Java編寫的C語言詞法分析器

1、這是java編寫的C語言詞法分析器，我將核心代碼整理起來，準備放在QQ空間和博客上，目的是互相學習借鑒，希望可以得到高手改進。這個詞法分析器實現(xiàn)的功能有打開文件、保存文件、打開幫助文檔、文本域內(nèi)容的剪切和復制和黏貼、進行詞法分析程序的項目結(jié)構(gòu)如圖，Word類和Unidentifiable類是兩個JavaBean類，存放的參數(shù)有兩個row(整型)、word(String)，row用于獲取行數(shù)，word用于獲取標識符，LexerFrame是詞法分析器的界面類，Analyze封裝了進行詞法分析的核心代碼 ，doc文件夾放一個幫助文檔，當用戶點擊幫助按鈕時可以彈出來以幫助用戶使用。/核心程序：pack

2、age com.lexer;import java.util.ArrayList;/*120號為關(guān)鍵字,用下標表示,i+1就是其機器碼;2140號為操作符,用下標表示,i+21就是其機器碼;4160號為分界符,* 用下標表示,i+41就是其機器碼;用戶自定義的標識符，其機器碼為51;常數(shù)的機器碼為52；不可以識別的標識符，其機器碼為0*/public class Analyze /關(guān)鍵字private Stringkeyword="int","long","char","if","else",&

3、quot;for","while","return","break","continue","switch","case","default","float","double","void","struct","static","do","short"/運算符private Stringoperator=&q

4、uot;+","-","*","/","%","=",">","<","!","=","!=",">=","<=","+","-","&","&&","|","","&qu

5、ot;/分界符private String delimiter=",","","(",")","","","'",""",":","#"public Analyze() /* 判斷是否是數(shù)字*/public boolean isDigit(char ch)if(ch>='0'&&ch<='9')return true;e

6、lsereturn false;/* 判斷是否是字母的函數(shù)*/public boolean isLetter(char ch)if(ch>='a'&&ch<='z')|(ch>='A'&&ch<='Z')return true;elsereturn false;/* 判斷是否由兩個運算符組成*/public boolean isTwoOperator(String str,char ch)char lc;int flag=0;if(str.length()>1|str.

7、length()=0)/字符數(shù)大于2和無字符的情況return false;else/字符數(shù)等于2的情況lc=str.charAt(str.length()-1);if(ch='='&&(lc='>'|lc='<'|lc='='|lc='!')else if(ch='+'&&lc='+')else if(ch='-'&&lc='-')else if(ch='|'&&am

8、p;lc='|')else if(ch='&'&&lc='&')elsereturn false;/否就返回falsereturn true;/其它符號的情況都返回true/* 獲取關(guān)鍵字的機器碼*/public int getKeywordOpcodes(String str)int i;for(i=0;i<keyword.length;i+)if(str.equals(keywordi)break;if(i<keyword.length)return i+1;/返回關(guān)鍵字的機器碼elsereturn

9、 0;/* 獲取操作符的機器碼*/public int getOperatorOpcodes(String str) int i;for(i=0;i<operator.length;i+)if(str.equals(operatori)break;if(i<operator.length)return i+21;/返回操作符的機器碼elsereturn 0;/* 獲取分界符的機器碼*/public int getDelimiterOpcodes(String str)int i;for(i=0;i<delimiter.length;i+)if(str.equals(delim

10、iteri)break;if(i<delimiter.length)return i+41;/返回分界符的機器碼elsereturn 0;/* 判斷字符是否可以識別*/public boolean isIdent(String str)char ch;int i;for(i=0;i<str.length();i+)ch=str.charAt(i);/非數(shù)字串的情況和非由英文字母組成的字符串if(i=0&&!isLetter(ch)|(!isDigit(ch)&&!isLetter(ch) break;if(i<str.length()retur

11、n false;elsereturn true;/* 預處理函數(shù)*/public String preFunction(String str)String ts=""int i;char ch,nc;/這里的i<str.length()-1for(i=0;i<str.length()-1;i+)ch=str.charAt(i);nc=str.charAt(i+1);if(ch='n')/如果字符是換行符,將n換成$ch='$'ts=ts+ch;else if(ch=' '|ch='r'|ch=

