Symaptic: os/textandloc/charconvfw/charconv_fw/src/charconv/sms7bit.cpp@260cb5ec6c19

     1 /*

     2 * Copyright (c) 1997-2009 Nokia Corporation and/or its subsidiary(-ies).

     3 * All rights reserved.

     4 * This component and the accompanying materials are made available

     5 * under the terms of "Eclipse Public License v1.0"

     6 * which accompanies this distribution, and is available

     7 * at the URL "http://www.eclipse.org/legal/epl-v10.html".

8 *

     9 * Initial Contributors:

    10 * Nokia Corporation - initial contribution.

    11 *

    12 * Contributors:

    13 *

    14 * Description:

    15 *

    16 */

    19 #include <e32std.h>

    20 #include <convdata.h>

    22 #define ARRAY_LENGTH(aArray) (sizeof(aArray)/sizeof((aArray)[0]))

    24 LOCAL_D const TUint16 keyedTables16OfIndexedTables16_indexedEntries_sms7BitToUnicode_1[]=

    25 	{

    26 	0x0040,

    27 	0x00a3,

    28 	0x0024,

    29 	0x00a5,

    30 	0x00e8,

    31 	0x00e9,

    32 	0x00f9,

    33 	0x00ec,

    34 	0x00f2,

    35 	0x00e7,

    36 	0x000a,

    37 	0x00d8,

    38 	0x00f8,

    39 	0x000d,

    40 	0x00c5,

    41 	0x00e5,

    42 	0x0394,

    43 	0x005f,

    44 	0x03a6,

    45 	0x0393,

    46 	0x039b,

    47 	0x03a9,

    48 	0x03a0,

    49 	0x03a8,

    50 	0x03a3,

    51 	0x0398,

    52 	0x039e

    53 	};

    55 LOCAL_D const TUint16 keyedTables16OfIndexedTables16_indexedEntries_sms7BitToUnicode_2[]=

    56 	{

    57 	0x00c6,

    58 	0x00e6,

    59 	0x00df,

    60 	0x00c9,

    61 	0x0020,

    62 	0x0021,

    63 	0x0022,

    64 	0x0023,

    65 	0x00a4

    66 	};

    68 LOCAL_D const TUint16 keyedTables16OfIndexedTables16_indexedEntries_sms7BitToUnicode_3[]=

    69 	{

    70 	0x00c4,

    71 	0x00d6,

    72 	0x00d1,

    73 	0x00dc,

    74 	0x00a7,

    75 	0x00bf

    76 	};

    78 LOCAL_D const TUint16 keyedTables16OfIndexedTables16_indexedEntries_sms7BitToUnicode_4[]=

    79 	{

    80 	0x00e4,

    81 	0x00f6,

    82 	0x00f1,

    83 	0x00fc,

    84 	0x00e0

    85 	};

    87 LOCAL_D const TUint16 keyedTables16OfIndexedTables16_indexedEntries_sms7BitToUnicode_5[]=

    88 	{

    89 	0x0040,

    90 	0x00a3,

    91 	0x0024,

    92 	0x00a5,

    93 	0x00e8,

    94 	0x00e9,

    95 	0x00f9,

    96 	0x00ec,

    97 	0x00f2,

    98 	0x00c7,

    99 	0x000c, // meant to be a "Page Break" - a FORM FEED character is the nearest thing in Unicode to this

   100 	0x00d8,

   101 	0x00f8,

   102 	0x000d,

   103 	0x00c5,

   104 	0x00e5,

   105 	0x0394,

   106 	0x005f,

   107 	0x03a6,

   108 	0x0393,

   109 	0x005e, // '^'

   110 	0x03a9,

   111 	0x03a0,

   112 	0x03a8,

   113 	0x03a3,

   114 	0x0398,

   115 	0x039e

   116 	};

   118 LOCAL_D const TUint16 keyedTables16OfIndexedTables16_indexedEntries_sms7BitToUnicode_6[]=

   119 	{

   120 	0x00c6,

   121 	0x00e6,

   122 	0x00df,

   123 	0x00c9,

   124 	0x0020,

   125 	0x0021,

   126 	0x0022,

   127 	0x0023,

   128 	0x00a4,

   129 	0x0025,

   130 	0x0026,

   131 	0x0027,

   132 	0x007b, // '{'

   133 	0x007d, // '}'

