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