http://blog.gregwym.info/tag/compiler/ http://blog.gregwym.info/compiler-ge-ge-bu-zhou-de-han-yi.html 2011-04-19T06:15:49+08:00 Step1: String -> Scanning (DFA) -> Tokens Scanner又叫Lexical analyzer或Lexer.其目的在于, 将需要compile的源代码逐字读入compiler, 并将每一个符合"词汇命名规则(Lexical Syntax)"的字段转换成token存储.换句话说, 就是一遍读, 一遍看每一个单词的拼写对不对. 对的就转成token, 拼错了就输出error.作业中对应: A6 P4(某些语言的compiler在Scanning之后还包含Preprocessing, 因为不在241讨论范围内, 不做解释)Step2: Tokens -> Parsing(LL/LR) -> Intermediate Format (WLI)Parsing又叫Syntactic Analysis.其目的在于, 将token与token联系在一起, 并将他们的转换成符合一定规范的"中间格式", 一般是某种树状结构, 例如241中定义的WLI.在Parsing过程中, 如果遇到不符合某种语言的"语法规则(Grammar)"时, 则输出error. 如果语法正确, 则输出对应格式.简单说, 就是看的说的话是不是人话, 有没有缺个标点少个括号.如果不是人话那就说明你该重新学语法去了.作业中对应: A8 P4Step3: Intermediate Format -> Semantic Analysis(Context-Sensitive Analysis) ->Semantic Analysis的目的在于, 检查程序是否存在语义上的冲突. 或者说, 上下文是不是相符.比如在C中, 如果没有declare过变量int a;, 则a = 3;就不合法.再比如, 如果a declare为int, 则a = 'b';就不合法, 因为a不能为char.作业中对应: A9 A10(Optimization为代码优化, 241没有涉及, 知道即可)Step4: -> Code Generation -> Code Fragment将Intermediate Format转换为另一种格式, 比如MIPS或者二进制文件, 可与上一步同步进行.作业中对应: A9 A10, A3, A4Step5: Code Fragements -> Linking -> Executable File(Single File)将多个compile好的多个文件链接在一起, 生成一个可执行的二进制文件(或仅生成合并以后的单个文件, 但文件本身可能无法执行)作业中对应: A5 P1, P2附: 09FALL 第二题答案 Scanning Parsing Semantic Analysis Parsing Semantic Analysis Linking Semantic Analysis Parsing 1A@#SR%$# + foo x http://blog.gregwym.info/lr-1--parser-example.html 2011-04-19T05:24:08+08:00 Terminal Symbols: 6个BOF, EOF, id, -, (, )Nonterminal Symbols: 3个S, expr, termStart Symbol:SProduction Rules: 5个 0 S BOF expr EOF 1 expr term 2 expr expr - term 3 term id 4 term ( expr ) Total State: 11个 (0 to 11)Total Transitions: 28个 State Symbol Action 0 BOF shift to state 6 1 ( shift to state 3 1 id shift to state 2 1 term shift to state 8 3 ( shift to state 3 3 expr shift to state 7 3 id shift to state 2 3 term shift to state 4 6 ( shift to state 3 6 expr shift to state 10 6 id shift to state 2 6 term shift to state 4 7 ) shift to state 9 7 - shift to state 1 10 - shift to state 1 10 EOF shift to state 5 2 ) reduce by rule 3 2 - reduce by rule 3 2 EOF reduce by rule 3 4 ) reduce by rule 1 4 - reduce by rule 1 4 EOF reduce by rule 1 8 ) reduce by rule 2 8 - reduce by rule 2 8 EOF reduce by rule 2 9 ) reduce by rule 4 9 - reduce by rule 4 9 EOF reduce by rule 4 String to Parse:BOF id - ( id ) - id EOFParse Step:蓝色的行是Shift, 灰色的行是ReduceReduce的次数取决于Production Rule RHS的长度当前Symbol所对应的State, State Stack (At EOL) Description S0 0 Start BOF S6 0 6 S0 BOF S6 id S2 0 6 2 S6 id S2 R3 - 0 6 S2 - Reduce by rule #3 (replace id with term) term S4 0 6 4 S6 term S4 R1 - 0 6 S4 - Reduce by rule #1 (replace term with expr) expr S10 0 6 10 S6 expr S10 - S1 0 6 10 1 S10 - S1 ( S3 0 6 10 1 3 S1 ( S3 id S2 0 6 10 1 3 2 S3 id S2 R3 ) 0 6 10 1 3 S2 ) Reduce by rule #3 (replace id with term) term S4 0 6 10 1 3 4 S3 term S4 R1 ) 0 6 10 1 3 S4 ) Reduce by rule #1 (replace term with expr) expr S7 0 6 10 1 3 7 S3 expr S7 ) S9 0 6 10 1 3 7 9 S7 ) S9 R4 R4 R4 - 0 6 10 1 S9 - Reduce by rule #4 (repace "( expr )" with term) term S8 0 6 10 1 8 S1 term S8 R2 R2 R2 - 0 6 S8 - Reduce by rule #2 (replace "expr - term" with expr) expr S10 0 6 10 S6 expr S10 - S1 0 6 10 1 S10 - S1 id S2 0 6 10 1 2 S1 id S2 R3 EOF 0 6 10 1 S2 EOF Reduce by rule #3 (replace id with term) term S8 0 6 10 1 8 S1 term S8 R2 R2 R2 EOF 0 6 S8 EOF Reduce by rule #2 (replace "expr - term" with expr) expr S10 0 6 10 S6 expr S10 EOF S5 Final S5 S10 EOF S5 原题来源: http://www.student.cs.uwaterloo.ca/~cs241/parsing/lr1.html