Mercurial > thymian

annotate 3rdparty/vmime/tests/parser/textTest.cpp @ 0:a4671277546c tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
created the repository for the thymian project
author ferencd
date Tue, 17 Aug 2021 11:19:54 +0200
parents
children
rev   line source
ferencd@0 1 //
ferencd@0 2 // VMime library (http://www.vmime.org)
ferencd@0 3 // Copyright (C) 2002-2013 Vincent Richard <vincent@vmime.org>
ferencd@0 4 //
ferencd@0 5 // This program is free software; you can redistribute it and/or
ferencd@0 6 // modify it under the terms of the GNU General Public License as
ferencd@0 7 // published by the Free Software Foundation; either version 3 of
ferencd@0 8 // the License, or (at your option) any later version.
ferencd@0 9 //
ferencd@0 10 // This program is distributed in the hope that it will be useful,
ferencd@0 11 // but WITHOUT ANY WARRANTY; without even the implied warranty of
ferencd@0 12 // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
ferencd@0 13 // General Public License for more details.
ferencd@0 14 //
ferencd@0 15 // You should have received a copy of the GNU General Public License along
ferencd@0 16 // with this program; if not, write to the Free Software Foundation, Inc.,
ferencd@0 17 // 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
ferencd@0 18 //
ferencd@0 19 // Linking this library statically or dynamically with other modules is making
ferencd@0 20 // a combined work based on this library. Thus, the terms and conditions of
ferencd@0 21 // the GNU General Public License cover the whole combination.
ferencd@0 22 //
ferencd@0 23
ferencd@0 24 #include "tests/testUtils.hpp"
ferencd@0 25
ferencd@0 26 #include <locale>
ferencd@0 27 #include <clocale>
ferencd@0 28
ferencd@0 29
ferencd@0 30 VMIME_TEST_SUITE_BEGIN(textTest)
ferencd@0 31
ferencd@0 32 VMIME_TEST_LIST_BEGIN
ferencd@0 33 VMIME_TEST(testConstructors)
ferencd@0 34 VMIME_TEST(testCopy)
ferencd@0 35 VMIME_TEST(testNewFromString)
ferencd@0 36 VMIME_TEST(testDisplayForm)
ferencd@0 37 VMIME_TEST(testParse)
ferencd@0 38 VMIME_TEST(testGenerate)
ferencd@0 39
ferencd@0 40 VMIME_TEST(testWordConstructors)
ferencd@0 41 VMIME_TEST(testWordParse)
ferencd@0 42 VMIME_TEST(testWordGenerate)
ferencd@0 43 VMIME_TEST(testWordGenerateSpace)
ferencd@0 44 VMIME_TEST(testWordGenerateSpace2)
ferencd@0 45 VMIME_TEST(testWordGenerateMultiBytes)
ferencd@0 46 VMIME_TEST(testWordGenerateQuote)
ferencd@0 47 VMIME_TEST(testWordGenerateSpecialCharsets)
ferencd@0 48 VMIME_TEST(testWordGenerateSpecials)
ferencd@0 49
ferencd@0 50 VMIME_TEST(testWhitespace)
ferencd@0 51 VMIME_TEST(testWhitespaceMBox)
ferencd@0 52
ferencd@0 53 VMIME_TEST(testFoldingAscii)
ferencd@0 54 VMIME_TEST(testForcedNonEncoding)
ferencd@0 55
ferencd@0 56 VMIME_TEST(testBugFix20110511)
ferencd@0 57
ferencd@0 58 VMIME_TEST(testInternationalizedEmail_specialChars)
ferencd@0 59 VMIME_TEST(testInternationalizedEmail_UTF8)
ferencd@0 60 VMIME_TEST(testInternationalizedEmail_nonUTF8)
ferencd@0 61 VMIME_TEST(testInternationalizedEmail_folding)
ferencd@0 62 VMIME_TEST_LIST_END
ferencd@0 63
ferencd@0 64
ferencd@0 65 static const vmime::string getDisplayText(const vmime::text& t)
ferencd@0 66 {
ferencd@0 67 return t.getWholeBuffer();
ferencd@0 68 }
ferencd@0 69
ferencd@0 70 static const vmime::string cleanGeneratedWords(const std::string& str)
ferencd@0 71 {
ferencd@0 72 std::istringstream iss(str);
ferencd@0 73
ferencd@0 74 std::string res;
ferencd@0 75 std::string x;
ferencd@0 76
ferencd@0 77 while (std::getline(iss, x))
ferencd@0 78 res += vmime::utility::stringUtils::trim(x);
ferencd@0 79
ferencd@0 80 return res;
ferencd@0 81 }
ferencd@0 82
ferencd@0 83
ferencd@0 84 void setUp()
ferencd@0 85 {
ferencd@0 86 // Set the global C and C++ locale to the user-configured locale.
ferencd@0 87 // The locale should use UTF-8 encoding for these tests to run successfully.
