2 This file is part of GNUnet
3 (C) 2012 Christian Grothoff (and other contributing authors)
5 GNUnet is free software; you can redistribute it and/or modify
6 it under the terms of the GNU General Public License as published
7 by the Free Software Foundation; either version 3, or (at your
8 option) any later version.
10 GNUnet is distributed in the hope that it will be useful, but
11 WITHOUT ANY WARRANTY; without even the implied warranty of
12 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 General Public License for more details.
15 You should have received a copy of the GNU General Public License
16 along with GNUnet; see the file COPYING. If not, write to the
17 Free Software Foundation, Inc., 59 Temple Place - Suite 330,
18 Boston, MA 02111-1307, USA.
21 * @file regex/test_regex_eval_api.c
22 * @brief test for regex.c
23 * @author Maximilian Szengel
28 #include "regex_internal_lib.h"
29 #include "regex_test_lib.h"
30 #include "regex_internal.h"
38 struct Regex_String_Pair
43 enum Match_Result expected_results[20];
48 * Random regex test. Generate a random regex as well as 'str_count' strings to
49 * match it against. Will match using GNUNET_REGEX implementation and compare
50 * the result to glibc regex result. 'rx_length' has to be smaller then
53 * @param rx_length length of the regular expression.
54 * @param max_str_len maximum length of the random strings.
55 * @param str_count number of generated random strings.
57 * @return 0 on success, non 0 otherwise.
60 test_random (unsigned int rx_length, unsigned int max_str_len,
61 unsigned int str_count)
69 int eval_canonical_check;
70 struct REGEX_INTERNAL_Automaton *dfa;
72 regmatch_t matchptr[1];
75 char *canonical_regex = NULL;
77 /* At least one string is needed for matching */
78 GNUNET_assert (str_count > 0);
79 /* The string should be at least as long as the regex itself */
80 GNUNET_assert (max_str_len >= rx_length);
82 /* Generate random regex and a string that matches the regex */
83 matching_str = GNUNET_malloc (rx_length + 1);
84 rand_rx = REGEX_TEST_generate_random_regex (rx_length, matching_str);
88 for (i = 0; i < str_count; i++)
92 matching_str = REGEX_TEST_generate_random_string (max_str_len);
95 /* Match string using DFA */
96 dfa = REGEX_INTERNAL_construct_dfa (rand_rx, strlen (rand_rx), 0);
99 GNUNET_log (GNUNET_ERROR_TYPE_ERROR, "Constructing DFA failed\n");
103 eval = REGEX_INTERNAL_eval (dfa, matching_str);
104 /* save the canonical regex for later comparison */
105 canonical_regex = GNUNET_strdup (REGEX_INTERNAL_get_canonical_regex (dfa));
106 REGEX_INTERNAL_automaton_destroy (dfa);
108 /* Match string using glibc regex */
109 if (0 != regcomp (&rx, rand_rx, REG_EXTENDED))
111 GNUNET_log (GNUNET_ERROR_TYPE_ERROR,
112 "Could not compile regex using regcomp: %s\n", rand_rx);
116 eval_check = regexec (&rx, matching_str, 1, matchptr, 0);
119 /* We only want to match the whole string, because that's what our DFA does,
121 if (eval_check == 0 &&
122 (matchptr[0].rm_so != 0 || matchptr[0].rm_eo != strlen (matching_str)))
125 /* Match canonical regex */
127 REGEX_INTERNAL_construct_dfa (canonical_regex, strlen (canonical_regex),
131 GNUNET_log (GNUNET_ERROR_TYPE_ERROR, "Constructing DFA failed\n");
135 eval_canonical = REGEX_INTERNAL_eval (dfa, matching_str);
136 REGEX_INTERNAL_automaton_destroy (dfa);
138 if (0 != regcomp (&rx, canonical_regex, REG_EXTENDED))
140 GNUNET_log (GNUNET_ERROR_TYPE_ERROR,
141 "Could not compile regex using regcomp: %s\n",
146 eval_canonical_check = regexec (&rx, matching_str, 1, matchptr, 0);
149 /* We only want to match the whole string, because that's what our DFA does,
151 if (eval_canonical_check == 0 &&
152 (matchptr[0].rm_so != 0 || matchptr[0].rm_eo != strlen (matching_str)))
153 eval_canonical_check = 1;
155 /* compare results */
156 if (eval_check != eval || eval_canonical != eval_canonical_check)
158 regerror (eval_check, &rx, error, sizeof error);
159 GNUNET_log (GNUNET_ERROR_TYPE_ERROR, "Unexpected result:\nregex: %s\ncanonical_regex: %s\n\
160 string: %s\ngnunet regex: %i\nglibc regex: %i\n\
161 canonical regex: %i\ncanonical regex glibc: %i\n\
162 glibc error: %s\n\n", rand_rx, canonical_regex, matching_str,
163 eval, eval_check, eval_canonical, eval_canonical_check, error);
166 GNUNET_free (canonical_regex);
167 GNUNET_free (matching_str);
168 canonical_regex = NULL;
172 GNUNET_free (rand_rx);
177 GNUNET_free_non_null (matching_str);
178 GNUNET_free_non_null (rand_rx);
179 GNUNET_free_non_null (canonical_regex);
184 * Automaton test that compares the result of matching regular expression 'rx'
185 * with the strings and expected results in 'rxstr' with the result of matching
186 * the same strings with glibc regex.
