| 1 |
/* Generated by re2c 0.12.0 on Fri Feb 8 14:52:45 2008 */ |
|---|
| 2 |
#line 1 "noit_tokenizer.re" |
|---|
| 3 |
#include <stdlib.h> |
|---|
| 4 |
#include <string.h> |
|---|
| 5 |
|
|---|
| 6 |
struct token { |
|---|
| 7 |
char *token; |
|---|
| 8 |
const char *start; |
|---|
| 9 |
const char *end; |
|---|
| 10 |
const char *next; |
|---|
| 11 |
enum { NT_IDENT, NT_DQSTRING, NT_SPACE, NT_UNKNOWN, NT_EOF } type; |
|---|
| 12 |
}; |
|---|
| 13 |
#define SET_TOKEN(t,a) (t)->next = (a) |
|---|
| 14 |
|
|---|
| 15 |
static void c_unescape(char *p, char *only) { |
|---|
| 16 |
char *bt = p; |
|---|
| 17 |
#define ASSIGN(a) *(bt++) = (a) |
|---|
| 18 |
while(p[0] != '\0') { |
|---|
| 19 |
if(p[0] == '\\' && p[1] != '\0' && (!only || p[1] == *only)) { |
|---|
| 20 |
switch(p[1]) { |
|---|
| 21 |
case ' ': ASSIGN(' '); p+=2; break; |
|---|
| 22 |
case '"': ASSIGN('"'); p+=2; break; |
|---|
| 23 |
case 'n': ASSIGN('\n'); p+=2; break; |
|---|
| 24 |
case 'r': ASSIGN('\r'); p+=2; break; |
|---|
| 25 |
case 't': ASSIGN('\t'); p+=2; break; |
|---|
| 26 |
case 'a': ASSIGN('\a'); p+=2; break; |
|---|
| 27 |
case 'b': ASSIGN('\b'); p+=2; break; |
|---|
| 28 |
case 'v': ASSIGN('\v'); p+=2; break; |
|---|
| 29 |
case 'f': ASSIGN('\f'); p+=2; break; |
|---|
| 30 |
case '0': ASSIGN('\0'); p+=2; break; |
|---|
| 31 |
case '\\': ASSIGN('\\'); p+=2; break; |
|---|
| 32 |
default: ASSIGN(*p); p++; ASSIGN(*p); p++; break; |
|---|
| 33 |
} |
|---|
| 34 |
} |
|---|
| 35 |
else { |
|---|
| 36 |
ASSIGN(*p); p++; |
|---|
| 37 |
} |
|---|
| 38 |
} |
|---|
| 39 |
*bt = '\0'; |
|---|
| 40 |
} |
|---|
| 41 |
|
|---|
| 42 |
#define BAIL_UNKNOWN do { t->type = NT_UNKNOWN; return -1; } while(0) |
|---|
| 43 |
static int token_scan(struct token *t) |
|---|
| 44 |
{ |
|---|
| 45 |
t->start = t->end = t->next; |
|---|
| 46 |
|
|---|
| 47 |
mainpattern: |
|---|
| 48 |
|
|---|
| 49 |
#line 50 "noit_tokenizer.c" |
|---|
| 50 |
{ |
|---|
| 51 |
unsigned char yych; |
|---|
| 52 |
|
|---|
| 53 |
yych = (unsigned char)*t->next; |
|---|
| 54 |
switch(yych) { |
|---|
| 55 |
case 0x00: goto yy10; |
|---|
| 56 |
case 0x09: |
|---|
| 57 |
case 0x0A: |
|---|
| 58 |
case 0x0D: |
|---|
| 59 |
case ' ': goto yy2; |
|---|
| 60 |
case '"': goto yy4; |
|---|
| 61 |
case '\'': goto yy6; |
|---|
| 62 |
default: goto yy8; |
|---|
| 63 |
} |
|---|
| 64 |
yy2: |
|---|
| 65 |
++t->next; |
|---|
| 66 |
yych = (unsigned char)*t->next; |
|---|
| 67 |
goto yy17; |
|---|
| 68 |
yy3: |
|---|
| 69 |
#line 53 "noit_tokenizer.re" |
|---|
| 70 |
{ t->token = NULL; |
|---|
| 71 |
t->end = t->next; |
|---|
| 72 |
t->type = NT_SPACE; |
|---|
| 73 |
return 1; } |
|---|
| 74 |
#line 75 "noit_tokenizer.c" |
|---|
| 75 |
yy4: |
|---|
| 76 |
++t->next; |
|---|
| 77 |
#line 57 "noit_tokenizer.re" |
|---|
| 78 |
{ t->type = NT_DQSTRING; |
|---|
| 79 |
if(t->start != t->end) { |
|---|
| 80 |
t->start++; |
|---|
| 81 |
t->end = t->next - 1; |
|---|
| 82 |
t->token = malloc(t->end-t->start + 1); |
|---|
| 83 |
strlcpy(t->token, t->start, t->end-t->start + 1); |
|---|
| 84 |
c_unescape(t->token, NULL); |
|---|
| 85 |
return 1; |
|---|
| 86 |
} |
|---|
| 87 |
else |
|---|
| 88 |
