1 | // © 2017 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 __NUMBER_PATTERNSTRING_H__ |
8 | #define __NUMBER_PATTERNSTRING_H__ |
9 | |
10 | |
11 | #include <cstdint> |
12 | #include "unicode/unum.h" |
13 | #include "unicode/unistr.h" |
14 | #include "number_types.h" |
15 | #include "number_decimalquantity.h" |
16 | #include "number_decimfmtprops.h" |
17 | #include "number_affixutils.h" |
18 | |
19 | U_NAMESPACE_BEGIN namespace number { |
20 | namespace impl { |
21 | |
22 | // Forward declaration |
23 | class PatternParser; |
24 | |
25 | // Exported as U_I18N_API because it is a public member field of exported ParsedSubpatternInfo |
26 | struct U_I18N_API Endpoints { |
27 | int32_t start = 0; |
28 | int32_t end = 0; |
29 | }; |
30 | |
31 | // Exported as U_I18N_API because it is a public member field of exported ParsedPatternInfo |
32 | struct U_I18N_API ParsedSubpatternInfo { |
33 | uint64_t groupingSizes = 0x0000ffffffff0000L; |
34 | int32_t integerLeadingHashSigns = 0; |
35 | int32_t integerTrailingHashSigns = 0; |
36 | int32_t integerNumerals = 0; |
37 | int32_t integerAtSigns = 0; |
38 | int32_t integerTotal = 0; // for convenience |
39 | int32_t fractionNumerals = 0; |
40 | int32_t fractionHashSigns = 0; |
41 | int32_t fractionTotal = 0; // for convenience |
42 | bool hasDecimal = false; |
43 | int32_t widthExceptAffixes = 0; |
44 | // Note: NullableValue causes issues here with std::move. |
45 | bool hasPadding = false; |
46 | UNumberFormatPadPosition paddingLocation = UNUM_PAD_BEFORE_PREFIX; |
47 | DecimalQuantity rounding; |
48 | bool exponentHasPlusSign = false; |
49 | int32_t exponentZeros = 0; |
50 | bool hasPercentSign = false; |
51 | bool hasPerMilleSign = false; |
52 | bool hasCurrencySign = false; |
53 | bool hasMinusSign = false; |
54 | bool hasPlusSign = false; |
55 | |
56 | Endpoints prefixEndpoints; |
57 | Endpoints suffixEndpoints; |
58 | Endpoints paddingEndpoints; |
59 | }; |
60 | |
61 | // Exported as U_I18N_API because it is needed for the unit test PatternStringTest |
62 | struct U_I18N_API ParsedPatternInfo : public AffixPatternProvider, public UMemory { |
63 | UnicodeString pattern; |
64 | ParsedSubpatternInfo positive; |
65 | ParsedSubpatternInfo negative; |
66 | |
67 | ParsedPatternInfo() |
68 | : state(this->pattern), currentSubpattern(nullptr) {} |
69 | |
70 | ~ParsedPatternInfo() U_OVERRIDE = default; |
71 | |
72 | // Need to declare this explicitly because of the destructor |
73 | ParsedPatternInfo& operator=(ParsedPatternInfo&& src) U_NOEXCEPT = default; |
74 | |
75 | static int32_t getLengthFromEndpoints(const Endpoints& endpoints); |
76 | |
77 | char16_t charAt(int32_t flags, int32_t index) const U_OVERRIDE; |
78 | |
79 | int32_t length(int32_t flags) const U_OVERRIDE; |
80 | |
81 | UnicodeString getString(int32_t flags) const U_OVERRIDE; |
82 | |
83 | bool positiveHasPlusSign() const U_OVERRIDE; |
84 | |
85 | bool hasNegativeSubpattern() const U_OVERRIDE; |
86 | |
87 | bool negativeHasMinusSign() const U_OVERRIDE; |
88 | |
89 | bool hasCurrencySign() const U_OVERRIDE; |
90 | |
91 | bool containsSymbolType(AffixPatternType type, UErrorCode& status) const U_OVERRIDE; |
92 | |
93 | bool hasBody() const U_OVERRIDE; |
94 | |
95 | private: |
96 | struct U_I18N_API ParserState { |
97 | const UnicodeString& pattern; // reference to the parent |
98 | int32_t offset = 0; |
99 | |
100 | explicit ParserState(const UnicodeString& _pattern) |
101 | : pattern(_pattern) {} |
102 | |
103 | ParserState& operator=(ParserState&& src) U_NOEXCEPT { |
104 | // Leave pattern reference alone; it will continue to point to the same place in memory, |
105 | // which gets overwritten by ParsedPatternInfo's implicit move assignment. |
106 | offset = src.offset; |
107 | return *this; |
108 | } |
109 | |
110 | UChar32 peek(); |
111 | |
112 | UChar32 next(); |
113 | |
114 | // TODO: We don't currently do anything with the message string. |
115 | // This method is here as a shell for Java compatibility. |
116 | inline void toParseException(const char16_t* message) { (void) message; } |
117 | } state; |
118 | |
119 | // NOTE: In Java, these are written as pure functions. |
120 | // In C++, they're written as methods. |
121 | // The behavior is the same. |
122 | |
123 | // Mutable transient pointer: |
