contrib/apr-util/uri/apr_uri.c

   1 /* Licensed to the Apache Software Foundation (ASF) under one or more
   2  * contributor license agreements.  See the NOTICE file distributed with
   3  * this work for additional information regarding copyright ownership.
   4  * The ASF licenses this file to You under the Apache License, Version 2.0
   5  * (the "License"); you may not use this file except in compliance with
   6  * the License.  You may obtain a copy of the License at
   7  *
   8  *     http://www.apache.org/licenses/LICENSE-2.0
   9  *
  10  * Unless required by applicable law or agreed to in writing, software
  11  * distributed under the License is distributed on an "AS IS" BASIS,
  12  * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  13  * See the License for the specific language governing permissions and
  14  * limitations under the License.
  15  */
  16
  17 /*
  18  * apr_uri.c: URI related utility things
  19  *
  20  */
  21
  22 #include <stdlib.h>
  23
  24 #include "apu.h"
  25 #include "apr.h"
  26 #include "apr_general.h"
  27 #include "apr_strings.h"
  28
  29 #define APR_WANT_STRFUNC
  30 #include "apr_want.h"
  31
  32 #include "apr_uri.h"
  33
  34 typedef struct schemes_t schemes_t;
  35
  36 /** Structure to store various schemes and their default ports */
  37 struct schemes_t {
  38     /** The name of the scheme */
  39     const char *name;
  40     /** The default port for the scheme */
  41     apr_port_t default_port;
  42 };
  43
  44 /* Some WWW schemes and their default ports; this is basically /etc/services */
  45 /* This will become global when the protocol abstraction comes */
  46 /* As the schemes are searched by a linear search, */
  47 /* they are sorted by their expected frequency */
  48 static schemes_t schemes[] =
  49 {
  50     {"http",     APR_URI_HTTP_DEFAULT_PORT},
  51     {"ftp",      APR_URI_FTP_DEFAULT_PORT},
  52     {"https",    APR_URI_HTTPS_DEFAULT_PORT},
  53     {"gopher",   APR_URI_GOPHER_DEFAULT_PORT},
  54     {"ldap",     APR_URI_LDAP_DEFAULT_PORT},
  55     {"nntp",     APR_URI_NNTP_DEFAULT_PORT},
  56     {"snews",    APR_URI_SNEWS_DEFAULT_PORT},
  57     {"imap",     APR_URI_IMAP_DEFAULT_PORT},
  58     {"pop",      APR_URI_POP_DEFAULT_PORT},
  59     {"sip",      APR_URI_SIP_DEFAULT_PORT},
  60     {"rtsp",     APR_URI_RTSP_DEFAULT_PORT},
  61     {"wais",     APR_URI_WAIS_DEFAULT_PORT},
  62     {"z39.50r",  APR_URI_WAIS_DEFAULT_PORT},
  63     {"z39.50s",  APR_URI_WAIS_DEFAULT_PORT},
  64     {"prospero", APR_URI_PROSPERO_DEFAULT_PORT},
  65     {"nfs",      APR_URI_NFS_DEFAULT_PORT},
  66     {"tip",      APR_URI_TIP_DEFAULT_PORT},
  67     {"acap",     APR_URI_ACAP_DEFAULT_PORT},
  68     {"telnet",   APR_URI_TELNET_DEFAULT_PORT},
  69     {"ssh",      APR_URI_SSH_DEFAULT_PORT},
  70     { NULL, 0xFFFF }     /* unknown port */
  71 };
  72
  73 APU_DECLARE(apr_port_t) apr_uri_port_of_scheme(const char *scheme_str)
  74 {
  75     schemes_t *scheme;
  76
  77     if (scheme_str) {
  78         for (scheme = schemes; scheme->name != NULL; ++scheme) {
  79             if (strcasecmp(scheme_str, scheme->name) == 0) {
  80                 return scheme->default_port;
  81             }
  82         }
  83     }
  84     return 0;
  85 }
  86
  87 /* Unparse a apr_uri_t structure to an URI string.
  88  * Optionally suppress the password for security reasons.
