_obsolete_dont_use/slre/unit_test.d

   1 import iv.slre;
   2
   3 int static_total_tests = 0;
   4 int static_failed_tests = 0;
   5
   6
   7 void FAIL (usize line, const(char)* str) {
   8   import core.stdc.stdio : printf;
   9   printf("Fail on line %d: [%s]\n", line, str);
  10   ++static_failed_tests;
  11 }
  12
  13
  14 void ASSERT(usize ln=__LINE__) (bool cond) {
  15   ++static_total_tests;
  16   if (!cond) FAIL(ln, "fucked");
  17 }
  18
  19
  20 // regex must have exactly one bracket pair
  21 char[] slre_replace (const(char)[] regex, const(char)[] buf, const(char)[] sub) {
  22   char[] s;
  23   Slre.Capture[1] cap;
  24   while (buf.length) {
  25     auto n = Slre.matchFirst(regex, buf[], cap[]);
  26     if (n < 0) {
  27       // no match
  28       s ~= buf[];
  29       break;
  30     }
  31     // if some bytes were skipped, add them verbatim
  32     if (cap[0].ofs > 0) s ~= buf[0..cap[0].ofs];
  33     // add replacement
  34     s ~= sub[];
  35     // if we have some bytes left, add them verbatim
  36     if (cap[0].ofs+cap[0].len < n) {
  37       //{ import std.stdio; writeln("n=", n, "; ofs=", cap[0].ofs, "; len=", cap[0].len, "; end=", cap[0].ofs+cap[0].len); }
  38       s ~= buf[cap[0].ofs+cap[0].len..n];
  39     }
  40     buf = buf[n..$];
  41   }
  42   return s;
  43 }
  44
  45
  46 void main () {
  47   Slre.Capture[10] caps;
  48
  49   int slre_match (const(char)[] re, const(char)[] str, int strlen, Slre.Capture* caps, int capslen, int flags) {
  50     return Slre.matchFirst(re, str[0..strlen], caps[0..capslen], flags);
  51   }
  52
  53   /* Metacharacters */
  54   ASSERT(slre_match("$", "abcd", 4, null, 0, 0) == 4);
  55   ASSERT(slre_match("^", "abcd", 4, null, 0, 0) == 0);
  56   ASSERT(slre_match("x|^", "abcd", 4, null, 0, 0) == 0);
  57   ASSERT(slre_match("x|$", "abcd", 4, null, 0, 0) == 4);
  58   ASSERT(slre_match("x", "abcd", 4, null, 0, 0) == Slre.Result.NoMatch);
  59   ASSERT(slre_match(".", "abcd", 4, null, 0, 0) == 1);
  60   ASSERT(slre_match("^.*\\\\.*$", "c:\\Tools", 8, null, 0, Slre.Flag.IgnoreCase) == 8);
  61   ASSERT(slre_match("\\", "a", 1, null, 0, 0) == Slre.Result.InvalidMetaChar);
  62   ASSERT(slre_match("\\x", "a", 1, null, 0, 0) == Slre.Result.InvalidMetaChar);
  63   ASSERT(slre_match("\\x1", "a", 1, null, 0, 0) == Slre.Result.InvalidMetaChar);
  64   ASSERT(slre_match("\\x20", " ", 1, null, 0, 0) == 1);
  65
  66   ASSERT(slre_match("^.+$", "", 0, null, 0, 0) == Slre.Result.NoMatch);
  67   ASSERT(slre_match("^(.+)$", "", 0, null, 0, 0) == Slre.Result.NoMatch);
  68   ASSERT(slre_match("^([\\+-]?)([\\d]+)$", "+", 1, caps.ptr, 10, Slre.Flag.IgnoreCase) == Slre.Result.NoMatch);
  69   ASSERT(slre_match("^([\\+-]?)([\\d]+)$", "+27", 3, caps.ptr, 10, Slre.Flag.IgnoreCase) == 3);
