C 语言编写的简单词法分析器 reference.c

#include <stdio.h>
#include <string.h>
#define MAX 500 
#ifdef __unix
#define fopen_s(pFile,filename,mode) ((*(pFile))=fopen((filename), (mode)))==NULL
#endif
int main()
{
FILE *in, *out;
char word[MAX]; 
char cp; 
int i;
if (( fopen_s(&in,"in.txt", "r")) != 0)//in=fopen("in.txt","r"))==NULL
{
printf("can not open in.txt! \n");
return 1;
}
else
{
printf("success open in.txt!\n");
}
if ((fopen_s(&out,"out.txt", "w")) != 0)//out = fopen("out.txt", "r")) == NULL
{
printf("Can not open out.txt!\n");
return 0;
}
else
{
printf("Successfully opened out.txt\n");
}
cp = fgetc(in);
while (cp != EOF)
{

while (cp == ' ' || cp == '\t' || cp == '\n')
{
cp = fgetc(in);
}

i = 0;

if (cp >= '0'&&cp <= '9')
{
word[i++] = cp;
cp = fgetc(in);
while (cp >= '0'&&cp <= '9')
{
word[i++] = cp;
cp = fgetc(in);
}
if (cp == ' '||cp=='\t'||cp=='\n'||cp==' = '||cp==' + '||cp==' - '||cp=='*'||cp==' / '||cp=='>'||cp=='<'||cp=='!'||cp=='; '||cp==':'||cp==', '||cp=='{'||cp=='}'||cp=='['||cp==']'||cp=='('||cp==')')
{
word[i] = '\0';
fprintf(out, "%c%c%s%c,%d%c\n", '(','"',word,'"',28,')');
}
else
{
while (cp != ' '&&cp!='\t'&&cp!='\n'&&cp!=' = '&&cp!=' + '&&cp!=' - '&&cp!='*'&&cp!=' / '&&cp!='>'&&cp!='<'&&cp!='!'&&cp!= ';'&&cp != ':'&&cp != ','&&cp != '{'&&cp != '}'&&cp != '['&&cp != ']'&&cp != '('&&cp != ')')
{
word[i++] = cp;
cp = fgetc(in);
}
word[i] = '\0';
fprintf(out, "%s%c%s%c%s\n", "error: ", '"', word, '"', " some thing wrong ");
}
continue;
}

if ((cp >= 'a'&&cp <= 'z') || (cp >= 'A'&&cp <= 'Z') || cp == '_')
{
while ((cp >= 'a'&&cp <= 'z') || (cp >= '0'&&cp <= '9') || (cp >= 'A'&&cp <= 'Z') || cp == '_')
{
word[i++] = cp;
cp = fgetc(in);
}
if (cp == ' '||cp=='\t'||cp=='\n'||cp==' = '||cp==' + '||cp==' - '||cp=='*'||cp==' / '||cp=='>'||cp=='<'||cp=='!'||cp=='; '||cp==':'||cp==', '||cp=='{'||cp=='}'||cp=='['||cp==']'||cp=='('||cp==')')
{
word[i] = '\0';
/* 鍩烘湰淇濈暀瀛楁娴?*/
if (strcmp(word, "main") == 0)
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 1, ')');
else if (strcmp(word, "void") == 0)
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 2, ')');
else if (strcmp(word, "int") == 0)
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 3, ')');
else if (strcmp(word, "float") == 0)
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 4, ')');
else if (strcmp(word, "double") == 0)
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 5, ')');
else if (strcmp(word, "char") == 0)
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 6, ')');
else if (strcmp(word, "struct") == 0)
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 7, ')');
else if (strcmp(word, "const") == 0)
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 8, ')');
else if (strcmp(word, "extern") == 0)
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 9, ')');
else if (strcmp(word, "register") == 0)
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 10, ')');
else if (strcmp(word, "static") == 0)
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 11, ')');
else if (strcmp(word, "if") == 0)
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 12, ')');
else if (strcmp(word, "else") == 0)
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 13, ')');
else if (strcmp(word, "switch") == 0)
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 14, ')');
else if(strcmp(word, "case") == 0)
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 15, ')');
else if (strcmp(word, "for") == 0)
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 16, ')');
else if (strcmp(word, "do") == 0)
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 17, ')');
else if (strcmp(word, "while") == 0)
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 18, ')');
else if (strcmp(word, "goto") == 0)
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 19, ')');
else if (strcmp(word, "continue") == 0)
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 20, ')');
else if (strcmp(word, "break") == 0)
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 21, ')');
else if (strcmp(word, "default") == 0)
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 22, ')');
else if (strcmp(word, "sizeof") == 0)
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 23, ')');
else if (strcmp(word, "return") == 0)
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 24, ')');
else
fprintf(out, " Custom identifier :%c%c%s%c,%d%c\n", '(', '"', word, '"', 29, ')');
}

else
{
while (cp != ' '&&cp!='\t'&&cp!='\n'&&cp!=' = '&&cp!=' + '&&cp!=' - '&&cp!='*'&&cp!=' / '&&cp!='>'&&cp!='<'&&cp!='!'&&cp!= ';'&&cp != ':'&&cp != ','&&cp != '{'&&cp != '}'&&cp != '['&&cp != ']'&&cp != '('&&cp != ')')
{
word[i++] = cp;
cp = fgetc(in);
}
word[i] = '\0';
fprintf(out, "%s%c%s%c%s\n", "error: ", '"', word, '"', " is not a legitimate identifier ");
}
continue;
}
/* 杩愮畻绗?*/
if (cp == '+')
{
word[i++] = cp;
word[i] = '\0';
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 31, ')');
cp = fgetc(in);
continue;
}
if (cp == '-')
{
word[i++] = cp;
word[i] = '\0';
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 32, ')');
cp = fgetc(in);
continue;
}
if (cp == '*')
{
word[i++] = cp;
word[i] = '\0';
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 33, ')');
cp = fgetc(in);
continue;
}
if (cp == '/')
{
word[i++] = cp;
cp = fgetc(in);
if (cp == '*') { word[--i] = '\0'; while (cp != '/') { cp = fgetc(in); }cp = fgetc(in); }
else
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 34, ')');
word[i] = '\0';
continue;
}
if (cp == '=')
{
word[i++] = cp;
cp = fgetc(in);
if (cp == '=')
{
word[i++] = cp;
word[i] = '\0';
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 39, ')');
cp = fgetc(in);
}
else
{
word[i] = '\0';
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 27, ')');
}
continue;
}
if (cp == '<')
{
word[i++] = cp;
cp = fgetc(in);
if (cp == '=')
{
word[i++] = cp;
word[i] = '\0';
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 36, ')');
cp = fgetc(in);
}
else
{
word[i] = '\0';
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 35, ')');
}
continue;
}
if (cp == '>')
{
word[i++] = cp;
cp = fgetc(in);
if (cp == '=')
{
word[i++] = cp;
word[i] = '\0';
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 38, ')');
cp = fgetc(in);
}
else
{
word[i] = '\0';
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 37, ')');
}
continue;
}
if (cp == '!')
{
word[i++] = cp;
cp = fgetc(in);
if (cp == '=')
{
word[i++] = cp;
word[i] = '\0';
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 40, ')');
cp = fgetc(in);
}
else
{
word[i] = '\0';
fprintf(out, "%s%c%s%c%s\n", "error: ", '"', word, '"', " is not a legitimate identifier ");
}
continue;
}
/* 鐣岀 */
if (cp == ';')
{
word[i++] = cp;
word[i] = '\0';
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 41, ')');
cp = fgetc(in);
continue;
}
if (cp == ':')
{
word[i++] = cp;
word[i] = '\0';
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 42, ')');
cp = fgetc(in);
continue;
}
if (cp == ',')
{
word[i++] = cp;
word[i] = '\0';
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 43, ')');
cp = fgetc(in);
continue;
}
if (cp == '{')
{
word[i++] = cp;
word[i] = '\0';
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 44, ')');
cp = fgetc(in);
continue;
}
if (cp == '}')
{
word[i++] = cp;
word[i] = '\0';
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 45, ')');
cp = fgetc(in);
continue;
}
if (cp == '[')
{
word[i++] = cp;
word[i] = '\0';
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 46, ')');
cp = fgetc(in);
continue;
}
if (cp == ']')
{
word[i++] = cp;
word[i] = '\0';
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 47, ')');
cp = fgetc(in);
continue;
}
if (cp == '(')
{
word[i++] = cp;
word[i] = '\0';
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 48, ')');
cp = fgetc(in);
continue;
}
if (cp == ')')
{
word[i++] = cp;
word[i] = '\0';
fprintf(out, "%c%c%s%c,%d%c\n", '(', '"', word, '"', 49, ')');
cp = fgetc(in);
continue;
}

word[i++] = cp;
cp = fgetc(in);
word[i] = '\0';
fprintf(out, "%s%c%s%c%s\n", "error: ", '"', word, '"', "not a id");
}
fclose(in);
fclose(out);
printf("Success! Please see out.txt\n");
return 0;
}

本栏目推荐文章
• 【C语言】
• 【C语言】形参和实参的一些注意点
• Go 语言为什么不支持并发读写 map？
• 【C语言】函数的递归调用
• 字节微服务HTTP框架Hertz使用与源码分析｜拥抱开
• 和利时MACS6通过dat文件实现任意时间范围与测点的历史数据提取与科研分析
• Scala编程语言day1
• NUS CS1101S：SICP JavaScript 描述：四、元语言抽象
• 学了这么多编程语言，你学会了几种“Hello world”呢？
• 各种语言版本的“Hello, world”程序汇总

词法 分析器 reference 语言词法 分析器reference语言 词法 分析器 语言lut 词法 分析器 原理lex 词法 分析器 一手 原理 词法 分析器 原理 代码 词法 语法 语言 程序 词法 自然语言 自然 语言 词法 自然语言 句法 语义 unexported语言cannot refer 分析器 语法 语言 程序