]>
Commit | Line | Data |
---|---|---|
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_currency.h" | |
14 | #include "ucurrimp.h" | |
15 | #include "unicode/errorcode.h" | |
16 | #include "numparse_utils.h" | |
17 | ||
18 | using namespace icu; | |
19 | using namespace icu::numparse; | |
20 | using namespace icu::numparse::impl; | |
21 | ||
22 | ||
23 | CombinedCurrencyMatcher::CombinedCurrencyMatcher(const CurrencySymbols& currencySymbols, const DecimalFormatSymbols& dfs, | |
24 | parse_flags_t parseFlags, UErrorCode& status) | |
25 | : fCurrency1(currencySymbols.getCurrencySymbol(status)), | |
26 | fCurrency2(currencySymbols.getIntlCurrencySymbol(status)), | |
27 | fUseFullCurrencyData(0 == (parseFlags & PARSE_FLAG_NO_FOREIGN_CURRENCY)), | |
28 | afterPrefixInsert(dfs.getPatternForCurrencySpacing(UNUM_CURRENCY_INSERT, false, status)), | |
29 | beforeSuffixInsert(dfs.getPatternForCurrencySpacing(UNUM_CURRENCY_INSERT, true, status)), | |
30 | fLocaleName(dfs.getLocale().getName(), -1, status) { | |
31 | utils::copyCurrencyCode(fCurrencyCode, currencySymbols.getIsoCode()); | |
32 | ||
33 | // Pre-load the long names for the current locale and currency | |
34 | // if we are parsing without the full currency data. | |
35 | if (!fUseFullCurrencyData) { | |
36 | for (int32_t i=0; i<StandardPlural::COUNT; i++) { | |
37 | auto plural = static_cast<StandardPlural::Form>(i); | |
38 | fLocalLongNames[i] = currencySymbols.getPluralName(plural, status); | |
39 | } | |
40 | } | |
41 | ||
42 | // TODO: Figure out how to make this faster and re-enable. | |
43 | // Computing the "lead code points" set for fastpathing is too slow to use in production. | |
44 | // See http://bugs.icu-project.org/trac/ticket/13584 | |
45 | // // Compute the full set of characters that could be the first in a currency to allow for | |
46 | // // efficient smoke test. | |
47 | // fLeadCodePoints.add(fCurrency1.char32At(0)); | |
48 | // fLeadCodePoints.add(fCurrency2.char32At(0)); | |
49 | // fLeadCodePoints.add(beforeSuffixInsert.char32At(0)); | |
50 | // uprv_currencyLeads(fLocaleName.data(), fLeadCodePoints, status); | |
51 | // // Always apply case mapping closure for currencies | |
52 | // fLeadCodePoints.closeOver(USET_ADD_CASE_MAPPINGS); | |
53 | // fLeadCodePoints.freeze(); | |
54 | } | |
55 | ||
56 | bool | |
57 | CombinedCurrencyMatcher::match(StringSegment& segment, ParsedNumber& result, UErrorCode& status) const { | |
58 | if (result.currencyCode[0] != 0) { | |
59 | return false; | |
60 | } | |
61 | ||
62 | // Try to match a currency spacing separator. | |
63 | int32_t initialOffset = segment.getOffset(); | |
64 | bool maybeMore = false; | |
65 | if (result.seenNumber() && !beforeSuffixInsert.isEmpty()) { | |
66 | int32_t overlap = segment.getCommonPrefixLength(beforeSuffixInsert); | |
67 | if (overlap == beforeSuffixInsert.length()) { | |
68 | segment.adjustOffset(overlap); | |
69 | // Note: let currency spacing be a weak match. Don't update chars consumed. | |
70 | } | |
71 | maybeMore = maybeMore || overlap == segment.length(); | |
72 | } | |
73 | ||
74 | // Match the currency string, and reset if we didn't find one. | |
75 | maybeMore = maybeMore || matchCurrency(segment, result, status); | |
76 | if (result.currencyCode[0] == 0) { | |
77 | segment.setOffset(initialOffset); | |
78 | return maybeMore; | |
79 | } | |
80 | ||
81 | // Try to match a currency spacing separator. | |
82 | if (!result.seenNumber() && !afterPrefixInsert.isEmpty()) { | |
83 | int32_t overlap = segment.getCommonPrefixLength(afterPrefixInsert); | |
84 | if (overlap == afterPrefixInsert.length()) { | |
85 | segment.adjustOffset(overlap); | |
