chiark - git - mdw - disorder/blob - lib/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 2 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, 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.
  14  *
  15  * You should have received a copy of the GNU General Public License
  16  * along with this program; if not, write to the Free Software
  17  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307
  18  * USA
  19  */
  20 #include "test.h"
  21
  22 static int test_multipart_callback(const char *s, void *u) {
  23   struct vector *parts = u;
  24
  25   vector_append(parts, (char *)s);
  26   return 0;
  27 }
  28
  29 static int header_callback(const char *name, const char *value,
  30                            void *u) {
  31   hash *const h = u;
  32
  33   hash_add(h, name, &value, HASH_INSERT);
  34   return 0;
  35 }
  36
  37 static void test_mime(void) {
  38   char *t, *n, *v;
  39   struct vector parts[1];
  40   struct kvp *k;
  41   const char *s, *cs, *enc;
  42   hash *h;
  43
  44   t = 0;
  45   k = 0;
  46   insist(!mime_content_type("text/plain", &t, &k));
  47   check_string(t, "text/plain");
  48   insist(k == 0);
  49
  50   insist(mime_content_type("TEXT ((broken) comment", &t, &k) < 0);
  51   insist(mime_content_type("TEXT ((broken) comment\\", &t, &k) < 0);
  52
  53   t = 0;
  54   k = 0;
  55   insist(!mime_content_type("TEXT ((nested)\\ comment) /plain", &t, &k));
  56   check_string(t, "text/plain");
  57   insist(k == 0);
  58
  59   t = 0;
  60   k = 0;
  61   insist(!mime_content_type(" text/plain ; Charset=\"utf-\\8\"", &t, &k));
  62   check_string(t, "text/plain");
  63   insist(k != 0);
  64   insist(k->next == 0);
  65   check_string(k->name, "charset");
  66   check_string(k->value, "utf-8");
  67
  68   t = 0;
  69   k = 0;
  70   insist(!mime_content_type("text/plain;charset = ISO-8859-1 ", &t, &k));
  71   insist(k != 0);
  72   insist(k->next == 0);
  73   check_string(t, "text/plain");
  74   check_string(k->name, "charset");
  75   check_string(k->value, "ISO-8859-1");
  76
  77   t = n = v = 0;
  78   insist(!mime_rfc2388_content_disposition("form-data; name=\"field1\"", &t, &n, &v));
  79   check_string(t, "form-data");
  80   check_string(n, "name");
  81   check_string(v, "field1");
  82
  83   insist(!mime_rfc2388_content_disposition("inline", &t, &n, &v));
  84   check_string(t, "inline");
  85   insist(n == 0);
  86   insist(v == 0);
  87
  88   /* Current versions of the code only understand a single arg to these
  89    * headers.  This is a bug at the level they work at but suffices for
  90    * DisOrder's current purposes. */
  91
  92   insist(!mime_rfc2388_content_disposition(
  93               "attachment; filename=genome.jpeg;\n"
  94               "modification-date=\"Wed, 12 Feb 1997 16:29:51 -0500\"",
  95          &t, &n, &v));
  96   check_string(t, "attachment");
  97   check_string(n, "filename");
  98   check_string(v, "genome.jpeg");
  99
 100   vector_init(parts);
 101   insist(mime_multipart("--outer\r\n"
 102                         "Content-Type: text/plain\r\n"
 103                         "Content-Disposition: inline\r\n"
 104                         "Content-Description: text-part-1\r\n"
 105                         "\r\n"
 106                         "Some text goes here\r\n"
 107                         "\r\n"
 108                         "--outer\r\n"
 109                         "Content-Type: multipart/mixed; boundary=inner\r\n"
 110                         "Content-Disposition: attachment\r\n"
 111                         "Content-Description: multipart-2\r\n"
 112                         "\r\n"
 113                         "--inner\r\n"
 114                         "Content-Type: text/plain\r\n"
 115                         "Content-Disposition: inline\r\n"
 116                         "Content-Description: text-part-2\r\n"
 117                         "\r\n"