   134 	0x002a,

   135 	0x002b,

   136 	0x002c,

   137 	0x002d,

   138 	0x002e,

   139 	0x005c, // '\'

   140 	0x0030,

   141 	0x0031,

   142 	0x0032,

   143 	0x0033,

   144 	0x0034,

   145 	0x0035,

   146 	0x0036,

   147 	0x0037,

   148 	0x0038,

   149 	0x0039,

   150 	0x003a,

   151 	0x003b,

   152 	0x005b, // '['

   153 	0x007e, // '~'

   154 	0x005d, // ']'

   155 	0x003f,

   156 	0x007c  // '|'

   157 	};

   159 LOCAL_D const TUint16 keyedTables16OfIndexedTables16_indexedEntries_sms7BitToUnicode_7[]=

   160 	{

   161 	0x00c4,

   162 	0x00d6,

   163 	0x00d1,

   164 	0x00dc,

   165 	0x00a7,

   166 	0x00bf,

   167 	0x0061,

   168 	0x0062,

   169 	0x0063,

   170 	0x0064,

   171 	0x20ac  // Euro symbol

   172 	};

   174 LOCAL_D const SCnvConversionData::SOneDirectionData::SRange::UData::SKeyedTable16OfIndexedTables16::SKeyedEntry keyedTables16OfIndexedTables16_keyedEntries_sms7BitToUnicode_1[]=

   175 	{

   176 		{

   177 		0x00,

   178 		0x1a,

   179 		keyedTables16OfIndexedTables16_indexedEntries_sms7BitToUnicode_1

   180 		},

   181 		{

   182 		0x1c,

   183 		0x24,

   184 		keyedTables16OfIndexedTables16_indexedEntries_sms7BitToUnicode_2

   185 		},

   186 		{

   187 		0x5b,

   188 		0x60,

   189 		keyedTables16OfIndexedTables16_indexedEntries_sms7BitToUnicode_3

   190 		},

   191 		{

   192 		0x7b,

   193 		0x7f,

   194 		keyedTables16OfIndexedTables16_indexedEntries_sms7BitToUnicode_4

   195 		}

   196 	};

   198 LOCAL_D const SCnvConversionData::SOneDirectionData::SRange::UData::SKeyedTable16OfIndexedTables16::SKeyedEntry keyedTables16OfIndexedTables16_keyedEntries_sms7BitToUnicode_2[]=

   199 	{

   200 		{

   201 		0x1b00,

   202 		0x1b1a,

   203 		keyedTables16OfIndexedTables16_indexedEntries_sms7BitToUnicode_5

   204 		},

   205 		{

   206 		0x1b1c,

   207 		0x1b40,

   208 		keyedTables16OfIndexedTables16_indexedEntries_sms7BitToUnicode_6

   209 		},

   210 		{

   211 		0x1b5b,

   212 		0x1b65,

   213 		keyedTables16OfIndexedTables16_indexedEntries_sms7BitToUnicode_7

   214 		},

   215 		{

   216 		0x1b7b,

   217 		0x1b7f,

   218 		keyedTables16OfIndexedTables16_indexedEntries_sms7BitToUnicode_4

   219 		}

   220 	};

   222 #if 0

   223 LOCAL_D const SCnvConversionData::SOneDirectionData::SRange::UData::SKeyedTable1616::SEntry keyedTable1616_sms7BitToUnicode_1[]=

   224 	{

   225 		{

   226 		0x1b0a,

   227 		0x000c

   228 		},

   229 		{

   230 		0x1b14,

   231 		0x005e

   232 		},

   233 		{

   234 		0x1b28,

   235 		0x007b

   236 		},

   237 		{

   238 		0x1b29,

   239 		0x007d

   240 		},

   241 		{

   242 		0x1b2f,

   243 		0x005c

   244 		},

   245 		{

   246 		0x1b3c,

   247 		0x005b

   248 		},

   249 		{

   250 		0x1b3d,

   251 		0x007e

   252 		},

   253 		{

   254 		0x1b3e,

   255 		0x005d

   256 		},

   257 		{

   258 		0x1b40,

   259 		0x007c

   260 		},

   261 		{

   262 		0x1b65,

   263 		0x20ac

   264 		}

   265 	};

