1 |
/* |
/* |
2 |
* Cisco 7200 (Predator) simulation platform. |
* Cisco router simulation platform. |
3 |
* Copyright (c) 2006 Christophe Fillot (cf@utc.fr) |
* Copyright (c) 2006 Christophe Fillot (cf@utc.fr) |
4 |
* |
* |
5 |
* Mini-parser. |
* Mini-parser. |
27 |
|
|
28 |
#define TOKEN_MAX_SIZE 512 |
#define TOKEN_MAX_SIZE 512 |
29 |
|
|
|
/* Parser states */ |
|
|
enum { |
|
|
PARSER_STATE_BLANK, |
|
|
PARSER_STATE_STRING, |
|
|
PARSER_STATE_QUOTED_STRING, |
|
|
}; |
|
|
|
|
30 |
/* Character types */ |
/* Character types */ |
31 |
enum { |
enum { |
32 |
PARSER_CHAR_BLANK, |
PARSER_CHAR_BLANK, |
37 |
}; |
}; |
38 |
|
|
39 |
/* Get a description given an error code */ |
/* Get a description given an error code */ |
40 |
char *parser_strerror(int error) |
char *parser_strerror(parser_context_t *ctx) |
41 |
{ |
{ |
42 |
printf("error = %d\n",error); |
printf("error = %d\n",ctx->error); |
43 |
|
|
44 |
switch(error) { |
switch(ctx->error) { |
45 |
|
case 0: |
46 |
|
return "no error"; |
47 |
case PARSER_ERROR_NOMEM: |
case PARSER_ERROR_NOMEM: |
48 |
return "insufficient memory"; |
return "insufficient memory"; |
49 |
case PARSER_ERROR_UNEXP_QUOTE: |
case PARSER_ERROR_UNEXP_QUOTE: |
55 |
} |
} |
56 |
} |
} |
57 |
|
|
58 |
/* Create a new token */ |
/* Dump a token list */ |
59 |
static int token_create(parser_token_t **head,parser_token_t **last, |
void parser_dump_tokens(parser_context_t *ctx) |
|
char *value) |
|
60 |
{ |
{ |
61 |
parser_token_t *t; |
parser_token_t *tok; |
62 |
|
|
63 |
if (!(t = malloc(sizeof(*t)))) |
for(tok=ctx->tok_head;tok;tok=tok->next) |
64 |
return(-1); |
printf("\"%s\" ",tok->value); |
65 |
|
} |
66 |
|
|
67 |
if (!(t->value = strdup(value))) { |
/* Map a token list to an array */ |
68 |
free(t); |
char **parser_map_array(parser_context_t *ctx) |
69 |
return(-1); |
{ |
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 |
t->next = NULL; |
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 (*last) { |
if (!new_str) |
97 |
(*last)->next = t; |
return(-1); |
98 |
*last = t; |
|
99 |
} else { |
ctx->tmp_tok = new_str; |
100 |
*head = *last = t; |
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); |
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 */ |
/* Free a token list */ |
154 |
void parser_free_tokens(parser_token_t *tok_list) |
void parser_free_tokens(parser_token_t *tok_list) |
155 |
{ |
{ |
156 |
parser_token_t *t,*next; |
parser_token_t *t,*next; |
157 |
|
|
158 |
for(t=tok_list;t;t=next) { |
for(t=tok_list;t;t=next) { |
159 |
next = t->next; |
next = t->next; |
160 |
free(t->value); |
free(t->value); |
162 |
} |
} |
163 |
} |
} |
164 |
|
|
165 |
/* Dump a token list */ |
/* Free memory used by a parser context */ |
166 |
void parser_dump_tokens(parser_token_t *tok_list) |
void parser_context_free(parser_context_t *ctx) |
|
{ |
|
|
parser_token_t *t; |
|
|
|
|
|
for(t=tok_list;t;t=t->next) |
|
|
printf("\"%s\" ",t->value); |
|
|
} |
|
|
|
|
|
/* Map a token list to an array */ |
|
|
char **parser_map_array(parser_token_t *tok_list,int tok_count) |
|
167 |
{ |
{ |
168 |
char **map; |
parser_free_tokens(ctx->tok_head); |
|
int i; |
|
|
|
|
|
if (tok_count <= 0) |
|
|
return NULL; |
|
169 |
|
|
170 |
if (!(map = calloc(tok_count,sizeof(char **)))) |
if (ctx->tmp_tok != NULL) |
171 |
return NULL; |
free(ctx->tmp_tok); |
|
|
|
|
for(i=0;(i<tok_count) && tok_list;i++,tok_list=tok_list->next) |
|
|
map[i] = tok_list->value; |
|
172 |
|
|
173 |
return map; |
parser_context_init(ctx); |
