编译原理(PL0编译程序源代码)

/*PL/0编译程序（C语言版）*编译和运行环境：*Visual C++6.0*WinXP/7*使用方法：*运行后输入PL/0源程序文件名*回答是否将虚拟机代码写入文件*回答是否将符号表写入文件*执行成功会产生四个文件（词法分析结果.txt符号表.txt虚拟代码.txt源程序和地址.txt）*/#include <stdio.h>#include"pl0.h"#include"string"#define stacksize 500//解释执行时使用的栈int main(){bool nxtlev[symnum];printf("请输入源程序文件名:");scanf("%s",fname);fin=fopen(fname,"r");//以只读方式打开pl0源程序文件cifa=fopen("词法分析结果.txt","w");fa1=fopen("源程序和地址.txt","w");//输出源文件及各行对应的首地址fprintf(fa1,"输入pl0源程序文件名：");fprintf(fa1,"%s\n",fname);if(fin){printf("是否将虚拟机代码写入文件?(Y/N)");//是否输出虚拟机代码scanf("%s",fname);listswitch=(fname[0]=='y'||fname[0]=='Y');printf("是否将符号表写入文件?(Y/N)");//是否输出符号表scanf("%s",fname);tableswitch=(fname[0]=='y'||fname[0]=='Y');init();//初始化err=0;cc=cx=ll=0;ch=' ';if(-1!=getsym()){fa=fopen("虚拟代码.txt","w");fas=fopen("符号表.txt","w");addset(nxtlev,declbegsys,statbegsys,symnum);nxtlev[period]=true;if(-1==block(0,0,nxtlev)){//调用编译程序fclose(fa);fclose(fa1);fclose(fas);fclose(fin);return 0;}if(sym!=period){error(9);//结尾丢失了句号}if(err!=0){printf("pl0源程序出现错误，退出编译!!!请从第一个错误处开始修改.\n\n");fprintf(cifa,"源程序出现错误，请检查!!!");fprintf(fa1,"源程序出现错误，请检查!!!");fprintf(fa,"源程序出现错误，请检查!!!");fprintf(fas,"源程序出现错误，请检查!!!");}fclose(fa);fclose(fa1);fclose(fas);}fclose(fin);}else{printf("Can't open file!\n");}fclose(cifa);//printf("\n");return 0;}void init(){//初始化int i;for(i=0;i<=255;i++)ssym[i]=nul;//设置单字符符号ssym['+']=plus;ssym['-']=minus;ssym['*']=times;ssym['/']=slash;ssym['(']=lparen;ssym[')']=rparen;ssym['=']=eql;ssym[',']=comma;ssym['.']=period;ssym['#']=neq;ssym[';']=semicolon;strcpy(&(word[0][0]),"begin");//保留字设置，以字母顺序排列便于折半查找strcpy(&(word[1][0]),"call");strcpy(&(word[2][0]),"const");strcpy(&(word[3][0]),"do");strcpy(&(word[4][0]),"end");strcpy(&(word[5][0]),"if");strcpy(&(word[6][0]),"odd");strcpy(&(word[7][0]),"procedure");strcpy(&(word[8][0]),"read");strcpy(&(word[9][0]),"then");strcpy(&(word[10][0]),"var");strcpy(&(word[11][0]),"while");strcpy(&(word[12][0]),"write");wsym[0]=beginsym;//设置保留字类别一字即一类wsym[1]=callsym;wsym[2]=constsym;wsym[3]=dosym;wsym[4]=endsym;wsym[5]=ifsym;wsym[6]=oddsym;wsym[7]=procsym;wsym[8]=readsym;wsym[9]=thensym;wsym[10]=varsym;wsym[11]=whilesym;wsym[12]=writesym;strcpy(&(mnemonic[lit][0]),"lit");//设置指令名称strcpy(&(mnemonic[opr][0]),"opr");strcpy(&(mnemonic[lod][0]),"lod");strcpy(&(mnemonic[sto][0]),"sto");strcpy(&(mnemonic[cal][0]),"cal");strcpy(&(mnemonic[inte][0]),"int");strcpy(&(mnemonic[jmp][0]),"jmp");strcpy(&(mnemonic[jpc][0]),"jpc");for(i=0;i<symnum;i++){//设置符号集declbegsys[i]=false;statbegsys[i]=false;facbegsys[i]=false;}declbegsys[constsym]=true;//设置声明开始符号集declbegsys[varsym]=true;declbegsys[procsym]=true;statbegsys[beginsym]=true;//设置语句开始符号集statbegsys[callsym]=true;statbegsys[ifsym]=true;statbegsys[whilesym]=true;facbegsys[ident]=true;//设置因子开始符号集facbegsys[number]=true;facbegsys[lparen]=true;}//用数组实现集合的集合运算int inset(int e,bool* s){return s[e];}int addset(bool*sr,bool* s1,bool* s2,int n){int i;for(i=0;i<n;i++)sr[i]=s1[i]||s2[i];return 0;}void error(int n){//出错处理，打印出错位置和错误编码char space[81];memset(space, 32,81);space[cc-1]=0;printf("error(%d)",n);fprintf(fa1,"error(%d)",n);switch(n){case 1:printf("\t\t常量说明中的“=”写成“:=”\n");fprintf(fa1,"\t\t常量说明中的“=”写成“:=”\n");break;case 2:printf("\t\t常量说明中的=后应该是数字\n");fprintf(fa1,"\t\t常量说明中的=后应该是数字\n");break;case 3:printf("\t\t常量说明符中的表示符应该是=\n" );fprintf(fa1,"\t\t常量说明符中的表示符应该是=\n");break;case 4:printf("\t\tconst,var,procedure后应为标识符\n" );fprintf(fa1,"\t\tconst,var,procedure后应为标识符\n");break;case 5:printf("\t\t漏掉了“,”或“;”\n" );fprintf(fa1,"\t\t漏掉了“,”或“;”\n" );break;case 6:printf("\t\t过程说明后的符号不正确\n" );fprintf(fa1,"\t\t过程说明后的符号不正确\n");break;case 7:printf("\t\t应是语句开始符\n" );fprintf(fa1,"\t\t应是语句开始符\n" );break;case 8:printf("\t\t程序体语句部分的后跟符不正确\n" );fprintf(fa1,"\t\t程序体语句部分的后跟符不正确\n" );break;case 9:printf("\t\t程序结尾丢了句号“.”\n\n" );fprintf(fa1,"\t\t程序结尾丢了句号“.”\n");break;case 10:printf("\t\t语句之间漏了“;”\n" );fprintf(fa1,"\t\t语句之间漏了“;”\n");break;case 11:printf("\t\t标识符拼写错误或未说明\n" );fprintf(fa1,"\t\t标识符拼写错误或未说明\n");break;case 12:printf("\t\t赋值语句中，赋值号左部标识符属性应是变量\n" );fprintf(fa1,"\t\t赋值语句中，赋值号左部标识符属性应是变量\n");break;case 13:printf("\t\t赋值语句左部标识符后应是复制号“:=”\n" );fprintf(fa1,"\t\t赋值语句左部标识符后应是复制号“:=”\n");break;case 14:printf("\t\tcall后应为标识符\n" );fprintf(fa1,"\t\tcall后应为标识符\n");break;case 15:printf("\t\tcall后标识符属性应为过程\n" );fprintf(fa1,"\t\tcall后标识符属性应为过程\n");break;case 16:printf("\t\t条件语句中丢了then\n" );fprintf(fa1,"\t\t条件语句中丢了then\n");break;case 17:printf("\t\t丢了“end”或“;”\n" );fprintf(fa1,"\t\t丢了“end”或“;”\n");break;case 18:printf("\t\twhile型循环语句中丢了“do”\n" );fprintf(fa1,"\t\twhile型循环语句中丢了“do”\n");break;case 19:printf("\t\t语句后的符号不正确\n" );fprintf(fa1,"\t\t语句后的符号不正确\n" );break;case 20:printf("\t\t应为关系运算符\n" );fprintf(fa1,"\t\t应为关系运算符\n");break;case 