]"
   << (GotoOperand ?) >>
#token 'ADCB'      "[aA][dD][cC][bB]"
   << (GotoOperand ?) >>
#token 'ADCD'      "[aA][dD][cC][dD]"
   << (GotoOperand ?) >>
#token 'ADD'       "[aA][dD][dD]"
   << (GotoABregister ?) >>
#token 'ADDA'      "[aA][dD][dD][aA]"
   << (GotoOperand ?) >>
#token 'ADDB'      "[aA][dD][dD][bB]"
   << (GotoOperand ?) >>
#token 'ADDD'      "[aA][dD][dD][dD]"
   << (GotoOperand ?) >>
#token 'AND'       "[aA][nN][dD]"
   << (GotoABregister ?) >>
#token 'ANDA'      "[aA][nN][dD][aA]"
   << (GotoOperand ?) >>
#token 'ANDB'      "[aA][nN][dD][bB]"
   << (GotoOperand ?) >>
#token 'ANDCC'     "[aA][nN][dD][cC][cC]"
   << (GotoRegister ?) >>
...[long list of opcodes snipped]

#token IDENTIFIER [STRING] "<identifier>"
  <<  (local (;; (= [TokenScan natural] 1) ; process all string characters
         (= [TokenLength natural] ?:TokenCharacterCount)=
         (= [TokenString (reference TokenBodyT)] (. ?:TokenCharacters))
         (= [Result (reference string)] (. ?:Lexeme:Literal:String:Value))
         [ThisCharacterCode natural]
         (define Ordinala #61)
         (define Ordinalf #66)
         (define OrdinalA #41)
         (define OrdinalF #46)
     );;
     (;; (= (@ Result) `') ; start with empty string
     (while (<= TokenScan TokenLength)
      (;;   (= ThisCharacterCode (coerce natural TokenString:TokenScan))  
        (+= TokenScan) ; bump past character
        (ifthen (>= ThisCharacterCode Ordinala)
           (-= ThisCharacterCode #20) ; fold to upper case
        )ifthen
        (= (@ Result) (append (@ Result) (coerce character ThisCharacterCode)))=

        );;
     )while
     );;
  )local
  (= ?:Lexeme:Literal:String:Format (LiteralFormat:MakeCompactStringLiteralFormat 0))  ; nothing interesting in string
  (GotoOperandField ?)
  >>

#token '#'   "\#" -- special constant introduction (FDB)
   << (GotoDataField ?) >>

#token NUMBER [NATURAL] "<decimal_number>"
  << (local [format LiteralFormat:NaturalLiteralFormat]
    (;; (= ?:Lexeme:Literal:Natural:Value (ConvertDecimalTokenStringToNatural (. format) ? 0 0))
    (= ?:Lexeme:Literal:Natural:Format (LiteralFormat:MakeCompactNaturalLiteralFormat format))
    );;
 )local
 (GotoOperandField ?)
  >>

#token NUMBER [NATURAL] "\$ <hexadecimal_digit>+"
  << (local [format LiteralFormat:NaturalLiteralFormat]
    (;; (= ?:Lexeme:Literal:Natural:Value (ConvertHexadecimalTokenStringToNatural (. format) ? 1 0))
    (= ?:Lexeme:Literal:Natural:Format (LiteralFormat:MakeCompactNaturalLiteralFormat format))
    );;
 )local
 (GotoOperandField ?)
  >>

#token NUMBER [NATURAL] "\% <binary_digit>+"
  << (local [format LiteralFormat:NaturalLiteralFormat]
    (;; (= ?:Lexeme:Literal:Natural:Value (ConvertBinaryTokenStringToNatural (. format) ? 1 0))
    (= ?:Lexeme:Literal:Natural:Format (LiteralFormat:MakeCompactNaturalLiteralFormat format))
    );;
 )local
 (GotoOperandField ?)
  >>

#token CHARACTER [CHARACTER] "<squoted_character>"
  <<  (= ?:Lexeme:Literal:Character:Value (TokenStringCharacter ? 2))
  (= ?:Lexeme:Literal:Character:Format (LiteralFormat:MakeCompactCharacterLiteralFormat 0 0)) ; nothing special about character
  (GotoOperandField ?)
  >>


