diff options
| -rw-r--r-- | src/mm-charsets.c | 77 | ||||
| -rw-r--r-- | src/mm-charsets.h | 10 | ||||
| -rw-r--r-- | src/tests/test-charsets.c | 205 |
3 files changed, 292 insertions, 0 deletions
diff --git a/src/mm-charsets.c b/src/mm-charsets.c index 8fea400e..d2b9a667 100644 --- a/src/mm-charsets.c +++ b/src/mm-charsets.c @@ -425,3 +425,80 @@ mm_charset_utf8_to_unpacked_gsm (const char *utf8, guint32 *out_len) return g_byte_array_free (gsm, FALSE); } +guint8 * +gsm_unpack (const guint8 *gsm, + guint32 gsm_len, + guint8 start_offset, /* in _bits_ */ + guint32 *out_unpacked_len) +{ + GByteArray *unpacked; + int i, nchars; + + nchars = ((gsm_len * 8) - start_offset) / 7; + unpacked = g_byte_array_sized_new (nchars + 1); + + for (i = 0; i < nchars; i++) { + guint8 bits_here, bits_in_next, octet, offset, c; + guint32 start_bit; + + start_bit = start_offset + (i * 7); /* Overall bit offset of char in buffer */ + offset = start_bit % 8; /* Offset to start of char in this byte */ + bits_here = offset ? (8 - offset) : 7; + bits_in_next = 7 - bits_here; + + /* Grab bits in the current byte */ + octet = gsm[start_bit / 8]; + c = (octet >> offset) & (0xFF >> (8 - bits_here)); + + /* Grab any bits that spilled over to next byte */ + if (bits_in_next) { + octet = gsm[(start_bit / 8) + 1]; + c |= (octet & (0xFF >> (8 - bits_in_next))) << bits_here; + } + g_byte_array_append (unpacked, &c, 1); + } + + *out_unpacked_len = unpacked->len; + return g_byte_array_free (unpacked, FALSE); +} + +guint8 * +gsm_pack (const guint8 *src, + guint32 src_len, + guint8 start_offset, + guint32 *out_packed_len) +{ + GByteArray *packed; + guint8 c, add_last = 0; + int i; + + packed = g_byte_array_sized_new (src_len); + + for (i = 0, c = 0; i < src_len; i++) { + guint8 bits_here, offset; + guint32 start_bit; + + start_bit = start_offset + (i * 7); /* Overall bit offset of char in buffer */ + offset = start_bit % 8; /* Offset to start of char in this byte */ + bits_here = offset ? (8 - offset) : 7; + + c |= (src[i] & 0x7F) << offset; + if (offset) { + /* Add this packed byte */ + g_byte_array_append (packed, &c, 1); + c = add_last = 0; + } + + /* Pack the rest of this char into the next byte */ + if (bits_here != 7) { + c = (src[i] & 0x7F) >> bits_here; + add_last = 1; + } + } + if (add_last) + g_byte_array_append (packed, &c, 1); + + *out_packed_len = packed->len; + return g_byte_array_free (packed, FALSE); +} + diff --git a/src/mm-charsets.h b/src/mm-charsets.h index ff39400e..661052da 100644 --- a/src/mm-charsets.h +++ b/src/mm-charsets.h @@ -52,5 +52,15 @@ guint8 *mm_charset_utf8_to_unpacked_gsm (const char *utf8, guint32 *out_len); guint8 *mm_charset_gsm_unpacked_to_utf8 (const guint8 *gsm, guint32 len); +guint8 *gsm_unpack (const guint8 *gsm, + guint32 gsm_len, + guint8 start_offset, /* in bits */ + guint32 *out_unpacked_len); + +guint8 *gsm_pack (const guint8 *src, + guint32 src_len, + guint8 start_offset, /* in bits */ + guint32 *out_packed_len); + #endif /* MM_CHARSETS_H */ diff --git a/src/tests/test-charsets.c b/src/tests/test-charsets.c index 80518dc5..2727345c 100644 --- a/src/tests/test-charsets.c +++ b/src/tests/test-charsets.c @@ -90,6 +90,202 @@ test_mixed_chars (void *f, gpointer d) g_free (utf8); } +static void +test_unpack_gsm7 (void *f, gpointer d) +{ + static const guint8 gsm[] = { 0xC8, 0xF7, 0x1D, 0x14, 0x96, 0x97, 0x41, 0xF9, 0x77, 0xFD, 0x07 }; + static const guint8 expected[] = { 0x48, 0x6f, 0x77, 0x20, 0x61, 0x72, 0x65, 0x20, 0x79, 0x6f, 0x75, 0x3f }; + guint8 *unpacked; + guint32 unpacked_len = 0; + + unpacked = gsm_unpack (gsm, sizeof (gsm), 0, &unpacked_len); + g_assert (unpacked); + g_assert_cmpint (unpacked_len, ==, sizeof (expected)); + g_assert_cmpint (memcmp (unpacked, expected, unpacked_len), ==, 0); + + g_free (unpacked); +} + +static void +test_unpack_gsm7_7_chars (void *f, gpointer d) +{ + static const guint8 gsm[] = { 0xF1, 0x7B, 0x59, 0x4E, 0xCF, 0xD7, 0x01 }; + static const guint8 expected[] = { 0x71, 0x77, 0x65, 0x72, 0x74, 0x79, 0x75, 0x00 }; + guint8 *unpacked; + guint32 unpacked_len = 0; + + /* Tests the edge case where there are 7 bits left in the packed + * buffer but those 7 bits do not contain a character. In this case + * we expect a trailing NULL byte and the caller must know enough about + * the intended message to remove it when required. + */ + + unpacked = gsm_unpack (gsm, sizeof (gsm), 0, &unpacked_len); + g_assert (unpacked); + g_assert_cmpint (unpacked_len, ==, sizeof (expected)); + g_assert_cmpint (memcmp (unpacked, expected, unpacked_len), ==, 0); + + g_free (unpacked); +} + +static void +test_unpack_gsm7_all_chars (void *f, gpointer d) +{ + /* Packed array of all chars in GSM default and extended charset */ + static const guint8 gsm[] = { + 0x80, 0x80, 0x60, 0x40, 0x28, 0x18, 0x0E, 0x88, 0x84, 0x62, 0xC1, 0x68, + 0x38, 0x1E, 0x90, 0x88, 0x64, 0x42, 0xA9, 0x58, 0x2E, 0x98, 0x8C, 0x66, + 0xC3, 0xE9, 0x78, 0x3E, 0xA0, 0x90, 0x68, 0x44, 0x2A, 0x99, 0x4E, 0xA8, + 0x94, 0x6A, 0xC5, 0x6A, 0xB9, 0x5E, 0xB0, 0x98, 0x6C, 0x46, 0xAB, 0xD9, + 0x6E, 0xB8, 0x9C, 0x6E, 0xC7, 0xEB, 0xF9, 0x7E, 0xC0, 0xA0, 0x70, 0x48, + 0x2C, 0x1A, 0x8F, 0xC8, 0xA4, 0x72, 0xC9, 0x6C, 0x3A, 0x9F, 0xD0, 0xA8, + 0x74, 0x4A, 0xAD, 0x5A, 0xAF, 0xD8, 0xAC, 0x76, 0xCB, 0xED, 0x7A, 0xBF, + 0xE0, 0xB0, 0x78, 0x4C, 0x2E, 0x9B, 0xCF, 0xE8, 0xB4, 0x7A, 0xCD, 0x6E, + 0xBB, 0xDF, 0xF0, 0xB8, 0x7C, 0x4E, 0xAF, 0xDB, 0xEF, 0xF8, 0xBC, 