124 | ParsedSubpatternInfo* currentSubpattern; |
125 | |
126 | // In Java, "negative == null" tells us whether or not we had a negative subpattern. |
127 | // In C++, we need to remember in another boolean. |
128 | bool fHasNegativeSubpattern = false; |
129 | |
130 | const Endpoints& getEndpoints(int32_t flags) const; |
131 | |
132 | /** Run the recursive descent parser. */ |
133 | void consumePattern(const UnicodeString& patternString, UErrorCode& status); |
134 | |
135 | void consumeSubpattern(UErrorCode& status); |
136 | |
137 | void consumePadding(PadPosition paddingLocation, UErrorCode& status); |
138 | |
139 | void consumeAffix(Endpoints& endpoints, UErrorCode& status); |
140 | |
141 | void consumeLiteral(UErrorCode& status); |
142 | |
143 | void consumeFormat(UErrorCode& status); |
144 | |
145 | void consumeIntegerFormat(UErrorCode& status); |
146 | |
147 | void consumeFractionFormat(UErrorCode& status); |
148 | |
149 | void consumeExponent(UErrorCode& status); |
150 | |
151 | friend class PatternParser; |
152 | }; |
153 | |
154 | enum IgnoreRounding { |
155 | IGNORE_ROUNDING_NEVER = 0, IGNORE_ROUNDING_IF_CURRENCY = 1, IGNORE_ROUNDING_ALWAYS = 2 |
156 | }; |
157 | |
158 | class U_I18N_API PatternParser { |
159 | public: |
160 | /** |
161 | * Runs the recursive descent parser on the given pattern string, returning a data structure with raw information |
162 | * about the pattern string. |
163 | * |
164 | * <p> |
165 | * To obtain a more useful form of the data, consider using {@link #parseToProperties} instead. |
166 | * |
167 | * TODO: Change argument type to const char16_t* instead of UnicodeString? |
168 | * |
169 | * @param patternString |
170 | * The LDML decimal format pattern (Excel-style pattern) to parse. |
171 | * @return The results of the parse. |
172 | */ |
173 | static void parseToPatternInfo(const UnicodeString& patternString, ParsedPatternInfo& patternInfo, |
174 | UErrorCode& status); |
175 | |
176 | /** |
177 | * Parses a pattern string into a new property bag. |
178 | * |
179 | * @param pattern |
180 | * The pattern string, like "#,##0.00" |
181 | * @param ignoreRounding |
182 | * Whether to leave out rounding information (minFrac, maxFrac, and rounding increment) when parsing the |
183 | * pattern. This may be desirable if a custom rounding mode, such as CurrencyUsage, is to be used |
184 | * instead. |
185 | * @return A property bag object. |
186 | * @throws IllegalArgumentException |
187 | * If there is a syntax error in the pattern string. |
188 | */ |
189 | static DecimalFormatProperties parseToProperties(const UnicodeString& pattern, |
190 | IgnoreRounding ignoreRounding, UErrorCode& status); |
191 | |
192 | static DecimalFormatProperties parseToProperties(const UnicodeString& pattern, UErrorCode& status); |
193 | |
194 | /** |
195 | * Parses a pattern string into an existing property bag. All properties that can be encoded into a pattern string |
196 | * will be overwritten with either their default value or with the value coming from the pattern string. Properties |
197 | * that cannot be encoded into a pattern string, such as rounding mode, are not modified. |
198 | * |
199 | * @param pattern |
200 | * The pattern string, like "#,##0.00" |
201 | * @param properties |
202 | * The property bag object to overwrite. |
203 | * @param ignoreRounding |
204 | * See {@link #parseToProperties(String pattern, int ignoreRounding)}. |
205 | * @throws IllegalArgumentException |
206 | * If there was a syntax error in the pattern string. |
207 | */ |
208 | static void parseToExistingProperties(const UnicodeString& pattern, |
209 | DecimalFormatProperties& properties, |
210 | IgnoreRounding ignoreRounding, UErrorCode& status); |
211 | |
212 | private: |
213 | static void parseToExistingPropertiesImpl(const UnicodeString& pattern, |
214 | DecimalFormatProperties& properties, |
215 | IgnoreRounding ignoreRounding, UErrorCode& status); |
216 | |
217 | /** Finalizes the temporary data stored in the ParsedPatternInfo to the Properties. */ |
218 | static void patternInfoToProperties(DecimalFormatProperties& properties, |
219 | ParsedPatternInfo& patternInfo, IgnoreRounding _ignoreRounding, |
220 | UErrorCode& status); |
221 | }; |
222 | |
223 | class U_I18N_API PatternStringUtils { |
224 | public: |
225 | /** |
