| 1 | // © 2018 and later: Unicode, Inc. and others. | 
|---|
| 2 | // License & terms of use: http://www.unicode.org/copyright.html | 
|---|
| 3 |  | 
|---|
| 4 | #include "unicode/utypes.h" | 
|---|
| 5 |  | 
|---|
| 6 | #if !UCONFIG_NO_FORMATTING | 
|---|
| 7 |  | 
|---|
| 8 | // Allow implicit conversion from char16_t* to UnicodeString for this file: | 
|---|
| 9 | // Helpful in toString methods and elsewhere. | 
|---|
| 10 | #define UNISTR_FROM_STRING_EXPLICIT | 
|---|
| 11 |  | 
|---|
| 12 | #include "numparse_types.h" | 
|---|
| 13 | #include "numparse_compositions.h" | 
|---|
| 14 | #include "string_segment.h" | 
|---|
| 15 | #include "unicode/uniset.h" | 
|---|
| 16 |  | 
|---|
| 17 | using namespace icu; | 
|---|
| 18 | using namespace icu::numparse; | 
|---|
| 19 | using namespace icu::numparse::impl; | 
|---|
| 20 |  | 
|---|
| 21 |  | 
|---|
| 22 | bool SeriesMatcher::match(StringSegment& segment, ParsedNumber& result, UErrorCode& status) const { | 
|---|
| 23 | ParsedNumber backup(result); | 
|---|
| 24 |  | 
|---|
| 25 | int32_t initialOffset = segment.getOffset(); | 
|---|
| 26 | bool maybeMore = true; | 
|---|
| 27 | for (auto* it = begin(); it < end();) { | 
|---|
| 28 | const NumberParseMatcher* matcher = *it; | 
|---|
| 29 | int matcherOffset = segment.getOffset(); | 
|---|
| 30 | if (segment.length() != 0) { | 
|---|
| 31 | maybeMore = matcher->match(segment, result, status); | 
|---|
| 32 | } else { | 
|---|
| 33 | // Nothing for this matcher to match; ask for more. | 
|---|
| 34 | maybeMore = true; | 
|---|
| 35 | } | 
|---|
| 36 |  | 
|---|
| 37 | bool success = (segment.getOffset() != matcherOffset); | 
|---|
| 38 | bool isFlexible = matcher->isFlexible(); | 
|---|
| 39 | if (success && isFlexible) { | 
|---|
| 40 | // Match succeeded, and this is a flexible matcher. Re-run it. | 
|---|
| 41 | } else if (success) { | 
|---|
| 42 | // Match succeeded, and this is NOT a flexible matcher. Proceed to the next matcher. | 
|---|
| 43 | it++; | 
|---|
| 44 | // Small hack: if there is another matcher coming, do not accept trailing weak chars. | 
|---|
| 45 | // Needed for proper handling of currency spacing. | 
|---|
| 46 | if (it < end() && segment.getOffset() != result.charEnd && result.charEnd > matcherOffset) { | 
|---|
| 47 | segment.setOffset(result.charEnd); | 
|---|
| 48 | } | 
|---|
| 49 | } else if (isFlexible) { | 
|---|
| 50 | // Match failed, and this is a flexible matcher. Try again with the next matcher. | 
|---|
| 51 | it++; | 
|---|
| 52 | } else { | 
|---|
| 53 | // Match failed, and this is NOT a flexible matcher. Exit. | 
|---|
| 54 | segment.setOffset(initialOffset); | 
|---|
| 55 | result = backup; | 
|---|
| 56 | return maybeMore; | 
|---|
| 57 | } | 
|---|
| 58 | } | 
|---|
| 59 |  | 
|---|
| 60 | // All matchers in the series succeeded. | 
|---|
| 61 | return maybeMore; | 
|---|
| 62 | } | 
|---|
| 63 |  | 
|---|
| 64 | bool SeriesMatcher::smokeTest(const StringSegment& segment) const { | 
|---|
| 65 | // NOTE: The range-based for loop calls the virtual begin() and end() methods. | 
|---|
| 66 | // NOTE: We only want the first element. Use the for loop for boundary checking. | 
|---|
| 67 | for (auto& matcher : *this) { | 
|---|
| 68 | // SeriesMatchers are never allowed to start with a Flexible matcher. | 
|---|
| 69 | U_ASSERT(!matcher->isFlexible()); | 
|---|
| 70 | return matcher->smokeTest(segment); | 
|---|
| 71 | } | 
|---|
| 72 | return false; | 
|---|
| 73 | } | 
|---|
| 74 |  | 
|---|
| 75 | void SeriesMatcher::postProcess(ParsedNumber& result) const { | 
|---|
| 76 | // NOTE: The range-based for loop calls the virtual begin() and end() methods. | 
|---|
| 77 | for (auto* matcher : *this) { | 
|---|
| 78 | matcher->postProcess(result); | 
|---|
| 79 | } | 
|---|
| 80 | } | 
|---|
| 81 |  | 
|---|
| 82 |  | 
|---|
| 83 | ArraySeriesMatcher::ArraySeriesMatcher() | 
|---|
| 84 | : fMatchersLen(0) { | 
|---|
| 85 | } | 
|---|
| 86 |  | 
|---|
| 87 | ArraySeriesMatcher::ArraySeriesMatcher(MatcherArray& matchers, int32_t matchersLen) | 
|---|
| 88 | : fMatchers(std::move(matchers)), fMatchersLen(matchersLen) { | 
|---|
| 89 | } | 
|---|
| 90 |  | 
|---|
| 91 | int32_t ArraySeriesMatcher::length() const { | 
|---|
| 92 | return fMatchersLen; | 
|---|
| 93 | } | 
|---|
| 94 |  | 
|---|
| 95 | const NumberParseMatcher* const* ArraySeriesMatcher::begin() const { | 
|---|
| 96 | return fMatchers.getAlias(); | 
|---|
| 97 | } | 
|---|
| 98 |  | 
|---|
| 99 | const NumberParseMatcher* const* ArraySeriesMatcher::end() const { | 
|---|
| 100 | return fMatchers.getAlias() + fMatchersLen; | 
|---|
| 101 | } | 
|---|
| 102 |  | 
|---|
| 103 | UnicodeString ArraySeriesMatcher::toString() const { | 
|---|
| 104 | return u "<ArraySeries>"; | 
|---|
| 105 | } | 
|---|
| 106 |  | 
|---|
| 107 |  | 
|---|
| 108 | #endif /* #if !UCONFIG_NO_FORMATTING */ | 
|---|
| 109 |  | 
|---|