%%OperandField

#skip "<hblanks>"
  << (GotoEOLComment ?) >>
#ifnotoken
  << (GotoEOLComment ?) >>

-- Tokens signalling switch to index register modes
#token ','   "\,"
   <<(GotoRegisterField ?)>>
#token '['   "\["
   <<(GotoRegisterField ?)>>

-- Operators for arithmetic syntax
#token '!!'  "\!\!"
#token '!'   "\!"
#token '##'  "\#\#"
#token '#'   "\#"
#token '&'   "\&"
#token '('   "\("
#token ')'   "\)"
#token '*'   "\*"
#token '+'   "\+"
#token '-'   "\-"
#token '/'   "\/"
#token '//'   "\/\/"
#token '<'   "\<"
#token '<'   "\<" 
#token '<<'  "\<\<"
#token '<='  "\<\="
#token '</'  "\<\/"
#token '='   "\="
#token '>'   "\>"
#token '>'   "\>"
#token '>='  "\>\="
#token '>>'  "\>\>"
#token '>/'  "\>\/"
#token '\\'  "\\"
#token '|'   "\|"
#token '||'  "\|\|"

#token NUMBER [NATURAL] "<decimal_number>"
  << (local [format LiteralFormat:NaturalLiteralFormat]
    (;; (= ?:Lexeme:Literal:Natural:Value (ConvertDecimalTokenStringToNatural (. format) ? 0 0))
    (= ?:Lexeme:Literal:Natural:Format (LiteralFormat:MakeCompactNaturalLiteralFormat format))
    );;
 )local
  >>

#token NUMBER [NATURAL] "\$ <hexadecimal_digit>+"
  << (local [format LiteralFormat:NaturalLiteralFormat]
    (;; (= ?:Lexeme:Literal:Natural:Value (ConvertHexadecimalTokenStringToNatural (. format) ? 1 0))
    (= ?:Lexeme:Literal:Natural:Format (LiteralFormat:MakeCompactNaturalLiteralFormat format))
    );;
 )local
  >>

#token NUMBER [NATURAL] "\% <binary_digit>+"
  << (local [format LiteralFormat:NaturalLiteralFormat]
    (;; (= ?:Lexeme:Literal:Natural:Value (ConvertBinaryTokenStringToNatural (. format) ? 1 0))
    (= ?:Lexeme:Literal:Natural:Format (LiteralFormat:MakeCompactNaturalLiteralFormat format))
    );;
 )local
  >>

-- Notice that an apparent register is accepted as a label in this mode
#token IDENTIFIER [STRING] "<identifier>"
  <<  (local (;; (= [TokenScan natural] 1) ; process all string characters
         (= [TokenLength natural] ?:TokenCharacterCount)=
         (= [TokenString (reference TokenBodyT)] (. ?:TokenCharacters))
         (= [Result (reference string)] (. ?:Lexeme:Literal:String:Value))
         [ThisCharacterCode natural]
         (define Ordinala #61)
         (define Ordinalf #66)
         (define OrdinalA #41)
         (define OrdinalF #46)
     );;
     (;; (= (@ Result) `') ; start with empty string
     (while (<= TokenScan TokenLength)
      (;;   (= ThisCharacterCode (coerce natural TokenString:TokenScan))  
        (+= TokenScan) ; bump past character
        (ifthen (>= ThisCharacterCode Ordinala)
           (-= ThisCharacterCode #20) ; fold to upper case
        )ifthen
        (= (@ Result) (append (@ Result) (coerce character ThisCharacterCode)))=

        );;
     )while
     );;
  )local
  (= ?:Lexeme:Literal:String:Format (LiteralFormat:MakeCompactStringLiteralFormat 0))  ; nothing interesting in string
  >>

%%Register -- operand field for TFR, ANDCC, ORCC, EXG opcodes

#skip "<hblanks>"
#ifnotoken << (GotoRegisterField ?) >>

%%RegisterField -- handles registers and indexing mode syntax
-- In this mode, names that look like registers are recognized as registers

