/src/icu/icu4c/source/i18n/regeximp.cpp
Line | Count | Source (jump to first uncovered line) |
1 | | // © 2016 and later: Unicode, Inc. and others. |
2 | | // License & terms of use: http://www.unicode.org/copyright.html |
3 | | // |
4 | | // Copyright (C) 2012 International Business Machines Corporation |
5 | | // and others. All rights reserved. |
6 | | // |
7 | | // file: regeximp.cpp |
8 | | // |
9 | | // ICU Regular Expressions, |
10 | | // miscellaneous implementation functions. |
11 | | // |
12 | | |
13 | | #include "unicode/utypes.h" |
14 | | |
15 | | #if !UCONFIG_NO_REGULAR_EXPRESSIONS |
16 | | #include "regeximp.h" |
17 | | #include "unicode/utf16.h" |
18 | | |
19 | | U_NAMESPACE_BEGIN |
20 | | |
21 | | CaseFoldingUTextIterator::CaseFoldingUTextIterator(UText &text) : |
22 | 0 | fUText(text), fFoldChars(nullptr), fFoldLength(0) { |
23 | 0 | } |
24 | | |
25 | 0 | CaseFoldingUTextIterator::~CaseFoldingUTextIterator() {} |
26 | | |
27 | 0 | UChar32 CaseFoldingUTextIterator::next() { |
28 | 0 | UChar32 foldedC; |
29 | 0 | UChar32 originalC; |
30 | 0 | if (fFoldChars == nullptr) { |
31 | | // We are not in a string folding of an earlier character. |
32 | | // Start handling the next char from the input UText. |
33 | 0 | originalC = UTEXT_NEXT32(&fUText); |
34 | 0 | if (originalC == U_SENTINEL) { |
35 | 0 | return originalC; |
36 | 0 | } |
37 | 0 | fFoldLength = ucase_toFullFolding(originalC, &fFoldChars, U_FOLD_CASE_DEFAULT); |
38 | 0 | if (fFoldLength >= UCASE_MAX_STRING_LENGTH || fFoldLength < 0) { |
39 | | // input code point folds to a single code point, possibly itself. |
40 | | // See comment in ucase.h for explanation of return values from ucase_toFullFoldings. |
41 | 0 | if (fFoldLength < 0) { |
42 | 0 | fFoldLength = ~fFoldLength; |
43 | 0 | } |
44 | 0 | foldedC = (UChar32)fFoldLength; |
45 | 0 | fFoldChars = nullptr; |
46 | 0 | return foldedC; |
47 | 0 | } |
48 | | // String foldings fall through here. |
49 | 0 | fFoldIndex = 0; |
50 | 0 | } |
51 | | |
52 | 0 | U16_NEXT(fFoldChars, fFoldIndex, fFoldLength, foldedC); |
53 | 0 | if (fFoldIndex >= fFoldLength) { |
54 | 0 | fFoldChars = nullptr; |
55 | 0 | } |
56 | 0 | return foldedC; |
57 | 0 | } |
58 | | |
59 | | |
60 | 0 | UBool CaseFoldingUTextIterator::inExpansion() { |
61 | 0 | return fFoldChars != nullptr; |
62 | 0 | } |
63 | | |
64 | | |
65 | | |
66 | | CaseFoldingUCharIterator::CaseFoldingUCharIterator(const char16_t *chars, int64_t start, int64_t limit) : |
67 | 0 | fChars(chars), fIndex(start), fLimit(limit), fFoldChars(nullptr), fFoldLength(0) { |
68 | 0 | } |
69 | | |
70 | | |
71 | 0 | CaseFoldingUCharIterator::~CaseFoldingUCharIterator() {} |
72 | | |
73 | | |
74 | 0 | UChar32 CaseFoldingUCharIterator::next() { |
75 | 0 | UChar32 foldedC; |
76 | 0 | UChar32 originalC; |
77 | 0 | if (fFoldChars == nullptr) { |
78 | | // We are not in a string folding of an earlier character. |
79 | | // Start handling the next char from the input UText. |
80 | 0 | if (fIndex >= fLimit) { |
81 | 0 | return U_SENTINEL; |
82 | 0 | } |
83 | 0 | U16_NEXT(fChars, fIndex, fLimit, originalC); |
84 | |
|
85 | 0 | fFoldLength = ucase_toFullFolding(originalC, &fFoldChars, U_FOLD_CASE_DEFAULT); |
86 | 0 | if (fFoldLength >= UCASE_MAX_STRING_LENGTH || fFoldLength < 0) { |
87 | | // input code point folds to a single code point, possibly itself. |
88 | | // See comment in ucase.h for explanation of return values from ucase_toFullFoldings. |
89 | 0 | if (fFoldLength < 0) { |
90 | 0 | fFoldLength = ~fFoldLength; |
91 | 0 | } |
92 | 0 | foldedC = (UChar32)fFoldLength; |
93 | 0 | fFoldChars = nullptr; |
94 | 0 | return foldedC; |
95 | 0 | } |
96 | | // String foldings fall through here. |
97 | 0 | fFoldIndex = 0; |
98 | 0 | } |
99 | | |
100 | 0 | U16_NEXT(fFoldChars, fFoldIndex, fFoldLength, foldedC); |
101 | 0 | if (fFoldIndex >= fFoldLength) { |
102 | 0 | fFoldChars = nullptr; |
103 | 0 | } |
104 | 0 | return foldedC; |
105 | 0 | } |
106 | | |
107 | | |
108 | 0 | UBool CaseFoldingUCharIterator::inExpansion() { |
109 | 0 | return fFoldChars != nullptr; |
110 | 0 | } |
111 | | |
112 | 0 | int64_t CaseFoldingUCharIterator::getIndex() { |
113 | 0 | return fIndex; |
114 | 0 | } |
115 | | |
116 | | |
117 | | U_NAMESPACE_END |
118 | | |
119 | | #endif |
120 | | |