  70   ASSERT(caps[0].ptr == "+");
  71   ASSERT(caps[1].ptr == "27");
  72
  73   ASSERT(slre_match("tel:\\+(\\d+[\\d-]+\\d)", "tel:+1-201-555-0123;a=b", 23, caps.ptr, 10, 0) == 19);
  74   ASSERT(caps[0].ptr == "1-201-555-0123");
  75
  76   // character sets
  77   ASSERT(slre_match("[abc]", "1c2", 3, null, 0, 0) == 2);
  78   ASSERT(slre_match("[abc]", "1C2", 3, null, 0, 0) == Slre.Result.NoMatch);
  79   ASSERT(slre_match("[abc]", "1C2", 3, null, 0, Slre.Flag.IgnoreCase) == 2);
  80   ASSERT(slre_match("[.2]", "1C2", 3, null, 0, 0) == 1);
  81   ASSERT(slre_match("[\\S]+", "ab cd", 5, null, 0, 0) == 2);
  82   ASSERT(slre_match("[\\S]+\\s+[tyc]*", "ab cd", 5, null, 0, 0) == 4);
  83   ASSERT(slre_match("[\\d]", "ab cd", 5, null, 0, 0) == Slre.Result.NoMatch);
  84   ASSERT(slre_match("[^\\d]", "ab cd", 5, null, 0, 0) == 1);
  85   ASSERT(slre_match("[^\\d]+", "abc123", 6, null, 0, 0) == 3);
  86   ASSERT(slre_match("[1-5]+", "123456789", 9, null, 0, 0) == 5);
  87   ASSERT(slre_match("[1-5a-c]+", "123abcdef", 9, null, 0, 0) == 6);
  88   ASSERT(slre_match("[1-5a-]+", "123abcdef", 9, null, 0, 0) == 4);
  89   ASSERT(slre_match("[1-5a-]+", "123a--2oo", 9, null, 0, 0) == 7);
  90   ASSERT(slre_match("[htps]+://", "https://", 8, null, 0, 0) == 8);
  91   ASSERT(slre_match("[^\\s]+", "abc def", 7, null, 0, 0) == 3);
  92   ASSERT(slre_match("[^fc]+", "abc def", 7, null, 0, 0) == 2);
  93   ASSERT(slre_match("[^d\\sf]+", "abc def", 7, null, 0, 0) == 3);
  94
  95   // flags: case sensitivity
  96   ASSERT(slre_match("FO", "foo", 3, null, 0, 0) == Slre.Result.NoMatch);
  97   ASSERT(slre_match("FO", "foo", 3, null, 0, Slre.Flag.IgnoreCase) == 2);
  98   ASSERT(slre_match("(?m)FO", "foo", 3, null, 0, 0) == Slre.Result.UnexpectedQuantifier);
  99   ASSERT(slre_match("(?m)x", "foo", 3, null, 0, 0) == Slre.Result.UnexpectedQuantifier);
 100
 101   ASSERT(slre_match("fo", "foo", 3, null, 0, 0) == 2);
 102   ASSERT(slre_match(".+", "foo", 3, null, 0, 0) == 3);
 103   ASSERT(slre_match(".+k", "fooklmn", 7, null, 0, 0) == 4);
 104   ASSERT(slre_match(".+k.", "fooklmn", 7, null, 0, 0) == 5);
 105   ASSERT(slre_match("p+", "fooklmn", 7, null, 0, 0) == Slre.Result.NoMatch);
 106   ASSERT(slre_match("ok", "fooklmn", 7, null, 0, 0) == 4);
 107   ASSERT(slre_match("lmno", "fooklmn", 7, null, 0, 0) == Slre.Result.NoMatch);
 108   ASSERT(slre_match("mn.", "fooklmn", 7, null, 0, 0) == Slre.Result.NoMatch);
 109   ASSERT(slre_match("o", "fooklmn", 7, null, 0, 0) == 2);
 110   ASSERT(slre_match("^o", "fooklmn", 7, null, 0, 0) == Slre.Result.NoMatch);
 111   ASSERT(slre_match("^", "fooklmn", 7, null, 0, 0) == 0);
 112   ASSERT(slre_match("n$", "fooklmn", 7, null, 0, 0) == 7);