ferencd@0 88 try
ferencd@0 89 {
ferencd@0 90 std::locale::global(std::locale(""));
ferencd@0 91 }
ferencd@0 92 catch (std::exception &)
ferencd@0 93 {
ferencd@0 94 std::setlocale(LC_ALL, "");
ferencd@0 95 }
ferencd@0 96 }
ferencd@0 97
ferencd@0 98 void tearDown()
ferencd@0 99 {
ferencd@0 100 // Restore default locale
ferencd@0 101 std::locale::global(std::locale("C"));
ferencd@0 102 }
ferencd@0 103
ferencd@0 104
ferencd@0 105 void testConstructors()
ferencd@0 106 {
ferencd@0 107 vmime::text t1;
ferencd@0 108
ferencd@0 109 VASSERT_EQ("1.1", 0, t1.getWordCount());
ferencd@0 110
ferencd@0 111 vmime::text t2("Test\xa9\xc3");
ferencd@0 112
ferencd@0 113 VASSERT_EQ("2.1", 1, t2.getWordCount());
ferencd@0 114 VASSERT_EQ("2.2", "Test\xa9\xc3", t2.getWordAt(0)->getBuffer());
ferencd@0 115 VASSERT_EQ("2.3", vmime::charset::getLocalCharset(), t2.getWordAt(0)->getCharset());
ferencd@0 116
ferencd@0 117 vmime::text t3("Test\xa9\xc3", vmime::charset(vmime::charsets::ISO8859_13));
ferencd@0 118
ferencd@0 119 VASSERT_EQ("3.1", 1, t3.getWordCount());
ferencd@0 120 VASSERT_EQ("3.2", "Test\xa9\xc3", t3.getWordAt(0)->getBuffer());
ferencd@0 121 VASSERT_EQ("3.3", vmime::charset(vmime::charsets::ISO8859_13), t3.getWordAt(0)->getCharset());
ferencd@0 122
ferencd@0 123 vmime::word w1("Test", vmime::charset(vmime::charsets::UTF_8));
ferencd@0 124 vmime::text t4(w1);
ferencd@0 125
ferencd@0 126 VASSERT_EQ("4.1", 1, t4.getWordCount());
ferencd@0 127 VASSERT_EQ("4.2", w1.getBuffer(), t4.getWordAt(0)->getBuffer());
ferencd@0 128 VASSERT_EQ("4.3", w1.getCharset(), t4.getWordAt(0)->getCharset());
ferencd@0 129
ferencd@0 130 vmime::word w2("Other", vmime::charset(vmime::charsets::US_ASCII));
ferencd@0 131 t4.appendWord(vmime::make_shared <vmime::word>(w2));
ferencd@0 132
ferencd@0 133 vmime::text t5(t4);
ferencd@0 134
ferencd@0 135 VASSERT_EQ("5.1", 2, t5.getWordCount());
ferencd@0 136 VASSERT_EQ("5.2", w1.getBuffer(), t5.getWordAt(0)->getBuffer());
ferencd@0 137 VASSERT_EQ("5.3", w1.getCharset(), t5.getWordAt(0)->getCharset());
ferencd@0 138 VASSERT_EQ("5.4", w2.getBuffer(), t5.getWordAt(1)->getBuffer());
ferencd@0 139 VASSERT_EQ("5.5", w2.getCharset(), t5.getWordAt(1)->getCharset());
ferencd@0 140 }
ferencd@0 141
ferencd@0 142 void testCopy()
ferencd@0 143 {
ferencd@0 144 vmime::text t1("Test: \xa9\xc3");
ferencd@0 145
ferencd@0 146 VASSERT("operator==", t1 == t1);
ferencd@0 147 VASSERT("clone", *vmime::clone(t1) == t1);
ferencd@0 148
ferencd@0 149 vmime::text t2;
ferencd@0 150 t2.copyFrom(t1);
ferencd@0 151
ferencd@0 152 VASSERT("copyFrom", t1 == t2);
ferencd@0 153 }
ferencd@0 154
ferencd@0 155 void testNewFromString()
ferencd@0 156 {
ferencd@0 157 vmime::string s1 = "only ASCII characters";
ferencd@0 158 vmime::charset c1("test");
ferencd@0 159 vmime::text t1;
ferencd@0 160
ferencd@0 161 t1.createFromString(s1, c1);
ferencd@0 162
ferencd@0 163 VASSERT_EQ("1.1", 1, t1.getWordCount());
ferencd@0 164 VASSERT_EQ("1.2", s1, t1.getWordAt(0)->getBuffer());
ferencd@0 165 VASSERT_EQ("1.3", vmime::charset(vmime::charsets::US_ASCII), t1.getWordAt(0)->getCharset());
ferencd@0 166
ferencd@0 167 vmime::string s2_1 = "some ASCII characters and special chars: ";
ferencd@0 168 vmime::string s2_2 = "\xf1\xf2\xf3\xf4 ";
ferencd@0 169 vmime::string s2_3 = "and then more ASCII chars.";
ferencd@0 170 vmime::string s2 = s2_1 + s2_2 + s2_3;
ferencd@0 171 vmime::charset c2("test");
ferencd@0 172 vmime::text t2;
ferencd@0 173
ferencd@0 174 t2.createFromString(s2, c2);
ferencd@0 175
ferencd@0 176 VASSERT_EQ("2.1", 3, t2.getWordCount());
ferencd@0 177 VASSERT_EQ("2.2", "some ASCII characters and special chars: ", t2.getWordAt(0)->getBuffer());
ferencd@0 178 VASSERT_EQ("2.3", vmime::charset(vmime::charsets::US_ASCII), t2.getWordAt(0)->getCharset());