#skip "<hblanks>"
  << (GotoEOLComment ?) >>
#ifnotoken
  << (GotoEOLComment ?) >>

#token '['   "\["
#token ']'   "\]"
#token '--'  "\-\-"
#token '++'  "\+\+"

#token 'A'      "[aA]"
#token 'B'      "[bB]"
#token 'CC'     "[cC][cC]"
#token 'DP'     "[dD][pP] | [dD][pP][rR]" -- DPR shouldnt be needed, but found one instance
#token 'D'      "[dD]"
#token 'Z'      "[zZ]"

-- Index register designations
#token 'X'      "[xX]"
#token 'Y'      "[yY]"
#token 'U'      "[uU]"
#token 'S'      "[sS]"
#token 'PCR'    "[pP][cC][rR]"
#token 'PC'     "[pP][cC]"

#token ','    "\,"

-- Operators for arithmetic syntax
#token '!!'  "\!\!"
#token '!'   "\!"
#token '##'  "\#\#"
#token '#'   "\#"
#token '&'   "\&"
#token '('   "\("
#token ')'   "\)"
#token '*'   "\*"
#token '+'   "\+"
#token '-'   "\-"
#token '/'   "\/"
#token '<'   "\<"
#token '<'   "\<" 
#token '<<'  "\<\<"
#token '<='  "\<\="
#token '<|'  "\<\|"
#token '='   "\="
#token '>'   "\>"
#token '>'   "\>"
#token '>='  "\>\="
#token '>>'  "\>\>"
#token '>|'  "\>\|"
#token '\\'  "\\"
#token '|'   "\|"
#token '||'  "\|\|"

#token NUMBER [NATURAL] "<decimal_number>"
  << (local [format LiteralFormat:NaturalLiteralFormat]
    (;; (= ?:Lexeme:Literal:Natural:Value (ConvertDecimalTokenStringToNatural (. format) ? 0 0))
    (= ?:Lexeme:Literal:Natural:Format (LiteralFormat:MakeCompactNaturalLiteralFormat format))
    );;
 )local
  >>

... [snip]

%% -- end M6809.lex

**************** PARSER **************

-- M6809.ATG: Motorola 6809 assembly code parser
-- (C) Copyright 1989;1999-2002 Ira D. Baxter; All Rights Reserved

m6809 = sourcelines ;

sourcelines = ;
sourcelines = sourcelines sourceline EOL ;
  <<PrettyPrinter>>: { V(CV(sourcelines[1]),H(sourceline,A<eol>(EOL))); }

-- leading opcode field symbol should be treated as keyword.

sourceline = ;
sourceline = labels ;
sourceline = optional_labels 'EQU' expression ;
  <<PrettyPrinter>>: { H(optional_labels,A<opcode>('EQU'),A<operand>(expression)); }
sourceline = LABEL 'SET' expression ;
  <<PrettyPrinter>>: { H(A<firstlabel>(LABEL),A<opcode>('SET'),A<operand>(expression)); }
sourceline = optional_label instruction ;
  <<PrettyPrinter>>: { H(optional_label,instruction); }
sourceline = optional_label optlabelleddirective ;
  <<PrettyPrinter>>: { H(optional_label,optlabelleddirective); }
sourceline = optional_label implicitdatadirective ;
  <<PrettyPrinter>>: { H(optional_label,implicitdatadirective); }
sourceline = unlabelleddirective ;
sourceline = '?ERROR' ;
  <<PrettyPrinter>>: { A<opcode>('?ERROR'); }

optional_label = labels ;
optional_label = LABEL ':' ;
  <<PrettyPrinter>>: { H(A<firstlabel>(LABEL),':'); }
optional_label = ;

optional_labels = ;
optional_labels = labels ;
labels = LABEL ;
  <<PrettyPrinter>>: { A<firstlabel>(LABEL); }
labels = labels ',' LABEL ;
  <<PrettyPrinter>>: { H(labels[1],',',A<otherlabels>(LABEL)); }

