1 | /*
|
---|
2 | * testRegexp.c: simple module for testing regular expressions
|
---|
3 | *
|
---|
4 | * See Copyright for the status of this software.
|
---|
5 | *
|
---|
6 | * Daniel Veillard <[email protected]>
|
---|
7 | */
|
---|
8 |
|
---|
9 | #include "libxml.h"
|
---|
10 | #ifdef LIBXML_REGEXP_ENABLED
|
---|
11 | #include <string.h>
|
---|
12 |
|
---|
13 | #include <libxml/tree.h>
|
---|
14 | #include <libxml/xmlregexp.h>
|
---|
15 |
|
---|
16 | static int repeat = 0;
|
---|
17 | static int debug = 0;
|
---|
18 |
|
---|
19 | static void testRegexp(xmlRegexpPtr comp, const char *value) {
|
---|
20 | int ret;
|
---|
21 |
|
---|
22 | ret = xmlRegexpExec(comp, (const xmlChar *) value);
|
---|
23 | if (ret == 1)
|
---|
24 | printf("%s: Ok\n", value);
|
---|
25 | else if (ret == 0)
|
---|
26 | printf("%s: Fail\n", value);
|
---|
27 | else
|
---|
28 | printf("%s: Error: %d\n", value, ret);
|
---|
29 | if (repeat) {
|
---|
30 | int j;
|
---|
31 | for (j = 0;j < 999999;j++)
|
---|
32 | xmlRegexpExec(comp, (const xmlChar *) value);
|
---|
33 | }
|
---|
34 | }
|
---|
35 |
|
---|
36 | static void
|
---|
37 | testRegexpFile(const char *filename) {
|
---|
38 | xmlRegexpPtr comp = NULL;
|
---|
39 | FILE *input;
|
---|
40 | char expression[5000];
|
---|
41 | int len;
|
---|
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 | while (fgets(expression, 4500, input) != NULL) {
|
---|
50 | len = strlen(expression);
|
---|
51 | len--;
|
---|
52 | while ((len >= 0) &&
|
---|
53 | ((expression[len] == '\n') || (expression[len] == '\t') ||
|
---|
54 | (expression[len] == '\r') || (expression[len] == ' '))) len--;
|
---|
55 | expression[len + 1] = 0;
|
---|
56 | if (len >= 0) {
|
---|
57 | if (expression[0] == '#')
|
---|
58 | continue;
|
---|
59 | if ((expression[0] == '=') && (expression[1] == '>')) {
|
---|
60 | char *pattern = &expression[2];
|
---|
61 |
|
---|
62 | if (comp != NULL) {
|
---|
63 | xmlRegFreeRegexp(comp);
|
---|
64 | comp = NULL;
|
---|
65 | }
|
---|
66 | printf("Regexp: %s\n", pattern) ;
|
---|
67 | comp = xmlRegexpCompile((const xmlChar *) pattern);
|
---|
68 | if (comp == NULL) {
|
---|
69 | printf(" failed to compile\n");
|
---|
70 | break;
|
---|
71 | }
|
---|
72 | } else if (comp == NULL) {
|
---|
73 | printf("Regexp: %s\n", expression) ;
|
---|
74 | comp = xmlRegexpCompile((const xmlChar *) expression);
|
---|
75 | if (comp == NULL) {
|
---|
76 | printf(" failed to compile\n");
|
---|
77 | break;
|
---|
78 | }
|
---|
79 | } else if (comp != NULL) {
|
---|
80 | testRegexp(comp, expression);
|
---|
81 | }
|
---|
82 | }
|
---|
83 | }
|
---|
84 | fclose(input);
|
---|
85 | if (comp != NULL)
|
---|
86 | xmlRegFreeRegexp(comp);
|
---|
87 | }
|
---|
88 |
|
---|
89 | #ifdef LIBXML_EXPR_ENABLED
|
---|
90 | static void
