lib/tld.c

   1 /* tld.c --- Handle TLD restriction checking.
   2  * Copyright (C) 2003, 2004  Free Software Foundation, Inc.
   3  *
   4  * Author: Thomas Jacob, Internet24.de
   5  *
   6  * This file is part of GNU Libidn.
   7  *
   8  * GNU Libidn is free software; you can redistribute it and/or
   9  * modify it under the terms of the GNU Lesser General Public
  10  * License as published by the Free Software Foundation; either
  11  * version 2.1 of the License, or (at your option) any later version.
  12  *
  13  * GNU Libidn is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  16  * Lesser General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU Lesser General Public
  19  * License along with GNU Libidn; if not, write to the Free Software
  20  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
  21  *
  22  */
  23
  24 /* Get stringprep_utf8_to_ucs4, stringprep_locale_to_utf8. */
  25 #include <stringprep.h>
  26
  27 /* Get strcmp(). */
  28 #include <string.h>
  29
  30 /* Get specifications. */
  31 #include <tld.h>
  32
  33 /* Array of built-in domain restriction structures.  See tlds.c.  */
  34 extern const Tld_table *_tld_tables[];
  35
  36 /**
  37  * tld_get_table:
  38  * @tld: TLD name (e.g. "com") as zero terminated ASCII byte string.
  39  * @tables: Zero terminated array of info-structures for TLDs.
  40  *
  41  * Get the TLD table for a named TLD by searching through the given
  42  * TLD table array.
  43  *
  44  * Return value: Return structure corresponding to TLD @tld by going
  45  * thru @tables, or return %NULL if no such structure is found.
  46  */
  47 const Tld_table *
  48 tld_get_table (const char *tld, const Tld_table ** tables)
  49 {
  50   const Tld_table **tldtable = NULL;
  51
  52   if (!tld || !tables)
  53     return NULL;
  54
  55   for (tldtable = tables; *tldtable; tldtable++)
  56     if (!strcmp ((*tldtable)->name, tld))
  57       return *tldtable;
  58
  59   return NULL;
  60 }
  61
  62 /**
  63  * tld_default_table:
  64  * @tld: TLD name (e.g. "com") as zero terminated ASCII byte string.
  65  * @overrides: Additional well-formed info-structures for TLDs, or %NULL
  66  * to only use library deault tables.
  67  *
  68  * Get the TLD table for a named TLD, using the internal defaults,
  69  * possibly overrided by the (optional) supplied tables.
  70  *
  71  * Return value: Return structure corresponding to TLD @tld_str, first
  72  * looking through @overrides then thru built-in list, or %NULL if no
  73  * such structure found.
  74  */
  75 const Tld_table *
  76 tld_default_table (const char *tld, const Tld_table ** overrides)
  77 {
  78   const Tld_table *tldtable = NULL;
  79
  80   if (!tld)
  81     return NULL;
  82
  83   if (overrides)
  84     tldtable = tld_get_table (tld, overrides);
  85
  86   if (!tldtable)
  87     tldtable = tld_get_table (tld, _tld_tables);
  88
  89   return tldtable;
  90 }
  91
  92 #define DOTP(c) ((c) == 0x002E || (c) == 0x3002 ||      \
  93                  (c) == 0xFF0E || (c) == 0xFF61)
  94
  95 /**
  96  * tld_get_4:
  97  * @in: Array of unicode code points to process. Does not need to be
  98  * zero terminated.
  99  * @inlen: Number of unicode code points.
 100  * @out: Zero terminated ascii result string pointer.
 101  *
 102  * Isolate the top-level domain of @in and return it as an ASCII
 103  * string in @out.
 104  *
 105  * Return value: Return %TLD_SUCCESS on success, the corresponding
 106  * error code otherwise.
 107  */
 108 int
 109 tld_get_4 (const uint32_t * in, size_t inlen, char **out)
 110 {
 111   const uint32_t *ipos;
 112   size_t olen;
 113
 114   *out = NULL;
 115   if (!in || inlen == 0)
 116     return TLD_NODATA;
 117
 118   ipos = &in[inlen - 1];
 119   olen = 0;
 120   /* Scan backwards for non(latin)letters. */
 121   while (ipos >= in && ((*ipos >= 0x41 && *ipos <= 0x5A) ||
 122                         (*ipos >= 0x61 && *ipos <= 0x7A)))
 123     ipos--, olen++;
 124
 125   if (olen > 0 && DOTP (*ipos)) /* Found something that appears a TLD. */
 126     {
 127       char *out_s = malloc (sizeof (char) * (olen + 1));
 128       char *opos = out_s;
 129
 130       if (!opos)
 131         return TLD_MALLOC_ERROR;
 132
 133       ipos++;
 134       /* Transcribe to lowercase ascii string. */
 135       for (; ipos < &in[inlen]; ipos++, opos++)
 136         *opos = *ipos > 0x5A ? *ipos : *ipos + 0x20;
 137       *opos = 0;
 138       *out = out_s;
 139       return TLD_SUCCESS;
 140     }
 141
 142   return TLD_NOTLD;
 143 }
 144
 145 /**
 146  * tld_get_4z:
 147  * @in: Zero terminated array of unicode code points to process.
