12
|
1 #include <stdio.h>
|
|
2 #include <stdlib.h>
|
|
3 #include <sys/mman.h>
|
|
4 #include <sys/types.h>
|
|
5 #include <sys/stat.h>
|
|
6 #include <fcntl.h>
|
|
7 #include <unistd.h>
|
|
8 #include <string.h>
|
|
9
|
|
10 typedef unsigned char UCHAR;
|
|
11 typedef unsigned char* UCHARP;
|
|
12
|
|
13 typedef enum BOOL {
|
|
14 FALSE = 0,
|
|
15 TRUE = 1
|
|
16 } BOOL;
|
|
17
|
|
18 typedef struct ENV {
|
|
19 BOOL print_count;
|
|
20 BOOL print_filename;
|
|
21 int count;
|
|
22 char *filename;
|
|
23 } *ENVP;
|
|
24
|
|
25 void filter(UCHARP beg, UCHARP buf, UCHARP end, ENVP env);
|
|
26 void predict(UCHARP beg, UCHARP buf, UCHARP end, ENVP env);
|
|
27 void reject(UCHARP beg, UCHARP buf, UCHARP end, ENVP env);
|
|
28 void accept(UCHARP beg, UCHARP buf, UCHARP end, ENVP env);
|
|
29 void s0(UCHARP beg, UCHARP buf, UCHARP end, ENVP env);
|
|
30 #define __REGEX__ "h.s"
|
|
31
|
|
32 UCHARP get_line_beg(UCHARP p, UCHARP beg) {
|
|
33 while(p > beg) {
|
|
34 if ((*--p) == '\n') return p+1;
|
|
35 }
|
|
36 return beg;
|
|
37 }
|
|
38
|
|
39 void print_line(UCHARP beg, UCHARP end) {
|
|
40 fwrite(beg, sizeof(char), (end - beg + 1), stdout);
|
|
41 }
|
|
42
|
|
43 void grep(int fd, ENVP env) {
|
|
44 caddr_t file_mmap;
|
|
45 UCHARP buf, end, beg;
|
|
46 off_t size;
|
|
47 struct stat sb;
|
|
48
|
|
49 if (fstat(fd, &sb)) {
|
|
50 fprintf(stderr, "can't fstat %s\n", env->filename);
|
|
51 exit(0);
|
|
52 }
|
|
53
|
|
54 size = sb.st_size;
|
|
55 file_mmap = mmap(NULL, size, PROT_READ, MAP_SHARED, fd, (off_t)0);
|
|
56
|
|
57 if (file_mmap == (caddr_t)-1) {
|
|
58 fprintf(stderr, "can't mmap %s\n", env->filename);
|
|
59 exit(0);
|
|
60 }
|
|
61
|
|
62 beg = buf = (UCHARP) file_mmap;
|
|
63 end = beg + size - 1;
|
|
64 env->count = 0;
|
|
65
|
|
66 s0(beg, beg, end, env);
|
|
67
|
|
68 munmap(file_mmap, size);
|
|
69 return;
|
|
70 }
|
|
71
|
|
72 void stdingrep(ENVP env) {
|
|
73 char buf_[BUFSIZ];
|
|
74 UCHARP buf, beg, end;
|
|
75 buf = beg = (UCHARP) buf_;
|
|
76
|
|
77 while (fgets(buf_, BUFSIZ, stdin) != NULL) {
|
|
78 env->count = 0;
|
|
79 end = buf + strlen(buf_) - 1;
|
|
80 s0(beg, buf, end, env);
|
|
81 }
|
|
82 }
|
|
83
|
|
84 int main(int argc, char* argv[]) {
|
|
85 int i, fd, opt, with_filename = 0;
|
|
86
|
|
87 ENVP env = (ENVP)calloc(sizeof(ENVP), 1);
|
|
88
|
|
89 while ((opt=getopt(argc, argv, "chH")) != -1) {
|
|
90 switch (opt) {
|
|
91 case 'c':
|
|
92 env->print_count = 1;
|
|
93 break;
|
|
94 case 'h':
|
|
95 with_filename = -1;
|
|
96 break;
|
|
97 case 'H':
|
|
98 with_filename = 1;
|
|
99 break;
|
|
100 default:
|