  89  */
  90 APU_DECLARE(char *) apr_uri_unparse(apr_pool_t *p,
  91                                     const apr_uri_t *uptr,
  92                                     unsigned flags)
  93 {
  94     char *ret = "";
  95
  96     /* If suppressing the site part, omit both user name & scheme://hostname */
  97     if (!(flags & APR_URI_UNP_OMITSITEPART)) {
  98
  99         /* Construct a "user:password@" string, honoring the passed
 100          * APR_URI_UNP_ flags: */
 101         if (uptr->user || uptr->password) {
 102             ret = apr_pstrcat(p,
 103                       (uptr->user     && !(flags & APR_URI_UNP_OMITUSER))
 104                           ? uptr->user : "",
 105                       (uptr->password && !(flags & APR_URI_UNP_OMITPASSWORD))
 106                           ? ":" : "",
 107                       (uptr->password && !(flags & APR_URI_UNP_OMITPASSWORD))
 108                           ? ((flags & APR_URI_UNP_REVEALPASSWORD)
 109                               ? uptr->password : "XXXXXXXX")
 110                           : "",
 111                       ((uptr->user     && !(flags & APR_URI_UNP_OMITUSER)) ||
 112                        (uptr->password && !(flags & APR_URI_UNP_OMITPASSWORD)))
 113                           ? "@" : "",
 114                       NULL);
 115         }
 116
 117         /* Construct scheme://site string */
 118         if (uptr->hostname) {
 119             int is_default_port;
 120             const char *lbrk = "", *rbrk = "";
 121
 122             if (strchr(uptr->hostname, ':')) { /* v6 literal */
 123                 lbrk = "[";
 124                 rbrk = "]";
 125             }
 126
 127             is_default_port =
 128                 (uptr->port_str == NULL ||
 129                  uptr->port == 0 ||
 130                  uptr->port == apr_uri_port_of_scheme(uptr->scheme));
 131
 132             ret = apr_pstrcat(p, "//", ret, lbrk, uptr->hostname, rbrk,
 133                         is_default_port ? "" : ":",
 134                         is_default_port ? "" : uptr->port_str,
 135                         NULL);
 136         }
 137         if (uptr->scheme) {
 138             ret = apr_pstrcat(p, uptr->scheme, ":", ret, NULL);
 139         }
 140     }
 141
 142     /* Should we suppress all path info? */
 143     if (!(flags & APR_URI_UNP_OMITPATHINFO)) {
 144         /* Append path, query and fragment strings: */
 145         ret = apr_pstrcat(p,
 146                           ret,
 147                           (uptr->path)
 148                               ? uptr->path : "",
 149                           (uptr->query    && !(flags & APR_URI_UNP_OMITQUERY))
 150                               ? "?" : "",
 151                           (uptr->query    && !(flags & APR_URI_UNP_OMITQUERY))
 152                               ? uptr->query : "",
 153                           (uptr->fragment && !(flags & APR_URI_UNP_OMITQUERY))
 154                               ? "#" : NULL,
 155                           (uptr->fragment && !(flags & APR_URI_UNP_OMITQUERY))
 156                               ? uptr->fragment : NULL,
 157                           NULL);
 158     }
 159     return ret;
 160 }
 161
 162 /* Here is the hand-optimized parse_uri_components().  There are some wild
 163  * tricks we could pull in assembly language that we don't pull here... like we
 164  * can do word-at-time scans for delimiter characters using the same technique
 165  * that fast memchr()s use.  But that would be way non-portable. -djg
 166  */
 167
 168 /* We have a apr_table_t that we can index by character and it tells us if the
 169  * character is one of the interesting delimiters.  Note that we even get
 170  * compares for NUL for free -- it's just another delimiter.
 171  */
 172
 173 #define T_SLASH           0x01        /* '/' */
 174 #define T_QUESTION        0x02        /* '?' */
 175 #define T_HASH            0x04        /* '#' */
 176 #define T_ALPHA           0x08        /* 'A' ... 'Z', 'a' ... 'z' */
 177 #define T_SCHEME          0x10        /* '0' ... '9', '-', '+', '.'
 178                                        * (allowed in scheme except first char)
 179                                        */
 180 #define T_NUL             0x80        /* '\0' */
 181
 182 #if APR_CHARSET_EBCDIC
 183 /* Delimiter table for the EBCDIC character set */
 184 static const unsigned char uri_delims[256] = {
 185     T_NUL,                      /* 0x00     */
 186     0,                          /* 0x01     */
 187     0,                          /* 0x02     */
 188     0,                          /* 0x03     */
 189     0,                          /* 0x04     */
 190     0,                          /* 0x05     */
 191     0,                          /* 0x06     */
 192     0,                          /* 0x07     */
 193     0,                          /* 0x08     */
 194     0,                          /* 0x09     */
 195     0,                          /* 0x0a     */
 196     0,                          /* 0x0b     */
 197     0,                          /* 0x0c     */
 198     0,                          /* 0x0d     */
 199     0,                          /* 0x0e     */
 200     0,                          /* 0x0f     */
 201     0,                          /* 0x10     */
 202     0,                          /* 0x11     */
 203     0,                          /* 0x12     */
 204     0,                          /* 0x13     */
 205     0,                          /* 0x14     */
 206     0,                          /* 0x15     */
 207     0,                          /* 0x16     */
 208     0,                          /* 0x17     */
 209     0,                          /* 0x18     */
 210     0,                          /* 0x19     */
 211     0,                          /* 0x1a     */
 212     0,                          /* 0x1b     */
 213     0,                          /* 0x1c     */
 214     0,                          /* 0x1d     */
 215     0,                          /* 0x1e     */
 216     0,                          /* 0x1f     */
 217     0,                          /* 0x20     */
 218     0,                          /* 0x21     */
 219     0,                          /* 0x22     */
 220     0,                          /* 0x23     */
 221     0,                          /* 0x24     */