86 | // Note: let currency spacing be a weak match. Don't update chars consumed. | |
87 | } | |
88 | maybeMore = maybeMore || overlap == segment.length(); | |
89 | } | |
90 | ||
91 | return maybeMore; | |
92 | } | |
93 | ||
94 | bool CombinedCurrencyMatcher::matchCurrency(StringSegment& segment, ParsedNumber& result, | |
95 | UErrorCode& status) const { | |
96 | bool maybeMore = false; | |
97 | ||
98 | int32_t overlap1; | |
99 | if (!fCurrency1.isEmpty()) { | |
100 | overlap1 = segment.getCaseSensitivePrefixLength(fCurrency1); | |
101 | } else { | |
102 | overlap1 = -1; | |
103 | } | |
104 | maybeMore = maybeMore || overlap1 == segment.length(); | |
105 | if (overlap1 == fCurrency1.length()) { | |
106 | utils::copyCurrencyCode(result.currencyCode, fCurrencyCode); | |
107 | segment.adjustOffset(overlap1); | |
108 | result.setCharsConsumed(segment); | |
109 | return maybeMore; | |
110 | } | |
111 | ||
112 | int32_t overlap2; | |
113 | if (!fCurrency2.isEmpty()) { | |
114 | overlap2 = segment.getCaseSensitivePrefixLength(fCurrency2); | |
115 | } else { | |
116 | overlap2 = -1; | |
117 | } | |
118 | maybeMore = maybeMore || overlap2 == segment.length(); | |
119 | if (overlap2 == fCurrency2.length()) { | |
120 | utils::copyCurrencyCode(result.currencyCode, fCurrencyCode); | |
121 | segment.adjustOffset(overlap2); | |
122 | result.setCharsConsumed(segment); | |
123 | return maybeMore; | |
124 | } | |
125 | ||
126 | if (fUseFullCurrencyData) { | |
127 | // Use the full currency data. | |
128 | // NOTE: This call site should be improved with #13584. | |
129 | const UnicodeString segmentString = segment.toTempUnicodeString(); | |
130 | ||
131 | // Try to parse the currency | |
132 | ParsePosition ppos(0); | |
133 | int32_t partialMatchLen = 0; | |
134 | uprv_parseCurrency( | |
135 | fLocaleName.data(), | |
136 | segmentString, | |
137 | ppos, | |
138 | UCURR_SYMBOL_NAME, // checks for both UCURR_SYMBOL_NAME and UCURR_LONG_NAME | |
139 | &partialMatchLen, | |
140 | result.currencyCode, | |
141 | status); | |
142 | maybeMore = maybeMore || partialMatchLen == segment.length(); | |
143 | ||
144 | if (U_SUCCESS(status) && ppos.getIndex() != 0) { | |
145 | // Complete match. | |
146 | // NOTE: The currency code should already be saved in the ParsedNumber. | |
147 | segment.adjustOffset(ppos.getIndex()); | |
148 | result.setCharsConsumed(segment); | |
149 | return maybeMore; | |
150 | } | |
151 | ||
152 | } else { | |
153 | // Use the locale long names. | |
154 | int32_t longestFullMatch = 0; | |
155 | for (int32_t i=0; i<StandardPlural::COUNT; i++) { | |
156 | const UnicodeString& name = fLocalLongNames[i]; | |
157 | int32_t overlap = segment.getCommonPrefixLength(name); | |
158 | if (overlap == name.length() && name.length() > longestFullMatch) { | |
159 | longestFullMatch = name.length(); | |
160 | } | |
161 | maybeMore = maybeMore || overlap > 0; | |
162 | } | |
163 | if (longestFullMatch > 0) { | |
164 | utils::copyCurrencyCode(result.currencyCode, fCurrencyCode); | |
165 | segment.adjustOffset(longestFullMatch); | |
166 | result.setCharsConsumed(segment); | |
167 | return maybeMore; | |
168 | } | |
169 | } | |
170 | ||
171 | // No match found. | |
172 | return maybeMore; | |
173 | } | |
174 | ||
175 | bool CombinedCurrencyMatcher::smokeTest(const StringSegment&) const { | |
176 | // TODO: See constructor | |
177 | return true; | |
178 | //return segment.startsWith(fLeadCodePoints); | |
179 | } | |
180 | ||
181 | UnicodeString CombinedCurrencyMatcher::toString() const { | |
182 | return u"<CombinedCurrencyMatcher>"; | |
183 | } | |
184 | ||
185 | ||
186 | #endif /* #if !UCONFIG_NO_FORMATTING */ |