1 |
/* |
2 |
* Cisco router simulation platform. |
3 |
* Copyright (c) 2006 Christophe Fillot (cf@utc.fr) |
4 |
* |
5 |
* Mini-parser. |
6 |
*/ |
7 |
|
8 |
#include <stdio.h> |
9 |
#include <stdlib.h> |
10 |
#include <unistd.h> |
11 |
#include <string.h> |
12 |
#include <sys/types.h> |
13 |
#include <sys/stat.h> |
14 |
#include <sys/mman.h> |
15 |
#include <signal.h> |
16 |
#include <fcntl.h> |
17 |
#include <errno.h> |
18 |
#include <assert.h> |
19 |
#include <stdarg.h> |
20 |
#include <sys/ioctl.h> |
21 |
#include <sys/types.h> |
22 |
#include <sys/socket.h> |
23 |
#include <arpa/inet.h> |
24 |
|
25 |
#include "utils.h" |
26 |
#include "parser.h" |
27 |
|
28 |
#define TOKEN_MAX_SIZE 512 |
29 |
|
30 |
/* Character types */ |
31 |
enum { |
32 |
PARSER_CHAR_BLANK, |
33 |
PARSER_CHAR_NEWLINE, |
34 |
PARSER_CHAR_COMMENT, |
35 |
PARSER_CHAR_QUOTE, |
36 |
PARSER_CHAR_OTHER, |
37 |
}; |
38 |
|
39 |
/* Get a description given an error code */ |
40 |
char *parser_strerror(parser_context_t *ctx) |
41 |
{ |
42 |
printf("error = %d\n",ctx->error); |
43 |
|
44 |
switch(ctx->error) { |
45 |
case 0: |
46 |
return "no error"; |
47 |
case PARSER_ERROR_NOMEM: |
48 |
return "insufficient memory"; |
49 |
case PARSER_ERROR_UNEXP_QUOTE: |
50 |
return "unexpected quote"; |
51 |
case PARSER_ERROR_UNEXP_EOL: |
52 |
return "unexpected end of line"; |
53 |
default: |
54 |
return "unknown error"; |
55 |
} |
56 |
} |
57 |
|
58 |
/* Dump a token list */ |
59 |
void parser_dump_tokens(parser_context_t *ctx) |
60 |
{ |
61 |
parser_token_t *tok; |
62 |
|
63 |
for(tok=ctx->tok_head;tok;tok=tok->next) |
64 |
printf("\"%s\" ",tok->value); |
65 |
} |
66 |
|
67 |
/* Map a token list to an array */ |
68 |
char **parser_map_array(parser_context_t *ctx) |
69 |
{ |
70 |
parser_token_t *tok; |
71 |
char **map; |
72 |
int i; |
73 |
|
74 |
if (ctx->tok_count <= 0) |
75 |
return NULL; |
76 |
|
77 |
if (!(map = calloc(ctx->tok_count,sizeof(char **)))) |
78 |
return NULL; |
79 |
|
80 |
for(i=0,tok=ctx->tok_head;(i<ctx->tok_count) && tok;i++,tok=tok->next) |
81 |
map[i] = tok->value; |
82 |
|
83 |
return map; |
84 |
} |
85 |
|
86 |
/* Add a character to temporary token (resize if necessary) */ |
87 |
static int tmp_token_add_char(parser_context_t *ctx,u_char c) |
88 |
{ |
89 |
size_t new_size; |
90 |
char *new_str; |
91 |
|
92 |
if (!ctx->tmp_tok || (ctx->tmp_cur_len == (ctx->tmp_tot_len - 1))) { |
93 |
new_size = ctx->tmp_tot_len + TOKEN_MAX_SIZE; |
94 |
new_str = realloc(ctx->tmp_tok,new_size); |
95 |
|
96 |
if (!new_str) |
97 |
return(-1); |
98 |
|
99 |
ctx->tmp_tok = new_str; |
100 |
ctx->tmp_tot_len = new_size; |
101 |
} |
102 |
|
103 |
ctx->tmp_tok[ctx->tmp_cur_len++] = c; |
104 |
ctx->tmp_tok[ctx->tmp_cur_len] = 0; |
105 |
return(0); |
106 |
} |
107 |
|
108 |
/* Move current token to the active token list */ |
109 |
static int parser_move_tmp_token(parser_context_t *ctx) |
110 |
{ |
111 |
parser_token_t *tok; |
112 |
|
113 |
/* no token ... */ |
114 |
if (!ctx->tmp_tok) |
115 |
return(0); |
116 |
|
117 |
if (!(tok = malloc(sizeof(*tok)))) |
118 |
return(-1); |
119 |
|
120 |
tok->value = ctx->tmp_tok; |
121 |