 113   ASSERT(slre_match("n$k", "fooklmn", 7, null, 0, 0) == Slre.Result.NoMatch);
 114   ASSERT(slre_match("l$", "fooklmn", 7, null, 0, 0) == Slre.Result.NoMatch);
 115   ASSERT(slre_match(".$", "fooklmn", 7, null, 0, 0) == 7);
 116   ASSERT(slre_match("a?", "fooklmn", 7, null, 0, 0) == 0);
 117   ASSERT(slre_match("^a*CONTROL", "CONTROL", 7, null, 0, 0) == 7);
 118   ASSERT(slre_match("^[a]*CONTROL", "CONTROL", 7, null, 0, 0) == 7);
 119   ASSERT(slre_match("^(a*)CONTROL", "CONTROL", 7, null, 0, 0) == 7);
 120   ASSERT(slre_match("^(a*)?CONTROL", "CONTROL", 7, null, 0, 0) == 7);
 121
 122   ASSERT(slre_match("\\_", "abc", 3, null, 0, 0) == Slre.Result.InvalidMetaChar);
 123   ASSERT(slre_match("+", "fooklmn", 7, null, 0, 0) == Slre.Result.UnexpectedQuantifier);
 124   ASSERT(slre_match("()+", "fooklmn", 7, null, 0, 0) == Slre.Result.NoMatch);
 125   ASSERT(slre_match("\\x", "12", 2, null, 0, 0) == Slre.Result.InvalidMetaChar);
 126   ASSERT(slre_match("\\xhi", "12", 2, null, 0, 0) == Slre.Result.InvalidMetaChar);
 127   ASSERT(slre_match("\\x20", "_ J", 3, null, 0, 0) == 2);
 128   ASSERT(slre_match("\\x4A", "_ J", 3, null, 0, 0) == 3);
 129   ASSERT(slre_match("\\d+", "abc123def", 9, null, 0, 0) == 6);
 130
 131   // balancing brackets
 132   ASSERT(slre_match("(x))", "fooklmn", 7, null, 0, 0) == Slre.Result.UnbalancedBrackets);
 133   ASSERT(slre_match("(", "fooklmn", 7, null, 0, 0) == Slre.Result.UnbalancedBrackets);
 134
 135   ASSERT(slre_match("klz?mn", "fooklmn", 7, null, 0, 0) == 7);
 136   ASSERT(slre_match("fa?b", "fooklmn", 7, null, 0, 0) == Slre.Result.NoMatch);
 137
 138   // brackets & capturing
 139   ASSERT(slre_match("^(te)", "tenacity subdues all", 20, caps.ptr, 10, 0) == 2);
 140   ASSERT(slre_match("(bc)", "abcdef", 6, caps.ptr, 10, 0) == 3);
 141   ASSERT(slre_match(".(d.)", "abcdef", 6, caps.ptr, 10, 0) == 5);
 142   ASSERT(slre_match(".(d.)\\)?", "abcdef", 6, caps.ptr, 10, 0) == 5);
 143   ASSERT(caps[0].ptr == "de");
 144   ASSERT(slre_match("(.+)", "123", 3, caps.ptr, 10, 0) == 3);
 145   ASSERT(slre_match("(2.+)", "123", 3, caps.ptr, 10, 0) == 3);
 146   ASSERT(caps[0].ptr == "23");
 147   ASSERT(slre_match("(.+2)", "123", 3, caps.ptr, 10, 0) == 2);
 148   ASSERT(caps[0].ptr == "12");
 149   ASSERT(slre_match("(.*(2.))", "123", 3, caps.ptr, 10, 0) == 3);
 150   ASSERT(slre_match("(.)(.)", "123", 3, caps.ptr, 10, 0) == 2);
 151   ASSERT(slre_match("(\\d+)\\s+(\\S+)", "12 hi", 5, caps.ptr, 10, 0) == 5);
 152   ASSERT(slre_match("ab(cd)+ef", "abcdcdef", 8, null, 0, 0) == 8);
 153   ASSERT(slre_match("ab(cd)*ef", "abcdcdef", 8, null, 0, 0) == 8);
 154   ASSERT(slre_match("ab(cd)+?ef", "abcdcdef", 8, null, 0, 0) == 8);
 155   ASSERT(slre_match("ab(cd)+?.", "abcdcdef", 8, null, 0, 0) == 5);
 156   ASSERT(slre_match("ab(cd)?", "abcdcdef", 8, null, 0, 0) == 4);