|
---|
91 | runFileTest(xmlExpCtxtPtr ctxt, const char *filename) {
|
---|
92 | xmlExpNodePtr expr = NULL, sub;
|
---|
93 | FILE *input;
|
---|
94 | char expression[5000];
|
---|
95 | int len;
|
---|
96 |
|
---|
97 | input = fopen(filename, "r");
|
---|
98 | if (input == NULL) {
|
---|
99 | xmlGenericError(xmlGenericErrorContext,
|
---|
100 | "Cannot open %s for reading\n", filename);
|
---|
101 | return;
|
---|
102 | }
|
---|
103 | while (fgets(expression, 4500, input) != NULL) {
|
---|
104 | len = strlen(expression);
|
---|
105 | len--;
|
---|
106 | while ((len >= 0) &&
|
---|
107 | ((expression[len] == '\n') || (expression[len] == '\t') ||
|
---|
108 | (expression[len] == '\r') || (expression[len] == ' '))) len--;
|
---|
109 | expression[len + 1] = 0;
|
---|
110 | if (len >= 0) {
|
---|
111 | if (expression[0] == '#')
|
---|
112 | continue;
|
---|
113 | if ((expression[0] == '=') && (expression[1] == '>')) {
|
---|
114 | char *str = &expression[2];
|
---|
115 |
|
---|
116 | if (expr != NULL) {
|
---|
117 | xmlExpFree(ctxt, expr);
|
---|
118 | if (xmlExpCtxtNbNodes(ctxt) != 0)
|
---|
119 | printf(" Parse/free of Expression leaked %d\n",
|
---|
120 | xmlExpCtxtNbNodes(ctxt));
|
---|
121 | expr = NULL;
|
---|
122 | }
|
---|
123 | printf("Expression: %s\n", str) ;
|
---|
124 | expr = xmlExpParse(ctxt, str);
|
---|
125 | if (expr == NULL) {
|
---|
126 | printf(" parsing Failed\n");
|
---|
127 | break;
|
---|
128 | }
|
---|
129 | } else if (expr != NULL) {
|
---|
130 | int expect = -1;
|
---|
131 | int nodes1, nodes2;
|
---|
132 |
|
---|
133 | if (expression[0] == '0')
|
---|
134 | expect = 0;
|
---|
135 | if (expression[0] == '1')
|
---|
136 | expect = 1;
|
---|
137 | printf("Subexp: %s", expression + 2) ;
|
---|
138 | nodes1 = xmlExpCtxtNbNodes(ctxt);
|
---|
139 | sub = xmlExpParse(ctxt, expression + 2);
|
---|
140 | if (sub == NULL) {
|
---|
141 | printf(" parsing Failed\n");
|
---|
142 | break;
|
---|
143 | } else {
|
---|
144 | int ret;
|
---|
145 |
|
---|
146 | nodes2 = xmlExpCtxtNbNodes(ctxt);
|
---|
147 | ret = xmlExpSubsume(ctxt, expr, sub);
|
---|
148 |
|
---|
149 | if ((expect == 1) && (ret == 1)) {
|
---|
150 | printf(" => accept, Ok\n");
|
---|
151 | } else if ((expect == 0) && (ret == 0)) {
|
---|
152 | printf(" => reject, Ok\n");
|
---|
153 | } else if ((expect == 1) && (ret == 0)) {
|
---|
154 | printf(" => reject, Failed\n");
|
---|
155 | } else if ((expect == 0) && (ret == 1)) {
|
---|
156 | printf(" => accept, Failed\n");
|
---|
157 | } else {
|
---|
158 | printf(" => fail internally\n");
|
---|
159 | }
|
---|
160 | if (xmlExpCtxtNbNodes(ctxt) > nodes2) {
|
---|
161 | printf(" Subsume leaked %d\n",
|
---|
162 | xmlExpCtxtNbNodes(ctxt) - nodes2);
|
---|
163 | nodes1 += xmlExpCtxtNbNodes(ctxt) - nodes2;
|
---|