188 * @param a automaton.
189 * @param rx compiled glibc regex.
190 * @param rxstr regular expression and strings with expected results to
193 * @return 0 on successfull, non 0 otherwise
196 test_automaton (struct REGEX_INTERNAL_Automaton *a, regex_t * rx,
197 struct Regex_String_Pair *rxstr)
203 regmatch_t matchptr[1];
208 GNUNET_log (GNUNET_ERROR_TYPE_ERROR, "Automaton was NULL\n");
214 for (i = 0; i < rxstr->string_count; i++)
216 eval = REGEX_INTERNAL_eval (a, rxstr->strings[i]);
217 eval_check = regexec (rx, rxstr->strings[i], 1, matchptr, 0);
219 /* We only want to match the whole string, because that's what our DFA does,
221 if (eval_check == 0 &&
222 (matchptr[0].rm_so != 0 ||
223 matchptr[0].rm_eo != strlen (rxstr->strings[i])))
226 if ((rxstr->expected_results[i] == match && (0 != eval || 0 != eval_check))
227 || (rxstr->expected_results[i] == nomatch &&
228 (0 == eval || 0 == eval_check)))
231 regerror (eval_check, rx, error, sizeof error);
232 GNUNET_log (GNUNET_ERROR_TYPE_ERROR,
233 "Unexpected result:\nregex: %s\ncanonical_regex: %s\n"
234 "string: %s\nexpected result: %i\n"
235 "gnunet regex: %i\nglibc regex: %i\nglibc error: %s\n"
236 "rm_so: %i\nrm_eo: %i\n\n", rxstr->regex,
237 REGEX_INTERNAL_get_canonical_regex (a), rxstr->strings[i],
238 rxstr->expected_results[i], eval, eval_check, error,
239 matchptr[0].rm_so, matchptr[0].rm_eo);
246 main (int argc, char *argv[])
248 GNUNET_log_setup ("test-regex", "WARNING", NULL);
250 struct REGEX_INTERNAL_Automaton *a;
258 struct Regex_String_Pair rxstr[19] = {
260 {"ababcd", "abab", "aabcd", "a", "abb"},
261 {match, nomatch, match, match, nomatch}},
262 {"ab(c|d)+c*(a(b|c)d)+", 5,
263 {"abcdcdcdcdddddabd", "abcd",
264 "abcddddddccccccccccccccccccccccccabdacdabd",
265 "abccccca", "abcdcdcdccdabdabd"},
266 {match, nomatch, match, nomatch, match}},
267 {"ab+c*(a(bx|c)d)+", 5,
268 {"abcdcdcdcdddddabd", "abcd",
269 "abcddddddccccccccccccccccccccccccabdacdabd",
270 "abccccca", "abcdcdcdccdabdabd"},
271 {nomatch, nomatch, nomatch, nomatch, nomatch}},
272 {"a+X*y+c|p|R|Z*K*y*R+w|Y*6+n+h*k*w+V*F|W*B*e*", 1,
273 {"kaXycQepRZKyRwY6nhkwVFWBegNVtLPj39XhJJ6bEifRSZRYZg"},
275 {"k|a+X*y+c|Q*e|p|R|Z*K*y*R+w|Y*6+n+h*k*w+V*F|W*B*e*g|N+V|t+L|P*j*3*9+X*h*J|J*6|b|E*i*f*R+S|Z|R|Y*Z|g*", 1,
276 {"kaXycQepRZKyRwY6nhkwVFWBegNVtLPj39XhJJ6bEifRSZRYZg"},
278 {"F?W+m+2*6*c*s|P?U?a|B|y*i+t+A|V|6*C*7*e?Z*n*i|J?5+g?W*V?7*j?p?1|r?B?C+E+3+6*i+W*P?K?0|D+7?y*m+3?g?K?", 1,
279 {"osfjsodfonONONOnosndfsdnfsd"},
281 {"V|M*o?x*p*d+h+b|E*m?h?Y*E*O?W*W*P+o?Z+H*M|I*q+C*a+5?5*9|b?z|G*y*k?R|p+u|8*h?B+l*H|e|L*O|1|F?v*0?5|C+", 1,
282 {"VMoxpdhbEmhYEOWWPoZHMIqCa559bzGykRpu8hBlHeLO1Fv05C"},
285 {"", "bla", "blabla", "bl", "la", "b", "l", "a"},
286 {match, match, match, nomatch, nomatch, nomatch, nomatch, nomatch}},
287 {"ab(c|d)+c*(a(b|c)+d)+(bla)(bla)*", 8,
288 {"ab", "abcabdbla", "abdcccccccccccabcbccdblablabla", "bl", "la", "b",
291 {nomatch, match, match, nomatch, nomatch, nomatch, nomatch, nomatch}},