 157   ASSERT(slre_match("a(b)(cd)", "abcdcdef", 8, caps.ptr, 1, 0) == Slre.Result.CapsArrayTooSmall);
 158   ASSERT(slre_match("(.+/\\d+\\.\\d+)\\.jpg$", "/foo/bar/12.34.jpg", 18, caps.ptr, 1, 0) == 18);
 159   ASSERT(slre_match("(ab|cd).*\\.(xx|yy)", "ab.yy", 5, null, 0, 0) == 5);
 160   ASSERT(slre_match(".*a", "abcdef", 6, null, 0, 0) == 1);
 161   ASSERT(slre_match("(.+)c", "abcdef", 6, null, 0, 0) == 3);
 162   ASSERT(slre_match("\\n", "abc\ndef", 7, null, 0, 0) == 4);
 163   ASSERT(slre_match("b.\\s*\\n", "aa\r\nbb\r\ncc\r\n\r\n", 14, caps.ptr, 10, 0) == 8);
 164
 165   // greedy vs non-greedy
 166   ASSERT(slre_match(".+c", "abcabc", 6, null, 0, 0) == 6);
 167   ASSERT(slre_match(".+?c", "abcabc", 6, null, 0, 0) == 3);
 168   ASSERT(slre_match(".*?c", "abcabc", 6, null, 0, 0) == 3);
 169   ASSERT(slre_match(".*c", "abcabc", 6, null, 0, 0) == 6);
 170   ASSERT(slre_match("bc.d?k?b+", "abcabc", 6, null, 0, 0) == 5);
 171
 172   // branching
 173   ASSERT(slre_match("|", "abc", 3, null, 0, 0) == 0);
 174   ASSERT(slre_match("|.", "abc", 3, null, 0, 0) == 1);
 175   ASSERT(slre_match("x|y|b", "abc", 3, null, 0, 0) == 2);
 176   ASSERT(slre_match("k(xx|yy)|ca", "abcabc", 6, null, 0, 0) == 4);
 177   ASSERT(slre_match("k(xx|yy)|ca|bc", "abcabc", 6, null, 0, 0) == 3);
 178   ASSERT(slre_match("(|.c)", "abc", 3, caps.ptr, 10, 0) == 3);
 179   ASSERT(caps[0].ptr == "bc");
 180   ASSERT(slre_match("a|b|c", "a", 1, null, 0, 0) == 1);
 181   ASSERT(slre_match("a|b|c", "b", 1, null, 0, 0) == 1);
 182   ASSERT(slre_match("a|b|c", "c", 1, null, 0, 0) == 1);
 183   ASSERT(slre_match("a|b|c", "d", 1, null, 0, 0) == Slre.Result.NoMatch);
 184
 185   // optional match at the end of the string
 186   ASSERT(slre_match("^.*c.?$", "abc", 3, null, 0, 0) == 3);
 187   ASSERT(slre_match("^.*C.?$", "abc", 3, null, 0, Slre.Flag.IgnoreCase) == 3);
 188   ASSERT(slre_match("bk?", "ab", 2, null, 0, 0) == 2);
 189   ASSERT(slre_match("b(k?)", "ab", 2, null, 0, 0) == 2);
 190   ASSERT(slre_match("b[k-z]*", "ab", 2, null, 0, 0) == 2);
 191   ASSERT(slre_match("ab(k|z|y)*", "ab", 2, null, 0, 0) == 2);
 192   ASSERT(slre_match("[b-z].*", "ab", 2, null, 0, 0) == 2);
 193   ASSERT(slre_match("(b|z|u).*", "ab", 2, null, 0, 0) == 2);
 194   ASSERT(slre_match("ab(k|z|y)?", "ab", 2, null, 0, 0) == 2);
 195   ASSERT(slre_match(".*", "ab", 2, null, 0, 0) == 2);
 196   ASSERT(slre_match(".*$", "ab", 2, null, 0, 0) == 2);
 197   ASSERT(slre_match("a+$", "aa", 2, null, 0, 0) == 2);
 198   ASSERT(slre_match("a*$", "aa", 2, null, 0, 0) == 2);
 199   ASSERT(slre_match( "a+$" ,"Xaa", 3, null, 0, 0) == 3);
 200   ASSERT(slre_match( "a*$" ,"Xaa", 3, null, 0, 0) == 3);
 201
 202   // ignore case flag