 148  * @out: Zero terminated ascii result string pointer.
 149  *
 150  * Isolate the top-level domain of @in and return it as an ASCII
 151  * string in @out.
 152  *
 153  * Return value: Returns %TLD_SUCCESS on success, the corresponding
 154  * error code otherwise.
 155  */
 156 int
 157 tld_get_4z (const uint32_t * in, char **out)
 158 {
 159   const uint32_t *ipos = in;
 160
 161   if (!in)
 162     return TLD_NODATA;
 163
 164   while (*ipos)
 165     ipos++;
 166
 167   return tld_get_4 (in, ipos - in, out);
 168 }
 169
 170 /**
 171  * tld_get_z:
 172  * @in: Zero terminated character array to process.
 173  * @out: Zero terminated ascii result string pointer.
 174  *
 175  * Isolate the top-level domain of @in and return it as an ASCII
 176  * string in @out.  The input string @in may be UTF-8, ISO-8859-1 or
 177  * any ASCII compatible character encoding.
 178  *
 179  * Return value: Returns %TLD_SUCCESS on success, the corresponding
 180  * error code otherwise.
 181  */
 182 int
 183 tld_get_z (const char *in, char **out)
 184 {
 185   uint32_t *iucs;
 186   size_t i, ilen;
 187   int rc;
 188
 189   ilen = strlen (in);
 190   iucs = calloc (ilen, sizeof (*iucs));
 191
 192   if (!iucs)
 193     return TLD_MALLOC_ERROR;
 194
 195   for (i = 0; i < ilen; i++)
 196     iucs[i] = in[i];
 197
 198   rc = tld_get_4 (iucs, ilen, out);
 199
 200   free (iucs);
 201
 202   return rc;
 203 }
 204
 205 /*
 206  * tld_checkchar:
 207  * @ch: 32 bit unicode character to check.
 208  * @tld: Tld_table data structure to check @ch against
 209  *
 210  * Verify if @ch is either in [a-z0-9-.] or mentioned as a valid
 211  * character in @tld.
 212  *
 213  * Return value: Return %TLD_SUCCESS if @ch is a valid character for
 214  * the TLD @tld or if @tld is %NULL, %TLD_INVALID if @ch is invalid as
 215  * defined by @tld.
 216  */
 217 static int
 218 _tld_checkchar (uint32_t ch, const Tld_table * tld)
 219 {
 220   const Tld_table_element *s, *e, *m;
 221
 222   if (!tld)
 223     return TLD_SUCCESS;
 224
 225   /* Check for [-a-z0-9.]. */
 226   if ((ch >= 0x61 && ch <= 0x7A) ||
 227       (ch >= 0x30 && ch <= 0x39) || ch == 0x2D || DOTP (ch))
 228     return TLD_SUCCESS;
 229
 230   s = tld->valid;
 231   e = s + tld->nvalid;
 232   while (s < e)
 233     {
 234       m = s + ((e - s) >> 1);
 235       if (ch < m->start)
 236         e = m;
 237       else if (ch > m->end)
 238         s = m + 1;
 239       else
 240         return TLD_SUCCESS;
 241     }
 242
 243   return TLD_INVALID;
 244 }
 245
 246 /**
 247  * tld_check_4t
 248  * @in: Array of unicode code points to process. Does not need to be
 249  * zero terminated.
 250  * @inlen: Number of unicode code points.
 251  * @errpos: Position of offending character is returned here.
 252  * @tld: Data structure representing the restrictions for
 253  * which the input should be tested.
 254  *
 255  * Test each of the code points in @in for whether or not
 256  * they are allowed by the data structure in @tld, return
 257  * the position of the first character for which this is not
 258  * the case in @errpos.
 259  *
 260  * Return value: Returns %TLD_SUCCESS if all code points
 261  * are valid or when @tld is null, %TLD_INVALID if a
 262  * character is not allowed, or additional error codes on
 263  * general failure conditions.