ferencd@0 179 VASSERT_EQ("2.4", "\xf1\xf2\xf3\xf4", t2.getWordAt(1)->getBuffer());
ferencd@0 180 VASSERT_EQ("2.5", c2, t2.getWordAt(1)->getCharset());
ferencd@0 181 VASSERT_EQ("2.6", "and then more ASCII chars.", t2.getWordAt(2)->getBuffer());
ferencd@0 182 VASSERT_EQ("2.7", vmime::charset(vmime::charsets::US_ASCII), t2.getWordAt(2)->getCharset());
ferencd@0 183 }
ferencd@0 184
ferencd@0 185 static const vmime::string parseText(const vmime::string& buffer)
ferencd@0 186 {
ferencd@0 187 vmime::text t;
ferencd@0 188 t.parse(buffer);
ferencd@0 189
ferencd@0 190 std::ostringstream oss;
ferencd@0 191 oss << t;
ferencd@0 192
ferencd@0 193 return (oss.str());
ferencd@0 194 }
ferencd@0 195
ferencd@0 196 void testParse()
ferencd@0 197 {
ferencd@0 198 // From RFC-2047
ferencd@0 199 VASSERT_EQ("1", "[text: [[word: charset=US-ASCII, buffer=Keith Moore]]]",
ferencd@0 200 parseText("=?US-ASCII?Q?Keith_Moore?="));
ferencd@0 201
ferencd@0 202 VASSERT_EQ("2", "[text: [[word: charset=ISO-8859-1, buffer=Keld J\xf8rn Simonsen]]]",
ferencd@0 203 parseText("=?ISO-8859-1?Q?Keld_J=F8rn_Simonsen?="));
ferencd@0 204
ferencd@0 205 VASSERT_EQ("3", "[text: [[word: charset=ISO-8859-1, buffer=Andr\xe9]," \
ferencd@0 206 "[word: charset=us-ascii, buffer= Pirard]]]",
ferencd@0 207 parseText("=?ISO-8859-1?Q?Andr=E9?= Pirard"));
ferencd@0 208
ferencd@0 209 VASSERT_EQ("4", "[text: [[word: charset=ISO-8859-1, buffer=If you can read this yo]," \
ferencd@0 210 "[word: charset=ISO-8859-2, buffer=u understand the example.]]]",
ferencd@0 211 parseText("=?ISO-8859-1?B?SWYgeW91IGNhbiByZWFkIHRoaXMgeW8=?=\r\n " \
ferencd@0 212 "=?ISO-8859-2?B?dSB1bmRlcnN0YW5kIHRoZSBleGFtcGxlLg==?="));
ferencd@0 213
ferencd@0 214 // Bugfix: in "=?charset?q?=XX=YY?=", the "?=" finish
ferencd@0 215 // sequence was not correctly found (should be the one
ferencd@0 216 // after '=YY' and not the one after '?q').
ferencd@0 217 VASSERT_EQ("5", "[text: [[word: charset=abc, buffer=\xe9\xe9]]]",
ferencd@0 218 parseText("=?abc?q?=E9=E9?="));
ferencd@0 219
ferencd@0 220 // Question marks (?) in the middle of the string
ferencd@0 221 VASSERT_EQ("6", "[text: [[word: charset=iso-8859-1, buffer=Know wh\xe4t? It works!]]]",
ferencd@0 222 parseText("=?iso-8859-1?Q?Know_wh=E4t?_It_works!?="));
ferencd@0 223
ferencd@0 224 // With language specifier
ferencd@0 225 VASSERT_EQ("7", "[text: [[word: charset=US-ASCII, buffer=Keith Moore, lang=EN]]]",
ferencd@0 226 parseText("=?US-ASCII*EN?Q?Keith_Moore?="));
ferencd@0 227 }
ferencd@0 228
ferencd@0 229 void testGenerate()
ferencd@0 230 {
ferencd@0 231 // TODO
ferencd@0 232
ferencd@0 233 // With language specifier
ferencd@0 234 vmime::word wlang1("Émeline", vmime::charset("UTF-8"), "FR");
ferencd@0 235 VASSERT_EQ("lang1", "=?UTF-8*FR?Q?=C3=89meline?=", wlang1.generate());
ferencd@0 236
ferencd@0 237 vmime::word wlang2("Keith Moore", vmime::charset("US-ASCII"), "EN");
ferencd@0 238 VASSERT_EQ("lang2", "=?US-ASCII*EN?Q?Keith_Moore?=", wlang2.generate());
ferencd@0 239 }
ferencd@0 240
ferencd@0 241 void testDisplayForm()
ferencd@0 242 {
ferencd@0 243 #define DISPLAY_FORM(x) getDisplayText(*vmime::text::decodeAndUnfold(x))
ferencd@0 244
ferencd@0 245 // From RFC-2047
ferencd@0 246 VASSERT_EQ("1", "a", DISPLAY_FORM("=?ISO-8859-1?Q?a?="));
ferencd@0 247 VASSERT_EQ("2", "a b", DISPLAY_FORM("=?ISO-8859-1?Q?a?= b"));
ferencd@0 248 VASSERT_EQ("3", "ab", DISPLAY_FORM("=?ISO-8859-1?Q?a?= =?ISO-8859-1?Q?b?="));
ferencd@0 249 VASSERT_EQ("4", "ab", DISPLAY_FORM("=?ISO-8859-1?Q?a?= \t =?ISO-8859-1?Q?b?="));
ferencd@0 250 VASSERT_EQ("5", "ab", DISPLAY_FORM("=?ISO-8859-1?Q?a?= \r\n \t =?ISO-8859-1?Q?b?="));
ferencd@0 251 VASSERT_EQ("6", "a b", DISPLAY_FORM("=?ISO-8859-1?Q?a_b?="));
ferencd@0 252 VASSERT_EQ("7", "a b", DISPLAY_FORM("=?ISO-8859-1?Q?a?= =?ISO-8859-2?Q?_b?="));
ferencd@0 253
ferencd@0 254 // Some more tests...