tok->next = NULL; |
122 |
|
123 |
/* add it to the token list */ |
124 |
if (ctx->tok_last != NULL) |
125 |
ctx->tok_last->next = tok; |
126 |
else |
127 |
ctx->tok_head = tok; |
128 |
|
129 |
ctx->tok_last = tok; |
130 |
ctx->tok_count++; |
131 |
|
132 |
/* start a new token */ |
133 |
ctx->tmp_tok = NULL; |
134 |
ctx->tmp_tot_len = ctx->tmp_cur_len = 0; |
135 |
return(0); |
136 |
} |
137 |
|
138 |
/* Initialize parser context */ |
139 |
void parser_context_init(parser_context_t *ctx) |
140 |
{ |
141 |
ctx->tok_head = ctx->tok_last = NULL; |
142 |
ctx->tok_count = 0; |
143 |
|
144 |
ctx->tmp_tok = NULL; |
145 |
ctx->tmp_tot_len = ctx->tmp_cur_len = 0; |
146 |
|
147 |
ctx->state = PARSER_STATE_BLANK; |
148 |
ctx->error = 0; |
149 |
|
150 |
ctx->consumed_len = 0; |
151 |
} |
152 |
|
153 |
/* Free a token list */ |
154 |
void parser_free_tokens(parser_token_t *tok_list) |
155 |
{ |
156 |
parser_token_t *t,*next; |
157 |
|
158 |
for(t=tok_list;t;t=next) { |
159 |
next = t->next; |
160 |
free(t->value); |
161 |
free(t); |
162 |
} |
163 |
} |
164 |
|
165 |
/* Free memory used by a parser context */ |
166 |
void parser_context_free(parser_context_t *ctx) |
167 |
{ |
168 |
parser_free_tokens(ctx->tok_head); |
169 |
|
170 |
if (ctx->tmp_tok != NULL) |
171 |
free(ctx->tmp_tok); |
172 |
|
173 |
parser_context_init(ctx); |
174 |
} |
175 |
|
176 |
/* Determine the type of the input character */ |
177 |
static int parser_get_char_type(u_char c) |
178 |
{ |
179 |
switch(c) { |
180 |
case '\n': |
181 |
case '\r': |
182 |
case 0: |
183 |
return(PARSER_CHAR_NEWLINE); |
184 |
case '\t': |
185 |
//case '\r': |
186 |
case ' ': |
187 |
return(PARSER_CHAR_BLANK); |
188 |
case '!': |
189 |
case '#': |
190 |
return(PARSER_CHAR_COMMENT); |
191 |
case '"': |
192 |
return(PARSER_CHAR_QUOTE); |
193 |
default: |
194 |
return(PARSER_CHAR_OTHER); |
195 |
} |
196 |
} |
197 |
|
198 |
/* Send a buffer to the tokenizer */ |
199 |
int parser_scan_buffer(parser_context_t *ctx,u_char *buf,size_t buf_size) |
200 |
{ |
201 |
int i,type; |
202 |
u_char c; |
203 |
|
204 |
for(i=0;(i<buf_size) && (ctx->state != PARSER_STATE_DONE);i++) |
205 |
{ |
206 |
ctx->consumed_len++; |
207 |
c = buf[i]; |
208 |
|
209 |
/* Determine character type */ |
210 |
type = parser_get_char_type(c); |
211 |
|
212 |
/* Basic finite state machine */ |
213 |
switch(ctx->state) { |
214 |
case PARSER_STATE_SKIP: |
215 |
if (type == PARSER_CHAR_NEWLINE) |
216 |
ctx->state = PARSER_STATE_DONE; |
217 |
|
218 |
/* Simply ignore character until we reach end of line */ |
219 |
break; |
220 |
|
221 |
case PARSER_STATE_BLANK: |
222 |
switch(type) { |
223 |
case PARSER_CHAR_BLANK: |
224 |
/* Eat space */ |
225 |
break; |
226 |
|
227 |
case PARSER_CHAR_COMMENT: |
228 |
ctx->state = PARSER_STATE_SKIP; |
229 |
break; |
230 |
|
231 |
case PARSER_CHAR_NEWLINE: |
232 |
ctx->state = PARSER_STATE_DONE; |
233 |
break; |
234 |
|
235 |
case PARSER_CHAR_QUOTE: |
236 |
ctx->state = PARSER_STATE_QUOTED_STRING; |
237 |
break; |
238 |
|
239 |
default: |
240 |
/* Begin a new string */ |
241 |
if (!tmp_token_add_char(ctx,c)) { |
242 |
ctx->state = PARSER_STATE_STRING; |