293 {"", "a", "aa", "aaa", "aaaa", "aaaaa"},
294 {nomatch, match, match, match, match, match}},
295 {"ab(c|d)+c*(a(b|c)+d)+(bla)+", 1,
299 {"b", "bb", "ac", "", "acb", "bacbacac", "acacac", "abc"},
300 {match, match, match, nomatch, match, match, match, nomatch}},
302 {"", "ab", "c", "abc", "ababcc", "acc", "abac"},
303 {nomatch, match, match, match, match, nomatch, nomatch}},
304 {"((j|2j)K|(j|2j)AK|(j|2j)(D|e|(j|2j)A(D|e))D*K)", 1,
305 {"", "2j2jADK", "j2jADK"},
306 {nomatch, match, match}},
307 {"((j|2j)K|(j|2j)(D|e|((j|2j)j|(j|2j)2j)A(D|e))D*K|(j|2j)AK)", 2,
308 {"", "2j2jjADK", "j2jADK"},
309 {nomatch, match, match}},
310 {"ab(c|d)+c*(a(b|c)d)+", 1,
319 {"GNVPN-0001-PAD(001110101001001010(0|1)*|001110101001001010000(0|1)*|001110101001001010001(0|1)*|001110101001001010010(0|1)*|001110101001001010011(0|1)*|001110101001001010100(0|1)*|001110101001001010101(0|1)*|001110101001001010110(0|1)*|001110101001001010111(0|1)*|0011101010110110(0|1)*|001110101011011000000(0|1)*|001110101011011000001(0|1)*|001110101011011000010(0|1)*|001110101011011000011(0|1)*|001110101011011000100(0|1)*|001110101011011000101(0|1)*|001110101011011000110(0|1)*|001110101011011000111(0|1)*|001110101011011001000(0|1)*|001110101011011001001(0|1)*|001110101011011001010(0|1)*|001110101011011001011(0|1)*|001110101011011001100(0|1)*|001110101011011001101(0|1)*|001110101011011001110(0|1)*|001110101011011001111(0|1)*|001110101011011010000(0|1)*|001110101011011010001(0|1)*|001110101011011010010(0|1)*|001110101011011010011(0|1)*|001110101011011010100(0|1)*|001110101011011010101(0|1)*|001110101011011010110(0|1)*|001110101011011010111(0|1)*|001110101011011011000(0|1)*|001110101011011011001(0|1)*|001110101011011011010(0|1)*|001110101011011011011(0|1)*|001110101011011011100(0|1)*|001110101011011011101(0|1)*|001110101011011011110(0|1)*|001110101011011011111(0|1)*|0011101110111101(0|1)*|001110111011110100000(0|1)*|001110111011110100001(0|1)*|001110111011110100010(0|1)*|001110111011110100011(0|1)*|001110111011110100100(0|1)*|001110111011110100101(0|1)*|001110111011110100110(0|1)*|001110111011110100111(0|1)*|001110111011110101000(0|1)*|001110111011110101001(0|1)*|001110111011110101010(0|1)*|001110111011110101011(0|1)*|001110111011110101100(0|1)*|001110111011110101101(0|1)*|001110111011110101110(0|1)*|001110111011110101111(0|1)*|001110111011110110000(0|1)*|001110111011110110001(0|1)*|001110111011110110010(0|1)*|001110111011110110011(0|1)*|001110111011110110100(0|1)*|001110111011110110101(0|1)*|001110111011110110110(0|1)*|001110111011110110111(0|1)*|001110111011110111000(0|1)*|001110111011110111001(0|1)*|001110111011110111010(0|1)*|001110111011110111011(0|1)*|001110111011110111100(0|1)*|001110111011110111101(0|1)*|001110111011110111110(0|1)*|0111010001010110(0|1)*|011101000101011000000(0|1)*|011101000101011000001(0|1)*|011101000101011000010(0|1)*|011101000101011000011(0|1)*|011101000101011000100(0|1)*|011101000101011000101(0|1)*|011101000101011000110(0|1)*|011101000101011000