contrib/libarchive/libarchive/test/test_pax_filename_encoding.c

   1 /*-
   2  * Copyright (c) 2003-2007 Tim Kientzle
   3  * All rights reserved.
   4  *
   5  * Redistribution and use in source and binary forms, with or without
   6  * modification, are permitted provided that the following conditions
   7  * are met:
   8  * 1. Redistributions of source code must retain the above copyright
   9  *    notice, this list of conditions and the following disclaimer.
  10  * 2. Redistributions in binary form must reproduce the above copyright
  11  *    notice, this list of conditions and the following disclaimer in the
  12  *    documentation and/or other materials provided with the distribution.
  13  *
  14  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR
  15  * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
  16  * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
  17  * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
  18  * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
  19  * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
  20  * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
  21  * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
  22  * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
  23  * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  24  */
  25 #include "test.h"
  26 __FBSDID("$FreeBSD$");
  27
  28 #include <locale.h>
  29
  30 /*
  31  * Pax interchange is supposed to encode filenames into
  32  * UTF-8.  Of course, that's not always possible.  This
  33  * test is intended to verify that filenames always get
  34  * stored and restored correctly, regardless of the encodings.
  35  */
  36
  37 /*
  38  * Read a manually-created archive that has filenames that are
  39  * stored in binary instead of UTF-8 and verify that we get
  40  * the right filename returned and that we get a warning only
  41  * if the header isn't marked as binary.
  42  */
  43 static void
  44 test_pax_filename_encoding_1(void)
  45 {
  46         static const char testname[] = "test_pax_filename_encoding.tar";
  47         /*
  48          * \314\214 is a valid 2-byte UTF-8 sequence.
  49          * \374 is invalid in UTF-8.
  50          */
  51         char filename[] = "abc\314\214mno\374xyz";
  52         struct archive *a;
  53         struct archive_entry *entry;
  54
  55         /*
  56          * Read an archive that has non-UTF8 pax filenames in it.
  57          */
  58         extract_reference_file(testname);
  59         a = archive_read_new();
  60         assertEqualInt(ARCHIVE_OK, archive_read_support_format_tar(a));
  61         assertEqualInt(ARCHIVE_OK, archive_read_support_compression_all(a));
  62         assertEqualInt(ARCHIVE_OK,
  63             archive_read_open_filename(a, testname, 10240));
  64         /*
  65          * First entry in this test archive has an invalid UTF-8 sequence
  66          * in it, but the header is not marked as hdrcharset=BINARY, so that
  67          * requires a warning.
  68          */
  69         failure("Invalid UTF8 in a pax archive pathname should cause a warning");
  70         assertEqualInt(ARCHIVE_WARN, archive_read_next_header(a, &entry));
  71         assertEqualString(filename, archive_entry_pathname(entry));
  72         /*
  73          * Second entry is identical except that it does have
  74          * hdrcharset=BINARY, so no warning should be generated.
  75          */
  76         failure("A pathname with hdrcharset=BINARY can have invalid UTF8\n"
  77             " characters in it without generating a warning");
  78         assertEqualInt(ARCHIVE_OK, archive_read_next_header(a, &entry));
  79         assertEqualString(filename, archive_entry_pathname(entry));
  80         archive_read_finish(a);
  81 }
  82
  83 /*
  84  * Set the locale and write a pathname containing invalid characters.
  85  * This should work; the underlying implementation should automatically
  86  * fall back to storing the pathname in binary.
  87  */
  88 static void
  89 test_pax_filename_encoding_2(void)
  90 {
  91         char filename[] = "abc\314\214mno\374xyz";
  92         struct archive *a;
  93         struct archive_entry *entry;
  94         char buff[65536];
  95         char longname[] = "abc\314\214mno\374xyz"
  96             "/abc\314\214mno\374xyz/abcdefghijklmnopqrstuvwxyz"
  97             "/abc\314\214mno\374xyz/abcdefghijklmnopqrstuvwxyz"
  98             "/abc\314\214mno\374xyz/abcdefghijklmnopqrstuvwxyz"
  99             "/abc\314\214mno\374xyz/abcdefghijklmnopqrstuvwxyz"
 100             "/abc\314\214mno\374xyz/abcdefghijklmnopqrstuvwxyz"
 101             "/abc\314\214mno\374xyz/abcdefghijklmnopqrstuvwxyz"
 102             ;
 103         size_t used;
 104
 105         /*
 106          * We need a starting locale which has invalid sequences.