|
101 fprintf(stderr, "invalid option.");
|
|
102 exit(0);
|
|
103 }
|
|
104 }
|
|
105
|
|
106 if (optind == argc) {
|
|
107 if (with_filename) {
|
|
108 /* handle -h/-H as help */
|
|
109 printf("grep pattern: %s\n", __REGEX__);
|
|
110 exit(1);
|
|
111 } else {
|
|
112 stdingrep(env);
|
|
113 }
|
|
114 } else {
|
|
115 switch (with_filename) {
|
|
116 case 0: //default
|
|
117 if (argc == optind + 1) {
|
|
118 /* just grep 1 file. */
|
|
119 env->print_filename = FALSE;
|
|
120 } else {
|
|
121 /* grep 1 more files. */
|
|
122 env->print_filename = TRUE;
|
|
123 }
|
|
124 break;
|
|
125 // option
|
|
126 case -1:
|
|
127 env->print_filename = FALSE;
|
|
128 break;
|
|
129 case 1:
|
|
130 env->print_filename = TRUE;
|
|
131 break;
|
|
132 default: break;
|
|
133 }
|
|
134 for (i = optind; i < argc; i++) {
|
|
135 fd = open(argv[i], O_RDONLY, 0666);
|
|
136 if (fd == 0) {
|
|
137 printf("%d, %s\n", i, argv[i]);
|
|
138 fprintf(stderr, "can't open %s:", argv[i]);
|
|
139 continue;
|
|
140 }
|
|
141 env->filename = argv[i];
|
|
142 grep(fd, env);
|
|
143 close(fd);
|
|
144 }
|
|
145 }
|
|
146
|
|
147 return 0;
|
|
148 }
|
|
149 void s1(UCHARP beg, UCHARP buf, UCHARP end, ENVP env);
|
|
150 void s2(UCHARP beg, UCHARP buf, UCHARP end, ENVP env);
|
|
151 void s3(UCHARP beg, UCHARP buf, UCHARP end, ENVP env);
|
|
152 void s4(UCHARP beg, UCHARP buf, UCHARP end, ENVP env);
|
|
153 void s5(UCHARP beg, UCHARP buf, UCHARP end, ENVP env);
|
|
154 void s6(UCHARP beg, UCHARP buf, UCHARP end, ENVP env);
|
|
155
|
|
156 void s0(UCHARP beg, UCHARP buf, UCHARP end, ENVP env) {
|
|
157 return filter(beg, buf, end, env);
|
|
158 }
|
|
159
|
|
160 void filter(UCHARP beg, UCHARP buf, UCHARP end, ENVP env) {
|
|
161 static const UCHARP const key = (UCHARP)"h";
|
|
162 int i, len = 1;
|
|
163 loop:
|
|
164 while ((buf = memchr(buf, key[0], end-buf))) {
|
|
165 for (i = 1; i < len; i++) {
|
|
166 if (buf[i] != key[i]) {
|
|
167 buf++;
|
|
168 goto loop;
|
|
169 }
|
|
170 }
|
|
171 goto next;
|
|
172 }
|
|
173 return reject(beg, end, end, env);
|
|
174 next:
|
|
175 return s1(beg, buf, end, env);
|
|
176 }
|
|
177
|
|
178 void s1(UCHARP beg, UCHARP buf, UCHARP end, ENVP env) {
|
|
179 static void (*tbl[256])(UCHARP, UCHARP, UCHARP, ENVP) = {[0 ... 255] = (void*)s1, /* NL */ [10] = (void*)reject, /* h */ [104] = (void*)s2};
|
|
180 if (buf > end) return tbl['\n'](beg, buf, end, env);
|
|
181 return tbl[*buf++](beg, buf, end, env);
|
|
182 }
|
|
183
|
|
184 void s2(UCHARP beg, UCHARP buf, UCHARP end, ENVP env) {
|
|