   266 #endif //0

   268 LOCAL_D const TUint16 keyedTables16OfIndexedTables16_indexedEntries_unicodeToSms7Bit_1[]=

   269 	{

   270 	0x60,

   271 	0x41,

   272 	0x41,

   273 	0x41,

   274 	0x41,

   275 	0x5b,

   276 	0x0e,

   277 	0x1c,

   278 	0x09,

   279 	0x45,

   280 	0x1f,

   281 	0x45,

   282 	0x45,

   283 	0x49,

   284 	0x49,

   285 	0x49,

   286 	0x49

   287 	};

   289 LOCAL_D const TUint16 keyedTables16OfIndexedTables16_indexedEntries_unicodeToSms7Bit_2[]=

   290 	{

   291 	0x5d,

   292 	0x4f,

   293 	0x4f,

   294 	0x4f,

   295 	0x4f,

   296 	0x5c

   297 	};

   299 LOCAL_D const TUint16 keyedTables16OfIndexedTables16_indexedEntries_unicodeToSms7Bit_3[]=

   300 	{

   301 	0x0b,

   302 	0x55,

   303 	0x55,

   304 	0x55,

   305 	0x5e,

   306 	0x59

   307 	};

   309 LOCAL_D const TUint16 keyedTables16OfIndexedTables16_indexedEntries_unicodeToSms7Bit_4[]=

   310 	{

   311 	0x1e,

   312 	0x7f,

   313 	0x61,

   314 	0x61,

   315 	0x61,

   316 	0x7b,

   317 	0x0f,

   318 	0x1d,

   319 	0x09,

   320 	0x04,

   321 	0x05,

   322 	0x65,

   323 	0x65,

   324 	0x07,

   325 	0x69,

   326 	0x69,

   327 	0x69

   328 	};

   330 LOCAL_D const TUint16 keyedTables16OfIndexedTables16_indexedEntries_unicodeToSms7Bit_5[]=

   331 	{

   332 	0x7d,

   333 	0x08,

   334 	0x6f,

   335 	0x6f,

   336 	0x6f,

   337 	0x7c

   338 	};

   340 LOCAL_D const TUint16 keyedTables16OfIndexedTables16_indexedEntries_unicodeToSms7Bit_6[]=

   341 	{

   342 	0x0c,

   343 	0x06,

   344 	0x75,

   345 	0x75,

   346 	0x7e,

   347 	0x79

   348 	};

   350 LOCAL_D const TUint16 keyedTables16OfIndexedTables16_indexedEntries_unicodeToSms7Bit_7[]=

   351 	{

   352 	0x49,

   353 	0x41,

   354 	0x42,

   355 	0x13,

   356 	0x10,

   357 	0x45,

   358 	0x5a,

   359 	0x48,

   360 	0x19,

   361 	0x49,

   362 	0x4b,

   363 	0x14,

   364 	0x4d,

   365 	0x4e,

   366 	0x1a,

   367 	0x4f

   368 	};

   370 LOCAL_D const TUint16 keyedTables16OfIndexedTables16_indexedEntries_unicodeToSms7Bit_8[]=

   371 	{

   372 	0x18,

   373 	0x54,

   374 	0x59,

   375 	0x12,

   376 	0x58,

   377 	0x17,

   378 	0x15,

   379 	0x49,

   380 	0x59,

   381 	0x41,

   382 	0x45,

   383 	0x48,

   384 	0x49

   385 	};

   387 LOCAL_D const TUint16 keyedTables16OfIndexedTables16_indexedEntries_unicodeToSms7Bit_9[]=

   388 	{

   389 	0x59,

   390 	0x41,

   391 	0x42,

   392 	0x13,

   393 	0x10,

   394 	0x45,

   395 	0x5a,

   396 	0x48,

   397 	0x19,

   398 	0x49,

   399 	0x4b,

   400 	0x14,

   401 	0x4d,

   402 	0x4e,

   403 	0x1a,

   404 	0x4f

   405 	};

   407 LOCAL_D const TUint16 keyedTables16OfIndexedTables16_indexedEntries_unicodeToSms7Bit_10[]=

   408 	{

   409 	0x16,

   410 	0x50,

   411 	0x18,

   412 	0x18,

   413 	0x54,

   414 	0x59,

   415 	0x12,

   416 	0x58,

   417 	0x17,

   418 	0x15,

   419 	0x49,

   420 	0x59,

   421 	0x4f,

   422 	0x59,

   423 	0x15

   424 	};