111(0|1)*|011101000101011001000(0|1)*|011101000101011001001(0|1)*|011101000101011001010(0|1)*|011101000101011001011(0|1)*|011101000101011001100(0|1)*|011101000101011001101(0|1)*|011101000101011001110(0|1)*|011101000101011001111(0|1)*|011101000101011010000(0|1)*|011101000101011010001(0|1)*|011101000101011010010(0|1)*|011101000101011010011(0|1)*|011101000101011010100(0|1)*|011101000101011010101(0|1)*|011101000101011010110(0|1)*|011101000101011010111(0|1)*|011101000101011011000(0|1)*|011101000101011011001(0|1)*|011101000101011011010(0|1)*|011101000101011011011(0|1)*|011101000101011011100(0|1)*|011101000101011011101(0|1)*|011101000101011011110(0|1)*|011101000101011011111(0|1)*|0111010001010111(0|1)*|011101000101011100000(0|1)*|011101000101011100001(0|1)*|011101000101011100010(0|1)*|011101000101011100011(0|1)*|011101000101011100100(0|1)*|011101000101011100101(0|1)*|011101000101011100110(0|1)*|011101000101011100111(0|1)*|011101000101011101000(0|1)*|011101000101011101001(0|1)*|011101000101011101010(0|1)*|011101000101011101011(0|1)*|011101000101011101100(0|1)*|011101000101011101101(0|1)*|011101000101011101110(0|1)*|011101000101011101111(0|1)*|011101000101011110000(0|1)*|011101000101011110001(0|1)*|011101000101011110010(0|1)*|011101000101011110011(0|1)*|011101000101011110100(0|1)*|011101000101011110101(0|1)*|011101000101011110110(0|1)*|011101000101011110111(0|1)*|011101000101011111000(0|1)*|011101000101011111001(0|1)*|011101000101011111010(0|1)*|011101000101011111011(0|1)*|011101000101011111100(0|1)*|011101000101011111101(0|1)*|011101000101011111110(0|1)*|011101000101011111111(0|1)*|0111010001011000(0|1)*|011101000101100000000(0|1)*|011101000101100000001(0|1)*|011101000101100000010(0|1)*|011101000101100000011(0|1)*|011101000101100000100(0|1)*|011101000101100000101(0|1)*|011101000101100000110(0|1)*|011101000101100000111(0|1)*|011101000101100001000(0|1)*|011101000101100001001(0|1)*|011101000101100001010(0|1)*|011101000101100001011(0|1)*|011101000101100001100(0|1)*|011101000101100001101(0|1)*|011101000101100001110(0|1)*|011101000101100001111(0|1)*|011101000101100010000(0|1)*|011101000101100010001(0|1)*|011101000101100010010(0|1)*|011101000101100010011(0|1)*|011101000101100010100(0|1)*|011101000101100010101(0|1)*|011101000101100010110(0|1)*|011101000101100010111(0|1)*|011101000101100011000(0|1)*|011101000101100011001(0|1)*|011101000101100011010(0|1)*|011101000101100011011(0|1)*|011101000101100011100(0|1)*|011101000101100011101(0|1)*|011101000101100011110(0|1)*|011101000101100011111(0|1)*|01110100010110010(0|1)*|011101000101100100000(0|1)*|011101000101100100001(0|1)*|011101000101100100010(0|1)*|011101000101100100011(0|1)*|011101000101100100100(0|1)*|011101000101100100101(0|1)*|011101000101100100110(0|1)*|011101000101100100111(0|1)*|011101000101100101000(0|1)*|011101000101100101001(0|1)*|011101000101100101010(0|1)*|011101000101100101011(0|1)*|011101000101100101100(0|1)*|011101000101100101101(0|1)*|011101000101100101110(0|1)*|011101000