unlabelleddirective = 'END' ;
  <<PrettyPrinter>>: { A<opcode>('END'); }
unlabelleddirective = 'END' expression ;
  <<PrettyPrinter>>: { H(A<opcode>('END'),A<operand>(expression)); }
unlabelleddirective = 'IF' expression EOL conditional ;
  <<PrettyPrinter>>: { V(H(A<opcode>('IF'),H(A<operand>(expression),A<eol>(EOL))),CV(conditional)); }
unlabelleddirective = 'IFDEF' IDENTIFIER EOL conditional ;
  <<PrettyPrinter>>: { V(H(A<opcode>('IFDEF'),H(A<operand>(IDENTIFIER),A<eol>(EOL))),CV(conditional)); }
unlabelleddirective = 'IFUND' IDENTIFIER EOL conditional ;
  <<PrettyPrinter>>: { V(H(A<opcode>('IFUND'),H(A<operand>(IDENTIFIER),A<eol>(EOL))),CV(conditional)); }
unlabelleddirective = 'INCLUDE' FILENAME ;
  <<PrettyPrinter>>: { H(A<opcode>('INCLUDE'),A<operand>(FILENAME)); }
unlabelleddirective = 'LIST' expression ;
  <<PrettyPrinter>>: { H(A<opcode>('LIST'),A<operand>(expression)); }
unlabelleddirective = 'NAME' IDENTIFIER ;
  <<PrettyPrinter>>: { H(A<opcode>('NAME'),A<operand>(IDENTIFIER)); }
unlabelleddirective = 'ORG' expression ;
  <<PrettyPrinter>>: { H(A<opcode>('ORG'),A<operand>(expression)); }
unlabelleddirective = 'PAGE' ;
  <<PrettyPrinter>>: { A<opcode>('PAGE'); }
unlabelleddirective = 'PAGE' HEADING ;
  <<PrettyPrinter>>: { H(A<opcode>('PAGE'),A<operand>(HEADING)); }
unlabelleddirective = 'PCA' expression ;
  <<PrettyPrinter>>: { H(A<opcode>('PCA'),A<operand>(expression)); }
unlabelleddirective = 'PCC' expression ;
  <<PrettyPrinter>>: { H(A<opcode>('PCC'),A<operand>(expression)); }
unlabelleddirective = 'PSR' expression ;
  <<PrettyPrinter>>: { H(A<opcode>('PSR'),A<operand>(expression)); }
unlabelleddirective = 'TABS' numberlist ;
  <<PrettyPrinter>>: { H(A<opcode>('TABS'),A<operand>(numberlist)); }
unlabelleddirective = 'TITLE' HEADING ;
  <<PrettyPrinter>>: { H(A<opcode>('TITLE'),A<operand>(HEADING)); }
unlabelleddirective = 'WITH' settings ;
  <<PrettyPrinter>>: { H(A<opcode>('WITH'),A<operand>(settings)); }

settings = setting ;
settings = settings ',' setting ;
  <<PrettyPrinter>>: { H*; }
setting = 'WI' '=' NUMBER ;
  <<PrettyPrinter>>: { H*; }
setting = 'DE' '=' NUMBER ;
  <<PrettyPrinter>>: { H*; }
setting = 'M6800' ;
setting = 'M6801' ;
setting = 'M6809' ;
setting = 'M6811' ;

-- collects lines of conditional code into blocks
conditional = 'ELSEIF' expression EOL conditional ;
  <<PrettyPrinter>>: { V(H(A<opcode>('ELSEIF'),H(A<operand>(expression),A<eol>(EOL))),CV(conditional[1])); }
conditional = 'ELSE' EOL else ;
  <<PrettyPrinter>>: { V(H(A<opcode>('ELSE'),A<eol>(EOL)),CV(else)); }
conditional = 'FIN' ;
  <<PrettyPrinter>>: { A<opcode>('FIN'); }
conditional = sourceline EOL conditional ;
  <<PrettyPrinter>>: { V(H(sourceline,A<eol>(EOL)),CV(conditional[1])); }

else = 'FIN' ;
  <<PrettyPrinter>>: { A<opcode>('FIN'); }
else = sourceline EOL else ;
  <<PrettyPrinter>>: { V(H(sourceline,A<eol>(EOL)),CV(else[1])); }