   426 LOCAL_D const TUint16 keyedTables16OfIndexedTables16_indexedEntries_unicodeToSms7Bit_11[]=

   427 	{

   428 	0x42,

   429 	0x19,

   430 	0x59,

   431 	0x59,

   432 	0x59,

   433 	0x12

   434 	};

   436 LOCAL_D const SCnvConversionData::SOneDirectionData::SRange::UData::SKeyedTable16OfIndexedTables16::SKeyedEntry keyedTables16OfIndexedTables16_keyedEntries_unicodeToSms7Bit_1[]=

   437 	{

   438 		{

   439 		0x00bf,

   440 		0x00cf,

   441 		keyedTables16OfIndexedTables16_indexedEntries_unicodeToSms7Bit_1

   442 		},

   443 		{

   444 		0x00d1,

   445 		0x00d6,

   446 		keyedTables16OfIndexedTables16_indexedEntries_unicodeToSms7Bit_2

   447 		},

   448 		{

   449 		0x00d8,

   450 		0x00dd,

   451 		keyedTables16OfIndexedTables16_indexedEntries_unicodeToSms7Bit_3

   452 		},

   453 		{

   454 		0x00df,

   455 		0x00ef,

   456 		keyedTables16OfIndexedTables16_indexedEntries_unicodeToSms7Bit_4

   457 		},

   458 		{

   459 		0x00f1,

   460 		0x00f6,

   461 		keyedTables16OfIndexedTables16_indexedEntries_unicodeToSms7Bit_5

   462 		},

   463 		{

   464 		0x00f8,

   465 		0x00fd,

   466 		keyedTables16OfIndexedTables16_indexedEntries_unicodeToSms7Bit_6

   467 		},

   468 		{

   469 		0x0390,

   470 		0x039f,

   471 		keyedTables16OfIndexedTables16_indexedEntries_unicodeToSms7Bit_7

   472 		},

   473 		{

   474 		0x03a3,

   475 		0x03af,

   476 		keyedTables16OfIndexedTables16_indexedEntries_unicodeToSms7Bit_8

   477 		},

   478 		{

   479 		0x03b0,

   480 		0x03bf,

   481 		keyedTables16OfIndexedTables16_indexedEntries_unicodeToSms7Bit_9

   482 		},

   483 		{

   484 		0x03c0,

   485 		0x03ce,

   486 		keyedTables16OfIndexedTables16_indexedEntries_unicodeToSms7Bit_10

   487 		},

   488 		{

   489 		0x03d0,

   490 		0x03d5,

   491 		keyedTables16OfIndexedTables16_indexedEntries_unicodeToSms7Bit_11

   492 		}

   493 	};

   495 LOCAL_D const SCnvConversionData::SOneDirectionData::SRange::UData::SKeyedTable1616::SEntry keyedTable1616_unicodeToSms7Bit_1[]=

