Z80 ASM BNF 结构...我在正确的轨道上吗？

Question

我正在尝试学习 BNF 并尝试组装一些 Z80 ASM 代码。由于我对这两个领域都是新手，所以我的问题是，我是否走在正确的轨道上？我正在尝试将 Z80 ASM 的格式编写为 EBNF，这样我就可以找出从那里去哪里从源代码创建机器代码。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" ;

如果我出错了，任何方向性的反馈都会非常好。

Answer 1

老式的汇编程序通常在汇编程序中手动编码，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程序代码。 当汇编语法很简单时（例如，始终为 OPCODE REG、OPERAND），这就足够了。

现代机器的指令集杂乱无章，带有大量指令变体和操作数，可能用复杂的语法表示，允许多个索引寄存器参与操作数表达式。使用各种类型的加法运算符允许具有固定和可重定位常量的复杂汇编时表达式使这一点复杂化。允许条件编译、宏、结构化数据声明等的复杂汇编器都增加了对语法的新要求。通过特殊方法处理所有这些语法非常困难，这也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。我已经为摩托罗拉 8 位机器（例如 6800/6809）构建了这样的汇编器，并且正准备为现代 x86 做同样的事情。我认为你正朝着正确的方向前进。

********** 编辑 **************** OP 要求提供例如词法分析器和解析器定义。 我在这里都提供了。

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。

为了节省空间，我已经删除了大部分暗角的复杂性 这就是这些定义的重点。 人们可能会对表面上的复杂性感到沮丧；这 关键是，有了这样的定义，你试图描述 语言的形状，而不是程序化的编码。 如果你将付出显着更高的复杂性 以特别的方式对所有这些进行编码，这将是很远的 不易维护。

了解这些定义也会有所帮助 与高端程序分析系统一起使用 将词法分析/解析工具作为子系统，称为 The DMS Software Reengineering Toolkit。 DMS 将自动从
解析器规范中的语法规则，这使其成为 更容易构建解析工具。最后， 解析器规范包含所谓的“prettyprinter” 声明，它允许 DMS 从 AST 重新生成源文本。 （语法的真正目的是让我们构建代表汇编程序的 AST 指令，然后将它们吐出以提供给真正的汇编器！）

需要注意的一点：词位和语法规则是如何表述的（metasyntxax！） 在不同的词法分析器/解析器生成器系统之间有所不同。这 基于 DMS 的规范的语法也不例外。 DMS 具有相对复杂的 它自己的语法规则，在此处可用的空间中解释起来确实不切实际。您必须接受其他系统使用类似符号的想法，因为 用于规则的 EBNF 和用于词位的正则表达式变体。

考虑到 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、...

********** 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 更普遍地用于结构化、嵌套的语言，如 Pascal、C++ 或真正源自 Algol 家族的任何语言（包括 C# 等现代语言）。如果我正在实现一个汇编器，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。我已经有一段时间没有使用 Z80 汇编语言了，但你可能会使用类似的东西：

/\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用逗号分隔的操作数组成的任何行。像这样提取汇编行后，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）。

我在上面使用正则表达式概述的解析器类型将被称为“临时”解析器，这实质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，通过文本行）。

Answer 3

我认为你不需要想太多。当您可以直接将整个事物（模大小写和空格）字符串匹配到一个操作码中时，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使您真正从解析和理解汇编程序 IMO 中受益匪浅。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，我只需要一对一的查找。