ferencd@0 255 VASSERT_EQ("8", "a b", DISPLAY_FORM(" a =?ISO-8859-1?Q?b?= "));
ferencd@0 256 VASSERT_EQ("9", "a b ", DISPLAY_FORM(" \t =?ISO-8859-1?Q?a?= b "));
ferencd@0 257 VASSERT_EQ("10", "a b", DISPLAY_FORM(" a\r\n\t b"));
ferencd@0 258
ferencd@0 259 VASSERT_EQ("11", "a b c", DISPLAY_FORM("a =?ISO-8859-1?Q?b?= c"));
ferencd@0 260 VASSERT_EQ("12", "a b c ", DISPLAY_FORM("a =?ISO-8859-1?Q?b?= c "));
ferencd@0 261 VASSERT_EQ("13", "a b c ", DISPLAY_FORM(" a =?ISO-8859-1?Q?b?= c "));
ferencd@0 262 VASSERT_EQ("14", "a b c d", DISPLAY_FORM("a =?ISO-8859-1?Q?b?= c =?ISO-8859-1?Q?d?= "));
ferencd@0 263 VASSERT_EQ("15", "a b c d e", DISPLAY_FORM("a =?ISO-8859-1?Q?b?= c =?ISO-8859-1?Q?d?= e"));
ferencd@0 264
ferencd@0 265 // Whitespaces and multiline
ferencd@0 266 VASSERT_EQ("16", "a b c d e", DISPLAY_FORM("=?ISO-8859-1?Q?a_b_?=c\n\t=?ISO-8859-1?Q?d_?=e"));
ferencd@0 267
ferencd@0 268 #undef DISPLAY_FORM
ferencd@0 269 }
ferencd@0 270
ferencd@0 271 void testWordConstructors()
ferencd@0 272 {
ferencd@0 273 VASSERT_EQ("1.1", vmime::charset::getLocalCharset(), vmime::word().getCharset());
ferencd@0 274 VASSERT_EQ("1.2", "", vmime::word().getBuffer());
ferencd@0 275
ferencd@0 276 VASSERT_EQ("2.1", vmime::charset::getLocalCharset(), vmime::word("foo").getCharset());
ferencd@0 277 VASSERT_EQ("2.2", "foo", vmime::word("foo").getBuffer());
ferencd@0 278
ferencd@0 279 VASSERT_EQ("3.1", "bar", vmime::word("foo", vmime::charset("bar")).getCharset().getName());
ferencd@0 280 VASSERT_EQ("3.2", "foo", vmime::word("foo", vmime::charset("bar")).getBuffer());
ferencd@0 281 }
ferencd@0 282
ferencd@0 283 void testWordParse()
ferencd@0 284 {
ferencd@0 285 // Simple encoded word
ferencd@0 286 vmime::word w1;
ferencd@0 287 w1.parse("=?foo?q?bar=E9 baz?=");
ferencd@0 288
ferencd@0 289 VASSERT_EQ("1.1", "foo", w1.getCharset().getName());
ferencd@0 290 VASSERT_EQ("1.2", "bar\xe9 baz", w1.getBuffer());
ferencd@0 291
ferencd@0 292 // Unencoded text
ferencd@0 293 vmime::word w2;
ferencd@0 294 w2.parse(" foo bar \tbaz...");
ferencd@0 295
ferencd@0 296 VASSERT_EQ("2.1", vmime::charset(vmime::charsets::US_ASCII), w2.getCharset());
ferencd@0 297 VASSERT_EQ("2.2", " foo bar \tbaz...", w2.getBuffer());
ferencd@0 298
ferencd@0 299 // Malformed word
ferencd@0 300 vmime::word w3;
ferencd@0 301 w3.parse("=?foo bar");
ferencd@0 302
ferencd@0 303 VASSERT_EQ("3.1", vmime::charset(vmime::charsets::US_ASCII), w3.getCharset());
ferencd@0 304 VASSERT_EQ("3.2", "=?foo bar", w3.getBuffer());
ferencd@0 305
ferencd@0 306 // Unknown encoding
ferencd@0 307 vmime::word w4;
ferencd@0 308 w4.parse("=?whatever?not_q_or_b?whatever?=");
ferencd@0 309
ferencd@0 310 VASSERT_EQ("4.1", vmime::charset(vmime::charsets::US_ASCII), w4.getCharset());
ferencd@0 311 VASSERT_EQ("4.2", "=?whatever?not_q_or_b?whatever?=", w4.getBuffer());
ferencd@0 312 }
ferencd@0 313
ferencd@0 314 void testWordGenerate()
ferencd@0 315 {
ferencd@0 316 VASSERT_EQ("1", "=?foo?Q?bar=E9_baz?=",
ferencd@0 317 vmime::word("bar\xe9 baz", vmime::charset("foo")).generate());
ferencd@0 318
ferencd@0 319 VASSERT_EQ("2", "=?foo?B?8fLz9PU=?=",
ferencd@0 320 vmime::word("\xf1\xf2\xf3\xf4\xf5", vmime::charset("foo")).generate());
ferencd@0 321 }
ferencd@0 322