   496 	{

   497 		{

   498 		0x000a,

   499 		0x0a

   500 		},

   501 		{

   502 		0x000d,

   503 		0x0d

   504 		},

   505 		{

   506 		0x001b,

   507 		0x1b

   508 		},

   509 		{

   510 		0x0020,

   511 		0x20

   512 		},

   513 		{

   514 		0x0021,

   515 		0x21

   516 		},

   517 		{

   518 		0x0022,

   519 		0x22

   520 		},

   521 		{

   522 		0x0023,

   523 		0x23

   524 		},

   525 		{

   526 		0x0024,

   527 		0x02

   528 		},

   529 		{

   530 		0x0040,

   531 		0x00

   532 		},

   533 		{

   534 		0x005f,

   535 		0x11

   536 		},

   537 		{

   538 		0x00a1,

   539 		0x40

   540 		},

   541 		{

   542 		0x00a3,

   543 		0x01

   544 		},

   545 		{

   546 		0x00a4,

   547 		0x24

   548 		},

   549 		{

   550 		0x00a5,

   551 		0x03

   552 		},

   553 		{

   554 		0x00a7,

   555 		0x5f

   556 		},

   557 		{

   558 		0x00ff,

   559 		0x79

   560 		},

   561 		{

   562 		0x0386,

   563 		0x41

   564 		},

   565 		{

   566 		0x0388,

   567 		0x45

   568 		},

   569 		{

   570 		0x0389,

   571 		0x48

   572 		},

   573 		{

   574 		0x038a,

   575 		0x49

   576 		},

   577 		{

   578 		0x038c,

   579 		0x4f

   580 		},

   581 		{

   582 		0x038e,

   583 		0x59

   584 		},

   585 		{

   586 		0x038f,

   587 		0x15

   588 		},

   589 		{

   590 		0x03a0,

   591 		0x16

   592 		},

   593 		{

   594 		0x03a1,

   595 		0x50

   596 		},

   597 		{

   598 		0x2126,

   599 		0x15

   600 		},

   601 		{

   602 		0x220f,

   603 		0x16

   604 		},

   605 		{

   606 		0x2211,

   607 		0x18

   608 		}

   609 	};

   611 LOCAL_D const SCnvConversionData::SOneDirectionData::SRange::UData::SKeyedTable1616::SEntry keyedTable1616_unicodeToSms7Bit_2[]=

   612 	{

   613 		{

   614 		0x000c,

   615 		0x1b0a

   616 		},

   617 		{

   618 		0x005b,

   619 		0x1b3c

   620 		},

   621 		{

   622 		0x005c,

   623 		0x1b2f

   624 		},

   625 		{

   626 		0x005d,

   627 		0x1b3e

   628 		},

   629 		{

   630 		0x005e,

   631 		0x1b14

   632 		},

   633 		{

   634 		0x007b,

   635 		0x1b28

   636 		},

   637 		{

   638 		0x007c,

   639 		0x1b40

   640 		},

   641 		{

   642 		0x007d,

   643 		0x1b29

   644 		},

   645 		{

   646 		0x007e,

   647 		0x1b3d

   648 		},

   649 		{

   650 		0x20ac,

   651 		0x1b65

   652 		}

   653 	};

   655 LOCAL_D const SCnvConversionData::SVariableByteData::SRange sms7BitVariableByteDataRanges[]=

   656 	{

   657 		{

   658 		0x00,

   659 		0x1a,

   660 		0,

   661 		0

   662 		},

   663 		{

   664 		0x1b,

   665 		0x1b,

   666 		1,

   667 		0

   668 		},

   669 		{

   670 		0x1c,

   671 		0xff,

   672 		0,

   673 		0

   674 		}

   675 	};