243 |
} else { |
244 |
ctx->state = PARSER_STATE_SKIP; |
245 |
ctx->error = PARSER_ERROR_NOMEM; |
246 |
} |
247 |
} |
248 |
break; |
249 |
|
250 |
case PARSER_STATE_STRING: |
251 |
switch(type) { |
252 |
case PARSER_CHAR_BLANK: |
253 |
if (!parser_move_tmp_token(ctx)) { |
254 |
ctx->state = PARSER_STATE_BLANK; |
255 |
} else { |
256 |
ctx->state = PARSER_STATE_SKIP; |
257 |
ctx->error = PARSER_ERROR_NOMEM; |
258 |
} |
259 |
break; |
260 |
|
261 |
case PARSER_CHAR_NEWLINE: |
262 |
if (parser_move_tmp_token(ctx) == -1) |
263 |
ctx->error = PARSER_ERROR_NOMEM; |
264 |
|
265 |
ctx->state = PARSER_STATE_DONE; |
266 |
break; |
267 |
|
268 |
case PARSER_CHAR_COMMENT: |
269 |
if (parser_move_tmp_token(ctx) == -1) |
270 |
ctx->error = PARSER_ERROR_NOMEM; |
271 |
|
272 |
ctx->state = PARSER_STATE_SKIP; |
273 |
break; |
274 |
|
275 |
case PARSER_CHAR_QUOTE: |
276 |
ctx->error = PARSER_ERROR_UNEXP_QUOTE; |
277 |
ctx->state = PARSER_STATE_SKIP; |
278 |
break; |
279 |
|
280 |
default: |
281 |
/* Add the character to the buffer */ |
282 |
if (tmp_token_add_char(ctx,c) == -1) { |
283 |
ctx->state = PARSER_STATE_SKIP; |
284 |
ctx->error = PARSER_ERROR_NOMEM; |
285 |
} |
286 |
} |
287 |
break; |
288 |
|
289 |
case PARSER_STATE_QUOTED_STRING: |
290 |
switch(type) { |
291 |
case PARSER_CHAR_NEWLINE: |
292 |
/* Unterminated string! */ |
293 |
ctx->error = PARSER_ERROR_UNEXP_EOL; |
294 |
ctx->state = PARSER_STATE_DONE; |
295 |
break; |
296 |
|
297 |
case PARSER_CHAR_QUOTE: |
298 |
if (!parser_move_tmp_token(ctx)) { |
299 |
ctx->state = PARSER_STATE_BLANK; |
300 |
} else { |
301 |
ctx->state = PARSER_STATE_SKIP; |
302 |
ctx->error = PARSER_ERROR_NOMEM; |
303 |
} |
304 |
break; |
305 |
|
306 |
default: |
307 |
/* Add the character to the buffer */ |
308 |
if (tmp_token_add_char(ctx,c) == -1) { |
309 |
ctx->state = PARSER_STATE_SKIP; |
310 |
ctx->error = PARSER_ERROR_NOMEM; |
311 |
} |
312 |
} |
313 |
break; |
314 |
} |
315 |
} |
316 |
|
317 |
return(ctx->state == PARSER_STATE_DONE); |
318 |
} |
319 |
|
320 |
/* Parser tests */ |
321 |
static char *parser_test_str[] = { |
322 |
"c7200 show_hardware R1", |
323 |
"c7200 show_hardware \"R1\"", |
324 |
" c7200 show_hardware \"R1\" ", |
325 |
"\"c7200\" \"show_hardware\" \"R1\"", |
326 |
"hypervisor set_working_dir \"C:\\Program Files\\Dynamips Test\"", |
327 |
"hypervisor # This is a comment set_working_dir \"C:\\Program Files\"", |
328 |
"\"c7200\" \"show_hardware\" \"R1", |
329 |
NULL, |
330 |
}; |
331 |
|
332 |
void parser_run_tests(void) |
333 |
{ |
334 |
parser_context_t ctx; |
335 |
int i,res; |
336 |
|
337 |
for(i=0;parser_test_str[i];i++) { |
338 |
parser_context_init(&ctx); |
339 |
|
340 |
res = parser_scan_buffer(&ctx,parser_test_str[i], |
341 |
strlen(parser_test_str[i])+1); |
342 |
|
343 |
printf("\n%d: Test string: [%s] => res=%d, state=%d\n", |
344 |
i,parser_test_str[i],res,ctx.state); |
345 |
|
346 |
if ((res != 0) && (ctx.error == 0)) { |
347 |
if (ctx.tok_head) { |
348 |
printf("Tokens: "); |
349 |
parser_dump_tokens(&ctx); |
350 |
printf("\n"); |
351 |
} |
352 |
} |
353 |
|
354 |
parser_context_free(&ctx); |
355 |
} |
356 |
} |