 222     0,                          /* 0x25     */
 223     0,                          /* 0x26     */
 224     0,                          /* 0x27     */
 225     0,                          /* 0x28     */
 226     0,                          /* 0x29     */
 227     0,                          /* 0x2a     */
 228     0,                          /* 0x2b     */
 229     0,                          /* 0x2c     */
 230     0,                          /* 0x2d     */
 231     0,                          /* 0x2e     */
 232     0,                          /* 0x2f     */
 233     0,                          /* 0x30     */
 234     0,                          /* 0x31     */
 235     0,                          /* 0x32     */
 236     0,                          /* 0x33     */
 237     0,                          /* 0x34     */
 238     0,                          /* 0x35     */
 239     0,                          /* 0x36     */
 240     0,                          /* 0x37     */
 241     0,                          /* 0x38     */
 242     0,                          /* 0x39     */
 243     0,                          /* 0x3a     */
 244     0,                          /* 0x3b     */
 245     0,                          /* 0x3c     */
 246     0,                          /* 0x3d     */
 247     0,                          /* 0x3e     */
 248     0,                          /* 0x3f     */
 249     0,                          /* 0x40 ' ' */
 250     0,                          /* 0x41     */
 251     0,                          /* 0x42     */
 252     0,                          /* 0x43     */
 253     0,                          /* 0x44     */
 254     0,                          /* 0x45     */
 255     0,                          /* 0x46     */
 256     0,                          /* 0x47     */
 257     0,                          /* 0x48     */
 258     0,                          /* 0x49     */
 259     0,                          /* 0x4a '[' */
 260     T_SCHEME,                   /* 0x4b '.' */
 261     0,                          /* 0x4c '<' */
 262     0,                          /* 0x4d '(' */
 263     T_SCHEME,                   /* 0x4e '+' */
 264     0,                          /* 0x4f '!' */
 265     0,                          /* 0x50 '&' */
 266     0,                          /* 0x51     */
 267     0,                          /* 0x52     */
 268     0,                          /* 0x53     */
 269     0,                          /* 0x54     */
 270     0,                          /* 0x55     */
 271     0,                          /* 0x56     */
 272     0,                          /* 0x57     */
 273     0,                          /* 0x58     */
 274     0,                          /* 0x59     */
 275     0,                          /* 0x5a ']' */
 276     0,                          /* 0x5b '$' */
 277     0,                          /* 0x5c '*' */
 278     0,                          /* 0x5d ')' */
 279     0,                          /* 0x5e ';' */
 280     0,                          /* 0x5f '^' */
 281     T_SCHEME,                   /* 0x60 '-' */
 282     T_SLASH,                    /* 0x61 '/' */
 283     0,                          /* 0x62     */
 284     0,                          /* 0x63     */
 285     0,                          /* 0x64     */
 286     0,                          /* 0x65     */
 287     0,                          /* 0x66     */
 288     0,                          /* 0x67     */
 289     0,                          /* 0x68     */
 290     0,                          /* 0x69     */
 291     0,                          /* 0x6a '|' */
 292     0,                          /* 0x6b ',' */
 293     0,                          /* 0x6c '%' */
 294     0,                          /* 0x6d '_' */
 295     0,                          /* 0x6e '>' */
 296     T_QUESTION,                 /* 0x6f '?' */
 297     0,                          /* 0x70     */
 298     0,                          /* 0x71     */
 299     0,                          /* 0x72     */
 300     0,                          /* 0x73     */
 301     0,                          /* 0x74     */
 302     0,                          /* 0x75     */
 303     0,                          /* 0x76     */
 304     0,                          /* 0x77     */
 305     0,                          /* 0x78     */
 306     0,                          /* 0x79 '`' */
 307     0,                          /* 0x7a ':' */
 308     T_HASH,                     /* 0x7b '#' */
 309     0,                          /* 0x7c '@' */
 310     0,                          /* 0x7d ''' */
 311     0,                          /* 0x7e '=' */
 312     0,                          /* 0x7f '"' */
 313     0,                          /* 0x80     */
 314     T_ALPHA,                    /* 0x81 'a' */
 315     T_ALPHA,                    /* 0x82 'b' */
 316     T_ALPHA,                    /* 0x83 'c' */
 317     T_ALPHA,                    /* 0x84 'd' */
 318     T_ALPHA,                    /* 0x85 'e' */
 319     T_ALPHA,                    /* 0x86 'f' */
 320     T_ALPHA,                    /* 0x87 'g' */
 321     T_ALPHA,                    /* 0x88 'h' */
 322     T_ALPHA,                    /* 0x89 'i' */
 323     0,                          /* 0x8a     */
 324     0,                          /* 0x8b     */
 325     0,                          /* 0x8c     */
 326     0,                          /* 0x8d     */
 327     0,                          /* 0x8e     */
 328     0,                          /* 0x8f     */
 329     0,                          /* 0x90     */
 330     T_ALPHA,                    /* 0x91 'j' */
 331     T_ALPHA,                    /* 0x92 'k' */
 332     T_ALPHA,                    /* 0x93 'l' */
 333     T_ALPHA,                    /* 0x94 'm' */
 334     T_ALPHA,                    /* 0x95 'n' */
 335     T_ALPHA,                    /* 0x96 'o' */
 336     T_ALPHA,                    /* 0x97 'p' */
 337     T_ALPHA,                    /* 0x98 'q' */
 338     T_ALPHA,                    /* 0x99 'r' */
 339     0,                          /* 0x9a     */
 340     0,                          /* 0x9b     */
 341     0,                          /* 0x9c     */
 342     0,                          /* 0x9d     */
 343     0,                          /* 0x9e     */