   677 LOCAL_D const SCnvConversionData::SOneDirectionData::SRange sms7BitToUnicodeDataRanges[]=

   678 	{

   679 		{

   680 		0x25,

   681 		0x3f,

   682 		SCnvConversionData::SOneDirectionData::SRange::EDirect,

   683 		0,

   684 		0,

   685 			{

   686 			0,

   687 			0

   688 			}

   689 		},

   690 		{

   691 		0x41,

   692 		0x5a,

   693 		SCnvConversionData::SOneDirectionData::SRange::EDirect,

   694 		0,

   695 		0,

   696 			{

   697 			0,

   698 			0

   699 			}

   700 		},

   701 		{

   702 		0x61,

   703 		0x7a,

   704 		SCnvConversionData::SOneDirectionData::SRange::EDirect,

   705 		0,

   706 		0,

   707 			{

   708 			0,

   709 			0

   710 			}

   711 		},

   712 		{

   713 		0x40,

   714 		0x40,

   715 		SCnvConversionData::SOneDirectionData::SRange::EOffset,

   716 		0,

   717 		0,

   718 			{

   719 			STATIC_CAST(TUint, 97),

   720 			0

   721 			}

   722 		},

   723 		{

   724 		0x00,

   725 		0x7f,

   726 		SCnvConversionData::SOneDirectionData::SRange::EKeyedTable16OfIndexedTables16,

   727 		0,

   728 		0,

   729 			{

   730 			UData_SKeyedTable16OfIndexedTables16(keyedTables16OfIndexedTables16_keyedEntries_sms7BitToUnicode_1)

   731 			}

   732 		},

   733 #if 0

   734 		{

   735 		0x1b0a,

   736 		0x1b65,

   737 		SCnvConversionData::SOneDirectionData::SRange::EKeyedTable1616,

   738 		0,

   739 		0,

   740 			{

   741 			UData_KeyedTable(keyedTable1616_sms7BitToUnicode_1)

   742 			}

   743 		}

   744 #else

   745 		{

   746 		0x1b41,

   747 		0x1b5a,

   748 		SCnvConversionData::SOneDirectionData::SRange::EOffset,

   749 		0,

   750 		0,

   751 			{

   752 			(TUint)-0x1b00,

   753 			0

   754 			}

   755 		},

   756 		{

   757 		0x1b66,

   758 		0x1b7a,

   759 		SCnvConversionData::SOneDirectionData::SRange::EOffset,

   760 		0,

   761 		0,

   762 			{

   763 			(TUint)-0x1b00,

   764 			0

   765 			}

   766 		},

   767 		{

   768 		0x1b00,

   769 		0x1b7f,

   770 		SCnvConversionData::SOneDirectionData::SRange::EKeyedTable16OfIndexedTables16,

   771 		0,

   772 		0,

   773 			{

   774 			UData_SKeyedTable16OfIndexedTables16(keyedTables16OfIndexedTables16_keyedEntries_sms7BitToUnicode_2)

   775 			}

   776 		}

   777 #endif

   778 	};

   780 LOCAL_D const SCnvConversionData::SOneDirectionData::SRange unicodeToSms7BitDataRanges[]=

   781 	{

   782 		{

   783 		0x0025,

   784 		0x003f,

   785 		SCnvConversionData::SOneDirectionData::SRange::EDirect,

   786 		1,

   787 		0,

   788 			{

   789 			0,

   790 			0

   791 			}

   792 		},

   793 		{

   794 		0x0041,

   795 		0x005a,

   796 		SCnvConversionData::SOneDirectionData::SRange::EDirect,

   797 		1,

   798 		0,

   799 			{

   800 			0,

   801 			0

   802 			}

   803 		},

   804 		{

   805 		0x0061,

   806 		0x007a,

   807 		SCnvConversionData::SOneDirectionData::SRange::EDirect,

   808 		1,

   809 		0,

   810 			{

   811 			0,

   812 			0

   813 			}

   814 		},

   815 		{

   816 		0x00bf,

   817 		0x03d5,

   818 		SCnvConversionData::SOneDirectionData::SRange::EKeyedTable16OfIndexedTables16,

   819 		1,

   820 		0,

   821 			{

   822 			UData_SKeyedTable16OfIndexedTables16(keyedTables16OfIndexedTables16_keyedEntries_unicodeToSms7Bit_1)

   823 			}

   824 		},

   825 		{

   826 		0x000a,

   827 		0x2211,

   828 		SCnvConversionData::SOneDirectionData::SRange::EKeyedTable1616,

   829 		1,

   830 		0,

   831 			{

   832 			UData_SKeyedTable1616(keyedTable1616_unicodeToSms7Bit_1)

   833 			}

   834 		},

   835 		{

   836 		0x000c,

   837 		0x20ac,

   838 		SCnvConversionData::SOneDirectionData::SRange::EKeyedTable1616,

   839 		2,

   840 		0,

   841 			{

   842 			UData_SKeyedTable1616(keyedTable1616_unicodeToSms7Bit_2)

   843 			}

   844 		}

   845 	};

   847 GLREF_D const SCnvConversionData sms7BitConversionData=

   848 	{

   849 	SCnvConversionData::EFixedBigEndian,

   850 		{

   851 		ARRAY_LENGTH(sms7BitVariableByteDataRanges),

   852 		sms7BitVariableByteDataRanges

   853 		},

   854 		{

   855 		ARRAY_LENGTH(sms7BitToUnicodeDataRanges),

   856 		sms7BitToUnicodeDataRanges

   857 		},

   858 		{

   859 		ARRAY_LENGTH(unicodeToSms7BitDataRanges),

   860 		unicodeToSms7BitDataRanges

   861 		},

   862 	NULL,

   863 	NULL

   864 	};

   866 GLREF_C void IsCharacterSetSMS7Bit(TInt& aConfidenceLevel, const TDesC8& aSample)

   867 	{

   868 	_LIT8(KHz1Esc,"\x7e\x7e");

   869 	_LIT8(KHz2Esc,"\x7e\x7b");

   870 	_LIT8(KHz3Esc,"\x7e\x7d");

   871 	_LIT8(KLit8EscapeSequenceForJisRoman, "\x1b\x28\x4a");

   872 	_LIT8(KLit8EscapeSequenceForAscii, "\x1b\x28\x42");

   873 //	_LIT8(KLit8EscapeSequenceForHalfWidthKatakana, "\x1b\x28\x49");