0x7E, + 0xCF, 0xEF, 0xFB, 0xFF, 0x1B, 0xC5, 0x86, 0xB2, 0x41, 0x6D, 0x52, 0x9B, + 0xD7, 0x86, 0xB7, 0xE9, 0x6D, 0x7C, 0x1B, 0xE0, 0xA6, 0x0C + }; + static const guint8 ext[] = { + 0x1B, 0x0A, 0x1B, 0x14, 0x1B, 0x28, 0x1B, 0x29, 0x1B, 0x2F, 0x1B, 0x3C, + 0x1B, 0x3D, 0x1B, 0x3E, 0x1B, 0x40, 0x1B, 0x65 + }; + guint8 *unpacked; + guint32 unpacked_len = 0; + int i; + + unpacked = gsm_unpack (gsm, sizeof (gsm), 0, &unpacked_len); + g_assert (unpacked); + g_assert_cmpint (unpacked_len, ==, 148); + + /* Test default chars */ + for (i = 0; i < 128; i++) + g_assert_cmpint (unpacked[i], ==, i); + + /* Text extended chars */ + g_assert_cmpint (memcmp ((guint8 *) (unpacked + 128), &ext[0], sizeof (ext)), ==, 0); + + g_free (unpacked); +} + +static void +test_pack_gsm7 (void *f, gpointer d) +{ + static const guint8 unpacked[] = { 0x48, 0x6f, 0x77, 0x20, 0x61, 0x72, 0x65, 0x20, 0x79, 0x6f, 0x75, 0x3f }; + static const guint8 expected[] = { 0xC8, 0xF7, 0x1D, 0x14, 0x96, 0x97, 0x41, 0xF9, 0x77, 0xFD, 0x07 }; + guint8 *packed; + guint32 packed_len = 0; + + packed = gsm_pack (unpacked, sizeof (unpacked), 0, &packed_len); + g_assert (packed); + g_assert_cmpint (packed_len, ==, sizeof (expected)); + g_assert_cmpint (memcmp (packed, expected, packed_len), ==, 0); + + g_free (packed); +} + +static void +test_pack_gsm7_7_chars (void *f, gpointer d) +{ + static const guint8 unpacked[] = { 0x71, 0x77, 0x65, 0x72, 0x74, 0x79, 0x75 }; + static const guint8 expected[] = { 0xF1, 0x7B, 0x59, 0x4E, 0xCF, 0xD7, 0x01 }; + guint8 *packed; + guint32 packed_len = 0; + + /* Tests the edge case where there are 7 bits left in the packed + * buffer but those 7 bits do not contain a character. In this case + * we expect a trailing NULL byte and the caller must know enough about + * the intended message to remove it when required. + */ + + packed = gsm_pack (unpacked, sizeof (unpacked), 0, &packed_len); + g_assert (packed); + g_assert_cmpint (packed_len, ==, sizeof (expected)); + g_assert_cmpint (memcmp (packed, expected, packed_len), ==, 0); + + g_free (packed); +} + +#if 0 +static void +print_array (const guint8 *array, guint32 len) +{ + int col, foo; + guint8 c; + + g_print ("\n"); + for (c = 0, col = 0, foo = 0; c < len; c++) { + g_print ("0x%02X, ", array[c] & 0xFF); + if (col++ == 11) { + col = 0; + g_print ("\n"); + } + foo = 1; + } + g_print ("\n"); +} +#endif + +static void +test_pack_gsm7_all_chars (void *f, gpointer d) +{ + /* Packed array of all chars in GSM default and extended charset */ + static const guint8 expected[] = { + 0x80, 0x80, 0x60, 0x40, 0x28, 0x18, 0x0E, 0x88, 0x84, 0x62, 0xC1, 0x68, + 0x38, 0x1E, 0x90, 0x88, 0x64, 0x42, 0xA9, 0x58, 0x2E, 0x98, 0x8C, 0x66, + 0xC3, 0xE9, 0x78, 