 344     0,                          /* 0x9f     */
 345     0,                          /* 0xa0     */
 346     0,                          /* 0xa1 '~' */
 347     T_ALPHA,                    /* 0xa2 's' */
 348     T_ALPHA,                    /* 0xa3 't' */
 349     T_ALPHA,                    /* 0xa4 'u' */
 350     T_ALPHA,                    /* 0xa5 'v' */
 351     T_ALPHA,                    /* 0xa6 'w' */
 352     T_ALPHA,                    /* 0xa7 'x' */
 353     T_ALPHA,                    /* 0xa8 'y' */
 354     T_ALPHA,                    /* 0xa9 'z' */
 355     0,                          /* 0xaa     */
 356     0,                          /* 0xab     */
 357     0,                          /* 0xac     */
 358     0,                          /* 0xad     */
 359     0,                          /* 0xae     */
 360     0,                          /* 0xaf     */
 361     0,                          /* 0xb0     */
 362     0,                          /* 0xb1     */
 363     0,                          /* 0xb2     */
 364     0,                          /* 0xb3     */
 365     0,                          /* 0xb4     */
 366     0,                          /* 0xb5     */
 367     0,                          /* 0xb6     */
 368     0,                          /* 0xb7     */
 369     0,                          /* 0xb8     */
 370     0,                          /* 0xb9     */
 371     0,                          /* 0xba     */
 372     0,                          /* 0xbb     */
 373     0,                          /* 0xbc     */
 374     0,                          /* 0xbd     */
 375     0,                          /* 0xbe     */
 376     0,                          /* 0xbf     */
 377     0,                          /* 0xc0 '{' */
 378     T_ALPHA,                    /* 0xc1 'A' */
 379     T_ALPHA,                    /* 0xc2 'B' */
 380     T_ALPHA,                    /* 0xc3 'C' */
 381     T_ALPHA,                    /* 0xc4 'D' */
 382     T_ALPHA,                    /* 0xc5 'E' */
 383     T_ALPHA,                    /* 0xc6 'F' */
 384     T_ALPHA,                    /* 0xc7 'G' */
 385     T_ALPHA,                    /* 0xc8 'H' */
 386     T_ALPHA,                    /* 0xc9 'I' */
 387     0,                          /* 0xca     */
 388     0,                          /* 0xcb     */
 389     0,                          /* 0xcc     */
 390     0,                          /* 0xcd     */
 391     0,                          /* 0xce     */
 392     0,                          /* 0xcf     */
 393     0,                          /* 0xd0 '}' */
 394     T_ALPHA,                    /* 0xd1 'J' */
 395     T_ALPHA,                    /* 0xd2 'K' */
 396     T_ALPHA,                    /* 0xd3 'L' */
 397     T_ALPHA,                    /* 0xd4 'M' */
 398     T_ALPHA,                    /* 0xd5 'N' */
 399     T_ALPHA,                    /* 0xd6 'O' */
 400     T_ALPHA,                    /* 0xd7 'P' */
 401     T_ALPHA,                    /* 0xd8 'Q' */
 402     T_ALPHA,                    /* 0xd9 'R' */
 403     0,                          /* 0xda     */
 404     0,                          /* 0xdb     */
 405     0,                          /* 0xdc     */
 406     0,                          /* 0xdd     */
 407     0,                          /* 0xde     */
 408     0,                          /* 0xdf     */
 409     0,                          /* 0xe0 '\' */
 410     0,                          /* 0xe1     */
 411     T_ALPHA,                    /* 0xe2 'S' */
 412     T_ALPHA,                    /* 0xe3 'T' */
 413     T_ALPHA,                    /* 0xe4 'U' */
 414     T_ALPHA,                    /* 0xe5 'V' */
 415     T_ALPHA,                    /* 0xe6 'W' */
 416     T_ALPHA,                    /* 0xe7 'X' */
 417     T_ALPHA,                    /* 0xe8 'Y' */
 418     T_ALPHA,                    /* 0xe9 'Z' */
 419     0,                          /* 0xea     */
 420     0,                          /* 0xeb     */
 421     0,                          /* 0xec     */
 422     0,                          /* 0xed     */
 423     0,                          /* 0xee     */
 424     0,                          /* 0xef     */
 425     T_SCHEME,                   /* 0xf0 '0' */
 426     T_SCHEME,                   /* 0xf1 '1' */
 427     T_SCHEME,                   /* 0xf2 '2' */
 428     T_SCHEME,                   /* 0xf3 '3' */
 429     T_SCHEME,                   /* 0xf4 '4' */
 430     T_SCHEME,                   /* 0xf5 '5' */
 431     T_SCHEME,                   /* 0xf6 '6' */
 432     T_SCHEME,                   /* 0xf7 '7' */
 433     T_SCHEME,                   /* 0xf8 '8' */
 434     T_SCHEME,                   /* 0xf9 '9' */
 435     0,                          /* 0xfa     */
 436     0,                          /* 0xfb     */
 437     0,                          /* 0xfc     */
 438     0,                          /* 0xfd     */
 439     0,                          /* 0xfe     */
 440     0                           /* 0xff     */
 441 };
 442 #else
 443 /* Delimiter table for the ASCII character set */
 444 static const unsigned char uri_delims[256] = {
 445     T_NUL,                      /* 0x00     */
 446     0,                          /* 0x01     */
 447     0,                          /* 0x02     */
 448     0,                          /* 0x03     */
 449     0,                          /* 0x04     */
 450     0,                          /* 0x05     */
 451     0,                          /* 0x06     */
 452     0,                          /* 0x07     */
 453     0,                          /* 0x08     */
 454     0,                          /* 0x09     */
 455     0,                          /* 0x0a     */
 456     0,                          /* 0x0b     */
 457     0,                          /* 0x0c     */
 458     0,                          /* 0x0d     */
 459     0,                          /* 0x0e     */
 460     0,                          /* 0x0f     */
 461     0,                          /* 0x10     */
 462     0,                          /* 0x11     */
 463     0,                          /* 0x12     */
 464     0,                          /* 0x13     */
 465     0,                          /* 0x14     */
 466     0,                          /* 0x15     */
