chiark - git - mdw - disorder/blob - libtests/t-mime.c

   1 /*
   2  * This file is part of DisOrder.
   3  * Copyright (C) 2005, 2007, 2008 Richard Kettlewell
   4  *
   5  * This program is free software: you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation, either version 3 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License
  16  * along with this program.  If not, see <http://www.gnu.org/licenses/>.
  17  */
  18 #include "test.h"
  19
  20 static int test_multipart_callback(const char *s, void *u) {
  21   struct vector *parts = u;
  22
  23   vector_append(parts, (char *)s);
  24   return 0;
  25 }
  26
  27 static int header_callback(const char *name, const char *value,
  28                            void *u) {
  29   hash *const h = u;
  30
  31   hash_add(h, name, &value, HASH_INSERT);
  32   return 0;
  33 }
  34
  35 static void test_mime(void) {
  36   char *t, *n, *v;
  37   struct vector parts[1];
  38   struct kvp *k;
  39   const char *s, *cs, *enc;
  40   hash *h;
  41
  42   t = 0;
  43   k = 0;
  44   insist(!mime_content_type("text/plain", &t, &k));
  45   check_string(t, "text/plain");
  46   insist(k == 0);
  47
  48   insist(mime_content_type("TEXT ((broken) comment", &t, &k) < 0);
  49   insist(mime_content_type("TEXT ((broken) comment\\", &t, &k) < 0);
  50
  51   t = 0;
  52   k = 0;
  53   insist(!mime_content_type("TEXT ((nested)\\ comment) /plain", &t, &k));
  54   check_string(t, "text/plain");
  55   insist(k == 0);
  56
  57   t = 0;
  58   k = 0;
  59   insist(!mime_content_type(" text/plain ; Charset=\"utf-\\8\"", &t, &k));
  60   check_string(t, "text/plain");
  61   insist(k != 0);
  62   insist(k->next == 0);
  63   check_string(k->name, "charset");
  64   check_string(k->value, "utf-8");
  65
  66   t = 0;
  67   k = 0;
  68   insist(!mime_content_type("text/plain;charset = ISO-8859-1 ", &t, &k));
  69   insist(k != 0);
  70   insist(k->next == 0);
  71   check_string(t, "text/plain");
  72   check_string(k->name, "charset");
  73   check_string(k->value, "ISO-8859-1");
  74
  75   t = n = v = 0;
  76   insist(!mime_rfc2388_content_disposition("form-data; name=\"field1\"", &t, &n, &v));
  77   check_string(t, "form-data");
  78   check_string(n, "name");
  79   check_string(v, "field1");
  80
  81   insist(!mime_rfc2388_content_disposition("inline", &t, &n, &v));
  82   check_string(t, "inline");
  83   insist(n == 0);
  84   insist(v == 0);
  85
  86   /* Current versions of the code only understand a single arg to these
  87    * headers.  This is a bug at the level they work at but suffices for
  88    * DisOrder's current purposes. */
  89
  90   insist(!mime_rfc2388_content_disposition(
  91               "attachment; filename=genome.jpeg;\n"
  92               "modification-date=\"Wed, 12 Feb 1997 16:29:51 -0500\"",
  93          &t, &n, &v));
  94   check_string(t, "attachment");
  95   check_string(n, "filename");
  96   check_string(v, "genome.jpeg");
  97
  98   vector_init(parts);
  99   insist(mime_multipart("--outer\r\n"
 100                         "Content-Type: text/plain\r\n"
 101                         "Content-Disposition: inline\r\n"
 102                         "Content-Description: text-part-1\r\n"
 103                         "\r\n"
 104                         "Some text goes here\r\n"
 105                         "\r\n"
 106                         "--outer\r\n"
 107                         "Content-Type: multipart/mixed; boundary=inner\r\n"
 108                         "Content-Disposition: attachment\r\n"
 109                         "Content-Description: multipart-2\r\n"
 110                         "\r\n"
 111                         "--inner\r\n"
 112                         "Content-Type: text/plain\r\n"
 113                         "Content-Disposition: inline\r\n"
 114                         "Content-Description: text-part-2\r\n"
 115                         "\r\n"
 116                         "Some more text here.\r\n"
