icuSources/i18n/numparse_affixes.h

   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 #ifndef __NUMPARSE_AFFIXES_H__
   8 #define __NUMPARSE_AFFIXES_H__
   9
  10 #include "numparse_types.h"
  11 #include "numparse_symbols.h"
  12 #include "numparse_currency.h"
  13 #include "number_affixutils.h"
  14 #include "number_currencysymbols.h"
  15
  16 #include <array>
  17
  18 U_NAMESPACE_BEGIN
  19 namespace numparse {
  20 namespace impl {
  21
  22 // Forward-declaration of implementation classes for friending
  23 class AffixPatternMatcherBuilder;
  24 class AffixPatternMatcher;
  25
  26 using ::icu::number::impl::AffixPatternProvider;
  27 using ::icu::number::impl::TokenConsumer;
  28 using ::icu::number::impl::CurrencySymbols;
  29
  30
  31 class CodePointMatcher : public NumberParseMatcher, public UMemory {
  32   public:
  33     CodePointMatcher() = default;  // WARNING: Leaves the object in an unusable state
  34
  35     CodePointMatcher(UChar32 cp);
  36
  37     bool match(StringSegment& segment, ParsedNumber& result, UErrorCode& status) const override;
  38
  39     bool smokeTest(const StringSegment& segment) const override;
  40
  41     UnicodeString toString() const override;
  42
  43   private:
  44     UChar32 fCp;
  45 };
  46
  47 } // namespace impl
  48 } // namespace numparse
  49
  50 // Export a explicit template instantiations of MaybeStackArray and CompactUnicodeString.
  51 // When building DLLs for Windows this is required even though no direct access leaks out of the i18n library.
  52 // (See digitlst.h, pluralaffix.h, datefmt.h, and others for similar examples.)
  53 // Note: These need to be outside of the impl::numparse namespace, or Clang will generate a compile error.
  54 #if U_PF_WINDOWS <= U_PLATFORM && U_PLATFORM <= U_PF_CYGWIN
  55 template class U_I18N_API MaybeStackArray<UChar, 4>;
  56 template class U_I18N_API MaybeStackArray<numparse::impl::CodePointMatcher*, 3>;
  57 template class U_I18N_API numparse::impl::CompactUnicodeString<4>;
  58 #endif
  59
  60 namespace numparse {
  61 namespace impl {
  62
  63 /**
  64  * A warehouse to retain ownership of CodePointMatchers.
  65  */
  66 // Exported as U_I18N_API for tests
  67 class U_I18N_API CodePointMatcherWarehouse : public UMemory {
  68   private:
  69     static constexpr int32_t CODE_POINT_STACK_CAPACITY = 5; // Number of entries directly on the stack
  70     static constexpr int32_t CODE_POINT_BATCH_SIZE = 10; // Number of entries per heap allocation
  71
  72   public:
  73     CodePointMatcherWarehouse();
  74
  75     // A custom destructor is needed to free the memory from MaybeStackArray.
  76     // A custom move constructor and move assignment seem to be needed because of the custom destructor.
  77
  78     ~CodePointMatcherWarehouse();
  79
  80     CodePointMatcherWarehouse(CodePointMatcherWarehouse&& src) U_NOEXCEPT;
  81
  82     CodePointMatcherWarehouse& operator=(CodePointMatcherWarehouse&& src) U_NOEXCEPT;
  83
  84     NumberParseMatcher& nextCodePointMatcher(UChar32 cp);
  85
  86   private:
  87     std::array<CodePointMatcher, CODE_POINT_STACK_CAPACITY> codePoints; // By value
  88     MaybeStackArray<CodePointMatcher*, 3> codePointsOverflow; // On heap in "batches"
  89     int32_t codePointCount; // Total for both the ones by value and on heap
  90     int32_t codePointNumBatches; // Number of batches in codePointsOverflow
  91 };
  92
  93
  94 struct AffixTokenMatcherSetupData {
  95     const CurrencySymbols& currencySymbols;
  96     const DecimalFormatSymbols& dfs;
  97     IgnorablesMatcher& ignorables;
  98     const Locale& locale;
  99     parse_flags_t parseFlags;
 100 };
 101
 102
 103 /**
 104  * Small helper class that generates matchers for individual tokens for AffixPatternMatcher.
 105  *
 106  * In Java, this is called AffixTokenMatcherFactory (a "factory"). However, in C++, it is called a
 107  * "warehouse", because in addition to generating the matchers, it also retains ownership of them. The
 108  * warehouse must stay in scope for the whole lifespan of the AffixPatternMatcher that uses matchers from
 109  * the warehouse.
 110  *
 111  * @author sffc
 112  */
 113 // Exported as U_I18N_API for tests
 114 class U_I18N_API AffixTokenMatcherWarehouse : public UMemory {
 115   public:
 116     AffixTokenMatcherWarehouse() = default;  // WARNING: Leaves the object in an unusable state
 117
 118     AffixTokenMatcherWarehouse(const AffixTokenMatcherSetupData* setupData);
 119
 120     NumberParseMatcher& minusSign();
 121
 122     NumberParseMatcher& plusSign();
 123
 124     NumberParseMatcher& percent();
 125
 126     NumberParseMatcher& permille();
 127
 128     NumberParseMatcher& currency(UErrorCode& status);
 129
 130     IgnorablesMatcher& ignorables();
 131
 132     NumberParseMatcher& nextCodePointMatcher(UChar32 cp);
 133
 134   private:
 135     // NOTE: The following field may be unsafe to access after construction is done!