 467     0,                          /* 0x16     */
 468     0,                          /* 0x17     */
 469     0,                          /* 0x18     */
 470     0,                          /* 0x19     */
 471     0,                          /* 0x1a     */
 472     0,                          /* 0x1b     */
 473     0,                          /* 0x1c     */
 474     0,                          /* 0x1d     */
 475     0,                          /* 0x1e     */
 476     0,                          /* 0x1f     */
 477     0,                          /* 0x20 ' ' */
 478     0,                          /* 0x21 '!' */
 479     0,                          /* 0x22 '"' */
 480     T_HASH,                     /* 0x23 '#' */
 481     0,                          /* 0x24 '$' */
 482     0,                          /* 0x25 '%' */
 483     0,                          /* 0x26 '&' */
 484     0,                          /* 0x27 ''' */
 485     0,                          /* 0x28 '(' */
 486     0,                          /* 0x29 ')' */
 487     0,                          /* 0x2a '*' */
 488     T_SCHEME,                   /* 0x2b '+' */
 489     0,                          /* 0x2c ',' */
 490     T_SCHEME,                   /* 0x2d '-' */
 491     T_SCHEME,                   /* 0x2e '.' */
 492     T_SLASH,                    /* 0x2f '/' */
 493     T_SCHEME,                   /* 0x30 '0' */
 494     T_SCHEME,                   /* 0x31 '1' */
 495     T_SCHEME,                   /* 0x32 '2' */
 496     T_SCHEME,                   /* 0x33 '3' */
 497     T_SCHEME,                   /* 0x34 '4' */
 498     T_SCHEME,                   /* 0x35 '5' */
 499     T_SCHEME,                   /* 0x36 '6' */
 500     T_SCHEME,                   /* 0x37 '7' */
 501     T_SCHEME,                   /* 0x38 '8' */
 502     T_SCHEME,                   /* 0x39 '9' */
 503     0,                          /* 0x3a ':' */
 504     0,                          /* 0x3b ';' */
 505     0,                          /* 0x3c '<' */
 506     0,                          /* 0x3d '=' */
 507     0,                          /* 0x3e '>' */
 508     T_QUESTION,                 /* 0x3f '?' */
 509     0,                          /* 0x40 '@' */
 510     T_ALPHA,                    /* 0x41 'A' */
 511     T_ALPHA,                    /* 0x42 'B' */
 512     T_ALPHA,                    /* 0x43 'C' */
 513     T_ALPHA,                    /* 0x44 'D' */
 514     T_ALPHA,                    /* 0x45 'E' */
 515     T_ALPHA,                    /* 0x46 'F' */
 516     T_ALPHA,                    /* 0x47 'G' */
 517     T_ALPHA,                    /* 0x48 'H' */
 518     T_ALPHA,                    /* 0x49 'I' */
 519     T_ALPHA,                    /* 0x4a 'J' */
 520     T_ALPHA,                    /* 0x4b 'K' */
 521     T_ALPHA,                    /* 0x4c 'L' */
 522     T_ALPHA,                    /* 0x4d 'M' */
 523     T_ALPHA,                    /* 0x4e 'N' */
 524     T_ALPHA,                    /* 0x4f 'O' */
 525     T_ALPHA,                    /* 0x50 'P' */
 526     T_ALPHA,                    /* 0x51 'Q' */
 527     T_ALPHA,                    /* 0x52 'R' */
 528     T_ALPHA,                    /* 0x53 'S' */
 529     T_ALPHA,                    /* 0x54 'T' */
 530     T_ALPHA,                    /* 0x55 'U' */
 531     T_ALPHA,                    /* 0x56 'V' */
 532     T_ALPHA,                    /* 0x57 'W' */
 533     T_ALPHA,                    /* 0x58 'X' */
 534     T_ALPHA,                    /* 0x59 'Y' */
 535     T_ALPHA,                    /* 0x5a 'Z' */
 536     0,                          /* 0x5b '[' */
 537     0,                          /* 0x5c '\' */
 538     0,                          /* 0x5d ']' */
 539     0,                          /* 0x5e '^' */
 540     0,                          /* 0x5f '_' */
 541     0,                          /* 0x60 '`' */
 542     T_ALPHA,                    /* 0x61 'a' */
 543     T_ALPHA,                    /* 0x62 'b' */
 544     T_ALPHA,                    /* 0x63 'c' */
 545     T_ALPHA,                    /* 0x64 'd' */
 546     T_ALPHA,                    /* 0x65 'e' */
 547     T_ALPHA,                    /* 0x66 'f' */
 548     T_ALPHA,                    /* 0x67 'g' */
 549     T_ALPHA,                    /* 0x68 'h' */
 550     T_ALPHA,                    /* 0x69 'i' */
 551     T_ALPHA,                    /* 0x6a 'j' */
 552     T_ALPHA,                    /* 0x6b 'k' */
 553     T_ALPHA,                    /* 0x6c 'l' */
 554     T_ALPHA,                    /* 0x6d 'm' */
 555     T_ALPHA,                    /* 0x6e 'n' */
 556     T_ALPHA,                    /* 0x6f 'o' */
 557     T_ALPHA,                    /* 0x70 'p' */
 558     T_ALPHA,                    /* 0x71 'q' */
 559     T_ALPHA,                    /* 0x72 'r' */
 560     T_ALPHA,                    /* 0x73 's' */
 561     T_ALPHA,                    /* 0x74 't' */
 562     T_ALPHA,                    /* 0x75 'u' */
 563     T_ALPHA,                    /* 0x76 'v' */
 564     T_ALPHA,                    /* 0x77 'w' */
 565     T_ALPHA,                    /* 0x78 'x' */
 566     T_ALPHA,                    /* 0x79 'y' */
 567     T_ALPHA,                    /* 0x7a 'z' */
 568     0,                          /* 0x7b '{' */
 569     0,                          /* 0x7c '|' */
 570     0,                          /* 0x7d '}' */
 571     0,                          /* 0x7e '~' */
 572     0,                          /* 0x7f     */
 573     0,                          /* 0x80     */
 574     0,                          /* 0x81     */
 575     0,                          /* 0x82     */
 576     0,                          /* 0x83     */
 577     0,                          /* 0x84     */
 578     0,                          /* 0x85     */
 579     0,                          /* 0x86     */
 580     0,                          /* 0x87     */
 581     0,                          /* 0x88     */
 582     0,                          /* 0x89     */
 583     0,                          /* 0x8a     */
 584     0,                          /* 0x8b     */
 585     0,                          /* 0x8c     */
 586     0,                          /* 0x8d     */
 587     0,                          /* 0x8e     */
 588     0,                          /* 0x8f     */
 589     0,                          /* 0x90     */