 264  */
 265 int
 266 tld_check_4t (const uint32_t * in, size_t inlen, size_t * errpos,
 267                const Tld_table * tld)
 268 {
 269   const uint32_t *ipos;
 270   int rc;
 271
 272   if (!tld)                     /* No data for TLD so everything is valid. */
 273     return TLD_SUCCESS;
 274
 275   ipos = in;
 276   while (ipos < &in[inlen])
 277     {
 278       rc = _tld_checkchar (*ipos, tld);
 279       if (rc != TLD_SUCCESS)
 280         {
 281           if (errpos)
 282             *errpos = ipos - in;
 283           return rc;
 284         }
 285       ipos++;
 286     }
 287   return TLD_SUCCESS;
 288 }
 289
 290 /**
 291  * tld_check_4tz
 292  * @in: Zero terminated array of unicode code points to process.
 293  * @errpos: Position of offending character is returned here.
 294  * @tld: Data structure representing the restrictions for
 295  * which the input should be tested.
 296  *
 297  * Test each of the code points in @in for whether or not
 298  * they are allowed by the data structure in @tld, return
 299  * the position of the first character for which this is not
 300  * the case in @errpos.
 301  *
 302  * Return value: Returns %TLD_SUCCESS if all code points
 303  * are valid or when @tld is null, %TLD_INVALID if a
 304  * character is not allowed, or additional error codes on
 305  * general failure conditions.
 306  */
 307 int
 308 tld_check_4tz (const uint32_t * in, size_t * errpos, const Tld_table * tld)
 309 {
 310   const uint32_t *ipos = in;
 311
 312   if (!ipos)
 313     return TLD_NODATA;
 314
 315   while (*ipos)
 316     ipos++;
 317
 318   return tld_check_4t (in, ipos - in, errpos, tld);
 319 }
 320
 321 /**
 322  * tld_check_4
 323  * @in: Array of unicode code points to process. Does not need to be
 324  * zero terminated.
 325  * @inlen: Number of unicode code points.
 326  * @errpos: Position of offending character is returned here.
 327  * @overrides: An array of additional domain restriction structures
 328  *  that complement and supersede the built-in information.
 329  *
 330  * Test each of the code points in @in for whether or not they are
 331  * allowed by the information in @overrides or by the built-in TLD
 332  * restriction data. When data for the same TLD is available both
 333  * internally and in @overrides, the information in @overrides takes
 334  * precedence. If several entries for a specific TLD are found, the
 335  * first one is used.  If @overrides is %NULL, only the built-in
 336  * information is used.  The position of the first offending character
 337  * is returned in @errpos.
 338  *
 339  * Return value: Returns %TLD_SUCCESS if all code points
 340  * are valid or when @tld is null, %TLD_INVALID if a
 341  * character is not allowed, or additional error codes on
 342  * general failure conditions.
 343  */
 344 int
 345 tld_check_4 (const uint32_t * in, size_t inlen, size_t * errpos,
 346              const Tld_table ** overrides)
 347 {
 348   const Tld_table *tld;
 349   char *domain;
 350   int rc;
 351
 352   if (errpos)
 353     *errpos = 0;
 354
 355   /* Get TLD name. */
 356   rc = tld_get_4 (in, inlen, &domain);
 357
 358   if (rc != TLD_SUCCESS)
 359     {
 360       if (rc == TLD_NOTLD)      /* No TLD, say OK */
 361         return TLD_SUCCESS;
 362       else
 363         return rc;
 364     }
 365
 366   /* Retrieve appropriate data structure. */
 367   tld = tld_default_table (domain, overrides);
 368   free (domain);
 369
 370   return tld_check_4t (in, inlen, errpos, tld);
 371 }
 372
 373 /**
 374  * tld_check_4z
 375  * @in: Zero-terminated array of unicode code points to process.
 376  * @errpos: Position of offending character is returned here.
 377  * @overrides: An array of additional domain restriction structures
 378  * that complement and supersede the built-in information.
 379  *
 380  * Test each of the code points in @in for whether or not they are
 381  * allowed by the information in @overrides or by the built-in TLD
 382  * restriction data. When data for the same TLD is available both
 383  * internally and in @overrides, the information in @overrides takes
 384  * precedence. If several entries for a specific TLD are found, the
 385  * first one is used.  If @overrides is %NULL, only the built-in
 386  * information is used.  The position of the first offending character
 387  * is returned in @errpos.