 136     const AffixTokenMatcherSetupData* fSetupData;
 137
 138     // NOTE: These are default-constructed and should not be used until initialized.
 139     MinusSignMatcher fMinusSign;
 140     PlusSignMatcher fPlusSign;
 141     PercentMatcher fPercent;
 142     PermilleMatcher fPermille;
 143     CombinedCurrencyMatcher fCurrency;
 144
 145     // Use a child class for code point matchers, since it requires non-default operators.
 146     CodePointMatcherWarehouse fCodePoints;
 147
 148     friend class AffixPatternMatcherBuilder;
 149     friend class AffixPatternMatcher;
 150 };
 151
 152
 153 class AffixPatternMatcherBuilder : public TokenConsumer, public MutableMatcherCollection {
 154   public:
 155     AffixPatternMatcherBuilder(const UnicodeString& pattern, AffixTokenMatcherWarehouse& warehouse,
 156                                IgnorablesMatcher* ignorables);
 157
 158     void consumeToken(::icu::number::impl::AffixPatternType type, UChar32 cp, UErrorCode& status) override;
 159
 160     /** NOTE: You can build only once! */
 161     AffixPatternMatcher build();
 162
 163   private:
 164     ArraySeriesMatcher::MatcherArray fMatchers;
 165     int32_t fMatchersLen;
 166     int32_t fLastTypeOrCp;
 167
 168     const UnicodeString& fPattern;
 169     AffixTokenMatcherWarehouse& fWarehouse;
 170     IgnorablesMatcher* fIgnorables;
 171
 172     void addMatcher(NumberParseMatcher& matcher) override;
 173 };
 174
 175
 176 // Exported as U_I18N_API for tests
 177 class U_I18N_API AffixPatternMatcher : public ArraySeriesMatcher {
 178   public:
 179     AffixPatternMatcher() = default;  // WARNING: Leaves the object in an unusable state
 180
 181     static AffixPatternMatcher fromAffixPattern(const UnicodeString& affixPattern,
 182                                                 AffixTokenMatcherWarehouse& warehouse,
 183                                                 parse_flags_t parseFlags, bool* success,
 184                                                 UErrorCode& status);
 185
 186     UnicodeString getPattern() const;
 187
 188     bool operator==(const AffixPatternMatcher& other) const;
 189
 190   private:
 191     CompactUnicodeString<4> fPattern;
 192
 193     AffixPatternMatcher(MatcherArray& matchers, int32_t matchersLen, const UnicodeString& pattern);
 194
 195     friend class AffixPatternMatcherBuilder;
 196 };
 197
 198
 199 class AffixMatcher : public NumberParseMatcher, public UMemory {
 200   public:
 201     AffixMatcher() = default;  // WARNING: Leaves the object in an unusable state
 202
 203     AffixMatcher(AffixPatternMatcher* prefix, AffixPatternMatcher* suffix, result_flags_t flags);
 204
 205     bool match(StringSegment& segment, ParsedNumber& result, UErrorCode& status) const override;
 206
 207     void postProcess(ParsedNumber& result) const override;
 208
 209     bool smokeTest(const StringSegment& segment) const override;
 210
 211     int8_t compareTo(const AffixMatcher& rhs) const;
 212
 213     UnicodeString toString() const override;
 214
 215   private:
 216     AffixPatternMatcher* fPrefix;
 217     AffixPatternMatcher* fSuffix;
 218     result_flags_t fFlags;
 219 };
 220
 221
 222 /**
 223  * A C++-only class to retain ownership of the AffixMatchers needed for parsing.
 224  */
 225 class AffixMatcherWarehouse {
 226   public:
 227     AffixMatcherWarehouse() = default;  // WARNING: Leaves the object in an unusable state
 228
 229     AffixMatcherWarehouse(AffixTokenMatcherWarehouse* tokenWarehouse);
 230
 231     void createAffixMatchers(const AffixPatternProvider& patternInfo, MutableMatcherCollection& output,
 232                              const IgnorablesMatcher& ignorables, parse_flags_t parseFlags,
 233                              UErrorCode& status);
 234
 235   private:
 236     // 9 is the limit: positive, zero, and negative, each with prefix, suffix, and prefix+suffix
 237     AffixMatcher fAffixMatchers[9];
 238     // 6 is the limit: positive, zero, and negative, a prefix and a suffix for each
 239     AffixPatternMatcher fAffixPatternMatchers[6];
 240     // Reference to the warehouse for tokens used by the AffixPatternMatchers
 241     AffixTokenMatcherWarehouse* fTokenWarehouse;
 242
 243     friend class AffixMatcher;
 244
 245     static bool isInteresting(const AffixPatternProvider& patternInfo, const IgnorablesMatcher& ignorables,
 246                               parse_flags_t parseFlags, UErrorCode& status);
 247 };
 248
 249
 250 } // namespace impl
 251 } // namespace numparse
 252 U_NAMESPACE_END
 253
 254 #endif //__NUMPARSE_AFFIXES_H__
 255 #endif /* #if !UCONFIG_NO_FORMATTING */