ferencd@0 323 void testWordGenerateSpace()
ferencd@0 324 {
ferencd@0 325 // No white-space between an unencoded word and a encoded one
ferencd@0 326 VASSERT_EQ("1", "Bonjour =?utf-8?Q?Fran=C3=A7ois?=",
ferencd@0 327 vmime::text::newFromString("Bonjour Fran\xc3\xa7ois",
ferencd@0 328 vmime::charset("utf-8"))->generate());
ferencd@0 329
ferencd@0 330 // White-space between two encoded words
ferencd@0 331 vmime::text txt;
ferencd@0 332 txt.appendWord(vmime::make_shared <vmime::word>("\xc3\x89t\xc3\xa9", "utf-8"));
ferencd@0 333 txt.appendWord(vmime::make_shared <vmime::word>("Fran\xc3\xa7ois", "utf-8"));
ferencd@0 334
ferencd@0 335 const vmime::string decoded = "\xc3\x89t\xc3\xa9""Fran\xc3\xa7ois";
ferencd@0 336 const vmime::string encoded = "=?utf-8?B?w4l0w6k=?= =?utf-8?Q?Fran=C3=A7ois?=";
ferencd@0 337
ferencd@0 338 // -- test encoding
ferencd@0 339 VASSERT_EQ("2", encoded, txt.generate());
ferencd@0 340
ferencd@0 341 // -- ensure no space is added when decoding
ferencd@0 342 vmime::text txt2;
ferencd@0 343 txt2.parse(encoded, 0, encoded.length());
ferencd@0 344
ferencd@0 345 VASSERT_EQ("3", decoded, txt2.getWholeBuffer());
ferencd@0 346
ferencd@0 347 // -- test rencoding
ferencd@0 348 VASSERT_EQ("4", encoded, txt2.generate());
ferencd@0 349 }
ferencd@0 350
ferencd@0 351 void testWordGenerateSpace2()
ferencd@0 352 {
ferencd@0 353 // White-space between two encoded words (#2)
ferencd@0 354 vmime::text txt;
ferencd@0 355 txt.appendWord(vmime::make_shared <vmime::word>("Facture ", "utf-8"));
ferencd@0 356 txt.appendWord(vmime::make_shared <vmime::word>("\xc3\xa0", "utf-8"));
ferencd@0 357 txt.appendWord(vmime::make_shared <vmime::word>(" envoyer ", "utf-8"));
ferencd@0 358 txt.appendWord(vmime::make_shared <vmime::word>("\xc3\xa0", "utf-8"));
ferencd@0 359 txt.appendWord(vmime::make_shared <vmime::word>(" Martine", "utf-8"));
ferencd@0 360
ferencd@0 361 const vmime::string decoded = "Facture ""\xc3\xa0"" envoyer ""\xc3\xa0"" Martine";
ferencd@0 362 const vmime::string encoded = "Facture =?utf-8?B?w6A=?= envoyer =?utf-8?B?w6A=?= Martine";
ferencd@0 363
ferencd@0 364 // -- test encoding
ferencd@0 365 VASSERT_EQ("1", encoded, txt.generate());
ferencd@0 366
ferencd@0 367 // -- ensure no space is added when decoding
ferencd@0 368 vmime::text txt2;
ferencd@0 369 txt2.parse(encoded, 0, encoded.length());
ferencd@0 370
ferencd@0 371 VASSERT_EQ("2", decoded, txt2.getWholeBuffer());
ferencd@0 372
ferencd@0 373 // -- test rencoding
ferencd@0 374 VASSERT_EQ("3", encoded, txt2.generate());
ferencd@0 375 }
ferencd@0 376
ferencd@0 377 void testWordGenerateMultiBytes()
ferencd@0 378 {
ferencd@0 379 // Ensure we don't encode a non-integral number of characters
ferencd@0 380 VASSERT_EQ("1", "=?utf-8?Q?aaa?==?utf-8?Q?=C3=A9?==?utf-8?Q?zzz?=",
ferencd@0 381 cleanGeneratedWords(vmime::word("aaa\xc3\xa9zzz", vmime::charset("utf-8")).generate(16)));
ferencd@0 382
ferencd@0 383 VASSERT_EQ("2", "=?utf-8?Q?aaa=C3=A9?==?utf-8?Q?zzz?=",
ferencd@0 384 cleanGeneratedWords(vmime::word("aaa\xc3\xa9zzz", vmime::charset("utf-8")).generate(17)));
ferencd@0 385 }
ferencd@0 386
ferencd@0 387 void testWordGenerateQuote()
ferencd@0 388 {
ferencd@0 389 std::string str;