21:printf("\t\t表达式标示符属性不能是过程\n" );fprintf(fa1,"\t\t表达式标示符属性不能是过程\n");break;case 22:printf("\t\t表达式漏掉了右括号\n" );fprintf(fa1,"\t\t表达式漏掉了右括号\n");break;case 23:printf("\t\t因子后的非法符号\n" );fprintf(fa1,"\t\t因子后的非法符号\n");break;case 24:printf("\t\t表达式的开始符不能是此符号\n" );fprintf(fa1,"\t\t表达式的开始符不能是此符号\n");break;case 25:printf("\t\t标识符越界\n" );fprintf(fa1,"\t\t标识符越界\n");break;case 26:printf("\t\t非法字符\n" );fprintf(fa1,"\t\t非法字符\n");break;case 31:printf("\t\t数越界\n");fprintf(fa1,"\t\t数越界\n");break;case 32:printf("\t\tread语句括号中的标识符不是变量\n" );fprintf(fa1,"\t\tread语句括号中的标识符不是变量\n");break;case 33:printf("\t\twrite()或read()中应为完整表达式\n" );fprintf(fa1,"\t\twrite()或read()中应为完整表达式\n");break;default:printf("\t\t出现未知错误\n" );fprintf(fa1,"\t\t出现未知错误\n");}err++;}//漏掉空格，读取一个字符，每次读一行，存入line缓冲区，line被getsym取空后再读一//行，被函数getsym调用int getch(){if(cc==ll){if(feof(fin)){printf("program incomplete");return-1;}ll=0;cc=0;printf("\n%d ",cx);fprintf(fa1,"\n%d ",cx);ch=' ';while(ch!=10){if(EOF==fscanf(fin,"%c",&ch)){line[ll]=0;break;}printf("%c",ch);fprintf(fa1,"%c",ch);line[ll]=ch;ll++;}fprintf(cifa,"\n");}ch=line[cc];cc++;return 0;}int getsym(){//词法分析int i,j,k,l;while(ch==' '||ch==10||ch==9)//忽略空格换行 TABgetchdo;if(ch>='a'&&ch<='z'){//以字母开头的为保留字或者标识符k=0,l=1;do{if(k<al){//al为标识符或保留字最大长度a[k]=ch;k++;}if(k==al&&l==1){error(25);l=0;}getchdo;}while(ch>='a'&&ch<='z'||ch>='0'&&ch<='9');a[k]=0;//末尾存零strcpy(id,a);i=0;j=norw-1;do{//开始折半查找k=(i+j)/2;if(strcmp(id,word[k])<=0){j=k-1;}if(strcmp(id,word[k])>=0){i=k+1;}}while(i<=j);if(i-1>j){//找到即为保留字sym=wsym[k];fprintf(cifa,"%s\t\t%ssym\n",id,id);//printf("%s\t\t%ssym\n",id,id);}else{//否则为标识符或数字sym=ident;fprintf(cifa,"%s\t\tident\n",id);//printf("%s\t\tident\n",id);}}else {if(ch>='0'&&ch<='9'){k=0;num=0;sym=number;do{num=10*num+ch-'0';//数字的数位处理k++;getchdo;}while(ch>='0'&&ch<='9');k--;if(k>nmax){//数字的长度限制fprintf(cifa,"0\t\tnumber\n");num=0;error(31);}elsefprintf(cifa,"%d\t\tnumber\n",num);//printf("%d\t\tnumber\n",num);}else{if(ch==':'){//检测赋值符号，:只能和=匹配，否则不能识别getchdo;if(ch=='='){sym=becomes;fprintf(cifa,":=\t\tbecomes\n");getchdo;}else{sym=nul;}}else{if(ch=='<'){getchdo;if(ch=='='){sym=leq;//小于等于fprintf(cifa,"<=\t\tleq\n");getchdo;}else{sym=lss;//小于fprintf(cifa,"<\t\tlss\n");}}else{if(ch=='>'){getchdo;if(ch=='='){sym=geq;//大于等于fprintf(cifa,">=\t\tgeq\n");getchdo;}else{sym=gtr;//大于fprintf(cifa,">\t\tgtr\n");}}else{sym=ssym[ch];//不满足上述条件时按单字//符处理switch(ch){case'+':fprintf(cifa,"%c\t\tplus\n",ch);break;case '-':fprintf(cifa,"%c\t\tminus\n",ch );break;case '*':fprintf(cifa,"%c\t\ttimes\n",ch);break;case '/':fprintf(cifa,"%c\t\tslash\n",ch);break;case '(':fprintf(cifa,"%c\t\tlparen\n",ch);break;case ')':fprintf(cifa,"%c\t\trparen\n",ch);break;case '=':fprintf(cifa,"%c\t\teql\n",ch);break;case ',':fprintf(cifa,"%c\t\tcomma\n",ch);break;case '#':fprintf(cifa,"%c\t\tneq\n",ch);break;case ';':fprintf(cifa,"%c\t\tsemicolon\n",ch);break;case '.':break;default :error(26);}if(sym!=period){//判断是否结束getchdo;}else{printf("\n");fprintf(cifa,".\t\tperiod\n");}}}}}}return 0;}//生成目标代码//目标代码的功能码，层差和位移量int gen(enum fct x,int y,int z){if(cx>=cxmax){//如果目标代码索引过大，报错printf("Program too long");return -1;}code[cx].f=x;code[cx].l=y;code[cx].a=z;cx++;return 0;}//测试字符串int test(bool* s1,bool* s2,int n){if(!inset(sym,s1)){//测试sym是否属于s1，不属于则报错nerror(n);while((!inset(sym,s1))&&(!inset(sym,s2))){//检测不通过时，不停获得符号，直到它属于需要或补救的集合getsymdo;}}return 0;}//编译程序主体//lev：当前分程序所在层，tx:名字表当前尾指针fsys：当前模块后跟符号集合int block(int lev,int tx,bool* fsys){int i;int dx;//名字分配到的相对地址int tx0;//保留初始txint cx0;//保留初始cxbool nxtlev[symnum];dx=3;//相对地址从3开始，前3个单元即0、1、2单元分别为 SL:静态链；//DL：动态链；RA：返回地址tx0=tx;//记录本层的初始位置table[tx].adr=cx;gendo(jmp,0,0);if(lev>levmax){//层数超过3error(32);}do{if(sym==constsym){//收到常量声明printf("该语句为常量定义语句\n");getsymdo;do{constdeclarationdo(&tx,lev,&dx);//常量声明处理，dx 会改//变所以使用指针while(sym==comma){//处理一次多常量定义getsymdo;constdeclarationdo(&tx,lev,&dx);}if(sym==semicolon){//常量声明处理结束getsymdo;}else{error(5);//漏掉了逗号或者分号（一般是分号）}}while(sym==ident);}if(sym==varsym){//收到变量声明printf("该语句为变量声明语句\n");getsymdo;do{vardeclarationdo(&tx,lev,&dx);//变量声明处理while(sym==comma){//处理一次多变量定义getsymdo;vardeclarationdo(&tx,lev,&dx);}if(sym==semicolon){//变量声明处理结束getsymdo;}else{error(5);//漏掉逗号或者分号}}while(sym==ident);}while(sym==procsym){//收到过程声明printf("该语句为过程声明语句\n");getsymdo;if(sym==ident){enter(procedur,&tx,lev,&dx);//记录过程名getsymdo;}else{error(4);//过程声明后应为标识符}if(sym==semicolon){getsymdo;}else{error(5);//漏掉了分号}memcpy(nxtlev,fsys,sizeof(bool)*symnum);nxtlev[semicolon]=true;if(-1==block(lev+1,tx,nxtlev)){return -1;}if(sym==semicolon){getsymdo;memcpy(nxtlev,statbegsys,sizeof(bool)*symnum);nxtlev[ident]=true;nxtlev[procsym]=true;testdo(nxtlev,fsys,6);}else{error(5);}}memcpy(nxtlev,statbegsys,sizeof(bool)*symnum);nxtlev[ident]=true;nxtlev[period]=true;testdo(nxtlev,declbegsys,7);}while(inset(sym,declbegsys));//直到没有声明符号code[table[tx0].adr].a=cx;//开始生成当前过程代码table[tx0].adr=cx;//当前过程代码地址table[tx0].size=dx;cx0=cx;gendo(inte,0,dx);//生成分配存代码if(tableswitch){//输出符号表if(tx0+1<tx){fprintf(fas,"TABLE:\n");//printf("NULL\n");}for(i=tx0+1;i<=tx;i++){switch(table[i].kind){case constant:fprintf(fas,"%d const %s ",i,table[i].name);fprintf(fas,"val=%d\n",table[i].val);break;case variable:fprintf(fas,"%d var %s ",i,table[i].name);fprintf(fas,"lev=%daddr=%d\n",table[i].level,table[i].adr);break;case procedur:fprintf(fas,"%d proc %s ",i,table[i].name);fprintf(fas,"lev=%daddr=%dsize=%d\n", table[i].level,table[i].adr,table[i].size);break;}}}memcpy(nxtlev,fsys,sizeof(bool)*symnum);//每个后跟符号集和都包含上层后跟符//号集和，以便补救nxtlev[semicolon]=true;nxtlev[endsym]=true;statementdo(nxtlev,&tx,lev);gendo(opr,0,0);//每个过程出口都要使用的释放数据指令memset(nxtlev,0,sizeof(bool)*symnum);//分程序没有补救集合testdo(fsys,nxtlev,8);//检测后跟符号集的正确性listcode(cx0);//输出代码return 0;}//k:const var procedure//ptx:符号表尾的指针//pdx:dx为当前应分配变量的相对地址//lev:符号名字所在的层次//往符号表中添加void enter(enum object k,int* ptx,int lev,int* pdx){(* ptx)++;strcpy(table[(*ptx)].name,id);table[(* ptx)].kind=k;switch(k){case constant:if(num>amax){error(31);num=0;}table[(*ptx)].val=num;break;case variable:table[(*ptx)].level=lev;table[(*ptx)].adr=(*pdx);(*pdx)++;break;case procedur:table[(*ptx)].level=lev;break;}}//寻找符号在符号表中的地址int position(char*idt,int tx){int i;strcpy(table[0].name,idt);i=tx;//符号表尾while(strcmp(table[i].name,idt)!=0){i--;}return i;}//常量声明处理int constdeclaration(int* ptx,int lev,int* pdx){if(sym==ident){getsymdo;if(sym==eql||sym==becomes){if(sym==becomes){error(1);}getsymdo;if(sym==number){enter(constant,ptx,lev,pdx);getsymdo;}else{error(2);}}else{error(3);}}else{error(4);}return 0;}//变量声明处理int vardeclaration (int* ptx,int lev,int* pdx){if(sym==ident){enter(variable,ptx,lev,pdx);getsymdo;}else{error(4);}return 0;}//输出目标代码清单void listcode(int cx0){int i;if(listswitch){for(i=cx0;i<cx;i++){fprintf(fa,"%d %s %d %d\n",i,mnemonic[code[i].f],code[i].l,code[i],a);}}}//语句处理int statement(bool* fsys,int* ptx,int lev){int i,cx1,cx2;bool nxtlev[symnum];if(sym==ident){i=position(id,*ptx);if(i==0){error(11);}else{if(table[i].kind!=variable){error(12);i=0;}else{getsymdo;if(sym==becomes){getsymdo;printf("该语句为赋值语句。