Z80 ASM BNF 结构...我走在正确的轨道上吗？

Question

我正在尝试学习 BNF 并尝试汇编一些 Z80 ASM 代码。由于我对这两个领域都是新手，我的问题是，我是否走在正确的道路上？我正在尝试将 Z80 ASM 的格式编写为 EBNF，以便我可以弄清楚从那里到哪里从源代码创建机器代码。目前我有以下几点：

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" ;

如果我出错了，任何方向反馈都会很好。

Answer 1

老式汇编程序通常在汇编程序中手工编码，并使用临时解析技术来处理汇编源代码行以生成实际的汇编代码。
当汇编语法很简单时（例如总是OPCODE REG、OPERAND），这种方法就足够好了。

现代机器具有混乱、令人讨厌的指令集，其中包含大量指令变体和操作数，这些指令集可以用复杂的语法来表达，允许多个索引寄存器参与操作数表达。允许复杂的汇编时表达式使用固定和可重定位常量以及各种类型的加法运算符使这一点变得复杂。允许条件编译、宏、结构化数据声明等的复杂汇编器都对语法提出了新的要求。通过临时方法处理所有这些语法非常困难，这就是发明解析器生成器的原因。

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 Z80 等传统处理器也是如此。我已经为 Motorola 8 位机器（例如 6800/6809）构建了这样的汇编器，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 要求提供词法分析器和解析器定义的示例。
我已经在这里提供了两者。

这些是 6809 汇编器真实规格的摘录。
完整的定义是此处样本大小的 2-3 倍。

为了减少篇幅，我删除了大部分暗角的复杂内容
这就是这些定义的要点。
人们可能会对其表面上的复杂性感到沮丧。这
要点是，通过这样的定义，您试图描述
语言的形状，而不是程序化的编码。
如果您这样做，您将付出更高的复杂性
以特别的方式对所有这些进行编码，这将是很远的事
不太可维护。

了解这些定义也会有一些帮助
与高端程序分析系统一起使用
有词法分析/解析工具作为子系统，称为
DMS 软件重新工程工具包。 DMS 将自动从
构建 AST
解析器规范中的语法规则，这使得它成为
构建解析工具要容易得多。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。
（语法的真正目的是让我们能够构建代表汇编程序的 AST
指令，然后将它们吐出以馈送到真正的汇编器！）

值得注意的一件事：词位和语法规则是如何陈述的（metasyntxax！）
不同的词法分析器/解析器生成器系统之间有所不同。这
基于 DMS 的规范语法也不例外。 DMS有比较完善的
它自己的语法规则，在此处可用的空间中实际上无法解释。您必须接受其他系统使用类似符号的想法，例如
EBNF 用于规则和词位的正则表达式变体。

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/解析器生成工具，例如 FLEX/YACC，
JAVACC、ANTLR、...

********** 词法分析器 **************

-- 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

**************** 解析器 * **************

-- 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 更广泛地用于结构化、嵌套语言，如 Pascal、C++，或者任何源自 Algol 家族（包括 C# 等现代语言）的语言。如果我要实现汇编程序，我可能会使用一些简单的正则表达式来对操作码和操作数进行模式匹配。我已经有一段时间没有使用 Z80 汇编语言了，但您可能会使用类似的内容：

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

这将匹配由两个或三个字母的操作码后跟一个或两个以逗号分隔的操作数组成的任何行。提取这样的汇编程序行后，您将查看操作码并生成指令的正确字节，包括操作数的值（如果适用）。

我上面概述的使用正则表达式的解析器类型称为“临时”解析器，这本质上意味着您在某种块的基础上分割和检查输入（在汇编语言的情况下，通过文本行）。

Answer 3

我觉得你没必要想太多当您可以将整个事物（模数大小写和空格）直接字符串匹配到一个操作码时，制作一个将“LD A,A”分解为加载操作、目标和源寄存器的解析器是没有意义的。

操作码并不多，而且它们的排列方式也没有让您从解析和理解汇编器（IMO）中真正获得很多好处。显然，您需要一个用于字节/地址/索引参数的解析器，但除此之外，我只需进行一对一的查找。