 203   ASSERT(slre_match("[a-h]+", "abcdefghxxx", 11, null, 0, 0) == 8);
 204   ASSERT(slre_match("[A-H]+", "ABCDEFGHyyy", 11, null, 0, 0) == 8);
 205   ASSERT(slre_match("[a-h]+", "ABCDEFGHyyy", 11, null, 0, 0) == Slre.Result.NoMatch);
 206   ASSERT(slre_match("[A-H]+", "abcdefghyyy", 11, null, 0, 0) == Slre.Result.NoMatch);
 207   ASSERT(slre_match("[a-h]+", "ABCDEFGHyyy", 11, null, 0, Slre.Flag.IgnoreCase) == 8);
 208   ASSERT(slre_match("[A-H]+", "abcdefghyyy", 11, null, 0, Slre.Flag.IgnoreCase) == 8);
 209
 210   {
 211     import core.stdc.stdio : printf;
 212     // example: HTTP request
 213     string request = " GET /index.html HTTP/1.0\r\n\r\n";
 214     //slre_cap[4] caps;
 215
 216     if (slre_match("^\\s*(\\S+)\\s+(\\S+)\\s+HTTP/(\\d)\\.(\\d)", request, cast(int)request.length, caps.ptr, 4, 0) > 0) {
 217       printf("Method: [%.*s], URI: [%.*s]\n", caps[0].len, caps[0].ptr.ptr, caps[1].len, caps[1].ptr.ptr);
 218     } else {
 219       printf("Error parsing [%.*s]\n", cast(uint)request.length, request.ptr);
 220     }
 221     ASSERT(caps[1].ptr == "/index.html");
 222   }
 223
 224   {
 225     import core.stdc.stdio : printf;
 226     // example: string replacement
 227     auto s = slre_replace("({{.+?}})", "Good morning, {{foo}}. How are you, {{bar}}?", "Bob");
 228     printf("%.*s\n", cast(uint)s.length, s.ptr);
 229     ASSERT(s == "Good morning, Bob. How are you, Bob?");
 230   }
 231
 232   {
 233     import core.stdc.stdio : printf;
 234     // example: find all URLs in a string
 235     string str = "<img src=\"HTTPS://FOO.COM/x?b#c=tab1\"/>   <a href=\"http://cesanta.com\">some link</a>";
 236     string regex = "((https?://)[^\\s/'\"<>]+/?[^\\s'\"<>]*)";
 237     //slre_cap[2] caps;
 238     int i, j = 0, str_len = cast(int)str.length;
 239     while (j < str_len && (i = slre_match(regex, str[j..$], str_len-j, caps.ptr, 2, Slre.Flag.IgnoreCase)) > 0) {
 240       printf("Found URL: [%.*s]\n", caps[0].len, caps[0].ptr.ptr);
 241       j += i;
 242     }
 243   }
 244
 245   {
 246     // example more complex regular expression
 247     string str = "aa 1234 xy\nxyz";
 248     string regex = "aa ([0-9]*) *([x-z]*)\\s+xy([yz])";
 249     //slre_cap[3] caps;
 250     ASSERT(slre_match(regex, str, cast(int)str.length, caps.ptr, 3, 0) > 0);
 251     ASSERT(caps[0].len == 4);
 252     ASSERT(caps[1].len == 2);
 253     ASSERT(caps[2].len == 1);
 254     ASSERT(caps[2].ptr[0] == 'z');
 255   }
 256
 257   {
 258     import core.stdc.stdio : printf;
 259     printf("Unit test %s (total test: %d, failed tests: %d)\n",
 260            static_failed_tests > 0 ? "FAILED".ptr : "PASSED".ptr,
 261            static_total_tests, static_failed_tests);
 262   }
 263
 264   //return static_failed_tests == 0 ? EXIT_SUCCESS : EXIT_FAILURE;
 265 }