 107          * de_DE.UTF-8 seems to be commonly supported.
 108          */
 109         /* If it doesn't exist, just warn and return. */
 110         if (LOCALE_UTF8 == NULL
 111             || NULL == setlocale(LC_ALL, LOCALE_UTF8)) {
 112                 skipping("invalid encoding tests require a suitable locale;"
 113                     " %s not available on this system", LOCALE_UTF8);
 114                 return;
 115         }
 116
 117         assert((a = archive_write_new()) != NULL);
 118         assertEqualIntA(a, 0, archive_write_set_format_pax(a));
 119         assertEqualIntA(a, 0, archive_write_set_compression_none(a));
 120         assertEqualIntA(a, 0, archive_write_set_bytes_per_block(a, 0));
 121         assertEqualInt(0,
 122             archive_write_open_memory(a, buff, sizeof(buff), &used));
 123
 124         assert((entry = archive_entry_new()) != NULL);
 125         /* Set pathname, gname, uname, hardlink to nonconvertible values. */
 126         archive_entry_copy_pathname(entry, filename);
 127         archive_entry_copy_gname(entry, filename);
 128         archive_entry_copy_uname(entry, filename);
 129         archive_entry_copy_hardlink(entry, filename);
 130         archive_entry_set_filetype(entry, AE_IFREG);
 131         failure("This should generate a warning for nonconvertible names.");
 132         assertEqualInt(ARCHIVE_WARN, archive_write_header(a, entry));
 133         archive_entry_free(entry);
 134
 135         assert((entry = archive_entry_new()) != NULL);
 136         /* Set path, gname, uname, and symlink to nonconvertible values. */
 137         archive_entry_copy_pathname(entry, filename);
 138         archive_entry_copy_gname(entry, filename);
 139         archive_entry_copy_uname(entry, filename);
 140         archive_entry_copy_symlink(entry, filename);
 141         archive_entry_set_filetype(entry, AE_IFLNK);
 142         failure("This should generate a warning for nonconvertible names.");
 143         assertEqualInt(ARCHIVE_WARN, archive_write_header(a, entry));
 144         archive_entry_free(entry);
 145
 146         assert((entry = archive_entry_new()) != NULL);
 147         /* Set pathname to a very long nonconvertible value. */
 148         archive_entry_copy_pathname(entry, longname);
 149         archive_entry_set_filetype(entry, AE_IFREG);
 150         failure("This should generate a warning for nonconvertible names.");
 151         assertEqualInt(ARCHIVE_WARN, archive_write_header(a, entry));
 152         archive_entry_free(entry);
 153
 154         assertEqualInt(0, archive_write_close(a));
 155         assertEqualInt(0, archive_write_finish(a));
 156
 157         /*
 158          * Now read the entries back.
 159          */
 160
 161         assert((a = archive_read_new()) != NULL);
 162         assertEqualInt(0, archive_read_support_format_tar(a));
 163         assertEqualInt(0, archive_read_open_memory(a, buff, used));
 164
 165         assertEqualInt(0, archive_read_next_header(a, &entry));
 166         assertEqualString(filename, archive_entry_pathname(entry));
 167         assertEqualString(filename, archive_entry_gname(entry));
 168         assertEqualString(filename, archive_entry_uname(entry));
 169         assertEqualString(filename, archive_entry_hardlink(entry));
 170
 171         assertEqualInt(0, archive_read_next_header(a, &entry));
 172         assertEqualString(filename, archive_entry_pathname(entry));
 173         assertEqualString(filename, archive_entry_gname(entry));
 174         assertEqualString(filename, archive_entry_uname(entry));
 175         assertEqualString(filename, archive_entry_symlink(entry));
 176
 177         assertEqualInt(0, archive_read_next_header(a, &entry));
 178         assertEqualString(longname, archive_entry_pathname(entry));
 179
 180         assertEqualInt(0, archive_read_close(a));
 181         assertEqualInt(0, archive_read_finish(a));
 182 }
 183
 184 /*
 185  * Create an entry starting from a wide-character Unicode pathname,
 186  * read it back into "C" locale, which doesn't support the name.