 590     0,                          /* 0x91     */
 591     0,                          /* 0x92     */
 592     0,                          /* 0x93     */
 593     0,                          /* 0x94     */
 594     0,                          /* 0x95     */
 595     0,                          /* 0x96     */
 596     0,                          /* 0x97     */
 597     0,                          /* 0x98     */
 598     0,                          /* 0x99     */
 599     0,                          /* 0x9a     */
 600     0,                          /* 0x9b     */
 601     0,                          /* 0x9c     */
 602     0,                          /* 0x9d     */
 603     0,                          /* 0x9e     */
 604     0,                          /* 0x9f     */
 605     0,                          /* 0xa0     */
 606     0,                          /* 0xa1     */
 607     0,                          /* 0xa2     */
 608     0,                          /* 0xa3     */
 609     0,                          /* 0xa4     */
 610     0,                          /* 0xa5     */
 611     0,                          /* 0xa6     */
 612     0,                          /* 0xa7     */
 613     0,                          /* 0xa8     */
 614     0,                          /* 0xa9     */
 615     0,                          /* 0xaa     */
 616     0,                          /* 0xab     */
 617     0,                          /* 0xac     */
 618     0,                          /* 0xad     */
 619     0,                          /* 0xae     */
 620     0,                          /* 0xaf     */
 621     0,                          /* 0xb0     */
 622     0,                          /* 0xb1     */
 623     0,                          /* 0xb2     */
 624     0,                          /* 0xb3     */
 625     0,                          /* 0xb4     */
 626     0,                          /* 0xb5     */
 627     0,                          /* 0xb6     */
 628     0,                          /* 0xb7     */
 629     0,                          /* 0xb8     */
 630     0,                          /* 0xb9     */
 631     0,                          /* 0xba     */
 632     0,                          /* 0xbb     */
 633     0,                          /* 0xbc     */
 634     0,                          /* 0xbd     */
 635     0,                          /* 0xbe     */
 636     0,                          /* 0xbf     */
 637     0,                          /* 0xc0     */
 638     0,                          /* 0xc1     */
 639     0,                          /* 0xc2     */
 640     0,                          /* 0xc3     */
 641     0,                          /* 0xc4     */
 642     0,                          /* 0xc5     */
 643     0,                          /* 0xc6     */
 644     0,                          /* 0xc7     */
 645     0,                          /* 0xc8     */
 646     0,                          /* 0xc9     */
 647     0,                          /* 0xca     */
 648     0,                          /* 0xcb     */
 649     0,                          /* 0xcc     */
 650     0,                          /* 0xcd     */
 651     0,                          /* 0xce     */
 652     0,                          /* 0xcf     */
 653     0,                          /* 0xd0     */
 654     0,                          /* 0xd1     */
 655     0,                          /* 0xd2     */
 656     0,                          /* 0xd3     */
 657     0,                          /* 0xd4     */
 658     0,                          /* 0xd5     */
 659     0,                          /* 0xd6     */
 660     0,                          /* 0xd7     */
 661     0,                          /* 0xd8     */
 662     0,                          /* 0xd9     */
 663     0,                          /* 0xda     */
 664     0,                          /* 0xdb     */
 665     0,                          /* 0xdc     */
 666     0,                          /* 0xdd     */
 667     0,                          /* 0xde     */
 668     0,                          /* 0xdf     */
 669     0,                          /* 0xe0     */
 670     0,                          /* 0xe1     */
 671     0,                          /* 0xe2     */
 672     0,                          /* 0xe3     */
 673     0,                          /* 0xe4     */
 674     0,                          /* 0xe5     */
 675     0,                          /* 0xe6     */
 676     0,                          /* 0xe7     */
 677     0,                          /* 0xe8     */
 678     0,                          /* 0xe9     */
 679     0,                          /* 0xea     */
 680     0,                          /* 0xeb     */
 681     0,                          /* 0xec     */
 682     0,                          /* 0xed     */
 683     0,                          /* 0xee     */
 684     0,                          /* 0xef     */
 685     0,                          /* 0xf0     */
 686     0,                          /* 0xf1     */
 687     0,                          /* 0xf2     */
 688     0,                          /* 0xf3     */
 689     0,                          /* 0xf4     */
 690     0,                          /* 0xf5     */
 691     0,                          /* 0xf6     */
 692     0,                          /* 0xf7     */
 693     0,                          /* 0xf8     */
 694     0,                          /* 0xf9     */
 695     0,                          /* 0xfa     */
 696     0,                          /* 0xfb     */
 697     0,                          /* 0xfc     */
 698     0,                          /* 0xfd     */
 699     0,                          /* 0xfe     */
 700     0                           /* 0xff     */
 701 };
 702 #endif
 703
 704
 705 /* it works like this:
 706     if (uri_delims[ch] & NOTEND_foobar) {
 707         then we're not at a delimiter for foobar
 708     }
 709 */
 710
 711 #define NOTEND_HOSTINFO   (T_SLASH | T_QUESTION | T_HASH | T_NUL)
 712 #define NOTEND_PATH       (T_QUESTION | T_HASH | T_NUL)
 713
 714 /* parse_uri_components():
 715  * Parse a given URI, fill in all supplied fields of a uri_components
 716  * structure. This eliminates the necessity of extracting host, port,
 717  * path, query info repeatedly in the modules.