 118                         "Some more text here.\r\n"
 119                         "\r\n"
 120                         "--inner\r\n"
 121                         "Content-Type: image/jpeg\r\n"
 122                         "Content-Disposition: attachment\r\n"
 123                         "Content-Description: jpeg-1\r\n"
 124                         "\r\n"
 125                         "<jpeg data>\r\n"
 126                         "--inner--\r\n"
 127                         "--outer--\r\n",
 128                         test_multipart_callback,
 129                         "outer",
 130                         parts) == 0);
 131   check_integer(parts->nvec, 2);
 132   check_string(parts->vec[0],
 133                "Content-Type: text/plain\r\n"
 134                "Content-Disposition: inline\r\n"
 135                "Content-Description: text-part-1\r\n"
 136                "\r\n"
 137                "Some text goes here\r\n");
 138   check_string(parts->vec[1],
 139                "Content-Type: multipart/mixed; boundary=inner\r\n"
 140                "Content-Disposition: attachment\r\n"
 141                "Content-Description: multipart-2\r\n"
 142                "\r\n"
 143                "--inner\r\n"
 144                "Content-Type: text/plain\r\n"
 145                "Content-Disposition: inline\r\n"
 146                "Content-Description: text-part-2\r\n"
 147                "\r\n"
 148                "Some more text here.\r\n"
 149                "\r\n"
 150                "--inner\r\n"
 151                "Content-Type: image/jpeg\r\n"
 152                "Content-Disposition: attachment\r\n"
 153                "Content-Description: jpeg-1\r\n"
 154                "\r\n"
 155                "<jpeg data>\r\n"
 156                "--inner--");
 157   /* No trailing CRLF is _correct_ - see RFC2046 5.1.1 note regarding CRLF
 158    * preceding the boundary delimiter line.  An implication of this is that we
 159    * must cope with partial lines at the end of the input when recursively
 160    * decomposing a multipart message. */
 161   vector_init(parts);
 162   insist(mime_multipart("--inner\r\n"
 163                         "Content-Type: text/plain\r\n"
 164                         "Content-Disposition: inline\r\n"
 165                         "Content-Description: text-part-2\r\n"
 166                         "\r\n"
 167                         "Some more text here.\r\n"
 168                         "\r\n"
 169                         "--inner\r\n"
 170                         "Content-Type: image/jpeg\r\n"
 171                         "Content-Disposition: attachment\r\n"
 172                         "Content-Description: jpeg-1\r\n"
 173                         "\r\n"
 174                         "<jpeg data>\r\n"
 175                         "--inner--",
 176                         test_multipart_callback,
 177                         "inner",
 178                         parts) == 0);
 179   check_integer(parts->nvec, 2);
 180   check_string(parts->vec[0],
 181                "Content-Type: text/plain\r\n"
 182                "Content-Disposition: inline\r\n"
 183                "Content-Description: text-part-2\r\n"
 184                "\r\n"
 185                "Some more text here.\r\n");
 186   check_string(parts->vec[1],
 187                "Content-Type: image/jpeg\r\n"
 188                "Content-Disposition: attachment\r\n"
 189                "Content-Description: jpeg-1\r\n"
 190                "\r\n"
 191                "<jpeg data>");
 192
 193   /* Bogus inputs to mime_multipart() */
 194   fprintf(stderr, "expect two mime_multipart errors:\n");
 195   insist(mime_multipart("--inner\r\n"
 196                         "Content-Type: text/plain\r\n"
 197                         "Content-Disposition: inline\r\n"
 198                         "Content-Description: text-part-2\r\n"
 199                         "\r\n"