goto dqstring; |
|---|
| 89 |
} |
|---|
| 90 |
#line 91 "noit_tokenizer.c" |
|---|
| 91 |
yy6: |
|---|
| 92 |
++t->next; |
|---|
| 93 |
#line 69 "noit_tokenizer.re" |
|---|
| 94 |
{ t->type = NT_IDENT; |
|---|
| 95 |
if(t->start != t->end) { |
|---|
| 96 |
t->start++; |
|---|
| 97 |
t->end = t->next - 1; |
|---|
| 98 |
t->token = malloc(t->end-t->start + 1); |
|---|
| 99 |
strlcpy(t->token, t->start, t->end-t->start + 1); |
|---|
| 100 |
return 1; |
|---|
| 101 |
} |
|---|
| 102 |
else |
|---|
| 103 |
goto sqstring; |
|---|
| 104 |
} |
|---|
| 105 |
#line 106 "noit_tokenizer.c" |
|---|
| 106 |
yy8: |
|---|
| 107 |
++t->next; |
|---|
| 108 |
yych = (unsigned char)*t->next; |
|---|
| 109 |
goto yy13; |
|---|
| 110 |
yy9: |
|---|
| 111 |
#line 81 "noit_tokenizer.re" |
|---|
| 112 |
{ char only = ' '; |
|---|
| 113 |
t->end = t->next; |
|---|
| 114 |
t->type = NT_IDENT; |
|---|
| 115 |
t->token = malloc(t->end-t->start + 1); |
|---|
| 116 |
strlcpy(t->token, t->start, t->end-t->start + 1); |
|---|
| 117 |
c_unescape(t->token, &only); |
|---|
| 118 |
return 1; |
|---|
| 119 |
} |
|---|
| 120 |
#line 121 "noit_tokenizer.c" |
|---|
| 121 |
yy10: |
|---|
| 122 |
++t->next; |
|---|
| 123 |
#line 89 "noit_tokenizer.re" |
|---|
| 124 |
{ t->token = NULL; |
|---|
| 125 |
t->type = NT_EOF; |
|---|
| 126 |
return 0; |
|---|
| 127 |
} |
|---|
| 128 |
#line 129 "noit_tokenizer.c" |
|---|
| 129 |
yy12: |
|---|
| 130 |
++t->next; |
|---|
| 131 |
yych = (unsigned char)*t->next; |
|---|
| 132 |
yy13: |
|---|
| 133 |
switch(yych) { |
|---|
| 134 |
case 0x00: |
|---|
| 135 |
case 0x09: |
|---|
| 136 |
case 0x0A: |
|---|
| 137 |
case 0x0D: |
|---|
| 138 |
case ' ': goto yy9; |
|---|
| 139 |
case '\\': goto yy14; |
|---|
| 140 |
default: goto yy12; |
|---|
| 141 |
} |
|---|
| 142 |
yy14: |
|---|
| 143 |
++t->next; |
|---|
| 144 |
yych = (unsigned char)*t->next; |
|---|
| 145 |
switch(yych) { |
|---|
| 146 |
case 0x00: |
|---|
| 147 |
case 0x09: |
|---|
| 148 |
case 0x0A: |
|---|
| 149 |
case 0x0D: goto yy9; |
|---|
| 150 |
case '\\': goto yy14; |
|---|
| 151 |
default: goto yy12; |
|---|
| 152 |
} |
|---|
| 153 |
yy16: |
|---|
| 154 |
++t->next; |
|---|
| 155 |
yych = (unsigned char)*t->next; |
|---|
| 156 |
yy17: |
|---|
| 157 |
switch(yych) { |
|---|
| 158 |
case 0x09: |
|---|
| 159 |
case 0x0A: |
|---|
| 160 |
case 0x0D: |
|---|
| 161 |
case ' ': goto yy16; |
|---|
| 162 |
default: goto yy3; |
|---|
| 163 |
} |
|---|
| 164 |
} |
|---|
| 165 |
#line 94 "noit_tokenizer.re" |
|---|
| 166 |
|
|---|
| 167 |
|
|---|
| 168 |
sqstring: |
|---|
| 169 |
|
|---|
| 170 |
#line 171 "noit_tokenizer.c" |
|---|
| 171 |
{ |
|---|
| 172 |
unsigned char yych; |
|---|
| 173 |
yych = (unsigned char)*t->next; |
|---|
| 174 |
switch(yych) { |
|---|
| 175 |
case 0x00: goto yy23; |
|---|
| 176 |
case '\'': goto yy20; |
|---|
| 177 |
default: goto yy21; |
|---|
| 178 |
} |
|---|
| 179 |
yy20: |
|---|
| 180 |
#line 98 "noit_tokenizer.re" |
|---|
| 181 |
{ t->end = t->next; |
|---|
| 182 |
goto mainpattern; } |
|---|
| 183 |
#line 184 "noit_tokenizer.c" |
|---|
| 184 |
yy21: |
|---|
| 185 |
++t->next; |
|---|
| 186 |
yych = (unsigned char)*t->next; |
|---|
