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
CombinedCurrencyMatcher(const CurrencySymbols & currencySymbols,const DecimalFormatSymbols & dfs,parse_flags_t parseFlags,UErrorCode & status)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
match(StringSegment & segment,ParsedNumber & result,UErrorCode & status) const57 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
matchCurrency(StringSegment & segment,ParsedNumber & result,UErrorCode & status) const94 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 // ISO codes should be accepted case-insensitive.
115 // https://unicode-org.atlassian.net/browse/ICU-13696
116 overlap2 = segment.getCommonPrefixLength(fCurrency2);
117 } else {
118 overlap2 = -1;
119 }
120 maybeMore = maybeMore || overlap2 == segment.length();
121 if (overlap2 == fCurrency2.length()) {
122 utils::copyCurrencyCode(result.currencyCode, fCurrencyCode);
123 segment.adjustOffset(overlap2);
124 result.setCharsConsumed(segment);
125 return maybeMore;
126 }
127
128 if (fUseFullCurrencyData) {
129 // Use the full currency data.
130 // NOTE: This call site should be improved with #13584.
131 const UnicodeString segmentString = segment.toTempUnicodeString();
132
133 // Try to parse the currency
134 ParsePosition ppos(0);
135 int32_t partialMatchLen = 0;
136 uprv_parseCurrency(
137 fLocaleName.data(),
138 segmentString,
139 ppos,
140 UCURR_SYMBOL_NAME, // checks for both UCURR_SYMBOL_NAME and UCURR_LONG_NAME
141 &partialMatchLen,
142 result.currencyCode,
143 status);
144 maybeMore = maybeMore || partialMatchLen == segment.length();
145
146 if (U_SUCCESS(status) && ppos.getIndex() != 0) {
147 // Complete match.
148 // NOTE: The currency code should already be saved in the ParsedNumber.
149 segment.adjustOffset(ppos.getIndex());
150 result.setCharsConsumed(segment);
151 return maybeMore;
152 }
153
154 } else {
155 // Use the locale long names.
156 int32_t longestFullMatch = 0;
157 for (int32_t i=0; i<StandardPlural::COUNT; i++) {
158 const UnicodeString& name = fLocalLongNames[i];
159 int32_t overlap = segment.getCommonPrefixLength(name);
160 if (overlap == name.length() && name.length() > longestFullMatch) {
161 longestFullMatch = name.length();
162 }
163 maybeMore = maybeMore || overlap > 0;
164 }
165 if (longestFullMatch > 0) {
166 utils::copyCurrencyCode(result.currencyCode, fCurrencyCode);
167 segment.adjustOffset(longestFullMatch);
168 result.setCharsConsumed(segment);
169 return maybeMore;
170 }
171 }
172
173 // No match found.
174 return maybeMore;
175 }
176
smokeTest(const StringSegment &) const177 bool CombinedCurrencyMatcher::smokeTest(const StringSegment&) const {
178 // TODO: See constructor
179 return true;
180 //return segment.startsWith(fLeadCodePoints);
181 }
182
toString() const183 UnicodeString CombinedCurrencyMatcher::toString() const {
184 return u"<CombinedCurrencyMatcher>";
185 }
186
187
188 #endif /* #if !UCONFIG_NO_FORMATTING */
189