 200                         "Some more text here.\r\n"
 201                         "\r\n"
 202                         "--inner\r\n"
 203                         "Content-Type: image/jpeg\r\n"
 204                         "Content-Disposition: attachment\r\n"
 205                         "Content-Description: jpeg-1\r\n"
 206                         "\r\n"
 207                         "<jpeg data>\r\n",
 208                         test_multipart_callback,
 209                         "inner",
 210                         parts) == -1);
 211   insist(mime_multipart("--wrong\r\n"
 212                         "Content-Type: text/plain\r\n"
 213                         "Content-Disposition: inline\r\n"
 214                         "Content-Description: text-part-2\r\n"
 215                         "\r\n"
 216                         "Some more text here.\r\n"
 217                         "\r\n"
 218                         "--inner--\r\n",
 219                         test_multipart_callback,
 220                         "inner",
 221                         parts) == -1);
 222
 223   /* XXX mime_parse */
 224
 225   check_string(mime_qp(""), "");
 226   check_string(mime_qp("foobar"), "foobar");
 227   check_string(mime_qp("foo=20bar"), "foo bar");
 228   check_string(mime_qp("x \r\ny"), "x\r\ny");
 229   check_string(mime_qp("x=\r\ny"), "xy");
 230   check_string(mime_qp("x= \r\ny"), "xy");
 231   check_string(mime_qp("x =\r\ny"), "x y");
 232   check_string(mime_qp("x = \r\ny"), "x y");
 233
 234   check_string(mime_to_qp(""), "");
 235   check_string(mime_to_qp("foobar\n"), "foobar\n");
 236   check_string(mime_to_qp("foobar \n"), "foobar=20\n");
 237   check_string(mime_to_qp("foobar\t\n"), "foobar=09\n");
 238   check_string(mime_to_qp("foobar \t \n"), "foobar=20=09=20\n");
 239   check_string(mime_to_qp(" foo=bar"), " foo=3Dbar\n");
 240   check_string(mime_to_qp("copyright \xC2\xA9"), "copyright =C2=A9\n");
 241   check_string(mime_to_qp("foo\nbar\nbaz\n"), "foo\nbar\nbaz\n");
 242   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");
 243
 244   /* from RFC2045 */
 245   check_string(mime_qp("Now's the time =\r\n"
 246 "for all folk to come=\r\n"
 247 " to the aid of their country."),
 248                "Now's the time for all folk to come to the aid of their country.");
 249
 250 #define check_base64(encoded, decoded) do {                     \
 251     size_t ns;                                                  \
 252     check_string(mime_base64(encoded, &ns), decoded);           \
 253     insist(ns == (sizeof decoded) - 1);                         \
 254     check_string(mime_to_base64((const uint8_t *)decoded,       \
 255                                          (sizeof decoded) - 1), \
 256                  encoded);                                      \
 257   } while(0)
 258
 259
 260   check_base64("",  "");
 261   check_base64("BBBB", "\x04\x10\x41");
 262   check_base64("////", "\xFF\xFF\xFF");
 263   check_base64("//BB", "\xFF\xF0\x41");
 264   check_base64("BBBB//BB////",
 265              "\x04\x10\x41" "\xFF\xF0\x41" "\xFF\xFF\xFF");
 266   check_base64("BBBBBA==",
 267                "\x04\x10\x41" "\x04");
 268   check_base64("BBBBBBA=",
 269                "\x04\x10\x41" "\x04\x10");
 270
 271   /* Check that decoding handles various kinds of rubbish OK */
 272   check_string(mime_base64("B B B B  / / B B / / / /", 0),
 273              "\x04\x10\x41" "\xFF\xF0\x41" "\xFF\xFF\xFF");
 274   check_string(mime_base64("B\r\nBBB.// B-B//~//", 0),
 275                "\x04\x10\x41" "\xFF\xF0\x41" "\xFF\xFF\xFF");
 276   check_string(mime_base64("BBBB BB==", 0),
 277                "\x04\x10\x41" "\x04");
 278   check_string(mime_base64("BBBB BB = =", 0),
 279                "\x04\x10\x41" "\x04");
 280   check_string(mime_base64("BBBB BBB=", 0),
 281                "\x04\x10\x41" "\x04\x10");
 282   check_string(mime_base64("BBBB BBB = ", 0),