   874 	_LIT8(KLit8EscapeSequenceForJisC6226_1978, "\x1b\x24\x40");

   875 	_LIT8(KLit8EscapeSequenceForJisX0208_1983, "\x1b\x24\x42");

   876 	_LIT8(KLit8EscapeSequenceForJisX0208_199x, "\x1b\x26\x40\x1b\x24\x42");

   877 	_LIT8(KLit8EscapeSequenceForJisX0212_1990, "\x1b\x24\x28\x44");

   879 	TInt hz1Result=0;

   880 	TInt hz2Result=0;

   881 	TInt hz3Result=0;

   882 	TInt JisRomanResult = 0;

   883 	TInt AsciiResult = 0;

   884 	TInt JisCResult = 0;

   885 	TInt JisX0208Result = 0;

   886 	TInt JisX0212Result = 0;

   887 	TInt JisX0208XResult = 0;

   888 	TInt currency = 0;

   889 	TInt email = 0;

   891 	TInt sampleLength = aSample.Length();

   892 	aConfidenceLevel =70;

   893 	if (sampleLength == 0)

   894 		return;

   896 	TInt HzEscSequences = 0;

   897 	TInt JISEscSequences = 0;

   899 	for (TInt i = 0; i < sampleLength; ++i)

   900 		{

   902 		if (i > hz1Result)

   903 			{

   904 			hz1Result=(aSample.Right(sampleLength-i)).Find(KHz1Esc);

   905 			if (hz1Result!=KErrNotFound) //aConfidenceLevel-=2;

   906 				HzEscSequences += 5;

   907 			}

   909 		if (i > hz2Result)

   910 			{

   911 			hz2Result=(aSample.Right(sampleLength-i)).Find(KHz2Esc);

   912 			if (hz2Result!=KErrNotFound) //aConfidenceLevel-=2;

   913 				HzEscSequences += 5;

   914 			}

   916 		if (i > hz3Result)

   917 			{

   918 			hz3Result=(aSample.Right(sampleLength-i)).Find(KHz3Esc);

   919 			if (hz3Result!=KErrNotFound) //aConfidenceLevel-=2;

   920 				HzEscSequences += 5;

   921 			}

   923 		if (i > JisRomanResult)

   924 			{

   925 			JisRomanResult = (aSample.Right(sampleLength-i)).Find(KLit8EscapeSequenceForJisRoman);

   926 			if (JisRomanResult!=KErrNotFound)

   927 				JISEscSequences +=10;

   928 			}

   930 		if (i > AsciiResult)

   931 			{

   932 			AsciiResult = (aSample.Right(sampleLength-i)).Find(KLit8EscapeSequenceForAscii);

   933 			if (AsciiResult!=KErrNotFound)

   934 				JISEscSequences +=10;

   935 			}

   937 		if (i > JisCResult)

   938 			{

   939 			JisCResult = (aSample.Right(sampleLength-i)).Find(KLit8EscapeSequenceForJisC6226_1978);

   940 			if (JisCResult!=KErrNotFound)

   941 				JISEscSequences +=5;

   942 			}

   944 		if (i > JisX0208Result)

   945 			{

   946 			JisX0208Result = (aSample.Right(sampleLength-i)).Find(KLit8EscapeSequenceForJisX0208_1983);

   947 			if (JisX0208Result!=KErrNotFound)

   948 				JISEscSequences +=5;

   949 			}

   951 		if (i > JisX0212Result)

   952 			{

   953 			JisX0212Result = (aSample.Right(sampleLength-i)).Find(KLit8EscapeSequenceForJisX0212_1990);

   954 			if (JisX0212Result!=KErrNotFound)

   955 				JISEscSequences +=5;

   956 			}

   958 		if (i > JisX0208XResult)

   959 			{

   960 			JisX0208XResult = (aSample.Right(sampleLength-i)).Find(KLit8EscapeSequenceForJisX0208_199x);

   961 			if (JisX0208XResult!=KErrNotFound)

   962 				JISEscSequences +=5;

   963 			}

   965 		if ((aSample[i]&0x80)!=0x00)

   966 			{

   967 			aConfidenceLevel = 0;

   968 			break;

   969 			}

   970 		else

   971 			{

   972 			TInt increment1 = i+1;

   973 			if (aSample[i] == 0x1b)