174 |
} |
} |
175 |
|
|
176 |
/* Add a character to a buffer */ |
/* Determine the type of the input character */ |
177 |
static int buffer_add_char(char *buffer,int *pos,char c) |
static int parser_get_char_type(u_char c) |
178 |
{ |
{ |
179 |
if (*pos >= TOKEN_MAX_SIZE) |
switch(c) { |
180 |
return(-1); |
case '\n': |
181 |
|
case '\r': |
182 |
buffer[(*pos)++] = c; |
case 0: |
183 |
buffer[*pos] = 0; |
return(PARSER_CHAR_NEWLINE); |
184 |
return(0); |
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 |
/* Tokenize a string */ |
/* Send a buffer to the tokenizer */ |
199 |
int parser_tokenize(char *str,struct parser_token **tokens,int *tok_count) |
int parser_scan_buffer(parser_context_t *ctx,u_char *buf,size_t buf_size) |
200 |
{ |
{ |
201 |
char buffer[TOKEN_MAX_SIZE+1]; |
int i,type; |
202 |
parser_token_t *tok_head,*tok_last; |
u_char c; |
|
int i,buf_pos,type; |
|
|
int state,error,done; |
|
|
size_t len; |
|
|
char c; |
|
|
|
|
|
len = strlen(str); |
|
|
tok_head = tok_last = NULL; |
|
|
*tokens = NULL; |
|
|
*tok_count = 0; |
|
|
state = PARSER_STATE_BLANK; |
|
|
done = FALSE; |
|
|
error = 0; |
|
|
buf_pos = 0; |
|
203 |
|
|
204 |
for(i=0;(i<len+1) && !error && !done;i++) |
for(i=0;(i<buf_size) && (ctx->state != PARSER_STATE_DONE);i++) |
205 |
{ |
{ |
206 |
c = str[i]; |
ctx->consumed_len++; |
207 |
|
c = buf[i]; |
208 |
|
|
209 |
/* Determine character type */ |
/* Determine character type */ |
210 |
switch(c) { |
type = parser_get_char_type(c); |
|
case '\n': |
|
|
case '\r': |
|
|
case 0: |
|
|
type = PARSER_CHAR_NEWLINE; |
|
|
break; |
|
|
case '\t': |
|
|
case ' ': |
|
|
type = PARSER_CHAR_BLANK; |
|
|
break; |
|
|
case '!': |
|
|
case '#': |
|
|
type = PARSER_CHAR_COMMENT; |
|
|
break; |
|
|
case '"': |
|
|
type = PARSER_CHAR_QUOTE; |
|
|
break; |
|
|
default: |
|
|
type = PARSER_CHAR_OTHER; |
|
|
} |
|
211 |
|
|
212 |
/* Basic finite state machine */ |
/* Basic finite state machine */ |
213 |
switch(state) { |
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: |
case PARSER_STATE_BLANK: |
222 |
switch(type) { |
switch(type) { |
223 |
case PARSER_CHAR_BLANK: |
case PARSER_CHAR_BLANK: |
224 |
/* Eat space */ |
/* Eat space */ |
225 |
break; |
break; |
226 |
|
|
|
case PARSER_CHAR_NEWLINE: |
|
227 |
case PARSER_CHAR_COMMENT: |
case PARSER_CHAR_COMMENT: |
228 |
done = TRUE; |
ctx->state = PARSER_STATE_SKIP; |
229 |
|
break; |
230 |
|
|
231 |
|
case PARSER_CHAR_NEWLINE: |
232 |
|
ctx->state = PARSER_STATE_DONE; |
233 |
break; |
break; |
234 |
|
|
235 |
case PARSER_CHAR_QUOTE: |
case PARSER_CHAR_QUOTE: |
236 |
state = PARSER_STATE_QUOTED_STRING; |
ctx->state = PARSER_STATE_QUOTED_STRING; |
|
buf_pos = 0; |
|
237 |
break; |
break; |
238 |
|
|
239 |
default: |
default: |
240 |
/* Begin a new string */ |
/* Begin a new string */ |
241 |
state = PARSER_STATE_STRING; |
if (!tmp_token_add_char(ctx,c)) { |
242 |
buf_pos = 0; |
ctx->state = PARSER_STATE_STRING; |
243 |
buffer_add_char(buffer,&buf_pos,c); |
} else { |
244 |
|
ctx->state = PARSER_STATE_SKIP; |
245 |
|
ctx->error = PARSER_ERROR_NOMEM; |
246 |
|
} |
247 |
} |
} |
248 |
break; |
break; |
249 |
|
|
250 |
case PARSER_STATE_STRING: |
case PARSER_STATE_STRING: |
251 |
switch(type) { |
switch(type) { |
252 |
case PARSER_CHAR_BLANK: |
case PARSER_CHAR_BLANK: |
253 |
if (token_create(&tok_head,&tok_last,buffer) == -1) |
if (!parser_move_tmp_token(ctx)) { |
254 |
error = PARSER_ERROR_NOMEM; |
ctx->state = PARSER_STATE_BLANK; |
255 |
|
} else { |
256 |
