intl/icu/source/i18n/regeximp.cpp@b8a032363ba2 (annotated)
intl/icu/source/i18n/regeximp.cpp
Thu, 22 Jan 2015 13:21:57 +0100
- author
- Michael Schloh von Bennewitz <michael@schloh.com>
- date
- Thu, 22 Jan 2015 13:21:57 +0100
- branch
- TOR_BUG_9701
- changeset 15
- b8a032363ba2
- permissions
- -rw-r--r--
Incorporate requested changes from Mozilla in review:
https://bugzilla.mozilla.org/show_bug.cgi?id=1123480#c6
| michael@0 | 1 | // |
| michael@0 | 2 | // Copyright (C) 2012 International Business Machines Corporation |
| michael@0 | 3 | // and others. All rights reserved. |
| michael@0 | 4 | // |
| michael@0 | 5 | // file: regeximp.cpp |
| michael@0 | 6 | // |
| michael@0 | 7 | // ICU Regular Expressions, |
| michael@0 | 8 | // miscellaneous implementation functions. |
| michael@0 | 9 | // |
| michael@0 | 10 | |
| michael@0 | 11 | #include "unicode/utypes.h" |
| michael@0 | 12 | |
| michael@0 | 13 | #if !UCONFIG_NO_REGULAR_EXPRESSIONS |
| michael@0 | 14 | #include "regeximp.h" |
| michael@0 | 15 | #include "unicode/utf16.h" |
| michael@0 | 16 | |
| michael@0 | 17 | U_NAMESPACE_BEGIN |
| michael@0 | 18 | |
| michael@0 | 19 | CaseFoldingUTextIterator::CaseFoldingUTextIterator(UText &text) : |
| michael@0 | 20 | fUText(text), fcsp(NULL), fFoldChars(NULL), fFoldLength(0) { |
| michael@0 | 21 | fcsp = ucase_getSingleton(); |
| michael@0 | 22 | } |
| michael@0 | 23 | |
| michael@0 | 24 | CaseFoldingUTextIterator::~CaseFoldingUTextIterator() {} |
| michael@0 | 25 | |
| michael@0 | 26 | UChar32 CaseFoldingUTextIterator::next() { |
| michael@0 | 27 | UChar32 foldedC; |
| michael@0 | 28 | UChar32 originalC; |
| michael@0 | 29 | if (fFoldChars == NULL) { |
| michael@0 | 30 | // We are not in a string folding of an earlier character. |
| michael@0 | 31 | // Start handling the next char from the input UText. |
| michael@0 | 32 | originalC = UTEXT_NEXT32(&fUText); |
| michael@0 | 33 | if (originalC == U_SENTINEL) { |
| michael@0 | 34 | return originalC; |
| michael@0 | 35 | } |
| michael@0 | 36 | fFoldLength = ucase_toFullFolding(fcsp, originalC, &fFoldChars, U_FOLD_CASE_DEFAULT); |
| michael@0 | 37 | if (fFoldLength >= UCASE_MAX_STRING_LENGTH || fFoldLength < 0) { |
| michael@0 | 38 | // input code point folds to a single code point, possibly itself. |
| michael@0 | 39 | // See comment in ucase.h for explanation of return values from ucase_toFullFoldings. |
| michael@0 | 40 | if (fFoldLength < 0) { |
| michael@0 | 41 | fFoldLength = ~fFoldLength; |
| michael@0 | 42 | } |
| michael@0 | 43 | foldedC = (UChar32)fFoldLength; |
| michael@0 | 44 | fFoldChars = NULL; |
| michael@0 | 45 | return foldedC; |
| michael@0 | 46 | } |
| michael@0 | 47 | // String foldings fall through here. |
| michael@0 | 48 | fFoldIndex = 0; |
| michael@0 | 49 | } |
| michael@0 | 50 | |
| michael@0 | 51 | U16_NEXT(fFoldChars, fFoldIndex, fFoldLength, foldedC); |
| michael@0 | 52 | if (fFoldIndex >= fFoldLength) { |