 187  * TODO: Figure out the "right" behavior here.
 188  */
 189 static void
 190 test_pax_filename_encoding_3(void)
 191 {
 192         wchar_t badname[] = L"xxxAyyyBzzz";
 193         const char badname_utf8[] = "xxx\xE1\x88\xB4yyy\xE5\x99\xB8zzz";
 194         struct archive *a;
 195         struct archive_entry *entry;
 196         char buff[65536];
 197         size_t used;
 198
 199         badname[3] = 0x1234;
 200         badname[7] = 0x5678;
 201
 202         /* If it doesn't exist, just warn and return. */
 203         if (NULL == setlocale(LC_ALL, "C")) {
 204                 skipping("Can't set \"C\" locale, so can't exercise "
 205                     "certain character-conversion failures");
 206                 return;
 207         }
 208
 209         /* If wctomb is broken, warn and return. */
 210         if (wctomb(buff, 0x1234) > 0) {
 211                 skipping("Cannot test conversion failures because \"C\" "
 212                     "locale on this system has no invalid characters.");
 213                 return;
 214         }
 215
 216         /* If wctomb is broken, warn and return. */
 217         if (wctomb(buff, 0x1234) > 0) {
 218                 skipping("Cannot test conversion failures because \"C\" "
 219                     "locale on this system has no invalid characters.");
 220                 return;
 221         }
 222
 223         /* Skip test if archive_entry_update_pathname_utf8() is broken. */
 224         /* In particular, this is currently broken on Win32 because
 225          * setlocale() does not set the default encoding for CP_ACP. */
 226         entry = archive_entry_new();
 227         if (archive_entry_update_pathname_utf8(entry, badname_utf8)) {
 228                 archive_entry_free(entry);
 229                 skipping("Cannot test conversion failures.");
 230                 return;
 231         }
 232         archive_entry_free(entry);
 233
 234         assert((a = archive_write_new()) != NULL);
 235         assertEqualIntA(a, 0, archive_write_set_format_pax(a));
 236         assertEqualIntA(a, 0, archive_write_set_compression_none(a));
 237         assertEqualIntA(a, 0, archive_write_set_bytes_per_block(a, 0));
 238         assertEqualInt(0,
 239             archive_write_open_memory(a, buff, sizeof(buff), &used));
 240
 241         assert((entry = archive_entry_new()) != NULL);
 242         /* Set pathname to non-convertible wide value. */
 243         archive_entry_copy_pathname_w(entry, badname);
 244         archive_entry_set_filetype(entry, AE_IFREG);
 245         assertEqualInt(ARCHIVE_OK, archive_write_header(a, entry));
 246         archive_entry_free(entry);
 247
 248         assert((entry = archive_entry_new()) != NULL);
 249         archive_entry_copy_pathname_w(entry, L"abc");
 250         /* Set gname to non-convertible wide value. */
 251         archive_entry_copy_gname_w(entry, badname);
 252         archive_entry_set_filetype(entry, AE_IFREG);
 253         assertEqualInt(ARCHIVE_OK, archive_write_header(a, entry));
 254         archive_entry_free(entry);
 255
 256         assert((entry = archive_entry_new()) != NULL);
 257         archive_entry_copy_pathname_w(entry, L"abc");
 258         /* Set uname to non-convertible wide value. */
 259         archive_entry_copy_uname_w(entry, badname);
 260         archive_entry_set_filetype(entry, AE_IFREG);
 261         assertEqualInt(ARCHIVE_OK, archive_write_header(a, entry));
 262         archive_entry_free(entry);
 263
 264         assert((entry = archive_entry_new()) != NULL);