 718  * Side effects:
 719  *  - fills in fields of uri_components *uptr
 720  *  - none on any of the r->* fields
 721  */
 722 APU_DECLARE(apr_status_t) apr_uri_parse(apr_pool_t *p, const char *uri,
 723                                         apr_uri_t *uptr)
 724 {
 725     const char *s;
 726     const char *s1;
 727     const char *hostinfo;
 728     char *endstr;
 729     int port;
 730     int v6_offset1 = 0, v6_offset2 = 0;
 731
 732     /* Initialize the structure. parse_uri() and parse_uri_components()
 733      * can be called more than once per request.
 734      */
 735     memset (uptr, '\0', sizeof(*uptr));
 736     uptr->is_initialized = 1;
 737
 738     /* We assume the processor has a branch predictor like most --
 739      * it assumes forward branches are untaken and backwards are taken.  That's
 740      * the reason for the gotos.  -djg
 741      */
 742     if (uri[0] == '/') {
 743         /* RFC2396 #4.3 says that two leading slashes mean we have an
 744          * authority component, not a path!  Fixing this looks scary
 745          * with the gotos here.  But if the existing logic is valid,
 746          * then presumably a goto pointing to deal_with_authority works.
 747          *
 748          * RFC2396 describes this as resolving an ambiguity.  In the
 749          * case of three or more slashes there would seem to be no
 750          * ambiguity, so it is a path after all.
 751          */
 752         if (uri[1] == '/' && uri[2] != '/') {
 753             s = uri + 2 ;
 754             goto deal_with_authority ;
 755         }
 756
 757 deal_with_path:
 758         /* we expect uri to point to first character of path ... remember
 759          * that the path could be empty -- http://foobar?query for example
 760          */
 761         s = uri;
 762         while ((uri_delims[*(unsigned char *)s] & NOTEND_PATH) == 0) {
 763             ++s;
 764         }
 765         if (s != uri) {
 766             uptr->path = apr_pstrmemdup(p, uri, s - uri);
 767         }
 768         if (*s == 0) {
 769             return APR_SUCCESS;
 770         }
 771         if (*s == '?') {
 772             ++s;
 773             s1 = strchr(s, '#');
 774             if (s1) {
 775                 uptr->fragment = apr_pstrdup(p, s1 + 1);
 776                 uptr->query = apr_pstrmemdup(p, s, s1 - s);
 777             }
 778             else {
 779                 uptr->query = apr_pstrdup(p, s);
 780             }
 781             return APR_SUCCESS;
 782         }
 783         /* otherwise it's a fragment */
 784         uptr->fragment = apr_pstrdup(p, s + 1);
 785         return APR_SUCCESS;
 786     }
 787
 788     /* find the scheme: */
 789     s = uri;
 790     /* first char must be letter */
 791     if (uri_delims[*(unsigned char *)s] & T_ALPHA) {
 792         ++s;
 793         while ((uri_delims[*(unsigned char *)s] & (T_ALPHA|T_SCHEME)))
 794             ++s;
 795     }
 796     /* scheme must be non-empty and followed by : */
 797     if (s != uri && s[0] == ':') {
 798         uptr->scheme = apr_pstrmemdup(p, uri, s - uri);
 799         s++;
 800     }
 801     else {
 802         /* No valid scheme, restart from the beginning */
 803         s = uri;
 804     }
 805
 806     if (s[0] != '/' || s[1] != '/') {
 807         if (uri == s) {
 808             /*
 809              * RFC 3986 3.3: If we have no scheme and no authority,
 810              * the leading segment of a relative path must not contain a ':'.