ferencd@0 390 vmime::utility::outputStreamStringAdapter os(str);
ferencd@0 391
ferencd@0 392 vmime::generationContext ctx;
ferencd@0 393 ctx.setMaxLineLength(1000);
ferencd@0 394
ferencd@0 395 // ASCII-only text is quotable
ferencd@0 396 str.clear();
ferencd@0 397 vmime::word("Quoted text").generate(ctx, os, 0, NULL, vmime::text::QUOTE_IF_POSSIBLE, NULL);
ferencd@0 398 VASSERT_EQ("1", "\"Quoted text\"", cleanGeneratedWords(str));
ferencd@0 399
ferencd@0 400 // Text with CR/LF is not quotable
ferencd@0 401 str.clear();
ferencd@0 402 vmime::word("Non-quotable\ntext", "us-ascii").generate(ctx, os, 0, NULL, vmime::text::QUOTE_IF_POSSIBLE, NULL);
ferencd@0 403 VASSERT_EQ("2", "=?us-ascii?Q?Non-quotable=0Atext?=", cleanGeneratedWords(str));
ferencd@0 404
ferencd@0 405 // Text with non-ASCII chars is not quotable
ferencd@0 406 str.clear();
ferencd@0 407 vmime::word("Non-quotable text \xc3\xa9").generate(ctx, os, 0, NULL, vmime::text::QUOTE_IF_POSSIBLE, NULL);
ferencd@0 408 VASSERT_EQ("3", "=?UTF-8?Q?Non-quotable_text_=C3=A9?=", cleanGeneratedWords(str));
ferencd@0 409 }
ferencd@0 410
ferencd@0 411 void testWordGenerateSpecialCharsets()
ferencd@0 412 {
ferencd@0 413 // ISO-2022-JP only uses 7-bit chars but should be encoded in Base64
ferencd@0 414 VASSERT_EQ("1", "=?iso-2022-jp?B?XlskQiVRITwlPSVKJWshJiU9JVUlSCUmJSclIl5bKEI=?=",
ferencd@0 415 cleanGeneratedWords(vmime::word("^[$B%Q!<%=%J%k!&%=%U%H%&%'%\"^[(B",
ferencd@0 416 vmime::charset("iso-2022-jp")).generate(100)));
ferencd@0 417 }
ferencd@0 418
ferencd@0 419 void testWordGenerateSpecials()
ferencd@0 420 {
ferencd@0 421 // In RFC-2047, quotation marks (ASCII 22h) should be encoded
ferencd@0 422 VASSERT_EQ("1", "=?UTF-8?Q?=22=C3=9Cml=C3=A4ute=22?=",
ferencd@0 423 vmime::word("\x22\xC3\x9Cml\xC3\xA4ute\x22", vmime::charset("UTF-8")).generate());
ferencd@0 424 }
ferencd@0 425
ferencd@0 426 void testWhitespace()
ferencd@0 427 {
ferencd@0 428 // Create
ferencd@0 429 vmime::text text;
ferencd@0 430 text.createFromString("Achim Br\xc3\xa4ndt", vmime::charsets::UTF_8);
ferencd@0 431
ferencd@0 432 VASSERT_EQ("1", 2, text.getWordCount());
ferencd@0 433 VASSERT_EQ("2", "Achim ", text.getWordAt(0)->getBuffer());
ferencd@0 434 VASSERT_EQ("3", "us-ascii", text.getWordAt(0)->getCharset());
ferencd@0 435 VASSERT_EQ("4", "Br\xc3\xa4ndt", text.getWordAt(1)->getBuffer());
ferencd@0 436 VASSERT_EQ("5", "utf-8", text.getWordAt(1)->getCharset());
ferencd@0 437
ferencd@0 438 // Generate
ferencd@0 439 VASSERT_EQ("6", "Achim =?utf-8?Q?Br=C3=A4ndt?=", text.generate());
ferencd@0 440
ferencd@0 441 // Parse
ferencd@0 442 text.parse("=?us-ascii?Q?Achim_?= =?utf-8?Q?Br=C3=A4ndt?=");
ferencd@0 443
ferencd@0 444 VASSERT_EQ("7", 2, text.getWordCount());
ferencd@0 445 VASSERT_EQ("8", "Achim ", text.getWordAt(0)->getBuffer());
ferencd@0 446 VASSERT_EQ("9", "us-ascii", text.getWordAt(0)->getCharset());
ferencd@0 447 VASSERT_EQ("10", "Br\xc3\xa4ndt", text.getWordAt(1)->getBuffer());
ferencd@0 448 VASSERT_EQ("11", "utf-8", text.getWordAt(1)->getCharset());
ferencd@0 449 }
ferencd@0 450
ferencd@0 451 void testWhitespaceMBox()
ferencd@0 452 {
ferencd@0 453 // Space MUST be encoded inside a word
ferencd@0 454 vmime::mailbox mbox(vmime::text("Achim Br\xc3\xa4ndt", vmime::charsets::UTF_8), "me@vmime.org");