164 | }
|
---|
165 | xmlExpFree(ctxt, sub);
|
---|
166 | if (xmlExpCtxtNbNodes(ctxt) > nodes1) {
|
---|
167 | printf(" Parse/free leaked %d\n",
|
---|
168 | xmlExpCtxtNbNodes(ctxt) - nodes1);
|
---|
169 | }
|
---|
170 | }
|
---|
171 |
|
---|
172 | }
|
---|
173 | }
|
---|
174 | }
|
---|
175 | if (expr != NULL) {
|
---|
176 | xmlExpFree(ctxt, expr);
|
---|
177 | if (xmlExpCtxtNbNodes(ctxt) != 0)
|
---|
178 | printf(" Parse/free of Expression leaked %d\n",
|
---|
179 | xmlExpCtxtNbNodes(ctxt));
|
---|
180 | }
|
---|
181 | fclose(input);
|
---|
182 | }
|
---|
183 |
|
---|
184 | static void
|
---|
185 | testReduce(xmlExpCtxtPtr ctxt, xmlExpNodePtr expr, const char *tst) {
|
---|
186 | xmlBufferPtr xmlExpBuf;
|
---|
187 | xmlExpNodePtr sub, deriv;
|
---|
188 | xmlExpBuf = xmlBufferCreate();
|
---|
189 |
|
---|
190 | sub = xmlExpParse(ctxt, tst);
|
---|
191 | if (sub == NULL) {
|
---|
192 | printf("Subset %s failed to parse\n", tst);
|
---|
193 | return;
|
---|
194 | }
|
---|
195 | xmlExpDump(xmlExpBuf, sub);
|
---|
196 | printf("Subset parsed as: %s\n",
|
---|
197 | (const char *) xmlBufferContent(xmlExpBuf));
|
---|
198 | deriv = xmlExpExpDerive(ctxt, expr, sub);
|
---|
199 | if (deriv == NULL) {
|
---|
200 | printf("Derivation led to an internal error, report this !\n");
|
---|
201 | return;
|
---|
202 | } else {
|
---|
203 | xmlBufferEmpty(xmlExpBuf);
|
---|
204 | xmlExpDump(xmlExpBuf, deriv);
|
---|
205 | if (xmlExpIsNillable(deriv))
|
---|
206 | printf("Resulting nillable derivation: %s\n",
|
---|
207 | (const char *) xmlBufferContent(xmlExpBuf));
|
---|
208 | else
|
---|
209 | printf("Resulting derivation: %s\n",
|
---|
210 | (const char *) xmlBufferContent(xmlExpBuf));
|
---|
211 | xmlExpFree(ctxt, deriv);
|
---|
212 | }
|
---|
213 | xmlExpFree(ctxt, sub);
|
---|
214 | }
|
---|
215 |
|
---|
216 | static void
|
---|
217 | exprDebug(xmlExpCtxtPtr ctxt, xmlExpNodePtr expr) {
|
---|
218 | xmlBufferPtr xmlExpBuf;
|
---|
219 | xmlExpNodePtr deriv;
|
---|
220 | const char *list[40];
|
---|
221 | int ret;
|
---|
222 |
|
---|
223 | xmlExpBuf = xmlBufferCreate();
|
---|
224 |
|
---|
225 | if (expr == NULL) {
|
---|
226 | printf("Failed to parse\n");
|
---|
227 | return;
|
---|
228 | }
|
---|
229 | xmlExpDump(xmlExpBuf, expr);
|
---|
230 | printf("Parsed as: %s\n", (const char *) xmlBufferContent(xmlExpBuf));
|
---|
231 | printf("Max token input = %d\n", xmlExpMaxToken(expr));
|
---|
232 | if (xmlExpIsNillable(expr) == 1)
|
---|
233 | printf("Is nillable\n");
|
---|
234 | ret = xmlExpGetLanguage(ctxt, expr, (const xmlChar **) &list[0], 40);
|
---|
235 | if (ret < 0)
|
---|
236 | printf("Failed to get list: %d\n", ret);
|
---|
237 | else {
|
---|
238 | int i;
|
---|
239 |
|
---|
240 | printf("Language has %d strings, testing string derivations\n", ret);