185 static void (*tbl[256])(UCHARP, UCHARP, UCHARP, ENVP) = {[0 ... 255] = (void*)s3, /* NL */ [10] = (void*)reject, /* h */ [104] = (void*)s4};
|
|
186 if (buf > end) return tbl['\n'](beg, buf, end, env);
|
|
187 return tbl[*buf++](beg, buf, end, env);
|
|
188 }
|
|
189
|
|
190 void s3(UCHARP beg, UCHARP buf, UCHARP end, ENVP env) {
|
|
191 static void (*tbl[256])(UCHARP, UCHARP, UCHARP, ENVP) = {[0 ... 255] = (void*)s1, /* NL */ [10] = (void*)reject, /* h */ [104] = (void*)s2, /* s */ [115] = (void*)s5};
|
|
192 if (buf > end) return tbl['\n'](beg, buf, end, env);
|
|
193 return tbl[*buf++](beg, buf, end, env);
|
|
194 }
|
|
195
|
|
196 void s4(UCHARP beg, UCHARP buf, UCHARP end, ENVP env) {
|
|
197 static void (*tbl[256])(UCHARP, UCHARP, UCHARP, ENVP) = {[0 ... 255] = (void*)s3, /* NL */ [10] = (void*)reject, /* h */ [104] = (void*)s4, /* s */ [115] = (void*)s6};
|
|
198 if (buf > end) return tbl['\n'](beg, buf, end, env);
|
|
199 return tbl[*buf++](beg, buf, end, env);
|
|
200 }
|
|
201
|
|
202 void s5(UCHARP beg, UCHARP buf, UCHARP end, ENVP env) {
|
|
203 return accept(beg, buf - 1, end, env);}
|
|
204
|
|
205 void s6(UCHARP beg, UCHARP buf, UCHARP end, ENVP env) {
|
|
206 return accept(beg, buf - 1, end, env);}
|
|
207
|
|
208 void
|
|
209 accept(UCHARP beg, UCHARP buf, UCHARP end, ENVP env)
|
|
210 {
|
|
211 if (buf < beg) buf = beg;
|
|
212 UCHARP ret = (buf < end) ? (UCHARP) memchr(buf, '\n', (end - buf)) : NULL;
|
|
213 if (env->print_count) {
|
|
214 env->count++;
|
|
215 if (ret == NULL) {
|
|
216 if (env->print_filename) {
|
|
217 printf("%s:", env->filename);
|
|
218 }
|
|
219 printf("%d\n", env->count);
|
|
220 return;
|
|
221 }
|
|
222 beg = buf = ret + 1;
|
|
223 } else {
|
|
224 beg = get_line_beg(buf, beg);
|
|
225 if (env->print_filename) {
|
|
226 printf("%s:", env->filename);
|
|
227 }
|
|
228 if (ret == NULL) {
|
|
229 print_line(beg, end);
|
|
230 return;
|
|
231 }
|
|
232 print_line(beg, ret);
|
|
233 beg = buf = ret + 1;
|
|
234 }
|
|
235 return s0(beg, buf, end, env);
|
|
236 }
|
|
237
|
|
238 void
|
|
239 reject(UCHARP beg, UCHARP buf, UCHARP end, ENVP env) {
|
|
240 UCHARP ret;
|
|
241 if (buf >= end) {
|
|
242 ret = NULL;
|
|
243 } else {
|
|
244 if (*(buf-1) == '\n') {
|
|
245 ret = buf-1;
|
|
246 } else {
|
|
247 ret = (UCHARP)memchr(buf, '\n', (end - buf));
|
|
248 }
|
|
249 }
|
|
250 if (ret == NULL) {
|
|
251 if (env->print_count) {
|
|
252 if (env->print_filename) {
|
|
253 printf("%s:", env->filename);
|
|
254 }
|
|
255 printf("%d\n", env->count);
|
|
256 }
|
|
257 return;
|
|
258 }
|
|
259 beg = buf = ret + 1;
|
|
260 return s0(beg, buf, end, env);
|
|
261 }
|