73 #include <f32file.h> |
73 #include <f32file.h> |
74 #include <collate.h> |
74 #include <collate.h> |
75 #include "collateimp.h" |
75 #include "collateimp.h" |
76 #include "CompareImp.h" |
76 #include "CompareImp.h" |
77 #include "u32std.h" |
77 #include "u32std.h" |
78 |
|
79 static inline TBool IsSupplementary(TUint aChar) |
|
80 /** |
|
81 @param aChar The 32-bit code point value of a Unicode character. |
|
82 |
|
83 @return True, if aChar is supplementary character; false, otherwise. |
|
84 */ |
|
85 { |
|
86 return (aChar > 0xFFFF); |
|
87 } |
|
88 |
|
89 static inline TText16 GetHighSurrogate(TUint aChar) |
|
90 /** |
|
91 Retrieve the high surrogate of a supplementary character. |
|
92 |
|
93 @param aChar The 32-bit code point value of a Unicode character. |
|
94 |
|
95 @return High surrogate of aChar, if aChar is a supplementary character; |
|
96 aChar itself, if aChar is not a supplementary character. |
|
97 |
|
98 @see TChar::GetLowSurrogate |
|
99 */ |
|
100 { |
|
101 return STATIC_CAST(TText16, 0xD7C0 + (aChar >> 10)); |
|
102 } |
|
103 |
|
104 static inline TText16 GetLowSurrogate(TUint aChar) |
|
105 /** |
|
106 Retrieve the low surrogate of a supplementary character. |
|
107 |
|
108 @param aChar The 32-bit code point value of a Unicode character. |
|
109 |
|
110 @return Low surrogate of aChar, if aChar is a supplementary character; |
|
111 zero, if aChar is not a supplementary character. |
|
112 |
|
113 @see TChar::GetHighSurrogate |
|
114 */ |
|
115 { |
|
116 return STATIC_CAST(TText16, 0xDC00 | (aChar & 0x3FF)); |
|
117 } |
|
118 |
|
119 |
78 |
120 ///***************** copied from locale euser source code *********************** |
79 ///***************** copied from locale euser source code *********************** |
121 static const TCollationMethod TheCollationMethod[] = |
80 static const TCollationMethod TheCollationMethod[] = |
122 { |
81 { |
123 { |
82 { |
1635 for(len=0;!lex.Eos();++len) |
1594 for(len=0;!lex.Eos();++len) |
1636 { |
1595 { |
1637 TUint32 code; |
1596 TUint32 code; |
1638 User::LeaveIfError(lex.Val(code, EHex)); |
1597 User::LeaveIfError(lex.Val(code, EHex)); |
1639 lex.Assign(lex.NextToken()); |
1598 lex.Assign(lex.NextToken()); |
1640 if (!IsSupplementary(code)) |
1599 if (!TChar::IsSupplementary(code)) |
1641 { |
1600 { |
1642 aStr[1+len] = (TUint16)code; |
1601 aStr[1+len] = (TUint16)code; |
1643 } |
1602 } |
1644 else |
1603 else |
1645 { |
1604 { |
1646 aStr[1+len] = GetHighSurrogate(code); |
1605 aStr[1+len] = TChar::GetHighSurrogate(code); |
1647 ++len; |
1606 ++len; |
1648 aStr[1+len] = GetLowSurrogate(code); |
1607 aStr[1+len] = TChar::GetLowSurrogate(code); |
1649 } |
1608 } |
1650 } |
1609 } |
1651 __ASSERT_ALWAYS(len > 0, User::Invariant()); |
1610 __ASSERT_ALWAYS(len > 0, User::Invariant()); |
1652 aStr.SetLength(1 + len); |
1611 aStr.SetLength(1 + len); |
1653 } |
1612 } |
1706 //"LATIN CAPITAL LETTER I WITH DOT ABOVE" - the searching algorithm does not work with it. |
1665 //"LATIN CAPITAL LETTER I WITH DOT ABOVE" - the searching algorithm does not work with it. |
1707 if(chCode == (TUint32)0x0130) |
1666 if(chCode == (TUint32)0x0130) |
1708 { |
1667 { |
1709 continue; |
1668 continue; |
1710 } |
1669 } |
1711 if (!IsSupplementary(chCode)) |
1670 if (!TChar::IsSupplementary(chCode)) |
1712 { |
1671 { |
1713 candidate[KChPos] = (TUint16)chCode; |
1672 candidate[KChPos] = (TUint16)chCode; |
1714 } |
1673 } |
1715 else |
1674 else |
1716 { |
1675 { |
1717 candidate[KChPos] = GetHighSurrogate(chCode); |
1676 candidate[KChPos] = TChar::GetHighSurrogate(chCode); |
1718 candidate[KChPos+1] = GetLowSurrogate(chCode); |
1677 candidate[KChPos+1] = TChar::GetLowSurrogate(chCode); |
1719 } |
1678 } |
1720 //"Character decomposition mapping" is the 5th field, starting from 0. |