| 187 |
switch(yych) { |
|---|
| 188 |
case 0x00: |
|---|
| 189 |
case '\'': goto yy20; |
|---|
| 190 |
default: goto yy21; |
|---|
| 191 |
} |
|---|
| 192 |
yy23: |
|---|
| 193 |
++t->next; |
|---|
| 194 |
#line 100 "noit_tokenizer.re" |
|---|
| 195 |
{ BAIL_UNKNOWN; } |
|---|
| 196 |
#line 197 "noit_tokenizer.c" |
|---|
| 197 |
} |
|---|
| 198 |
#line 101 "noit_tokenizer.re" |
|---|
| 199 |
|
|---|
| 200 |
|
|---|
| 201 |
dqstring: |
|---|
| 202 |
|
|---|
| 203 |
#line 204 "noit_tokenizer.c" |
|---|
| 204 |
{ |
|---|
| 205 |
unsigned char yych; |
|---|
| 206 |
yych = (unsigned char)*t->next; |
|---|
| 207 |
switch(yych) { |
|---|
| 208 |
case 0x00: goto yy33; |
|---|
| 209 |
case '"': goto yy29; |
|---|
| 210 |
case '\\': goto yy27; |
|---|
| 211 |
default: goto yy31; |
|---|
| 212 |
} |
|---|
| 213 |
yy27: |
|---|
| 214 |
yych = (unsigned char)*++t->next; |
|---|
| 215 |
switch(yych) { |
|---|
| 216 |
case 0x00: goto yy28; |
|---|
| 217 |
case '"': |
|---|
| 218 |
case '0': |
|---|
| 219 |
case '\\': |
|---|
| 220 |
case 'a': |
|---|
| 221 |
case 'b': |
|---|
| 222 |
case 'f': |
|---|
| 223 |
case 'n': |
|---|
| 224 |
case 'r': |
|---|
| 225 |
case 't': |
|---|
| 226 |
case 'v': goto yy37; |
|---|
| 227 |
default: goto yy35; |
|---|
| 228 |
} |
|---|
| 229 |
yy28: |
|---|
| 230 |
yy29: |
|---|
| 231 |
++t->next; |
|---|
| 232 |
#line 109 "noit_tokenizer.re" |
|---|
| 233 |
{ t->end = t->next--; |
|---|
| 234 |
goto mainpattern; |
|---|
| 235 |
} |
|---|
| 236 |
#line 237 "noit_tokenizer.c" |
|---|
| 237 |
yy31: |
|---|
| 238 |
++t->next; |
|---|
| 239 |
#line 112 "noit_tokenizer.re" |
|---|
| 240 |
{ goto dqstring; } |
|---|
| 241 |
#line 242 "noit_tokenizer.c" |
|---|
| 242 |
yy33: |
|---|
| 243 |
++t->next; |
|---|
| 244 |
#line 113 "noit_tokenizer.re" |
|---|
| 245 |
{ BAIL_UNKNOWN; } |
|---|
| 246 |
#line 247 "noit_tokenizer.c" |
|---|
| 247 |
yy35: |
|---|
| 248 |
++t->next; |
|---|
| 249 |
#line 108 "noit_tokenizer.re" |
|---|
| 250 |
{ goto dqstring; } |
|---|
| 251 |
#line 252 "noit_tokenizer.c" |
|---|
| 252 |
yy37: |
|---|
| 253 |
++t->next; |
|---|
| 254 |
#line 106 "noit_tokenizer.re" |
|---|
| 255 |
{ goto dqstring; } |
|---|
| 256 |
#line 257 "noit_tokenizer.c" |
|---|
| 257 |
} |
|---|
| 258 |
#line 114 "noit_tokenizer.re" |
|---|
| 259 |
|
|---|
| 260 |
} |
|---|
| 261 |
|
|---|
| 262 |
int noit_tokenize(const char *input, char **vector, int *cnt) { |
|---|
| 263 |
struct token t; |
|---|
| 264 |
int i = 0; |
|---|
| 265 |
|
|---|
| 266 |
SET_TOKEN(&t, input); |
|---|
| 267 |
while(token_scan(&t) != -1) { |
|---|
| 268 |
switch(t.type) { |
|---|
| 269 |
case NT_IDENT: |
|---|
| 270 |
case NT_DQSTRING: |
|---|
| 271 |
if(i<*cnt) vector[i] = t.token; |
|---|
| 272 |
i++; |
|---|
| 273 |
break; |
|---|
| 274 |
case NT_SPACE: |
|---|
| 275 |
break; |
|---|
| 276 |
case NT_EOF: |
|---|
| 277 |
if(i<*cnt) *cnt = i; |
|---|
| 278 |
return i; |
|---|
| 279 |
case NT_UNKNOWN: |
|---|
| 280 |
/* UNREACHED */ |
|---|
| 281 |
goto failure; |
|---|
| 282 |
} |
|---|
| 283 |
} |
|---|
| 284 |
failure: |
|---|
| 285 |
if(i<*cnt) *cnt = i; |
|---|
| 286 |
return input - t.next; |
|---|
| 287 |
} |
|---|