 117                         "\r\n"
 118                         "--inner\r\n"
 119                         "Content-Type: image/jpeg\r\n"
 120                         "Content-Disposition: attachment\r\n"
 121                         "Content-Description: jpeg-1\r\n"
 122                         "\r\n"
 123                         "<jpeg data>\r\n"
 124                         "--inner--\r\n"
 125                         "--outer--\r\n",
 126                         test_multipart_callback,
 127                         "outer",
 128                         parts) == 0);
 129   check_integer(parts->nvec, 2);
 130   check_string(parts->vec[0],
 131                "Content-Type: text/plain\r\n"
 132                "Content-Disposition: inline\r\n"
 133                "Content-Description: text-part-1\r\n"
 134                "\r\n"
 135                "Some text goes here\r\n");
 136   check_string(parts->vec[1],
 137                "Content-Type: multipart/mixed; boundary=inner\r\n"
 138                "Content-Disposition: attachment\r\n"
 139                "Content-Description: multipart-2\r\n"
 140                "\r\n"
 141                "--inner\r\n"
 142                "Content-Type: text/plain\r\n"
 143                "Content-Disposition: inline\r\n"
 144                "Content-Description: text-part-2\r\n"
 145                "\r\n"
 146                "Some more text here.\r\n"
 147                "\r\n"
 148                "--inner\r\n"
 149                "Content-Type: image/jpeg\r\n"
 150                "Content-Disposition: attachment\r\n"
 151                "Content-Description: jpeg-1\r\n"
 152                "\r\n"
 153                "<jpeg data>\r\n"
 154                "--inner--");
 155   /* No trailing CRLF is _correct_ - see RFC2046 5.1.1 note regarding CRLF
 156    * preceding the boundary delimiter line.  An implication of this is that we
 157    * must cope with partial lines at the end of the input when recursively
 158    * decomposing a multipart message. */
 159   vector_init(parts);
 160   insist(mime_multipart("--inner\r\n"
 161                         "Content-Type: text/plain\r\n"
 162                         "Content-Disposition: inline\r\n"
 163                         "Content-Description: text-part-2\r\n"
 164                         "\r\n"
 165                         "Some more text here.\r\n"
 166                         "\r\n"
 167                         "--inner\r\n"
 168                         "Content-Type: image/jpeg\r\n"
 169                         "Content-Disposition: attachment\r\n"
 170                         "Content-Description: jpeg-1\r\n"
 171                         "\r\n"
 172                         "<jpeg data>\r\n"
 173                         "--inner--",
 174                         test_multipart_callback,
 175                         "inner",
 176                         parts) == 0);
 177   check_integer(parts->nvec, 2);
 178   check_string(parts->vec[0],
 179                "Content-Type: text/plain\r\n"
 180                "Content-Disposition: inline\r\n"
 181                "Content-Description: text-part-2\r\n"
 182                "\r\n"
 183                "Some more text here.\r\n");
 184   check_string(parts->vec[1],
 185                "Content-Type: image/jpeg\r\n"
 186                "Content-Disposition: attachment\r\n"
 187                "Content-Description: jpeg-1\r\n"
 188                "\r\n"
 189                "<jpeg data>");
 190
 191   /* Bogus inputs to mime_multipart() */
 192   fprintf(stderr, "expect two mime_multipart errors:\n");
 193   insist(mime_multipart("--inner\r\n"
 194                         "Content-Type: text/plain\r\n"
 195                         "Content-Disposition: inline\r\n"
 196                         "Content-Description: text-part-2\r\n"
 197                         "\r\n"
 198                         "Some more text here.\r\n"
 199                         "\r\n"