 283                "\x04\x10\x41" "\x04\x10");
 284   check_string(mime_base64("BBBB=", 0),
 285                "\x04\x10\x41");
 286   check_string(mime_base64("BBBBBB==", 0),
 287                "\x04\x10\x41" "\x04");
 288   check_string(mime_base64("BBBBBBB=", 0),
 289                "\x04\x10\x41" "\x04\x10");
 290   /* Not actually valid base64 */
 291   check_string(mime_base64("BBBBx=", 0),
 292                "\x04\x10\x41");
 293
 294   h = hash_new(sizeof (char *));
 295   s = mime_parse("From: sender@example.com\r\n"
 296                  "To: rcpt@example.com\r\n"
 297                  "Subject: test #1\r\n"
 298                  "\r\n"
 299                  "body\r\n",
 300                  header_callback, h);
 301   insist(s != 0);
 302   check_string(*(char **)hash_find(h, "from"), "sender@example.com");
 303   check_string(*(char **)hash_find(h, "to"), "rcpt@example.com");
 304   check_string(*(char **)hash_find(h, "subject"), "test #1");
 305   check_string(s, "body\r\n");
 306
 307   h = hash_new(sizeof (char *));
 308   s = mime_parse("FROM: sender@example.com\r\n"
 309                  "TO: rcpt@example.com\r\n"
 310                  "SUBJECT: test #1\r\n"
 311                  "CONTENT-TRANSFER-ENCODING: 7bit\r\n"
 312                  "\r\n"
 313                  "body\r\n",
 314                  header_callback, h);
 315   insist(s != 0);
 316   check_string(*(char **)hash_find(h, "from"), "sender@example.com");
 317   check_string(*(char **)hash_find(h, "to"), "rcpt@example.com");
 318   check_string(*(char **)hash_find(h, "subject"), "test #1");
 319   check_string(*(char **)hash_find(h, "content-transfer-encoding"), "7bit");
 320   check_string(s, "body\r\n");
 321
 322   h = hash_new(sizeof (char *));
 323   s = mime_parse("From: sender@example.com\r\n"
 324                  "To:    \r\n"
 325                  "     rcpt@example.com\r\n"
 326                  "Subject: test #1\r\n"
 327                  "MIME-Version: 1.0\r\n"
 328                  "Content-Type: text/plain\r\n"
 329                  "Content-Transfer-Encoding: BASE64\r\n"
 330                  "\r\n"
 331                  "d2liYmxlDQo=\r\n",
 332                  header_callback, h);
 333   insist(s != 0);
 334   check_string(*(char **)hash_find(h, "from"), "sender@example.com");
 335   check_string(*(char **)hash_find(h, "to"), "rcpt@example.com");
 336   check_string(*(char **)hash_find(h, "subject"), "test #1");
 337   check_string(*(char **)hash_find(h, "mime-version"), "1.0");
 338   check_string(*(char **)hash_find(h, "content-type"), "text/plain");
 339   check_string(*(char **)hash_find(h, "content-transfer-encoding"), "BASE64");
 340   check_string(s, "wibble\r\n");
 341
 342 #define CHECK_QUOTE(INPUT, EXPECT) do {                 \
 343   s = quote822(INPUT, 0);                               \
 344   insist(s != 0);                                       \
 345   check_string(s, EXPECT);                              \
 346   s = mime_parse_word(s, &t, mime_http_separator);      \
 347   check_string(t, INPUT);                               \
 348 } while(0)
 349   CHECK_QUOTE("wibble", "wibble");
 350   CHECK_QUOTE("wibble spong", "\"wibble spong\"");
 351   CHECK_QUOTE("wibble\\spong", "\"wibble\\\\spong\"");
 352   CHECK_QUOTE("wibble\"spong", "\"wibble\\\"spong\"");
 353   CHECK_QUOTE("(wibble)", "\"(wibble)\"");
 354
 355   s = mime_encode_text("wibble\n", &cs, &enc);
 356   insist(s != 0);
 357   check_string(s, "wibble\n");
 358   check_string(cs, "us-ascii");
 359   check_string(enc, "7bit");
 360
 361   s = mime_encode_text("wibble\xC3\xB7\n", &cs, &enc);
 362   insist(s != 0);
 363   check_string(s, "wibble=C3=B7\n");
 364   check_string(cs, "utf-8");
 365   check_string(enc, "quoted-printable");
 366 }
 367
 368 TEST(mime);
 369
 370 /*
 371 Local Variables:
 372 c-basic-offset:2
 373 comment-column:40
 374 fill-column:79
 375 indent-tabs-mode:nil
 376 End:
 377 */