226 | * Determine whether a given roundingIncrement should be ignored for formatting |
227 | * based on the current maxFrac value (maximum fraction digits). For example a |
228 | * roundingIncrement of 0.01 should be ignored if maxFrac is 1, but not if maxFrac |
229 | * is 2 or more. Note that roundingIncrements are rounded up in significance, so |
230 | * a roundingIncrement of 0.006 is treated like 0.01 for this determination, i.e. |
231 | * it should not be ignored if maxFrac is 2 or more (but a roundingIncrement of |
232 | * 0.005 is treated like 0.001 for significance). |
233 | * |
234 | * This test is needed for both NumberPropertyMapper::oldToNew and |
235 | * PatternStringUtils::propertiesToPatternString. In Java it cannot be |
236 | * exported by NumberPropertyMapper (package provate) so it is in |
237 | * PatternStringUtils, do the same in C. |
238 | * |
239 | * @param roundIncr |
240 | * The roundingIncrement to be checked. Must be non-zero. |
241 | * @param maxFrac |
242 | * The current maximum fraction digits value. |
243 | * @return true if roundIncr should be ignored for formatting. |
244 | */ |
245 | static bool ignoreRoundingIncrement(double roundIncr, int32_t maxFrac); |
246 | |
247 | /** |
248 | * Creates a pattern string from a property bag. |
249 | * |
250 | * <p> |
251 | * Since pattern strings support only a subset of the functionality available in a property bag, a new property bag |
252 | * created from the string returned by this function may not be the same as the original property bag. |
253 | * |
254 | * @param properties |
255 | * The property bag to serialize. |
256 | * @return A pattern string approximately serializing the property bag. |
257 | */ |
258 | static UnicodeString propertiesToPatternString(const DecimalFormatProperties& properties, |
259 | UErrorCode& status); |
260 | |
261 | |
262 | /** |
263 | * Converts a pattern between standard notation and localized notation. Localized notation means that instead of |
264 | * using generic placeholders in the pattern, you use the corresponding locale-specific characters instead. For |
265 | * example, in locale <em>fr-FR</em>, the period in the pattern "0.000" means "decimal" in standard notation (as it |
266 | * does in every other locale), but it means "grouping" in localized notation. |
267 | * |
268 | * <p> |
269 | * A greedy string-substitution strategy is used to substitute locale symbols. If two symbols are ambiguous or have |
270 | * the same prefix, the result is not well-defined. |
271 | * |
272 | * <p> |
273 | * Locale symbols are not allowed to contain the ASCII quote character. |
274 | * |
275 | * <p> |
276 | * This method is provided for backwards compatibility and should not be used in any new code. |
277 | * |
278 | * TODO(C++): This method is not yet implemented. |
279 | * |
280 | * @param input |
281 | * The pattern to convert. |
282 | * @param symbols |
283 | * The symbols corresponding to the localized pattern. |
284 | * @param toLocalized |
285 | * true to convert from standard to localized notation; false to convert from localized to standard |
286 | * notation. |
287 | * @return The pattern expressed in the other notation. |
288 | */ |
289 | static UnicodeString convertLocalized(const UnicodeString& input, const DecimalFormatSymbols& symbols, |
290 | bool toLocalized, UErrorCode& status); |
291 | |
292 | /** |
293 | * This method contains the heart of the logic for rendering LDML affix strings. It handles |
294 | * sign-always-shown resolution, whether to use the positive or negative subpattern, permille |
295 | * substitution, and plural forms for CurrencyPluralInfo. |
296 | */ |
297 | static void patternInfoToStringBuilder(const AffixPatternProvider& patternInfo, bool isPrefix, |
298 | Signum signum, UNumberSignDisplay signDisplay, |
299 | StandardPlural::Form plural, bool perMilleReplacesPercent, |
300 | UnicodeString& output); |
301 | |
302 | private: |
303 | /** @return The number of chars inserted. */ |
304 | static int escapePaddingString(UnicodeString input, UnicodeString& output, int startIndex, |
305 | UErrorCode& status); |
306 | }; |
307 | |
308 | } // namespace impl |
309 | } // namespace number |
310 | U_NAMESPACE_END |
311 | |
312 | |
313 | #endif //__NUMBER_PATTERNSTRING_H__ |
314 | |
315 | #endif /* #if !UCONFIG_NO_FORMATTING */ |
316 | |