101100101111(0|1)*|011101000101100101111000(0|1)*|1100101010011100(0|1)*|110010101001110000000(0|1)*|110010101001110000000001(0|1)*|110010101001110000000010(0|1)*|110010101001110000000110(0|1)*|110010101001110000001(0|1)*|110010101001110000001000(0|1)*|110010101001110000001001(0|1)*|110010101001110000001010(0|1)*|110010101001110000001011(0|1)*|110010101001110000001101(0|1)*|110010101001110000001110(0|1)*|110010101001110000010(0|1)*|110010101001110000011(0|1)*|110010101001110000100(0|1)*|110010101001110000101(0|1)*|110010101001110000110(0|1)*|110010101001110000111(0|1)*|110010101001110001000(0|1)*|110010101001110001001(0|1)*|110010101001110001010(0|1)*|110010101001110001011(0|1)*|110010101001110001100(0|1)*|110010101001110001101(0|1)*|110010101001110001110(0|1)*|110010101001110001111(0|1)*|110010101001110010000(0|1)*|110010101001110010001(0|1)*|110010101001110010010(0|1)*|110010101001110010011(0|1)*|110010101001110010100(0|1)*|110010101001110010101(0|1)*|110010101001110010110(0|1)*|110010101001110010111(0|1)*|110010101001110011000(0|1)*|110010101001110011001(0|1)*|110010101001110011010(0|1)*|110010101001110011011(0|1)*|110010101001110011100(0|1)*|110010101001110011101(0|1)*|110010101001110011110(0|1)*|110010101001110011111(0|1)*|1101101010111010(0|1)*|110110101011101000000(0|1)*|110110101011101000000001(0|1)*|110110101011101000001000(0|1)*|110110101011101000001001(0|1)*|110110101011101000001010(0|1)*|110110101011101000001011(0|1)*|110110101011101000001100(0|1)*|110110101011101000001110(0|1)*|110110101011101000001111(0|1)*|110110101011101000010(0|1)*|110110101011101000010000(0|1)*|110110101011101000010001(0|1)*|110110101011101000010010(0|1)*|110110101011101000010011(0|1)*|110110101011101000011(0|1)*|110110101011101000100(0|1)*|110110101011101000101(0|1)*|110110101011101000110(0|1)*|110110101011101000111(0|1)*|110110101011101001000(0|1)*|110110101011101001001(0|1)*|110110101011101001010(0|1)*|110110101011101001011(0|1)*|110110101011101001100(0|1)*|110110101011101001101(0|1)*|110110101011101001110(0|1)*|110110101011101001111(0|1)*|110110101011101010000(0|1)*|110110101011101010001(0|1)*|110110101011101010010(0|1)*|110110101011101010011(0|1)*|110110101011101010100(0|1)*|110110101011101010101(0|1)*|110110101011101010110(0|1)*|110110101011101010111(0|1)*|110110101011101011000(0|1)*|110110101011101011001(0|1)*|110110101011101011010(0|1)*|110110101011101011011(0|1)*|110110101011101011100(0|1)*|110110101011101011101(0|1)*|110110101011101011110(0|1)*|110110101011101011111(0|1)*|1101101011010100(0|1)*|110110101101010000000(0|1)*|110110101101010000001(0|1)*|110110101101010000010(0|1)*|110110101101010000011(0|1)*|110110101101010000100(0|1)*|110110101101010000101(0|1)*|110110101101010000110(0|1)*|110110101101010000111(0|1)*|110110101101010001000(0|1)*|110110101101010001001(0|1)*|110110101101010001010(0|1)*|110110101101010001011(0|1)*|110110101101010001100(0|1)*|110110101101010001101(0|1)*|110110101101010001110(0|1)*|110110101101010001111