 811              */
 812             char *first_slash = strchr(uri, '/');
 813             if (first_slash) {
 814                 while (s < first_slash) {
 815                     if (s[0] == ':')
 816                         return APR_EGENERAL;
 817                     ++s;
 818                 }
 819                 /* no scheme but relative path, e.g. '../image.jpg' */
 820             }
 821             else {
 822                 if (strchr(uri, ':') != NULL)
 823                     return APR_EGENERAL;
 824                 /* no scheme, no slash, but relative path, e.g. 'image.jpg' */
 825             }
 826             goto deal_with_path;
 827         }
 828         /* scheme and relative path */
 829         uri = s;
 830         goto deal_with_path;
 831     }
 832
 833     s += 2;
 834
 835 deal_with_authority:
 836     hostinfo = s;
 837     while ((uri_delims[*(unsigned char *)s] & NOTEND_HOSTINFO) == 0) {
 838         ++s;
 839     }
 840     uri = s;        /* whatever follows hostinfo is start of uri */
 841     uptr->hostinfo = apr_pstrmemdup(p, hostinfo, uri - hostinfo);
 842
 843     /* If there's a username:password@host:port, the @ we want is the last @...
 844      * too bad there's no memrchr()... For the C purists, note that hostinfo
 845      * is definitely not the first character of the original uri so therefore
 846      * &hostinfo[-1] < &hostinfo[0] ... and this loop is valid C.
 847      */
 848     do {
 849         --s;
 850     } while (s >= hostinfo && *s != '@');
 851     if (s < hostinfo) {
 852         /* again we want the common case to be fall through */
 853 deal_with_host:
 854         /* We expect hostinfo to point to the first character of
 855          * the hostname.  If there's a port it is the first colon,
 856          * except with IPv6.
 857          */
 858         if (*hostinfo == '[') {
 859             v6_offset1 = 1;
 860             v6_offset2 = 2;
 861             s = memchr(hostinfo, ']', uri - hostinfo);
 862             if (s == NULL) {
 863                 return APR_EGENERAL;
 864             }
 865             if (*++s != ':') {
 866                 s = NULL; /* no port */
 867             }
 868         }
 869         else {
 870             s = memchr(hostinfo, ':', uri - hostinfo);
 871         }
 872         if (s == NULL) {
 873             /* we expect the common case to have no port */
 874             uptr->hostname = apr_pstrmemdup(p,
 875                                             hostinfo + v6_offset1,
 876                                             uri - hostinfo - v6_offset2);
 877             goto deal_with_path;
 878         }
 879         uptr->hostname = apr_pstrmemdup(p,
 880                                         hostinfo + v6_offset1,
 881                                         s - hostinfo - v6_offset2);
 882         ++s;
 883         uptr->port_str = apr_pstrmemdup(p, s, uri - s);
 884         if (uri != s) {
 885             port = strtol(uptr->port_str, &endstr, 10);
 886             uptr->port = port;
 887             if (*endstr == '\0') {
 888                 goto deal_with_path;
 889             }
 890             /* Invalid characters after ':' found */
 891             return APR_EGENERAL;
 892         }
 893         uptr->port = apr_uri_port_of_scheme(uptr->scheme);
 894         goto deal_with_path;
 895     }
 896
 897     /* first colon delimits username:password */
 898     s1 = memchr(hostinfo, ':', s - hostinfo);
 899     if (s1) {
 900         uptr->user = apr_pstrmemdup(p, hostinfo, s1 - hostinfo);
 901         ++s1;
 902         uptr->password = apr_pstrmemdup(p, s1, s - s1);
 903     }
 904     else {
 905         uptr->user = apr_pstrmemdup(p, hostinfo, s - hostinfo);
 906     }
 907     hostinfo = s + 1;
 908     goto deal_with_host;
 909 }
 910
 911 /* Special case for CONNECT parsing: it comes with the hostinfo part only */
 912 /* See the INTERNET-DRAFT document "Tunneling SSL Through a WWW Proxy"
 913  * currently at http://www.mcom.com/newsref/std/tunneling_ssl.html
 914  * for the format of the "CONNECT host:port HTTP/1.0" request
 915  */
 916 APU_DECLARE(apr_status_t) apr_uri_parse_hostinfo(apr_pool_t *p,
 917                                                  const char *hostinfo,
 918                                                  apr_uri_t *uptr)
 919 {
 920     const char *s;
 921     char *endstr;
 922     const char *rsb;
 923     int v6_offset1 = 0;
 924
 925     /* Initialize the structure. parse_uri() and parse_uri_components()
 926      * can be called more than once per request.
 927      */
 928     memset(uptr, '\0', sizeof(*uptr));
 929     uptr->is_initialized = 1;
 930     uptr->hostinfo = apr_pstrdup(p, hostinfo);
 931
 932     /* We expect hostinfo to point to the first character of
 933      * the hostname.  There must be a port, separated by a colon
 934      */
 935     if (*hostinfo == '[') {
 936         if ((rsb = strchr(hostinfo, ']')) == NULL ||
 937             *(rsb + 1) != ':') {
 938             return APR_EGENERAL;
 939         }
 940         /* literal IPv6 address */
 941         s = rsb + 1;
 942         ++hostinfo;
 943         v6_offset1 = 1;
 944     }
 945     else {
 946         s = strchr(hostinfo, ':');
 947     }
 948     if (s == NULL) {
 949         return APR_EGENERAL;
 950     }
 951     uptr->hostname = apr_pstrndup(p, hostinfo, s - hostinfo - v6_offset1);
 952     ++s;
 953     uptr->port_str = apr_pstrdup(p, s);
 954     if (*s != '\0') {
 955         uptr->port = (unsigned short) strtol(uptr->port_str, &endstr, 10);
 956         if (*endstr == '\0') {
 957             return APR_SUCCESS;
 958         }
 959         /* Invalid characters after ':' found */
 960     }
 961     return APR_EGENERAL;
 962 }