 200                         "--inner\r\n"
 201                         "Content-Type: image/jpeg\r\n"
 202                         "Content-Disposition: attachment\r\n"
 203                         "Content-Description: jpeg-1\r\n"
 204                         "\r\n"
 205                         "<jpeg data>\r\n",
 206                         test_multipart_callback,
 207                         "inner",
 208                         parts) == -1);
 209   insist(mime_multipart("--wrong\r\n"
 210                         "Content-Type: text/plain\r\n"
 211                         "Content-Disposition: inline\r\n"
 212                         "Content-Description: text-part-2\r\n"
 213                         "\r\n"
 214                         "Some more text here.\r\n"
 215                         "\r\n"
 216                         "--inner--\r\n",
 217                         test_multipart_callback,
 218                         "inner",
 219                         parts) == -1);
 220
 221   /* XXX mime_parse */
 222
 223   check_string(mime_qp(""), "");
 224   check_string(mime_qp("foobar"), "foobar");
 225   check_string(mime_qp("foo=20bar"), "foo bar");
 226   check_string(mime_qp("x \r\ny"), "x\r\ny");
 227   check_string(mime_qp("x=\r\ny"), "xy");
 228   check_string(mime_qp("x= \r\ny"), "xy");
 229   check_string(mime_qp("x =\r\ny"), "x y");
 230   check_string(mime_qp("x = \r\ny"), "x y");
 231
 232   check_string(mime_to_qp(""), "");
 233   check_string(mime_to_qp("foobar\n"), "foobar\n");
 234   check_string(mime_to_qp("foobar \n"), "foobar=20\n");
 235   check_string(mime_to_qp("foobar\t\n"), "foobar=09\n");
 236   check_string(mime_to_qp("foobar \t \n"), "foobar=20=09=20\n");
 237   check_string(mime_to_qp(" foo=bar"), " foo=3Dbar\n");
 238   check_string(mime_to_qp("copyright \xC2\xA9"), "copyright =C2=A9\n");
 239   check_string(mime_to_qp("foo\nbar\nbaz\n"), "foo\nbar\nbaz\n");
 240   check_string(mime_to_qp("wibble wobble wibble wobble wibble wobble wibble wobble wibble wobble wibble"), "wibble wobble wibble wobble wibble wobble wibble wobble wibble wobble wibb=\nle\n");
 241
 242   /* from RFC2045 */
 243   check_string(mime_qp("Now's the time =\r\n"
 244 "for all folk to come=\r\n"
 245 " to the aid of their country."),
 246                "Now's the time for all folk to come to the aid of their country.");
 247
 248 #define check_base64(encoded, decoded) do {                     \
 249     size_t ns;                                                  \
 250     check_string(mime_base64(encoded, &ns), decoded);           \
 251     insist(ns == (sizeof decoded) - 1);                         \
 252     check_string(mime_to_base64((const uint8_t *)decoded,       \
 253                                          (sizeof decoded) - 1), \
 254                  encoded);                                      \
 255   } while(0)
 256
 257
 258   check_base64("",  "");
 259   check_base64("BBBB", "\x04\x10\x41");
 260   check_base64("////", "\xFF\xFF\xFF");
 261   check_base64("//BB", "\xFF\xF0\x41");
 262   check_base64("BBBB//BB////",
 263              "\x04\x10\x41" "\xFF\xF0\x41" "\xFF\xFF\xFF");
 264   check_base64("BBBBBA==",
 265                "\x04\x10\x41" "\x04");
 266   check_base64("BBBBBBA=",
 267                "\x04\x10\x41" "\x04\x10");
 268
 269   /* Check that decoding handles various kinds of rubbish OK */
 270   check_string(mime_base64("B B B B  / / B B / / / /", 0),
 271              "\x04\x10\x41" "\xFF\xF0\x41" "\xFF\xFF\xFF");
 272   check_string(mime_base64("B\r\nBBB.// B-B//~//", 0),
 273                "\x04\x10\x41" "\xFF\xF0\x41" "\xFF\xFF\xFF");
 274   check_string(mime_base64("BBBB BB==", 0),
 275                "\x04\x10\x41" "\x04");
 276   check_string(mime_base64("BBBB BB = =", 0),
 277                "\x04\x10\x41" "\x04");
 278   check_string(mime_base64("BBBB BBB=", 0),
 279                "\x04\x10\x41" "\x04\x10");
 280   check_string(mime_base64("BBBB BBB = ", 0),