1679 //"Character decomposition mapping" is the 5th field, starting from 0. |
1721 TPtrC8 decomp(GetUnicodeDataField(stmt, 5)); |
1680 TPtrC8 decomp(GetUnicodeDataField(stmt, 5)); |
1722 if(decomp.Length() > 1 && decomp[0] != '<') |
1681 if(decomp.Length() > 1 && decomp[0] != '<') |
1723 { |
1682 { |
1760 //"LATIN CAPITAL LETTER I WITH DOT ABOVE" - the searching algorithm does not work with it. |
1719 //"LATIN CAPITAL LETTER I WITH DOT ABOVE" - the searching algorithm does not work with it. |
1761 if(chCode == (TUint32)0x0130) |
1720 if(chCode == (TUint32)0x0130) |
1762 { |
1721 { |
1763 continue; |
1722 continue; |
1764 } |
1723 } |
1765 if (!IsSupplementary(chCode)) |
1724 if (!TChar::IsSupplementary(chCode)) |
1766 { |
1725 { |
1767 candidate[KChPos] = (TUint16)chCode; |
1726 candidate[KChPos] = (TUint16)chCode; |
1768 candidate.SetLength(2); |
1727 candidate.SetLength(2); |
1769 } |
1728 } |
1770 else |
1729 else |
1771 { |
1730 { |
1772 candidate[KChPos] = GetHighSurrogate(chCode); |
1731 candidate[KChPos] = TChar::GetHighSurrogate(chCode); |
1773 candidate.SetLength(3); |
1732 candidate.SetLength(3); |
1774 candidate[KChPos+1] = GetLowSurrogate(chCode); |
1733 candidate[KChPos+1] = TChar::GetLowSurrogate(chCode); |
1775 } |
1734 } |
1776 //"Character decomposition mapping" is the 5th field, starting from 0. |
1735 //"Character decomposition mapping" is the 5th field, starting from 0. |
1777 TPtrC8 decomp(GetUnicodeDataField(stmt, 5)); |
1736 TPtrC8 decomp(GetUnicodeDataField(stmt, 5)); |
1778 if(decomp.Length() > 1 && decomp[0] != '<') |
1737 if(decomp.Length() > 1 && decomp[0] != '<') |
1779 { |
1738 { |
2552 test( 0 == buf01c5.MatchF( K01c6() ) ); |
2511 test( 0 == buf01c5.MatchF( K01c6() ) ); |
2553 |
2512 |
2554 test.Next(_L("MatchSurrogate")); |
2513 test.Next(_L("MatchSurrogate")); |
2555 for (ii=0;ii<KTestsSurrogate;++ii) |
2514 for (ii=0;ii<KTestsSurrogate;++ii) |
2556 { |
2515 { |
2557 TInt r=TPtrC16(TestsSurrogate[ii].iLeft).MatchF(TPtrC16(TestsSurrogate[ii].iRight)); |
2516 TInt r=TPtrC16(TestsSurrogate[ii].iLeft).Match2(TPtrC16(TestsSurrogate[ii].iRight)); |
|
2517 RDebug::Printf(" ii=%d, expect=%d, result=%d", ii, TestsSurrogate[ii].iResult, r); |
|
2518 test (r==TestsSurrogate[ii].iResult); |
|
2519 r=TPtrC16(TestsSurrogate[ii].iLeft).MatchF(TPtrC16(TestsSurrogate[ii].iRight)); |
2558 test (r==TestsSurrogate[ii].iResult); |
2520 test (r==TestsSurrogate[ii].iResult); |
2559 } |
2521 } |
2560 |
2522 |
2561 _LIT( KD800, "\xd800" ); |
2523 _LIT( KD800, "\xd800" ); |
2562 _LIT( KQuestion, "?" ); |
2524 _LIT( KQuestion, "?" ); |
2563 _LIT( KDC00, "\xdc00" ); |
2525 _LIT( KDC00, "\xdc00" ); |
2564 _LIT( KDFFF, "\xdfff" ); |
2526 _LIT( KDFFF, "\xdfff" ); |
|
2527 test( KErrCorruptSurrogateFound == TPtrC16( KD800() ).Match2( TPtrC16( KQuestion() ) ) ); |
|
2528 test( KErrCorruptSurrogateFound == TPtrC16( KD800() ).Match2( TPtrC16( KD800() ) ) ); |
|
2529 test( KErrCorruptSurrogateFound == TPtrC16( KDC00() ).Match2( TPtrC16( KQuestion() ) ) ); |
|
2530 test( KErrCorruptSurrogateFound == TPtrC16( KDFFF() ).Match2( TPtrC16( KQuestion() ) ) ); |
2565 |
2531 |
2566 test( KErrNotFound == TPtrC16( KD800() ).MatchF( TPtrC16( KQuestion() ) ) ); |
2532 test( KErrNotFound == TPtrC16( KD800() ).MatchF( TPtrC16( KQuestion() ) ) ); |
2567 test( 0 == TPtrC16( KD800() ).MatchF( TPtrC16( KD800() ) ) ); |
2533 test( 0 == TPtrC16( KD800() ).MatchF( TPtrC16( KD800() ) ) ); |
2568 test( KErrNotFound == TPtrC16( KDC00() ).MatchF( TPtrC16( KQuestion() ) ) ); |
2534 test( KErrNotFound == TPtrC16( KDC00() ).MatchF( TPtrC16( KQuestion() ) ) ); |
2569 test( KErrNotFound == TPtrC16( KDFFF() ).MatchF( TPtrC16( KQuestion() ) ) ); |
2535 test( KErrNotFound == TPtrC16( KDFFF() ).MatchF( TPtrC16( KQuestion() ) ) ); |