Daniel Veillard | 4255d50 | 2002-04-16 15:50:10 +0000 | [diff] [blame] | 1 | /* |
| 2 | * testRegexp.c: simple module for testing regular expressions |
| 3 | * |
| 4 | * See Copyright for the status of this software. |
| 5 | * |
| 6 | * Daniel Veillard <veillard@redhat.com> |
| 7 | */ |
| 8 | |
| 9 | #include <string.h> |
| 10 | #include "libxml.h" |
| 11 | #ifdef LIBXML_AUTOMATA_ENABLED |
| 12 | |
| 13 | #include <libxml/xmlautomata.h> |
| 14 | |
| 15 | static int scanNumber(char **ptr) { |
| 16 | int ret = 0; |
| 17 | char *cur; |
| 18 | |
| 19 | cur = *ptr; |
| 20 | while ((*cur >= '0') && (*cur <= '9')) { |
| 21 | ret = ret * 10 + (*cur - '0'); |
| 22 | cur++; |
| 23 | } |
| 24 | *ptr = cur; |
| 25 | return(ret); |
| 26 | } |
| 27 | |
| 28 | static void |
| 29 | testRegexpFile(const char *filename) { |
| 30 | FILE *input; |
| 31 | char exp[5000]; |
| 32 | int len; |
| 33 | int ret; |
| 34 | int i; |
| 35 | xmlAutomataPtr am; |
| 36 | xmlAutomataStatePtr states[1000]; |
| 37 | xmlRegexpPtr regexp = NULL; |
| 38 | xmlRegExecCtxtPtr exec; |
| 39 | |
| 40 | for (i = 0;i<1000;i++) |
| 41 | states[i] = NULL; |
| 42 | |
| 43 | input = fopen(filename, "r"); |
| 44 | if (input == NULL) { |
| 45 | xmlGenericError(xmlGenericErrorContext, |
| 46 | "Cannot open %s for reading\n", filename); |
| 47 | return; |
| 48 | } |
| 49 | |
| 50 | am = xmlNewAutomata(); |
| 51 | if (am == NULL) { |
| 52 | xmlGenericError(xmlGenericErrorContext, |
| 53 | "Cannot create automata\n"); |
| 54 | fclose(input); |
| 55 | } |
| 56 | states[0] = xmlAutomataGetInitState(am); |
| 57 | if (states[0] == NULL) { |
| 58 | xmlGenericError(xmlGenericErrorContext, |
| 59 | "Cannot get start state\n"); |
| 60 | xmlFreeAutomata(am); |
| 61 | fclose(input); |
| 62 | } |
| 63 | ret = 0; |
| 64 | |
| 65 | while (fgets(exp, 4500, input) != NULL) { |
| 66 | if (exp[0] == '#') |
| 67 | continue; |
| 68 | len = strlen(exp); |
| 69 | len--; |
| 70 | while ((len >= 0) && |
| 71 | ((exp[len] == '\n') || (exp[len] == '\t') || |
| 72 | (exp[len] == '\r') || (exp[len] == ' '))) len--; |
| 73 | exp[len + 1] = 0; |
| 74 | if (len >= 0) { |
| 75 | if ((am != NULL) && (exp[0] == 't') && (exp[1] == ' ')) { |
| 76 | char *ptr = &exp[2]; |
| 77 | int from, to; |
| 78 | |
| 79 | from = scanNumber(&ptr); |
| 80 | if (*ptr != ' ') { |
| 81 | xmlGenericError(xmlGenericErrorContext, |
| 82 | "Bad line %s\n", exp); |
| 83 | break; |
| 84 | } |
| 85 | if (states[from] == NULL) |
| 86 | states[from] = xmlAutomataNewState(am); |
| 87 | ptr++; |
| 88 | to = scanNumber(&ptr); |
| 89 | if (*ptr != ' ') { |
| 90 | xmlGenericError(xmlGenericErrorContext, |
| 91 | "Bad line %s\n", exp); |
| 92 | break; |
| 93 | } |
| 94 | if (states[to] == NULL) |
| 95 | states[to] = xmlAutomataNewState(am); |
| 96 | ptr++; |
| 97 | xmlAutomataNewTransition(am, states[from], states[to], |
| 98 | BAD_CAST ptr, NULL); |
| 99 | } else if ((am != NULL) && (exp[0] == 'e') && (exp[1] == ' ')) { |
| 100 | char *ptr = &exp[2]; |
| 101 | int from, to; |
| 102 | |
| 103 | from = scanNumber(&ptr); |
| 104 | if (*ptr != ' ') { |
| 105 | xmlGenericError(xmlGenericErrorContext, |
| 106 | "Bad line %s\n", exp); |
| 107 | break; |
| 108 | } |
| 109 | if (states[from] == NULL) |
