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 "static_unicode_sets.h"
13 #include "umutex.h"
14 #include "ucln_cmn.h"
15 #include "unicode/uniset.h"
16 #include "uresimp.h"
17 #include "cstring.h"
18 #include "uassert.h"
19
20 using namespace icu;
21 using namespace icu::unisets;
22
23
24 namespace {
25
26 UnicodeSet* gUnicodeSets[COUNT] = {};
27
28 // Save the empty instance in static memory to have well-defined behavior if a
29 // regular UnicodeSet cannot be allocated.
30 alignas(UnicodeSet)
31 char gEmptyUnicodeSet[sizeof(UnicodeSet)];
32
33 // Whether the gEmptyUnicodeSet is initialized and ready to use.
34 UBool gEmptyUnicodeSetInitialized = FALSE;
35
getImpl(Key key)36 inline UnicodeSet* getImpl(Key key) {
37 UnicodeSet* candidate = gUnicodeSets[key];
38 if (candidate == nullptr) {
39 return reinterpret_cast<UnicodeSet*>(gEmptyUnicodeSet);
40 }
41 return candidate;
42 }
43
computeUnion(Key k1,Key k2)44 UnicodeSet* computeUnion(Key k1, Key k2) {
45 UnicodeSet* result = new UnicodeSet();
46 if (result == nullptr) {
47 return nullptr;
48 }
49 result->addAll(*getImpl(k1));
50 result->addAll(*getImpl(k2));
51 result->freeze();
52 return result;
53 }
54
computeUnion(Key k1,Key k2,Key k3)55 UnicodeSet* computeUnion(Key k1, Key k2, Key k3) {
56 UnicodeSet* result = new UnicodeSet();
57 if (result == nullptr) {
58 return nullptr;
59 }
60 result->addAll(*getImpl(k1));
61 result->addAll(*getImpl(k2));
62 result->addAll(*getImpl(k3));
63 result->freeze();
64 return result;
65 }
66
67
saveSet(Key key,const UnicodeString & unicodeSetPattern,UErrorCode & status)68 void saveSet(Key key, const UnicodeString& unicodeSetPattern, UErrorCode& status) {
69 // assert unicodeSets.get(key) == null;
70 gUnicodeSets[key] = new UnicodeSet(unicodeSetPattern, status);
71 }
72
73 class ParseDataSink : public ResourceSink {
74 public:
put(const char * key,ResourceValue & value,UBool,UErrorCode & status)75 void put(const char* key, ResourceValue& value, UBool /*noFallback*/, UErrorCode& status) U_OVERRIDE {
76 ResourceTable contextsTable = value.getTable(status);
77 if (U_FAILURE(status)) { return; }
78 for (int i = 0; contextsTable.getKeyAndValue(i, key, value); i++) {
79 if (uprv_strcmp(key, "date") == 0) {
80 // ignore
81 } else {
82 ResourceTable strictnessTable = value.getTable(status);
83 if (U_FAILURE(status)) { return; }
84 for (int j = 0; strictnessTable.getKeyAndValue(j, key, value); j++) {
85 bool isLenient = (uprv_strcmp(key, "lenient") == 0);
86 ResourceArray array = value.getArray(status);
87 if (U_FAILURE(status)) { return; }
88 for (int k = 0; k < array.getSize(); k++) {
89 array.getValue(k, value);
90 UnicodeString str = value.getUnicodeString(status);
91 if (U_FAILURE(status)) { return; }
92 // There is both lenient and strict data for comma/period,
93 // but not for any of the other symbols.
94 if (str.indexOf(u'.') != -1) {
95 saveSet(isLenient ? PERIOD : STRICT_PERIOD, str, status);
96 } else if (str.indexOf(u',') != -1) {
97 saveSet(isLenient ? COMMA : STRICT_COMMA, str, status);
98 } else if (str.indexOf(u'+') != -1) {
99 saveSet(PLUS_SIGN, str, status);
100 } else if (str.indexOf(u'‒') != -1) {
101 saveSet(MINUS_SIGN, str, status);
102 } else if (str.indexOf(u'$') != -1) {
103 saveSet(DOLLAR_SIGN, str, status);
104 } else if (str.indexOf(u'£') != -1) {
105 saveSet(POUND_SIGN, str, status);
106 } else if (str.indexOf(u'₨') != -1) {
107 saveSet(RUPEE_SIGN, str, status);
108 }
109 if (U_FAILURE(status)) { return; }
110 }
111 }
112 }
113 }
114 }
115 };
116
117
118 icu::UInitOnce gNumberParseUniSetsInitOnce = U_INITONCE_INITIALIZER;
119
cleanupNumberParseUniSets()120 UBool U_CALLCONV cleanupNumberParseUniSets() {
121 if (gEmptyUnicodeSetInitialized) {
122 reinterpret_cast<UnicodeSet*>(gEmptyUnicodeSet)->~UnicodeSet();
123 gEmptyUnicodeSetInitialized = FALSE;
124 }
125 for (int32_t i = 0; i < COUNT; i++) {
126 delete gUnicodeSets[i];
127 gUnicodeSets[i] = nullptr;
128 }
129 gNumberParseUniSetsInitOnce.reset();
130 return TRUE;
131 }
132
initNumberParseUniSets(UErrorCode & status)133 void U_CALLCONV initNumberParseUniSets(UErrorCode& status) {
134 ucln_common_registerCleanup(UCLN_COMMON_NUMPARSE_UNISETS, cleanupNumberParseUniSets);
135
136 // Initialize the empty instance for well-defined fallback behavior
137 new(gEmptyUnicodeSet) UnicodeSet();
138 reinterpret_cast<UnicodeSet*>(gEmptyUnicodeSet)->freeze();
139 gEmptyUnicodeSetInitialized = TRUE;
140
141 // These sets were decided after discussion with icu-design@. See tickets #13084 and #13309.