ferencd@0 455 VASSERT_EQ("generate1", "=?us-ascii?Q?Achim_?= =?utf-8?Q?Br=C3=A4ndt?= <me@vmime.org>", mbox.generate());
ferencd@0 456
ferencd@0 457 vmime::text txt;
ferencd@0 458 txt.appendWord(vmime::make_shared <vmime::word>("Achim ", "us-ascii"));
ferencd@0 459 txt.appendWord(vmime::make_shared <vmime::word>("Br\xc3\xa4ndt", "utf-8"));
ferencd@0 460 mbox = vmime::mailbox(txt, "me@vmime.org");
ferencd@0 461 VASSERT_EQ("generate2", "=?us-ascii?Q?Achim_?= =?utf-8?Q?Br=C3=A4ndt?= <me@vmime.org>", mbox.generate());
ferencd@0 462
ferencd@0 463 mbox.parse("=?us-ascii?Q?Achim?= =?utf-8?Q?Br=C3=A4ndt?= <me@vmime.org>");
ferencd@0 464 VASSERT_EQ("parse.name.count", 2, mbox.getName().getWordCount());
ferencd@0 465 VASSERT_EQ("parse.name.word1.buffer", "Achim", mbox.getName().getWordAt(0)->getBuffer());
ferencd@0 466 VASSERT_EQ("parse.name.word1.charset", "us-ascii", mbox.getName().getWordAt(0)->getCharset());
ferencd@0 467 VASSERT_EQ("parse.name.word2.buffer", "Br\xc3\xa4ndt", mbox.getName().getWordAt(1)->getBuffer());
ferencd@0 468 VASSERT_EQ("parse.name.word2.charset", "utf-8", mbox.getName().getWordAt(1)->getCharset());
ferencd@0 469
ferencd@0 470 VASSERT_EQ("parse.email", "me@vmime.org", mbox.getEmail());
ferencd@0 471 }
ferencd@0 472
ferencd@0 473 void testFoldingAscii()
ferencd@0 474 {
ferencd@0 475 // In this test, no encoding is needed, but line should be folded anyway
ferencd@0 476 vmime::word w("01234567890123456789012345678901234567890123456789"
ferencd@0 477 "01234567890123456789012345678901234567890123456789", vmime::charset("us-ascii"));
ferencd@0 478
ferencd@0 479 VASSERT_EQ("fold.ascii",
ferencd@0 480 "=?us-ascii?Q?01234567890123456789012345678901234?=\r\n"
ferencd@0 481 " =?us-ascii?Q?5678901234567890123456789012345678?=\r\n"
ferencd@0 482 " =?us-ascii?Q?9012345678901234567890123456789?=", w.generate(50));
ferencd@0 483 }
ferencd@0 484
ferencd@0 485 void testForcedNonEncoding()
ferencd@0 486 {
ferencd@0 487 // Testing long unbreakable and unencodable header
ferencd@0 488 vmime::relay r;
ferencd@0 489 r.parse(" from User (Ee9GMqZQ8t7IQwftfAFHd2KyScCYRrFSJ50tKEoXv2bVCG4HcPU80GGWiFabAvG77FekpGgF1h@[127.0.0.1]) by servername.hostname.com\n\t"
ferencd@0 490 "with esmtp id 1NGTS9-2C0sqG0; Fri, 4 Dec 2009 09:23:49 +0100");
ferencd@0 491
ferencd@0 492 VASSERT_EQ("received.long", "from User\r\n (Ee9GMqZQ8t7IQwftfAFHd2KyScCYRrFSJ50tKEoXv2bVCG4HcPU80GGWiFabAvG77FekpGgF1h@[127.0.0.1])\r\n by servername.hostname.com with esmtp id 1NGTS9-2C0sqG0; Fri, 4 Dec 2009\r\n 09:23:49 +0100", r.generate(78));
ferencd@0 493 }
ferencd@0 494
ferencd@0 495 void testBugFix20110511()
ferencd@0 496 {
ferencd@0 497 /*
ferencd@0 498
ferencd@0 499 Using the latest version of vmime (0.9.1), encoding the following string: Jean
ferencd@0 500 Gwenaël Dutourd will result in:
ferencd@0 501 Jean =?utf-8?Q?Gwena=C3=ABl_?= Dutourd
ferencd@0 502 However, decoding this will result in Jean Gwenaël Dutourd (notice two spaces
ferencd@0 503 between the last 2 words). The encoder adds a _ after the second word, but
ferencd@0 504 since the last word is not encoded, the space between them is not ignored, and
ferencd@0 505 is decoded into an additional space.