0x3E, 0xA0, 0x90, 0x68, 0x44, 0x2A, 0x99, 0x4E, 0xA8, + 0x94, 0x6A, 0xC5, 0x6A, 0xB9, 0x5E, 0xB0, 0x98, 0x6C, 0x46, 0xAB, 0xD9, + 0x6E, 0xB8, 0x9C, 0x6E, 0xC7, 0xEB, 0xF9, 0x7E, 0xC0, 0xA0, 0x70, 0x48, + 0x2C, 0x1A, 0x8F, 0xC8, 0xA4, 0x72, 0xC9, 0x6C, 0x3A, 0x9F, 0xD0, 0xA8, + 0x74, 0x4A, 0xAD, 0x5A, 0xAF, 0xD8, 0xAC, 0x76, 0xCB, 0xED, 0x7A, 0xBF, + 0xE0, 0xB0, 0x78, 0x4C, 0x2E, 0x9B, 0xCF, 0xE8, 0xB4, 0x7A, 0xCD, 0x6E, + 0xBB, 0xDF, 0xF0, 0xB8, 0x7C, 0x4E, 0xAF, 0xDB, 0xEF, 0xF8, 0xBC, 0x7E, + 0xCF, 0xEF, 0xFB, 0xFF, 0x1B, 0xC5, 0x86, 0xB2, 0x41, 0x6D, 0x52, 0x9B, + 0xD7, 0x86, 0xB7, 0xE9, 0x6D, 0x7C, 0x1B, 0xE0, 0xA6, 0x0C + }; + static const guint8 ext[] = { + 0x1B, 0x0A, 0x1B, 0x14, 0x1B, 0x28, 0x1B, 0x29, 0x1B, 0x2F, 0x1B, 0x3C, + 0x1B, 0x3D, 0x1B, 0x3E, 0x1B, 0x40, 0x1B, 0x65 + }; + guint8 *packed, c; + guint32 packed_len = 0; + GByteArray *unpacked; + + unpacked = g_byte_array_sized_new (148); + for (c = 0; c < 128; c++) + g_byte_array_append (unpacked, &c, 1); + for (c = 0; c < sizeof (ext); c++) + g_byte_array_append (unpacked, &ext[c], 1); + + packed = gsm_pack (unpacked->data, unpacked->len, 0, &packed_len); + g_assert (packed); + g_assert_cmpint (packed_len, ==, sizeof (expected)); + g_assert_cmpint (memcmp (packed, expected, packed_len), ==, 0); + + g_free (packed); + g_byte_array_free (unpacked, TRUE); +} + +static void +test_pack_gsm7_24_chars (void *f, gpointer d) +{ + static const guint8 unpacked[] = { + 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 0x08, 0x09, 0x0A, 0x0B, + 0x0C, 0x0D, 0x0E, 0x0F, 0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16, 0x17 + }; + guint8 *packed; + guint32 packed_len = 0; + + /* Tests that no empty trailing byte is added when all the 7-bit characters + * are packed into an exact number of bytes. + */ + + packed = gsm_pack (unpacked, sizeof (unpacked), 0, &packed_len); + g_assert (packed); + g_assert_cmpint (packed_len, ==, 21); + + g_free (packed); +} + #if GLIB_CHECK_VERSION(2,25,12) typedef GTestFixtureFunc TCFunc; @@ -112,6 +308,15 @@ int main (int argc, char **argv) g_test_suite_add (suite, TESTCASE (test_esc_chars, NULL)); g_test_suite_add (suite, TESTCASE (test_mixed_chars, NULL)); + g_test_suite_add (suite, TESTCASE (test_unpack_gsm7, NULL)); + g_test_suite_add (suite, TESTCASE (test_unpack_gsm7_7_chars, NULL)); + g_test_suite_add (suite, TESTCASE (test_unpack_gsm7_all_chars, NULL)); + + g_test_suite_add (suite, TESTCASE (test_pack_gsm7, NULL)); + g_test_suite_add (suite, TESTCASE (test_pack_gsm7_7_chars, NULL)); + g_test_suite_add (suite, TESTCASE (test_pack_gsm7_all_chars, NULL)); + g_test_suite_add (suite, TESTCASE (test_pack_gsm7_24_chars, NULL)); + result = g_test_run (); return result; |