(0|1)*|110110101101010010000(0|1)*|110110101101010010001(0|1)*|110110101101010010010(0|1)*|110110101101010010011(0|1)*|110110101101010010100(0|1)*|1101101011010100101000(0|1)*|110110101101010010101(0|1)*|110110101101010010110(0|1)*|110110101101010010111(0|1)*|110110101101010011000(0|1)*|110110101101010011010(0|1)*|110110101101010011011(0|1)*|110110101101010011100(0|1)*|110110101101010011101(0|1)*|110110101101010011110(0|1)*|110110101101010011111(0|1)*|1101111010100100(0|1)*|110111101010010000000(0|1)*|110111101010010000001(0|1)*|110111101010010000010(0|1)*|110111101010010000011(0|1)*|110111101010010000100(0|1)*|110111101010010000101(0|1)*|110111101010010000110(0|1)*|110111101010010000111(0|1)*|110111101010010001000(0|1)*|110111101010010001001(0|1)*|110111101010010001010(0|1)*|110111101010010001011(0|1)*|110111101010010001100(0|1)*|110111101010010001101(0|1)*|110111101010010001110(0|1)*|110111101010010001111(0|1)*|110111101010010010000(0|1)*|110111101010010010001(0|1)*|110111101010010010010(0|1)*|110111101010010010011(0|1)*|110111101010010010100(0|1)*|110111101010010010101(0|1)*|110111101010010010110(0|1)*|110111101010010010111(0|1)*|110111101010010011000(0|1)*|110111101010010011001(0|1)*|110111101010010011010(0|1)*|110111101010010011011(0|1)*|110111101010010011100(0|1)*|110111101010010011101(0|1)*|110111101010010011110(0|1)*|110111101010010011111(0|1)*|11011110101001010(0|1)*|110111101010010100000(0|1)*|110111101010010100001(0|1)*|110111101010010100010(0|1)*|110111101010010100011(0|1)*|110111101010010100100(0|1)*|110111101010010100101(0|1)*|110111101010010100110(0|1)*|110111101010010100111(0|1)*|110111101010010101000(0|1)*|110111101010010101001(0|1)*|110111101010010101010(0|1)*|110111101010010101011(0|1)*|110111101010010101100(0|1)*|110111101010010101101(0|1)*|110111101010010101110(0|1)*|110111101010010101111(0|1)*)",
321 {"GNVPN-0001-PAD1101111010100101011101010101010101",
322 "GNVPN-0001-PAD11001010100111000101101010101"},
330 for (i = 0; i < 19; i++)
332 if (0 != regcomp (&rx, rxstr[i].regex, REG_EXTENDED))
334 GNUNET_log (GNUNET_ERROR_TYPE_ERROR,
335 "Could not compile regex using regcomp()\n");
340 a = REGEX_INTERNAL_construct_nfa (rxstr[i].regex, strlen (rxstr[i].regex));
341 check_nfa += test_automaton (a, &rx, &rxstr[i]);
342 REGEX_INTERNAL_automaton_destroy (a);
345 a = REGEX_INTERNAL_construct_dfa (rxstr[i].regex, strlen (rxstr[i].regex), 0);
346 check_dfa += test_automaton (a, &rx, &rxstr[i]);
347 check_proof = GNUNET_strdup (REGEX_INTERNAL_get_canonical_regex (a));
348 REGEX_INTERNAL_automaton_destroy (a);
350 a = REGEX_INTERNAL_construct_dfa (check_proof, strlen (check_proof), 0);
351 check_dfa += test_automaton (a, &rx, &rxstr[i]);
352 REGEX_INTERNAL_automaton_destroy (a);
354 GNUNET_log (GNUNET_ERROR_TYPE_DEBUG, "check_proof: %s\n", check_proof);
355 GNUNET_free_non_null (check_proof);
362 for (i = 0; i < 20; i++)
363 check_rand += test_random (50, 60, 10);
365 return check_nfa + check_dfa + check_rand;