-- keyword-less directive, generates data tables

implicitdatadirective = implicitdatadirective ',' implicitdataitem ;
  <<PrettyPrinter>>: { H*; }
implicitdatadirective = implicitdataitem ;

implicitdataitem = '#' expression ;
  <<PrettyPrinter>>: { A<operand>(H('#',expression)); }
implicitdataitem = '+' expression ;
  <<PrettyPrinter>>: { A<operand>(H('+',expression)); }
implicitdataitem = '-' expression ;
  <<PrettyPrinter>>: { A<operand>(H('-',expression)); }
implicitdataitem = expression ;
  <<PrettyPrinter>>: { A<operand>(expression); }
implicitdataitem = STRING ;
  <<PrettyPrinter>>: { A<operand>(STRING); }

-- instructions valid for m680C (see Software Dynamics ASM manual)
instruction = 'ABA' ;
  <<PrettyPrinter>>: { A<opcode>('ABA'); }
instruction = 'ABX' ;
  <<PrettyPrinter>>: { A<opcode>('ABX'); }

instruction = 'ADC' 'A' operandfetch ;
  <<PrettyPrinter>>: { H(A<opcode>(H('ADC','A')),A<operand>(operandfetch)); }
instruction = 'ADC' 'B' operandfetch ;
  <<PrettyPrinter>>: { H(A<opcode>(H('ADC','B')),A<operand>(operandfetch)); }
instruction = 'ADCA' operandfetch ;
  <<PrettyPrinter>>: { H(A<opcode>('ADCA'),A<operand>(operandfetch)); }
instruction = 'ADCB' operandfetch ;
  <<PrettyPrinter>>: { H(A<opcode>('ADCB'),A<operand>(operandfetch)); }
instruction = 'ADCD' operandfetch ;
  <<PrettyPrinter>>: { H(A<opcode>('ADCD'),A<operand>(operandfetch)); }

instruction = 'ADD' 'A' operandfetch ;
  <<PrettyPrinter>>: { H(A<opcode>(H('ADD','A')),A<operand>(operandfetch)); }
instruction = 'ADD' 'B' operandfetch ;
  <<PrettyPrinter>>: { H(A<opcode>(H('ADD','B')),A<operand>(operandfetch)); }
instruction = 'ADDA' operandfetch ;
  <<PrettyPrinter>>: { H(A<opcode>('ADDA'),A<operand>(operandfetch)); }

[..snip...]

-- condition code mask for ANDCC and ORCC
conditionmask = '#' expression ;
  <<PrettyPrinter>>: { H*; }
conditionmask = expression ;

target = expression ;

operandfetch = '#' expression ; --immediate
  <<PrettyPrinter>>: { H*; }

operandfetch = memoryreference ;

operandstore = memoryreference ;

memoryreference = '[' indexedreference ']' ;
  <<PrettyPrinter>>: { H*; }
memoryreference = indexedreference ;

indexedreference = offset ;
indexedreference = offset ',' indexregister ;
  <<PrettyPrinter>>: { H*; }
indexedreference = ',' indexregister ;
  <<PrettyPrinter>>: { H*; }
indexedreference = ',' '--' indexregister ;
  <<PrettyPrinter>>: { H*; }
indexedreference = ',' '-' indexregister ;
  <<PrettyPrinter>>: { H*; }
indexedreference = ',' indexregister '++' ;
  <<PrettyPrinter>>: { H*; }
indexedreference = ',' indexregister '+' ;
  <<PrettyPrinter>>: { H*; }

offset = '>' expression ; -- page zero ref
  <<PrettyPrinter>>: { H*; }
offset = '<' expression ; -- long reference
  <<PrettyPrinter>>: { H*; }
offset = expression ;
offset = 'A' ;
offset = 'B' ;
offset = 'D' ;

registerlist = registername ;
registerlist = registerlist ',' registername ;
  <<PrettyPrinter>>: { H*; }