| michael@0 | 53 | fFoldChars = NULL; |
| michael@0 | 54 | } |
| michael@0 | 55 | return foldedC; |
| michael@0 | 56 | } |
| michael@0 | 57 | |
| michael@0 | 58 | |
| michael@0 | 59 | UBool CaseFoldingUTextIterator::inExpansion() { |
| michael@0 | 60 | return fFoldChars != NULL; |
| michael@0 | 61 | } |
| michael@0 | 62 | |
| michael@0 | 63 | |
| michael@0 | 64 | |
| michael@0 | 65 | CaseFoldingUCharIterator::CaseFoldingUCharIterator(const UChar *chars, int64_t start, int64_t limit) : |
| michael@0 | 66 | fChars(chars), fIndex(start), fLimit(limit), fcsp(NULL), fFoldChars(NULL), fFoldLength(0) { |
| michael@0 | 67 | fcsp = ucase_getSingleton(); |
| michael@0 | 68 | } |
| michael@0 | 69 | |
| michael@0 | 70 | |
| michael@0 | 71 | CaseFoldingUCharIterator::~CaseFoldingUCharIterator() {} |
| michael@0 | 72 | |
| michael@0 | 73 | |
| michael@0 | 74 | UChar32 CaseFoldingUCharIterator::next() { |
| michael@0 | 75 | UChar32 foldedC; |
| michael@0 | 76 | UChar32 originalC; |
| michael@0 | 77 | if (fFoldChars == NULL) { |
| michael@0 | 78 | // We are not in a string folding of an earlier character. |
| michael@0 | 79 | // Start handling the next char from the input UText. |
| michael@0 | 80 | if (fIndex >= fLimit) { |
| michael@0 | 81 | return U_SENTINEL; |
| michael@0 | 82 | } |
| michael@0 | 83 | U16_NEXT(fChars, fIndex, fLimit, originalC); |
| michael@0 | 84 | |
| michael@0 | 85 | fFoldLength = ucase_toFullFolding(fcsp, originalC, &fFoldChars, U_FOLD_CASE_DEFAULT); |
| michael@0 | 86 | if (fFoldLength >= UCASE_MAX_STRING_LENGTH || fFoldLength < 0) { |
| michael@0 | 87 | // input code point folds to a single code point, possibly itself. |
| michael@0 | 88 | // See comment in ucase.h for explanation of return values from ucase_toFullFoldings. |
| michael@0 | 89 | if (fFoldLength < 0) { |
| michael@0 | 90 | fFoldLength = ~fFoldLength; |
| michael@0 | 91 | } |
| michael@0 | 92 | foldedC = (UChar32)fFoldLength; |
| michael@0 | 93 | fFoldChars = NULL; |
| michael@0 | 94 | return foldedC; |
| michael@0 | 95 | } |
| michael@0 | 96 | // String foldings fall through here. |
| michael@0 | 97 | fFoldIndex = 0; |
| michael@0 | 98 | } |
| michael@0 | 99 | |
| michael@0 | 100 | U16_NEXT(fFoldChars, fFoldIndex, fFoldLength, foldedC); |
| michael@0 | 101 | if (fFoldIndex >= fFoldLength) { |
| michael@0 | 102 | fFoldChars = NULL; |
| michael@0 | 103 | } |
| michael@0 | 104 | return foldedC; |
| michael@0 | 105 | } |
| michael@0 | 106 | |
| michael@0 | 107 | |
| michael@0 | 108 | UBool CaseFoldingUCharIterator::inExpansion() { |
| michael@0 | 109 | return fFoldChars != NULL; |
| michael@0 | 110 | } |
| michael@0 | 111 | |
| michael@0 | 112 | int64_t CaseFoldingUCharIterator::getIndex() { |
| michael@0 | 113 | return fIndex; |
| michael@0 | 114 | } |
| michael@0 | 115 | |
| michael@0 | 116 | |
| michael@0 | 117 | U_NAMESPACE_END |
| michael@0 | 118 | |
| michael@0 | 119 | #endif |
| michael@0 | 120 |