12、9;t')if(nc=' '|nc='r'|ch='t')continue;/連續(xù)' '或者't'或者'r'的情況，直接跳過 elsech=' '/一個' '或者't'或者'r'的情況，將這些字符換成' ' ts=ts+ch;elsets=ts+ch;/將字符連起來ch=str.charAt(str.length()-1);if(ch!=' '&&ch!='r'&a

13、mp;&ch!='t'&&ch!='n')ts=ts+ch;return ts;/* 將字符串分成一個個單詞，存放在數(shù)組列表*/public ArrayList<Word> divide(String str)ArrayList<Word> list=new ArrayList<Word>();String s=""char ch;int i;int row=1;for(i=0;i<str.length();i+)ch=str.charAt(i);if(i=0&&

14、;ch=' ')/字符串的第一個字符continue;if(ch=' ')/' '或者't'或者'r'的情況if(s!="")list.add(new Word(row, s);s=""/置空elsecontinue;else if(isDigit(ch)|isLetter(ch)if(s=""|isDigit(s.charAt(s.length()-1)|isLetter(s.charAt(s.length()-1)s = s + ch;elselist

15、.add(new Word(row, s);s = ""s=s + ch;elseif(isTwoOperator(s, ch)/兩個運算符的情況s = s + ch;elseif(s=""&&ch!='$')s = s + ch;else if(s=""&&ch='$')/若檢測到$符號，就換行row+;/行數(shù)加一elselist.add(new Word(row, s);s = ""if(ch!='$')s=s + ch;elser

16、ow+;if(s!="")list.add(new Word(row, s);return list;/* 判斷字符串是數(shù)字串，單個字符，還是一個字符串 */public int check(String str)char ch;ch=str.charAt(0);if(ch>='0'&&ch<='9')return 1;/數(shù)字串if(str.length()=1)return 2;/單個字符elsereturn 3;/一個字符串/* 檢查字符串是否為數(shù)字串，返回其機器碼 */public int checkDigi

17、t(String str)int i;char ch;for(i=0;i<str.length();i+)ch=str.charAt(i);if(ch>'9'|ch<'0')break;if(i<str.length()return 0;/不可識別的情況elsereturn 52;/常數(shù)/* 檢查字符串是否為單個字符，返回其機器碼 */public int checkChar(String str)if(getOperatorOpcodes(str)!=0)/操作符return getOperatorOpcodes(str);else i

18、f(getDelimiterOpcodes(str)!=0)/分界符return getDelimiterOpcodes(str);else if(isIdent(str)return 51;/用戶自定義標識符的機器碼elsereturn 0;/不可以被識別的標識符，機器碼為0/* 檢查字符串是否為字符串，返回其機器碼*/public int checkString(String str)if(getOperatorOpcodes(str)!=0)/操作符return getOperatorOpcodes(str);else if(getKeywordOpcodes(str)!=0)/關(guān)鍵字r

19、eturn getKeywordOpcodes(str);else if(isIdent(str)return 51;/用戶自定義標識符的機器碼elsereturn 0;/不可以被識別的標識符，機器碼為0/*在界面類寫這個方法來調(diào)用方法/* 詞法分析*/public void doTokenizing()consoleTextArea.setText(null);ArrayList<Word> wlist=new ArrayList<Word>();ArrayList<Unidentifiable> ulist=new ArrayList<Uniden

20、tifiable>();String s,ts,str;Word word;int i;int opcodes=-1;int errorNum=0;int count=0;s=fileContentTextArea.getText();if(s.length()>1)ts=analyze.preFunction(s);wlist=analyze.divide(ts);values=new Stringwlist.size()3;while(wlist.size()>0)word=(Word)wlist.remove(0);str=word.getWord();i=analyze.check(str);switch (i) case 1:opcodes=analyze.checkDigit(str);break;case 2:opcodes=analyze.checkChar(str);break;case 3:opcodes=analyze.checkString(str);break;if(opcodes=0)Unidentifiable u=new Unidentifiable(word.getRow(), str); ulist.add(u);errorNum+;