icuSources/i18n/numparse_compositions.cpp

   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         bool startCurrencyIsEmpty = (result.currencyCode[0]==0); // Apple fix for <rdar://problem/46915356>
  30         int matcherOffset = segment.getOffset();
  31         if (segment.length() != 0
  32                 || (startCurrencyIsEmpty && result.seenNumber())) { // Apple <rdar://problem/51938595>
  33             maybeMore = matcher->match(segment, result, status);
  34         } else {
  35             // Nothing for this matcher to match; ask for more.
  36             maybeMore = true;
  37         }
  38
  39         bool addedCurrency = (startCurrencyIsEmpty && result.currencyCode[0]!=0); // Apple <rdar://problem/51938595>
  40         bool success = ((segment.getOffset() != matcherOffset) || addedCurrency); // Apple fix for <rdar://problem/46915356>
  41         bool isFlexible = matcher->isFlexible();
  42         if (success && isFlexible) {
  43             // Match succeeded, and this is a flexible matcher. Re-run it.
  44         } else if (success) {
  45             // Match succeeded, and this is NOT a flexible matcher. Proceed to the next matcher.
  46             it++;
  47             // Small hack: if there is another matcher coming, do not accept trailing weak chars.
  48             // Needed for proper handling of currency spacing.
  49             if (it < end() && segment.getOffset() != result.charEnd && (result.charEnd > matcherOffset || addedCurrency)) { // Apple <rdar://problem/51938595>
  50                 segment.setOffset(result.charEnd);
  51             }
  52         } else if (isFlexible) {
  53             // Match failed, and this is a flexible matcher. Try again with the next matcher.
  54             it++;
  55         } else {
  56             // Match failed, and this is NOT a flexible matcher. Exit.
  57             segment.setOffset(initialOffset);
  58             result = backup;
  59             return maybeMore;
  60         }
  61     }
  62
  63     // All matchers in the series succeeded.
  64     return maybeMore;
  65 }
  66
  67 bool SeriesMatcher::smokeTest(const StringSegment& segment) const {
  68     // NOTE: The range-based for loop calls the virtual begin() and end() methods.
  69     // NOTE: We only want the first element. Use the for loop for boundary checking.
  70     for (auto& matcher : *this) {
  71         // SeriesMatchers are never allowed to start with a Flexible matcher.
  72         U_ASSERT(!matcher->isFlexible());
  73         return matcher->smokeTest(segment);
  74     }
  75     return false;
  76 }
  77
  78 void SeriesMatcher::postProcess(ParsedNumber& result) const {
  79     // NOTE: The range-based for loop calls the virtual begin() and end() methods.
  80     for (auto* matcher : *this) {
  81         matcher->postProcess(result);
  82     }
  83 }
  84
  85
  86 ArraySeriesMatcher::ArraySeriesMatcher()
  87         : fMatchersLen(0) {
  88 }
  89
  90 ArraySeriesMatcher::ArraySeriesMatcher(MatcherArray& matchers, int32_t matchersLen)
  91         : fMatchers(std::move(matchers)), fMatchersLen(matchersLen) {
  92 }
  93
  94 int32_t ArraySeriesMatcher::length() const {
  95     return fMatchersLen;
  96 }
  97
  98 const NumberParseMatcher* const* ArraySeriesMatcher::begin() const {
  99     return fMatchers.getAlias();
 100 }
 101
 102 const NumberParseMatcher* const* ArraySeriesMatcher::end() const {
 103     return fMatchers.getAlias() + fMatchersLen;
 104 }
 105
 106 UnicodeString ArraySeriesMatcher::toString() const {
 107     return u"<ArraySeries>";
 108 }
 109
 110
 111 #endif /* #if !UCONFIG_NO_FORMATTING */