 388  *
 389  * Return value: Returns %TLD_SUCCESS if all code points
 390  * are valid or when @tld is null, %TLD_INVALID if a
 391  * character is not allowed, or additional error codes on
 392  * general failure conditions.
 393  */
 394 int
 395 tld_check_4z (const uint32_t * in, size_t * errpos,
 396               const Tld_table ** overrides)
 397 {
 398   const uint32_t *ipos = in;
 399
 400   if (!ipos)
 401     return TLD_NODATA;
 402
 403   while (*ipos)
 404     ipos++;
 405
 406   return tld_check_4 (in, ipos - in, errpos, overrides);
 407 }
 408
 409 /**
 410  * tld_check_8z
 411  * @in: Zero-terminated UTF8 string to process.
 412  * @errpos: Position of offending character is returned here.
 413  * @overrides: An array of additional domain restriction structures
 414  * that complement and supersede the built-in information.
 415  *
 416  * Test each of the characters in @in for whether or not they are
 417  * allowed by the information in @overrides or by the built-in TLD
 418  * restriction data. When data for the same TLD is available both
 419  * internally and in @overrides, the information in @overrides takes
 420  * precedence. If several entries for a specific TLD are found, the
 421  * first one is used.  If @overrides is %NULL, only the built-in
 422  * information is used.  The position of the first offending character
 423  * is returned in @errpos.  Note that the error position refers to the
 424  * decoded character offset rather than the byte position in the
 425  * string.
 426  *
 427  * Return value: Returns %TLD_SUCCESS if all characters
 428  * are valid or when @tld is null, %TLD_INVALID if a
 429  * character is not allowed, or additional error codes on
 430  * general failure conditions.
 431  */
 432 int
 433 tld_check_8z (const char *in, size_t * errpos, const Tld_table ** overrides)
 434 {
 435   uint32_t *iucs;
 436   size_t ilen;
 437   int rc;
 438
 439   if (!in)
 440     return TLD_NODATA;
 441
 442   iucs = stringprep_utf8_to_ucs4 (in, -1, &ilen);
 443
 444   if (!iucs)
 445     return TLD_MALLOC_ERROR;
 446
 447   rc = tld_check_4 (iucs, ilen, errpos, overrides);
 448
 449   free (iucs);
 450
 451   return rc;
 452 }
 453
 454 /**
 455  * tld_check_lz
 456  * @in: Zero-terminated string in the current locales encoding to process.
 457  * @errpos: Position of offending character is returned here.
 458  * @overrides: An array of additional domain restriction structures
 459  * that complement and supersede the built-in information.
 460  *
 461  * Test each of the characters in @in for whether or not they are
 462  * allowed by the information in @overrides or by the built-in TLD
 463  * restriction data. When data for the same TLD is available both
 464  * internally and in @overrides, the information in @overrides takes
 465  * precedence. If several entries for a specific TLD are found, the
 466  * first one is used.  If @overrides is %NULL, only the built-in
 467  * information is used.  The position of the first offending character
 468  * is returned in @errpos.  Note that the error position refers to the
 469  * decoded character offset rather than the byte position in the
 470  * string.
 471  *
 472  * Return value: Returns %TLD_SUCCESS if all characters
 473  * are valid or when @tld is null, %TLD_INVALID if a
 474  * character is not allowed, or additional error codes on
 475  * general failure conditions.
 476  */
 477 int
 478 tld_check_lz (const char *in, size_t * errpos, const Tld_table ** overrides)
 479 {
 480   char *utf8;
 481   int rc;
 482
 483   if (!in)
 484     return TLD_NODATA;
 485
 486   utf8 = stringprep_locale_to_utf8 (in);
 487   if (!utf8)
 488     return TLD_ICONV_ERROR;
 489
 490
 491   rc = tld_check_8z (utf8, errpos, overrides);
 492
 493   free (utf8);
 494
 495   return rc;
 496 }
 497
 498 /**
 499  * Tld_rc:
 500  * @TLD_SUCCESS: Successful operation.  This value is guaranteed to
 501  * always be zero, the remaining ones are only guaranteed to hold
 502  * non-zero values, for logical comparison purposes.
 503  * @TLD_INVALID: Invalid character found.
 504  * @TLD_NODATA: No input data was provided.
 505  * @TLD_MALLOC_ERROR: Error during memory allocation.
 506  * @TLD_ICONV_ERROR: Error during iconv string conversion.
 507  * @TLD_NOTLD: No top-level domain found in domain string.
 508  *
 509  * Enumerated return codes of the TLD checking functions.
 510  * The value 0 is guaranteed to always correspond to success.
 511  */