142 // Zs+TAB is "horizontal whitespace" according to UTS #18 (blank property).
143 gUnicodeSets[DEFAULT_IGNORABLES] = new UnicodeSet(
144 u"[[:Zs:][\\u0009][:Bidi_Control:][:Variation_Selector:]]", status);
145 gUnicodeSets[STRICT_IGNORABLES] = new UnicodeSet(u"[[:Bidi_Control:]]", status);
146
147 LocalUResourceBundlePointer rb(ures_open(nullptr, "root", &status));
148 if (U_FAILURE(status)) { return; }
149 ParseDataSink sink;
150 ures_getAllItemsWithFallback(rb.getAlias(), "parse", sink, status);
151 if (U_FAILURE(status)) { return; }
152
153 // NOTE: It is OK for these assertions to fail if there was a no-data build.
154 U_ASSERT(gUnicodeSets[COMMA] != nullptr);
155 U_ASSERT(gUnicodeSets[STRICT_COMMA] != nullptr);
156 U_ASSERT(gUnicodeSets[PERIOD] != nullptr);
157 U_ASSERT(gUnicodeSets[STRICT_PERIOD] != nullptr);
158
159 gUnicodeSets[OTHER_GROUPING_SEPARATORS] = new UnicodeSet(
160 u"['٬‘’＇\\u0020\\u00A0\\u2000-\\u200A\\u202F\\u205F\\u3000]", status);
161 gUnicodeSets[ALL_SEPARATORS] = computeUnion(COMMA, PERIOD, OTHER_GROUPING_SEPARATORS);
162 gUnicodeSets[STRICT_ALL_SEPARATORS] = computeUnion(
163 STRICT_COMMA, STRICT_PERIOD, OTHER_GROUPING_SEPARATORS);
164
165 U_ASSERT(gUnicodeSets[MINUS_SIGN] != nullptr);
166 U_ASSERT(gUnicodeSets[PLUS_SIGN] != nullptr);
167
168 gUnicodeSets[PERCENT_SIGN] = new UnicodeSet(u"[%٪]", status);
169 gUnicodeSets[PERMILLE_SIGN] = new UnicodeSet(u"[‰؉]", status);
170 gUnicodeSets[INFINITY_KEY] = new UnicodeSet(u"[∞]", status);
171
172 U_ASSERT(gUnicodeSets[DOLLAR_SIGN] != nullptr);
173 U_ASSERT(gUnicodeSets[POUND_SIGN] != nullptr);
174 U_ASSERT(gUnicodeSets[RUPEE_SIGN] != nullptr);
175 gUnicodeSets[YEN_SIGN] = new UnicodeSet(u"[¥\\uffe5]", status);
176
177 gUnicodeSets[DIGITS] = new UnicodeSet(u"[:digit:]", status);
178
179 gUnicodeSets[DIGITS_OR_ALL_SEPARATORS] = computeUnion(DIGITS, ALL_SEPARATORS);
180 gUnicodeSets[DIGITS_OR_STRICT_ALL_SEPARATORS] = computeUnion(DIGITS, STRICT_ALL_SEPARATORS);
181
182 for (auto* uniset : gUnicodeSets) {
183 if (uniset != nullptr) {
184 uniset->freeze();
185 }
186 }
187 }
188
189 }
190
get(Key key)191 const UnicodeSet* unisets::get(Key key) {
192 UErrorCode localStatus = U_ZERO_ERROR;
193 umtx_initOnce(gNumberParseUniSetsInitOnce, &initNumberParseUniSets, localStatus);
194 if (U_FAILURE(localStatus)) {
195 return reinterpret_cast<UnicodeSet*>(gEmptyUnicodeSet);
196 }
197 return getImpl(key);
198 }
199
chooseFrom(UnicodeString str,Key key1)200 Key unisets::chooseFrom(UnicodeString str, Key key1) {
201 return get(key1)->contains(str) ? key1 : NONE;
202 }
203
chooseFrom(UnicodeString str,Key key1,Key key2)204 Key unisets::chooseFrom(UnicodeString str, Key key1, Key key2) {
205 return get(key1)->contains(str) ? key1 : chooseFrom(str, key2);
206 }
207
208 //Key unisets::chooseCurrency(UnicodeString str) {
209 // if (get(DOLLAR_SIGN)->contains(str)) {
210 // return DOLLAR_SIGN;
211 // } else if (get(POUND_SIGN)->contains(str)) {
212 // return POUND_SIGN;
213 // } else if (get(RUPEE_SIGN)->contains(str)) {
214 // return RUPEE_SIGN;
215 // } else if (get(YEN_SIGN)->contains(str)) {
216 // return YEN_SIGN;
217 // } else {
218 // return NONE;
219 // }
220 //}
221
222
223 #endif /* #if !UCONFIG_NO_FORMATTING */
224