   974 				{

   975 				static const TInt defaultExtensionTable[10] =

   976 					{0x0a, 0x14, 0x1b, 0x29, 0x2f, 0x3c, 0x3d, 0x3e, 0x40, 0x65};

   977 				if (increment1 >= sampleLength)

   978 					break;

   979 				for (TInt j =0; j < 10; j++)

   980 					{

   981 					if (aSample[increment1] == defaultExtensionTable[j])

   982 						{

   983 						aConfidenceLevel+=5;

   984 						break;

   985 						}

   986 					}

   987 				}

   988 			else if (currency<100 && (aSample[i]==0x01 || aSample[i]==0x02 || aSample[i]==0x03))

   989 				{

   990 				if (increment1 >= sampleLength)

   991 					break;

   992 				if ((aSample[increment1] >= 0x30) && (aSample[increment1] <= 0x39))

   993 					{

   994 					currency ? currency *= 3 : currency = 3;

   995 					}

   996 				}

   997 			else if (email<100 && aSample[i]==0x00) // @

   998 				{

   999 					TBool valid = ETrue;

  1000 					TInt pos = i-1;

  1001 					while(valid && pos>=0) // check before @

  1002 						{

  1003 							TInt8 act = aSample[pos];

  1004 							if(act==' ' || act=='\n' || act=='\t')

  1005 								break;

  1006 							if( !( ( act >= 'a' && act <= 'z' ) 	||

  1007 								   ( act >= 'A' && act <= 'Z' ) 	||

  1008 								   ( act >= '0' && act <= '9' ) 	||

  1009 								   act == '.' 						||

  1010 								   act == '-' 						||

  1011 								   act == /*'_'*/0x11 ) )

  1012 								{

  1013 								valid = EFalse;

  1014 								}

  1015 							pos--;

  1016 						}

  1018 					TInt space = aSample.Mid(i+1).Find((const unsigned char*)" ",1);

  1019 					TInt lf = aSample.Mid(i+1).Find((const unsigned char*)"\n",1);

  1020 					TInt end = aSample.Length() - (i+1);

  1021 					if(space != KErrNotFound)

  1022 						{

  1023 						end = space;

  1024 						}

  1025 					else if(lf != KErrNotFound)

  1026 						{

  1027 						end = lf;

  1028 						}

  1029 					TPtrC8 string = aSample.Mid(i+1,end);

  1030 					TInt commat = string.Find((const unsigned char*)"\x0",1);

  1031 					TInt period = string.Find((const unsigned char*)".",1);

  1032 					if(commat!=KErrNotFound || period==KErrNotFound)

  1033 						{

  1034 						valid = EFalse;

  1035 						}

  1036 					if(valid)

  1037 						{

  1038 						email ? email *= 5 : email = 5;

  1039 						}

  1040 					else

  1041 						{

  1042 						aConfidenceLevel += 5; // not valid address but still valid '@'

  1043 						}

  1044 				}

  1045 			//next test the occurance of values 00-2f (except 0a & 0d) next to another character

  1046 			// These are the ASCII control codes and therefore low chances of these values occuring

  1047 			// next to the ASCII value

  1048 			else if (aSample[i] < 0x20 && aSample[i] != '\r' && aSample[i] != '\n')

  1049 				{

  1050 				if (increment1 >= sampleLength)

  1051 					break;

  1052 				if (((aSample[increment1] >= 0x20) && (aSample[increment1] <= 0x5a)) ||

  1053 					((aSample[increment1] >= 0x61) && (aSample[increment1] <= 0x7a)))

  1054 					{

  1055 					aConfidenceLevel+=2;

  1056 					}

  1057 				}

  1058 			// other possible tests to include... Greek matching...

  1059 			}

  1060 		} // for loop

  1061 	aConfidenceLevel += currency + email;

  1062 	aConfidenceLevel = aConfidenceLevel - ((HzEscSequences*100)/sampleLength);

  1063 	aConfidenceLevel = aConfidenceLevel - ((JISEscSequences*100)/sampleLength);

  1064 	aConfidenceLevel =(aConfidenceLevel >0)? ((aConfidenceLevel > 100)? 100: aConfidenceLevel): 0;

  1065 	}

author	sl
	Tue, 10 Jun 2014 14:32:02 +0200
changeset 1	260cb5ec6c19
permissions	-rw-r--r--