registername = 'A' ;
registername = 'B' ;
registername = 'CC' ;
registername = 'DP' ;
registername = 'D' ;
registername = 'Z' ;
registername = indexregister ;

indexregister = 'X' ;
indexregister = 'Y' ;
indexregister = 'U' ;  -- not legal on M6811
indexregister = 'S' ;
indexregister = 'PCR' ;
indexregister = 'PC' ;

expression = sum '=' sum ;
  <<PrettyPrinter>>: { H*; }
expression = sum '<<' sum ;
  <<PrettyPrinter>>: { H*; }
expression = sum '</' sum ;
  <<PrettyPrinter>>: { H*; }
expression = sum '<=' sum ;
  <<PrettyPrinter>>: { H*; }
expression = sum '<' sum ;
  <<PrettyPrinter>>: { H*; }
expression = sum '>>' sum ;
  <<PrettyPrinter>>: { H*; }
expression = sum '>/' sum ;
  <<PrettyPrinter>>: { H*; }
expression = sum '>=' sum ;
  <<PrettyPrinter>>: { H*; }
expression = sum '>' sum ;
  <<PrettyPrinter>>: { H*; }
expression = sum '#' sum ;
  <<PrettyPrinter>>: { H*; }
expression = sum ;

sum = product ;
sum = sum '+' product ;
  <<PrettyPrinter>>: { H*; }
sum = sum '-' product ;
  <<PrettyPrinter>>: { H*; }
sum = sum '!' product ;
  <<PrettyPrinter>>: { H*; }
sum = sum '!!' product ;
  <<PrettyPrinter>>: { H*; }

product = term '*' product ;
  <<PrettyPrinter>>: { H*; }
product = term '||' product ; -- wrong?
  <<PrettyPrinter>>: { H*; }
product = term '/' product ;
  <<PrettyPrinter>>: { H*; }
product = term '//' product ;
  <<PrettyPrinter>>: { H*; }
product = term '&' product ;
  <<PrettyPrinter>>: { H*; }
product = term '##' product ;
  <<PrettyPrinter>>: { H*; }
product = term ;

term = '+' term ;
  <<PrettyPrinter>>: { H*; }
term = '-' term ; 
  <<PrettyPrinter>>: { H*; }
term = '\\' term ; -- complement
  <<PrettyPrinter>>: { H*; }
term = '&' term ; -- not

term = IDENTIFIER ;
term = NUMBER ;
term = CHARACTER ;
term = '*' ;
term = '(' expression ')' ;
  <<PrettyPrinter>>: { H*; }

numberlist = NUMBER ;
numberlist = numberlist ',' NUMBER ;
  <<PrettyPrinter>>: { H*; }

Answer 2

BNF更常用于结构化的嵌套语言，如Pascal，C ++，或者实际上来自Algol家族的任何东西（包括像C＃这样的现代语言）。如果我实现了汇编程序，我可能会使用一些简单的正则表达式来模式匹配操作码和操作数。我使用Z80汇编语言已经有一段时间了，但你可能会使用类似的东西：

/\s*(\w{2,3})\s+((\w+)(,\w+)?)?/

这将匹配任何由两个或三个字母的操作码组成的行，后跟一个或两个以逗号分隔的操作数。在提取这样的汇编程序行之后，您将查看操作码并为指令生成正确的字节，包括操作数的值（如果适用）。

我上面使用正则表达式概述的解析器类型将被称为“ad hoc”解析器，这实际上意味着您在某种块的基础上拆分和检查输入（在汇编语言的情况下，通过文本线）。

Answer 3

我认为你不需要过度思考它。没有必要使解析器将“LD A，A”拆分为加载操作，目标和源寄存器，只需将整个事物（模数字和空格）直接匹配到一个操作码即可。

没有那么多操作码，并且它们的排列方式不会让您从解析和理解汇编程序IMO中获得很多好处。显然你需要一个字节/地址/索引参数的解析器，但除此之外，我只需要一对一的查找。

Z80 ASM BNF结构......我在正确的轨道上吗？

3 个答案: