前段时间编译原理的实验..写个简单的词法分析程序.下面是代码
1 #include<iostream>
2 #include<fstream>
3 #include<string>
4 #include<vector>
5 #include<stdlib.h>
6
7 using namespace std;
8
9 string src="";
10 unsigned int begin = 0;
11 int variable = 0; //记录单词种别
12
13 vector<string> rew; // 保留字
14 vector<char> dl; //界符
15 vector<char> op; //操作符
16 vector<string> num; //常数
17 vector<string> ide; //标识符
18
19 //将文件中的保留字导出到rew容器中.
20 void inputReservedWord()
21 {
22 ifstream infile;
23 infile.open("表2-reservedWord.txt");
24 if(!infile)
25 {
26 cout << "Cannot open the named file." << endl;
27 return;
28 }
29 string s="";
30 while(!infile.eof())
31 {
32 infile>>s;
33 rew.push_back(s);
34 }
35 infile.close();
36 }
37
38 //将文件中的标识符导出到ide容器中.
39 void inputIdentifier()
40 {
41 ifstream infile;
42 infile.open("表1-identifier.txt");
43 if(!infile)
44 {
45 cout << "Cannot open the named file." << endl;
46 return;
47 }
48 string s="";
49 while(!infile.eof())
50 {
51 infile>>s;
52 ide.push_back(s);
53 }
54 infile.close();
55 }
56
57 //将文件中的常数导出到num容器中.
58 void inputNumber()
59 {
60 ifstream infile;
61 infile.open("表3-number.txt");
62 if(!infile)
63 {
64 cout << "Cannot open the named file." << endl;
65 return;
66 }
67 string s="";
68 while(!infile.eof())
69 {
70 infile>>s;
71 num.push_back(s);
72 }
73 infile.close();
74 }
75
76 //将文件中的运算符导出到op容器中.
77 void inputOperator()
78 {
79 ifstream infile;
80 infile.open("表4-operator.txt");
81 if(!infile)
82 {
83 cout << "Cannot open the named file." << endl;
84 return;
85 }
86 char c;
87 while(!infile.eof())
88 {
89 infile.get(c);
90 op.push_back(c);
91 }
92 infile.close();
93 }
94
95 //将文件中的界符导出到dl容器中.
96 void inputDelimiter()
97 {
98 ifstream infile;
99 infile.open("表5-delimiter.txt");
100 if(!infile)
101 {
102 cout << "Cannot open the named file." << endl;
103 return;
104 }
105 char c;
106 while(!infile.eof())
107 {
108 infile.get(c);
109 dl.push_back(c);
110 }
111 infile.close();
112 }
113
114 //判断是不是保留字
115 bool isReservedWord(string s)
116 {
117 bool flag = false;
118 for(vector<string>::size_type i = 0;i < rew.size();++i)
119 {
120 if(!s.compare(rew[i]))
121 {
122 variable = 2;
123 flag = true;
124 break;
125 }
126 }
127 return flag;
128 }
129
130 //判断标识符或者保留字的开头字符
131 bool isIdentifierOrReservedWordChar(char c)
132 {
133 if((c>='a' && c<='z') || (c>='A' && c<='Z') || c=='_')
134 return true;
135 else return false;
136 }
137
138 //获得以下标i开头的标识符或者保留字
139 string getIdentifierOrReservedWordChar(int i)
140 {
141 if((src[i]>='a' && src[i]<='z') || (src[i]>='A' && src[i]<='Z') || src[i]=='_')
142 {
143 ++i;
144 }
145 while((src[i]>='a' && src[i]<='z') || (src[i]>='A' && src[i]<='Z') ||
146 (src[i]>='0' && src[i]<='9'))
147 {
148 ++i;
149 }
150 string word=src.substr(begin,(i-begin));
151 begin=i-1;
152 return word;
153 }
154
155 //判断是不是标识符
156 bool isIdentifier(char c)
157 {
158 if((c>='a' && c<='z') || (c>='A' && c<='Z'))
159 {
160 // variable = 1;
161 return true;
162 }
163 else return false;
164 }
165
166 //判断是不是常数
167 bool isNumber(char c)
168 {
169 if(c>='0' && c<='9')
170 {
171 variable = 3;
172 return true;
173 }
174 else return false;
175 }
176
177 //获得以下标i开始的常数，并判断中间是否出现字符
178 string getNumber(int i)
179 {
180 while(src[i]>='0' && src[i]<='9')
181 {
182 ++i;
183 }
184 string word=src.substr(begin,(i-begin));
185 begin = i-1;
186 return word;
187 }
188
189 //判断是不是操作符
190 bool isOperator(char c)
191 {
192 bool flag=false;
193 for(vector<char>::size_type i=0;i<op.size();++i)
194 {
195 if(c==op[i])
196 {
197 variable = 4;
198 flag=true;
199 break;
200 }
201 }
202 return flag;
203 }
204
205 //判断是不是界符
206 bool isDelimiter(char c)
207 {
208 bool flag=false;
209 for(vector<char>::size_type i=0;i<dl.size();++i)
210 {
211 if(c == dl[i])
212 {
213 variable = 5;
214 flag = true;
215 break;
216 }
217 }
218 return flag;
219 }
220
221 void main()
222 {
223 ifstream infile;
224 ofstream outfile;
225 ofstream result;
226 result.open("result.txt");
227 if(!result)
228 {
229 cout<<"Cannot open the named file."<<endl;
230 }
231 inputReservedWord();
232 inputIdentifier();
233 inputNumber();
234 inputOperator();
235 inputDelimiter();
236 infile.open("test.txt");
237 if(!infile)
238 {
239 cout << "Cannot open the named file." << endl;
240 return;
241 }
242 char ch = ' ';
243 string s;
244 char csrc;
245 while (!infile.eof()) //得到源程序文本
246 {
247 infile.get(csrc);
248 s+= csrc;
249 csrc = ' ';
250 }
251 src = s;
252 infile.close();
253 cout<<src<<endl;
254
255 ch = ' ';
256 while(begin<src.length())
257 {
258 while(src[begin]==' ' || src[begin]=='\\n') //消除空格和回车
259 {
260 ++begin;
261 }
262 ch = src[begin];
263 if(isIdentifierOrReservedWordChar(ch))
264 {
265 string s = getIdentifierOrReservedWordChar(begin);
266 if(isReservedWord(s)) //s为保留字
267 {
268 vector<string>::size_type i= 0;
269 while(s!=rew[i] && i<rew.size())
270 {
271 i++;
272 }
273 cout<<s<<" "<<variable<<" "<<i<<endl;
274 result<<s<<" "<<variable<<" "<<i<<endl;
275 }
276 else //s为标识符
277 {
278 variable = 1;
279 vector<string>::size_type i=1;
280 while(s!=ide[i-1] && i<ide.size())
281 {
282 ++i;
283 }
284 if(i>=ide.size() || ide.empty())
285 {
286 outfile.open("表1-identifier.txt",ios::app);
287 outfile.write(s.c_str(),s.length());
288 outfile<<endl;
289 ide.push_back(s);
290 i=ide.size();
291 outfile.close();
292 }
293 cout<<s<<" "<<variable<<" "<<i<<endl;
294 result<<s<<" "<<variable<<" "<<i<<endl;
295 }
296 }
297 else if(isNumber(ch))
298 {
299 vector<string>::size_type i = 1;
300 string s = getNumber(begin);
301 while(s!=num[i-1] && i<num.size())
302 {
303 ++i;
304 }
305 if(i>=num.size() || num.empty())
306 {
307 outfile.open("表3-number.txt",ios::app);
308 outfile.write(s.c_str(),s.length());
309 outfile<<endl;
310 ide.push_back(s);
311 i=ide.size();
312 outfile.close();
313 }
314 cout<<s<<" "<<variable<<" "<<i<<endl;
315 result<<s<<" "<<variable<<" "<<i<<endl;
316 }
317 else if(isOperator(ch))
318 {
319 int i = 0;
320 while(ch!=op[i])
321 {
322 i++;
323 }
324 cout<<ch<<" "<<variable<<" "<<i<<endl;
325 result<<ch<<" "<<variable<<" "<<i<<endl;
326 }
327 else if(isDelimiter(ch))
328 {
329 int i = 0;
330 while(ch!=dl[i])
331 {
332 i++;
333 }
334 cout<<ch<<" "<<variable<<" "<<i<<endl;
335 result<<ch<<" "<<variable<<" "<<i<<endl;
336 }
337 else
338 {
339 cout<<"Unkown char!"<<endl;
340 }
341 ++begin;
342 }
343 }
需包含文件 表2-reservedWord.txt(存放保留字表) ,表1-identifier.txt(存放标识符表),表3-number.txt(存放常量表),表4-operator.txt(存放操作符表),表5-delimiter.txt(存放界符表),test.txt(存放测试代码),result.txt(存放分析结果)
代码缺点太多..欢迎大家大力批评..
在VC++6.0上运行会有4个warning.不影响结果.