| 110 | states[from] = xmlAutomataNewState(am); |
| 111 | ptr++; |
| 112 | to = scanNumber(&ptr); |
| 113 | if (states[to] == NULL) |
| 114 | states[to] = xmlAutomataNewState(am); |
| 115 | xmlAutomataNewEpsilon(am, states[from], states[to]); |
| 116 | } else if ((am != NULL) && (exp[0] == 'f') && (exp[1] == ' ')) { |
| 117 | char *ptr = &exp[2]; |
| 118 | int state; |
| 119 | |
| 120 | state = scanNumber(&ptr); |
| 121 | if (states[state] == NULL) { |
| 122 | xmlGenericError(xmlGenericErrorContext, |
| 123 | "Bad state %d : %s\n", state, exp); |
| 124 | break; |
| 125 | } |
| 126 | xmlAutomataSetFinalState(am, states[state]); |
| 127 | } else if ((am != NULL) && (exp[0] == 'c') && (exp[1] == ' ')) { |
| 128 | char *ptr = &exp[2]; |
| 129 | int from, to; |
| 130 | int min, max; |
| 131 | |
| 132 | from = scanNumber(&ptr); |
| 133 | if (*ptr != ' ') { |
| 134 | xmlGenericError(xmlGenericErrorContext, |
| 135 | "Bad line %s\n", exp); |
| 136 | break; |
| 137 | } |
| 138 | if (states[from] == NULL) |
| 139 | states[from] = xmlAutomataNewState(am); |
| 140 | ptr++; |
| 141 | to = scanNumber(&ptr); |
| 142 | if (*ptr != ' ') { |
| 143 | xmlGenericError(xmlGenericErrorContext, |
| 144 | "Bad line %s\n", exp); |
| 145 | break; |
| 146 | } |
| 147 | if (states[to] == NULL) |
| 148 | states[to] = xmlAutomataNewState(am); |
| 149 | ptr++; |
| 150 | min = scanNumber(&ptr); |
| 151 | if (*ptr != ' ') { |
| 152 | xmlGenericError(xmlGenericErrorContext, |
| 153 | "Bad line %s\n", exp); |
| 154 | break; |
| 155 | } |
| 156 | ptr++; |
| 157 | max = scanNumber(&ptr); |
| 158 | if (*ptr != ' ') { |
| 159 | xmlGenericError(xmlGenericErrorContext, |
| 160 | "Bad line %s\n", exp); |
| 161 | break; |
| 162 | } |
| 163 | ptr++; |
| 164 | xmlAutomataNewCountTrans(am, states[from], states[to], |
| 165 | BAD_CAST ptr, min, max, NULL); |
| 166 | } else if ((am != NULL) && (exp[0] == '-') && (exp[1] == '-')) { |
| 167 | /* end of the automata */ |
| 168 | regexp = xmlAutomataCompile(am); |
| 169 | xmlFreeAutomata(am); |
| 170 | am = NULL; |
| 171 | if (regexp == NULL) { |
| 172 | xmlGenericError(xmlGenericErrorContext, |
| 173 | "Failed to compile the automata"); |
| 174 | break; |
| 175 | } |
| 176 | } else if ((exp[0] == '=') && (exp[1] == '>')) { |
| 177 | if (regexp == NULL) { |
| 178 | printf("=> failed not compiled\n"); |
| 179 | } else { |
| 180 | if (exec == NULL) |
| 181 | exec = xmlRegNewExecCtxt(regexp, NULL, NULL); |
| 182 | if (ret == 0) { |
| 183 | ret = xmlRegExecPushString(exec, NULL, NULL); |
| 184 | } |
| 185 | if (ret == 1) |
| 186 | printf("=> Passed\n"); |
| 187 | else if ((ret == 0) || (ret == -1)) |
| 188 | printf("=> Failed\n"); |
| 189 | else if (ret < 0) |
| 190 | printf("=> Error\n"); |
| 191 | xmlRegFreeExecCtxt(exec); |
| 192 | exec = NULL; |
| 193 | } |
| 194 | ret = 0; |
| 195 | } else if (regexp != NULL) { |
| 196 | if (exec == NULL) |
| 197 | exec = xmlRegNewExecCtxt(regexp, NULL, NULL); |
| 198 | ret = xmlRegExecPushString(exec, BAD_CAST exp, NULL); |
| 199 | } else { |
| 200 | xmlGenericError(xmlGenericErrorContext, |
| 201 | "Unexpected line %s\n", exp); |
| 202 | } |
| 203 | } |
| 204 | } |
| 205 | fclose(input); |
| 206 | if (regexp != NULL) |