 265         archive_entry_copy_pathname_w(entry, L"abc");
 266         /* Set hardlink to non-convertible wide value. */
 267         archive_entry_copy_hardlink_w(entry, badname);
 268         archive_entry_set_filetype(entry, AE_IFREG);
 269         assertEqualInt(ARCHIVE_OK, archive_write_header(a, entry));
 270         archive_entry_free(entry);
 271
 272         assert((entry = archive_entry_new()) != NULL);
 273         archive_entry_copy_pathname_w(entry, L"abc");
 274         /* Set symlink to non-convertible wide value. */
 275         archive_entry_copy_symlink_w(entry, badname);
 276         archive_entry_set_filetype(entry, AE_IFLNK);
 277         assertEqualInt(ARCHIVE_OK, archive_write_header(a, entry));
 278         archive_entry_free(entry);
 279
 280         assertEqualInt(0, archive_write_close(a));
 281         assertEqualInt(0, archive_write_finish(a));
 282
 283         /*
 284          * Now read the entries back.
 285          */
 286
 287         assert((a = archive_read_new()) != NULL);
 288         assertEqualInt(0, archive_read_support_format_tar(a));
 289         assertEqualInt(0, archive_read_open_memory(a, buff, used));
 290
 291         failure("A non-convertible pathname should cause a warning.");
 292         assertEqualInt(ARCHIVE_WARN, archive_read_next_header(a, &entry));
 293         assertEqualWString(badname, archive_entry_pathname_w(entry));
 294         failure("If native locale can't convert, we should get UTF-8 back.");
 295         assertEqualString(badname_utf8, archive_entry_pathname(entry));
 296
 297         failure("A non-convertible gname should cause a warning.");
 298         assertEqualInt(ARCHIVE_WARN, archive_read_next_header(a, &entry));
 299         assertEqualWString(badname, archive_entry_gname_w(entry));
 300         failure("If native locale can't convert, we should get UTF-8 back.");
 301         assertEqualString(badname_utf8, archive_entry_gname(entry));
 302
 303         failure("A non-convertible uname should cause a warning.");
 304         assertEqualInt(ARCHIVE_WARN, archive_read_next_header(a, &entry));
 305         assertEqualWString(badname, archive_entry_uname_w(entry));
 306         failure("If native locale can't convert, we should get UTF-8 back.");
 307         assertEqualString(badname_utf8, archive_entry_uname(entry));
 308
 309         failure("A non-convertible hardlink should cause a warning.");
 310         assertEqualInt(ARCHIVE_WARN, archive_read_next_header(a, &entry));
 311         assertEqualWString(badname, archive_entry_hardlink_w(entry));
 312         failure("If native locale can't convert, we should get UTF-8 back.");
 313         assertEqualString(badname_utf8, archive_entry_hardlink(entry));
 314
 315         failure("A non-convertible symlink should cause a warning.");
 316         assertEqualInt(ARCHIVE_WARN, archive_read_next_header(a, &entry));
 317         assertEqualWString(badname, archive_entry_symlink_w(entry));
 318         assertEqualWString(NULL, archive_entry_hardlink_w(entry));
 319         failure("If native locale can't convert, we should get UTF-8 back.");
 320         assertEqualString(badname_utf8, archive_entry_symlink(entry));
 321
 322         assertEqualInt(ARCHIVE_EOF, archive_read_next_header(a, &entry));
 323
 324         assertEqualInt(0, archive_read_close(a));
 325         assertEqualInt(0, archive_read_finish(a));
 326 }
 327
 328 DEFINE_TEST(test_pax_filename_encoding)
 329 {
 330         test_pax_filename_encoding_1();
 331         test_pax_filename_encoding_2();
 332         test_pax_filename_encoding_3();
 333 }