ferencd@0 506
ferencd@0 507 See: http://sourceforge.net/projects/vmime/forums/forum/237357/topic/4531365
ferencd@0 508
ferencd@0 509 */
ferencd@0 510
ferencd@0 511 const std::string DECODED_TEXT = "Jean Gwenaël Dutourd";
ferencd@0 512 const std::string ENCODED_TEXT = "Jean =?utf-8?Q?Gwena=C3=ABl?= Dutourd";
ferencd@0 513
ferencd@0 514 // Encode
ferencd@0 515 VASSERT_EQ("encode", ENCODED_TEXT,
ferencd@0 516 vmime::text::newFromString(DECODED_TEXT, vmime::charset("utf-8"))->generate());
ferencd@0 517
ferencd@0 518 // Decode
ferencd@0 519 vmime::text t;
ferencd@0 520 t.parse(ENCODED_TEXT);
ferencd@0 521
ferencd@0 522 // -- words
ferencd@0 523 std::ostringstream oss; oss << t;
ferencd@0 524 VASSERT_EQ("decode1",
ferencd@0 525 "[text: [[word: charset=us-ascii, buffer=Jean ],"
ferencd@0 526 "[word: charset=utf-8, buffer=Gwenaël],"
ferencd@0 527 "[word: charset=us-ascii, buffer= Dutourd]]]", oss.str());
ferencd@0 528
ferencd@0 529 // -- getWholeBuffer
ferencd@0 530 VASSERT_EQ("decode2", DECODED_TEXT, t.getWholeBuffer());
ferencd@0 531 }
ferencd@0 532
ferencd@0 533 void testInternationalizedEmail_specialChars()
ferencd@0 534 {
ferencd@0 535 vmime::generationContext ctx(vmime::generationContext::getDefaultContext());
ferencd@0 536 ctx.setInternationalizedEmailSupport(true);
ferencd@0 537
ferencd@0 538 vmime::generationContext::switcher <vmime::generationContext> contextSwitcher(ctx);
ferencd@0 539
ferencd@0 540 // Special sequence/chars should still be encoded
ferencd@0 541 VASSERT_EQ("1", "=?us-ascii?Q?Test=3D=3Frfc2047_sequence?=",
ferencd@0 542 vmime::word("Test=?rfc2047 sequence", vmime::charset("us-ascii")).generate());
ferencd@0 543
ferencd@0 544 VASSERT_EQ("2", "=?us-ascii?Q?Line_One=0ALine_Two?=",
ferencd@0 545 vmime::word("Line One\nLine Two", vmime::charset("us-ascii")).generate());
ferencd@0 546 }
ferencd@0 547
ferencd@0 548 void testInternationalizedEmail_UTF8()
ferencd@0 549 {
ferencd@0 550 vmime::generationContext ctx(vmime::generationContext::getDefaultContext());
ferencd@0 551 ctx.setInternationalizedEmailSupport(true);
ferencd@0 552
ferencd@0 553 vmime::generationContext::switcher <vmime::generationContext> contextSwitcher(ctx);
ferencd@0 554
ferencd@0 555 // Already UTF-8 encoded text should be left as is
ferencd@0 556 VASSERT_EQ("1", "Achim Br\xc3\xa4ndt",
ferencd@0 557 vmime::word("Achim Br\xc3\xa4ndt", vmime::charset("utf-8")).generate());
ferencd@0 558 }
ferencd@0 559
ferencd@0 560 void testInternationalizedEmail_nonUTF8()
ferencd@0 561 {
ferencd@0 562 vmime::generationContext ctx(vmime::generationContext::getDefaultContext());
ferencd@0 563 ctx.setInternationalizedEmailSupport(true);
ferencd@0 564
ferencd@0 565 vmime::generationContext::switcher <vmime::generationContext> contextSwitcher(ctx);
ferencd@0 566
ferencd@0 567 // Non UTF-8 encoded text should first be converted to UTF-8
ferencd@0 568 VASSERT_EQ("1", "Achim Br\xc3\xa4ndt",
ferencd@0 569 vmime::word("Achim Br\xe4ndt", vmime::charset("iso-8859-1")).generate());
ferencd@0 570 }
ferencd@0 571
ferencd@0 572 void testInternationalizedEmail_folding()
ferencd@0 573 {
ferencd@0 574 vmime::generationContext ctx(vmime::generationContext::getDefaultContext());
ferencd@0 575 ctx.setInternationalizedEmailSupport(true);
ferencd@0 576
ferencd@0 577 vmime::generationContext::switcher <vmime::generationContext> contextSwitcher(ctx);
ferencd@0 578
ferencd@0 579 // RFC-2047 encoding must be performed, as line folding is needed
ferencd@0 580 vmime::word w1("01234567890123456789\xc3\xa0x012345678901234567890123456789"
ferencd@0 581 "01234567890123456789\xc3\xa0x012345678901234567890123456789", vmime::charset("utf-8"));
ferencd@0 582
ferencd@0 583 VASSERT_EQ("1",
ferencd@0 584 "=?utf-8?Q?01234567890123456789=C3=A0x01234567890?=\r\n"
ferencd@0 585 " =?utf-8?Q?1234567890123456789012345678901234567?=\r\n"
ferencd@0 586 " =?utf-8?Q?89=C3=A0x0123456789012345678901234567?=\r\n"
ferencd@0 587 " =?utf-8?Q?89?=", w1.generate(50));
ferencd@0 588
ferencd@0 589 // RFC-2047 encoding will not be forced, as words can be wrapped in a new line
ferencd@0 590 vmime::word w2("bla bla bla This is some '\xc3\xa0\xc3\xa7' UTF-8 encoded text", vmime::charset("utf-8"));
ferencd@0 591
ferencd@0 592 VASSERT_EQ("2",
ferencd@0 593 "bla bla bla This is\r\n"
ferencd@0 594 " some '\xc3\xa0\xc3\xa7' UTF-8\r\n"
ferencd@0 595 " encoded text", w2.generate(20));
ferencd@0 596 }
ferencd@0 597
ferencd@0 598 VMIME_TEST_SUITE_END
ferencd@0 599