|
---|
241 | for (i = 0;i < ret;i++) {
|
---|
242 | deriv = xmlExpStringDerive(ctxt, expr, BAD_CAST list[i], -1);
|
---|
243 | if (deriv == NULL) {
|
---|
244 | printf(" %s -> derivation failed\n", list[i]);
|
---|
245 | } else {
|
---|
246 | xmlBufferEmpty(xmlExpBuf);
|
---|
247 | xmlExpDump(xmlExpBuf, deriv);
|
---|
248 | printf(" %s -> %s\n", list[i],
|
---|
249 | (const char *) xmlBufferContent(xmlExpBuf));
|
---|
250 | }
|
---|
251 | xmlExpFree(ctxt, deriv);
|
---|
252 | }
|
---|
253 | }
|
---|
254 | xmlBufferFree(xmlExpBuf);
|
---|
255 | }
|
---|
256 | #endif
|
---|
257 |
|
---|
258 | static void usage(const char *name) {
|
---|
259 | fprintf(stderr, "Usage: %s [flags]\n", name);
|
---|
260 | fprintf(stderr, "Testing tool for libxml2 string and pattern regexps\n");
|
---|
261 | fprintf(stderr, " --debug: switch on debugging\n");
|
---|
262 | fprintf(stderr, " --repeat: loop on the operation\n");
|
---|
263 | #ifdef LIBXML_EXPR_ENABLED
|
---|
264 | fprintf(stderr, " --expr: test xmlExp and not xmlRegexp\n");
|
---|
265 | #endif
|
---|
266 | fprintf(stderr, " --input filename: use the given filename for regexp\n");
|
---|
267 | fprintf(stderr, " --input filename: use the given filename for exp\n");
|
---|
268 | }
|
---|
269 |
|
---|
270 | int main(int argc, char **argv) {
|
---|
271 | xmlRegexpPtr comp = NULL;
|
---|
272 | #ifdef LIBXML_EXPR_ENABLED
|
---|
273 | xmlExpNodePtr expr = NULL;
|
---|
274 | int use_exp = 0;
|
---|
275 | xmlExpCtxtPtr ctxt = NULL;
|
---|
276 | #endif
|
---|
277 | const char *pattern = NULL;
|
---|
278 | char *filename = NULL;
|
---|
279 | int i;
|
---|
280 |
|
---|
281 | xmlInitMemory();
|
---|
282 |
|
---|
283 | if (argc <= 1) {
|
---|
284 | usage(argv[0]);
|
---|
285 | return(1);
|
---|
286 | }
|
---|
287 | for (i = 1; i < argc ; i++) {
|
---|
288 | if (!strcmp(argv[i], "-"))
|
---|
289 | break;
|
---|
290 |
|
---|
291 | if (argv[i][0] != '-')
|
---|
292 | continue;
|
---|
293 | if (!strcmp(argv[i], "--"))
|
---|
294 | break;
|
---|
295 |
|
---|
296 | if ((!strcmp(argv[i], "-debug")) || (!strcmp(argv[i], "--debug"))) {
|
---|
297 | debug++;
|
---|
298 | } else if ((!strcmp(argv[i], "-repeat")) ||
|
---|
299 | (!strcmp(argv[i], "--repeat"))) {
|
---|
300 | repeat++;
|
---|
301 | #ifdef LIBXML_EXPR_ENABLED
|
---|
302 | } else if ((!strcmp(argv[i], "-expr")) ||
|
---|
303 | (!strcmp(argv[i], "--expr"))) {
|
---|
304 | use_exp++;
|
---|
305 | #endif
|
---|
306 | } else if ((!strcmp(argv[i], "-i")) || (!strcmp(argv[i], "-f")) ||
|
---|
307 | (!strcmp(argv[i], "--input")))
|
---|
308 | filename = argv[++i];
|
---|
309 | else {
|
---|
310 | fprintf(stderr, "Unknown option %s\n", argv[i]);
|
---|
311 | usage(argv[0]);
|
---|
312 | }
|
---|
313 | }
|
---|
314 |
|
---|