 281                "\x04\x10\x41" "\x04\x10");
 282   check_string(mime_base64("BBBB=", 0),
 283                "\x04\x10\x41");
 284   check_string(mime_base64("BBBBBB==", 0),
 285                "\x04\x10\x41" "\x04");
 286   check_string(mime_base64("BBBBBBB=", 0),
 287                "\x04\x10\x41" "\x04\x10");
 288   /* Not actually valid base64 */
 289   check_string(mime_base64("BBBBx=", 0),
 290                "\x04\x10\x41");
 291
 292   h = hash_new(sizeof (char *));
 293   s = mime_parse("From: sender@example.com\r\n"
 294                  "To: rcpt@example.com\r\n"
 295                  "Subject: test #1\r\n"
 296                  "\r\n"
 297                  "body\r\n",
 298                  header_callback, h);
 299   insist(s != 0);
 300   check_string(*(char **)hash_find(h, "from"), "sender@example.com");
 301   check_string(*(char **)hash_find(h, "to"), "rcpt@example.com");
 302   check_string(*(char **)hash_find(h, "subject"), "test #1");
 303   check_string(s, "body\r\n");
 304
 305   h = hash_new(sizeof (char *));
 306   s = mime_parse("FROM: sender@example.com\r\n"
 307                  "TO: rcpt@example.com\r\n"
 308                  "SUBJECT: test #1\r\n"
 309                  "CONTENT-TRANSFER-ENCODING: 7bit\r\n"
 310                  "\r\n"
 311                  "body\r\n",
 312                  header_callback, h);
 313   insist(s != 0);
 314   check_string(*(char **)hash_find(h, "from"), "sender@example.com");
 315   check_string(*(char **)hash_find(h, "to"), "rcpt@example.com");
 316   check_string(*(char **)hash_find(h, "subject"), "test #1");
 317   check_string(*(char **)hash_find(h, "content-transfer-encoding"), "7bit");
 318   check_string(s, "body\r\n");
 319
 320   h = hash_new(sizeof (char *));
 321   s = mime_parse("From: sender@example.com\r\n"
 322                  "To:    \r\n"
 323                  "     rcpt@example.com\r\n"
 324                  "Subject: test #1\r\n"
 325                  "MIME-Version: 1.0\r\n"
 326                  "Content-Type: text/plain\r\n"
 327                  "Content-Transfer-Encoding: BASE64\r\n"
 328                  "\r\n"
 329                  "d2liYmxlDQo=\r\n",
 330                  header_callback, h);
 331   insist(s != 0);
 332   check_string(*(char **)hash_find(h, "from"), "sender@example.com");
 333   check_string(*(char **)hash_find(h, "to"), "rcpt@example.com");
 334   check_string(*(char **)hash_find(h, "subject"), "test #1");
 335   check_string(*(char **)hash_find(h, "mime-version"), "1.0");
 336   check_string(*(char **)hash_find(h, "content-type"), "text/plain");
 337   check_string(*(char **)hash_find(h, "content-transfer-encoding"), "BASE64");
 338   check_string(s, "wibble\r\n");
 339
 340 #define CHECK_QUOTE(INPUT, EXPECT) do {                 \
 341   s = quote822(INPUT, 0);                               \
 342   insist(s != 0);                                       \
 343   check_string(s, EXPECT);                              \
 344   s = mime_parse_word(s, &t, mime_http_separator);      \
 345   check_string(t, INPUT);                               \
 346 } while(0)
 347   CHECK_QUOTE("wibble", "wibble");
 348   CHECK_QUOTE("wibble spong", "\"wibble spong\"");
 349   CHECK_QUOTE("wibble\\spong", "\"wibble\\\\spong\"");
 350   CHECK_QUOTE("wibble\"spong", "\"wibble\\\"spong\"");
 351   CHECK_QUOTE("(wibble)", "\"(wibble)\"");
 352
 353   s = mime_encode_text("wibble\n", &cs, &enc);
 354   insist(s != 0);
 355   check_string(s, "wibble\n");
 356   check_string(cs, "us-ascii");
 357   check_string(enc, "7bit");
 358
 359   s = mime_encode_text("wibble\xC3\xB7\n", &cs, &enc);
 360   insist(s != 0);
 361   check_string(s, "wibble=C3=B7\n");
 362   check_string(cs, "utf-8");
 363   check_string(enc, "quoted-printable");
 364 }
 365
 366 TEST(mime);
 367
 368 /*
 369 Local Variables:
 370 c-basic-offset:2
 371 comment-column:40
 372 fill-column:79
 373 indent-tabs-mode:nil
 374 End:
 375 */