| 207 | xmlRegFreeRegexp(regexp); |
| 208 | if (exec != NULL) |
| 209 | xmlRegFreeExecCtxt(exec); |
| 210 | if (am != NULL) |
| 211 | xmlFreeAutomata(am); |
| 212 | } |
| 213 | |
| 214 | int main(int argc, char **argv) { |
| 215 | |
| 216 | xmlInitMemory(); |
| 217 | |
| 218 | if (argc == 1) { |
| 219 | int ret; |
| 220 | xmlAutomataPtr am; |
| 221 | xmlAutomataStatePtr start, cur; |
| 222 | xmlRegexpPtr regexp; |
| 223 | xmlRegExecCtxtPtr exec; |
| 224 | |
| 225 | am = xmlNewAutomata(); |
| 226 | start = xmlAutomataGetInitState(am); |
| 227 | |
| 228 | /* generate a[ba]*a */ |
| 229 | cur = xmlAutomataNewTransition(am, start, NULL, BAD_CAST"a", NULL); |
| 230 | xmlAutomataNewTransition(am, cur, cur, BAD_CAST"b", NULL); |
| 231 | xmlAutomataNewTransition(am, cur, cur, BAD_CAST"a", NULL); |
| 232 | cur = xmlAutomataNewCountTrans(am, cur, NULL, BAD_CAST"a", 2, 3, NULL); |
| 233 | xmlAutomataSetFinalState(am, cur); |
| 234 | |
| 235 | /* compile it in a regexp and free the automata */ |
| 236 | regexp = xmlAutomataCompile(am); |
| 237 | xmlFreeAutomata(am); |
| 238 | |
| 239 | /* test the regexp */ |
| 240 | xmlRegexpPrint(stdout, regexp); |
| 241 | exec = xmlRegNewExecCtxt(regexp, NULL, NULL); |
| 242 | ret = xmlRegExecPushString(exec, BAD_CAST"a", NULL); |
| 243 | if (ret == 1) |
| 244 | printf("final\n"); |
| 245 | else if (ret < 0) |
| 246 | printf("error\n"); |
| 247 | ret =xmlRegExecPushString(exec, BAD_CAST"a", NULL); |
| 248 | if (ret == 1) |
| 249 | printf("final\n"); |
| 250 | else if (ret < 0) |
| 251 | printf("error\n"); |
| 252 | ret =xmlRegExecPushString(exec, BAD_CAST"b", NULL); |
| 253 | if (ret == 1) |
| 254 | printf("final\n"); |
| 255 | else if (ret < 0) |
| 256 | printf("error\n"); |
| 257 | ret =xmlRegExecPushString(exec, BAD_CAST"a", NULL); |
| 258 | if (ret == 1) |
| 259 | printf("final\n"); |
| 260 | else if (ret < 0) |
| 261 | printf("error\n"); |
| 262 | ret =xmlRegExecPushString(exec, BAD_CAST"a", NULL); |
| 263 | if (ret == 1) |
| 264 | printf("final\n"); |
| 265 | else if (ret < 0) |
| 266 | printf("error\n"); |
| 267 | ret =xmlRegExecPushString(exec, BAD_CAST"a", NULL); |
| 268 | if (ret == 1) |
| 269 | printf("final\n"); |
| 270 | else if (ret < 0) |
| 271 | printf("error\n"); |
| 272 | ret =xmlRegExecPushString(exec, BAD_CAST"a", NULL); |
| 273 | if (ret == 1) |
| 274 | printf("final\n"); |
| 275 | else if (ret < 0) |
| 276 | printf("error\n"); |
| 277 | if (ret == 0) { |
| 278 | ret = xmlRegExecPushString(exec, NULL, NULL); |
| 279 | if (ret == 1) |
| 280 | printf("final\n"); |
| 281 | else if (ret < 0) |
| 282 | printf("error\n"); |
| 283 | } |
| 284 | xmlRegFreeExecCtxt(exec); |
| 285 | |
| 286 | /* free the regexp */ |
| 287 | xmlRegFreeRegexp(regexp); |
| 288 | } else { |
| 289 | int i; |
| 290 | |
| 291 | for (i = 1;i < argc;i++) |
| 292 | testRegexpFile(argv[i]); |
| 293 | } |
| 294 | |
| 295 | xmlCleanupParser(); |
| 296 | xmlMemoryDump(); |
| 297 | return(0); |
| 298 | } |
| 299 | |
| 300 | #else |
| 301 | #include <stdio.h> |
| 302 | int main(int argc, char **argv) { |
| 303 | printf("%s : Automata support not compiled in\n", argv[0]); |
| 304 | return(0); |
| 305 | } |
| 306 | #endif /* LIBXML_AUTOMATA_ENABLED */ |