315 | #ifdef LIBXML_EXPR_ENABLED
|
---|
316 | if (use_exp)
|
---|
317 | ctxt = xmlExpNewCtxt(0, NULL);
|
---|
318 | #endif
|
---|
319 |
|
---|
320 | if (filename != NULL) {
|
---|
321 | #ifdef LIBXML_EXPR_ENABLED
|
---|
322 | if (use_exp)
|
---|
323 | runFileTest(ctxt, filename);
|
---|
324 | else
|
---|
325 | #endif
|
---|
326 | testRegexpFile(filename);
|
---|
327 | } else {
|
---|
328 | int data = 0;
|
---|
329 | #ifdef LIBXML_EXPR_ENABLED
|
---|
330 |
|
---|
331 | if (use_exp) {
|
---|
332 | for (i = 1; i < argc ; i++) {
|
---|
333 | if (strcmp(argv[i], "--") == 0)
|
---|
334 | data = 1;
|
---|
335 | else if ((argv[i][0] != '-') || (strcmp(argv[i], "-") == 0) ||
|
---|
336 | (data == 1)) {
|
---|
337 | if (pattern == NULL) {
|
---|
338 | pattern = argv[i];
|
---|
339 | printf("Testing expr %s:\n", pattern);
|
---|
340 | expr = xmlExpParse(ctxt, pattern);
|
---|
341 | if (expr == NULL) {
|
---|
342 | printf(" failed to compile\n");
|
---|
343 | break;
|
---|
344 | }
|
---|
345 | if (debug) {
|
---|
346 | exprDebug(ctxt, expr);
|
---|
347 | }
|
---|
348 | } else {
|
---|
349 | testReduce(ctxt, expr, argv[i]);
|
---|
350 | }
|
---|
351 | }
|
---|
352 | }
|
---|
353 | if (expr != NULL) {
|
---|
354 | xmlExpFree(ctxt, expr);
|
---|
355 | expr = NULL;
|
---|
356 | }
|
---|
357 | } else
|
---|
358 | #endif
|
---|
359 | {
|
---|
360 | for (i = 1; i < argc ; i++) {
|
---|
361 | if (strcmp(argv[i], "--") == 0)
|
---|
362 | data = 1;
|
---|
363 | else if ((argv[i][0] != '-') || (strcmp(argv[i], "-") == 0) ||
|
---|
364 | (data == 1)) {
|
---|
365 | if (pattern == NULL) {
|
---|
366 | pattern = argv[i];
|
---|
367 | printf("Testing %s:\n", pattern);
|
---|
368 | comp = xmlRegexpCompile((const xmlChar *) pattern);
|
---|
369 | if (comp == NULL) {
|
---|
370 | printf(" failed to compile\n");
|
---|
371 | break;
|
---|
372 | }
|
---|
373 | if (debug)
|
---|
374 | xmlRegexpPrint(stdout, comp);
|
---|
375 | } else {
|
---|
376 | testRegexp(comp, argv[i]);
|
---|
377 | }
|
---|
378 | }
|
---|
379 | }
|
---|
380 | if (comp != NULL)
|
---|
381 | xmlRegFreeRegexp(comp);
|
---|
382 | }
|
---|
383 | }
|
---|
384 | #ifdef LIBXML_EXPR_ENABLED
|
---|
385 | if (ctxt != NULL) {
|
---|
386 | printf("Ops: %d nodes, %d cons\n",
|
---|
387 | xmlExpCtxtNbNodes(ctxt), xmlExpCtxtNbCons(ctxt));
|
---|
388 | xmlExpFreeCtxt(ctxt);
|
---|
389 | }
|
---|
390 | #endif
|
---|
391 | xmlCleanupParser();
|
---|
392 | xmlMemoryDump();
|
---|
393 | return(0);
|
---|
394 | }
|
---|
395 |
|
---|
396 | #else
|
---|
397 | #include <stdio.h>
|
---|
398 | int main(int argc ATTRIBUTE_UNUSED, char **argv ATTRIBUTE_UNUSED) {
|
---|
399 | printf("%s : Regexp support not compiled in\n", argv[0]);
|
---|
400 | return(0);
|
---|
401 | }
|
---|
402 | #endif /* LIBXML_REGEXP_ENABLED */
|
---|