(*tok_count)++; |
ctx->state = PARSER_STATE_SKIP; |
257 |
state = PARSER_STATE_BLANK; |
ctx->error = PARSER_ERROR_NOMEM; |
258 |
|
} |
259 |
break; |
break; |
260 |
|
|
261 |
case PARSER_CHAR_NEWLINE: |
case PARSER_CHAR_NEWLINE: |
262 |
if (token_create(&tok_head,&tok_last,buffer) == -1) |
if (parser_move_tmp_token(ctx) == -1) |
263 |
error = PARSER_ERROR_NOMEM; |
ctx->error = PARSER_ERROR_NOMEM; |
264 |
|
|
265 |
(*tok_count)++; |
ctx->state = PARSER_STATE_DONE; |
|
done = TRUE; |
|
266 |
break; |
break; |
267 |
|
|
268 |
case PARSER_CHAR_COMMENT: |
case PARSER_CHAR_COMMENT: |
269 |
done = TRUE; |
if (parser_move_tmp_token(ctx) == -1) |
270 |
|
ctx->error = PARSER_ERROR_NOMEM; |
271 |
|
|
272 |
|
ctx->state = PARSER_STATE_SKIP; |
273 |
break; |
break; |
274 |
|
|
275 |
case PARSER_CHAR_QUOTE: |
case PARSER_CHAR_QUOTE: |
276 |
error = PARSER_ERROR_UNEXP_QUOTE; |
ctx->error = PARSER_ERROR_UNEXP_QUOTE; |
277 |
|
ctx->state = PARSER_STATE_SKIP; |
278 |
break; |
break; |
279 |
|
|
280 |
default: |
default: |
281 |
/* Add the character to the buffer */ |
/* Add the character to the buffer */ |
282 |
buffer_add_char(buffer,&buf_pos,c); |
if (tmp_token_add_char(ctx,c) == -1) { |
283 |
|
ctx->state = PARSER_STATE_SKIP; |
284 |
|
ctx->error = PARSER_ERROR_NOMEM; |
285 |
|
} |
286 |
} |
} |
287 |
break; |
break; |
288 |
|
|
290 |
switch(type) { |
switch(type) { |
291 |
case PARSER_CHAR_NEWLINE: |
case PARSER_CHAR_NEWLINE: |
292 |
/* Unterminated string! */ |
/* Unterminated string! */ |
293 |
error = PARSER_ERROR_UNEXP_EOL; |
ctx->error = PARSER_ERROR_UNEXP_EOL; |
294 |
|
ctx->state = PARSER_STATE_DONE; |
295 |
break; |
break; |
296 |
|
|
297 |
case PARSER_CHAR_QUOTE: |
case PARSER_CHAR_QUOTE: |
298 |
if (token_create(&tok_head,&tok_last,buffer) == -1) |
if (!parser_move_tmp_token(ctx)) { |
299 |
error = PARSER_ERROR_NOMEM; |
ctx->state = PARSER_STATE_BLANK; |
300 |
|
} else { |
301 |
(*tok_count)++; |
ctx->state = PARSER_STATE_SKIP; |
302 |
state = PARSER_STATE_BLANK; |
ctx->error = PARSER_ERROR_NOMEM; |
303 |
|
} |
304 |
break; |
break; |
305 |
|
|
306 |
default: |
default: |
307 |
/* Add the character to the buffer */ |
/* Add the character to the buffer */ |
308 |
buffer_add_char(buffer,&buf_pos,c); |
if (tmp_token_add_char(ctx,c) == -1) { |
309 |
|
ctx->state = PARSER_STATE_SKIP; |
310 |
|
ctx->error = PARSER_ERROR_NOMEM; |
311 |
|
} |
312 |
} |
} |
313 |
break; |
break; |
314 |
} |
} |
315 |
} |
} |
316 |
|
|
317 |
if (error) { |
return(ctx->state == PARSER_STATE_DONE); |
|
parser_free_tokens(tok_head); |
|
|
return(error); |
|
|
} |
|
|
|
|
|
*tokens = tok_head; |
|
|
return(0); |
|
318 |
} |
} |
319 |
|
|
320 |
/* Parser tests */ |
/* Parser tests */ |
331 |
|
|
332 |
void parser_run_tests(void) |
void parser_run_tests(void) |
333 |
{ |
{ |
334 |
struct parser_token *tok_list; |
parser_context_t ctx; |
335 |
int i,res,tok_count; |
int i,res; |
336 |
|
|
337 |
for(i=0;parser_test_str[i];i++) { |
for(i=0;parser_test_str[i];i++) { |
338 |
res = parser_tokenize(parser_test_str[i],&tok_list,&tok_count); |
parser_context_init(&ctx); |
339 |
|
|
340 |
printf("\n%d: Test string: [%s] => res=%d\n", |
res = parser_scan_buffer(&ctx,parser_test_str[i], |
341 |
i,parser_test_str[i],res); |
strlen(parser_test_str[i])+1); |
|
|
|
|
if (tok_list) { |
|
|
printf("Tokens: "); |
|
|
parser_dump_tokens(tok_list); |
|
|
printf("\n"); |
|
342 |
|
|
343 |
parser_free_tokens(tok_list); |
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 |
} |
} |