]> git.saurik.com Git - apple/icu.git/blob - icuSources/i18n/numparse_scientific.cpp
ICU-64232.0.1.tar.gz
[apple/icu.git] / icuSources / i18n / numparse_scientific.cpp
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_scientific.h"
14 #include "static_unicode_sets.h"
15
16 using namespace icu;
17 using namespace icu::numparse;
18 using namespace icu::numparse::impl;
19
20
21 namespace {
22
23 inline const UnicodeSet& minusSignSet() {
24 return *unisets::get(unisets::MINUS_SIGN);
25 }
26
27 inline const UnicodeSet& plusSignSet() {
28 return *unisets::get(unisets::PLUS_SIGN);
29 }
30
31 inline const UnicodeSet& ignorablesSet() { // <rdar://problem/39156484>
32 return *unisets::get(unisets::STRICT_IGNORABLES);
33 }
34
35 } // namespace
36
37
38 ScientificMatcher::ScientificMatcher(const DecimalFormatSymbols& dfs, const Grouper& grouper)
39 : fExponentSeparatorString(dfs.getConstSymbol(DecimalFormatSymbols::kExponentialSymbol)),
40 fExponentMatcher(dfs, grouper, PARSE_FLAG_INTEGER_ONLY | PARSE_FLAG_GROUPING_DISABLED) {
41
42 const UnicodeString& minusSign = dfs.getConstSymbol(DecimalFormatSymbols::kMinusSignSymbol);
43 if (minusSignSet().contains(minusSign)) {
44 fCustomMinusSign.setToBogus();
45 } else {
46 fCustomMinusSign = minusSign;
47 }
48
49 const UnicodeString& plusSign = dfs.getConstSymbol(DecimalFormatSymbols::kPlusSignSymbol);
50 if (plusSignSet().contains(plusSign)) {
51 fCustomPlusSign.setToBogus();
52 } else {
53 fCustomPlusSign = plusSign;
54 }
55 }
56
57 bool ScientificMatcher::match(StringSegment& segment, ParsedNumber& result, UErrorCode& status) const {
58 // Only accept scientific notation after the mantissa.
59 if (!result.seenNumber()) {
60 return false;
61 }
62
63 // Only accept one exponent per string.
64 if (0 != (result.flags & FLAG_HAS_EXPONENT)) {
65 return false;
66 }
67
68 // First match the scientific separator, and then match another number after it.
69 // NOTE: This is guarded by the smoke test; no need to check fExponentSeparatorString length again.
70 int overlap1 = segment.getCommonPrefixLength(fExponentSeparatorString);
71 if (overlap1 == fExponentSeparatorString.length()) {
72 // Full exponent separator match.
73 int32_t exponentStart = segment.getOffset(); // <rdar://problem/39156484>
74
75 // First attempt to get a code point, returning true if we can't get one.
76 if (segment.length() == overlap1) {
77 return true;
78 }
79 segment.adjustOffset(overlap1);
80
81 // Allow a sign, and then try to match digits.
82 while (segment.length() > 0 && segment.startsWith(ignorablesSet())) { // <rdar://problem/39156484>
83 segment.adjustOffsetByCodePoint();
84 }
85 int8_t exponentSign = 1;
86 if (segment.startsWith(minusSignSet())) {
87 exponentSign = -1;
88 segment.adjustOffsetByCodePoint();
89 } else if (segment.startsWith(plusSignSet())) {
90 segment.adjustOffsetByCodePoint();
91 } else if (segment.startsWith(fCustomMinusSign)) {
92 // Note: call site is guarded with startsWith, which returns false on empty string
93 int32_t overlap2 = segment.getCommonPrefixLength(fCustomMinusSign);
94 if (overlap2 != fCustomMinusSign.length()) {
95 // Partial custom sign match; un-match the exponent separator.
96 segment.setOffset(exponentStart);
97 return true;
98 }
99 exponentSign = -1;
100 segment.adjustOffset(overlap2);
101 } else if (segment.startsWith(fCustomPlusSign)) {
102 // Note: call site is guarded with startsWith, which returns false on empty string
103 int32_t overlap2 = segment.getCommonPrefixLength(fCustomPlusSign);
104 if (overlap2 != fCustomPlusSign.length()) {
105 // Partial custom sign match; un-match the exponent separator.
106 segment.setOffset(exponentStart);
107 return true;
108 }
109 segment.adjustOffset(overlap2);
110 }
111 while (segment.length() > 0 && segment.startsWith(ignorablesSet())) { // <rdar://problem/39156484>
112 segment.adjustOffsetByCodePoint();
113 }
114
115 // We are supposed to accept E0 after NaN, so we need to make sure result.quantity is available.
116 bool wasBogus = result.quantity.bogus;
117 result.quantity.bogus = false;
118 int digitsOffset = segment.getOffset();
119 bool digitsReturnValue = fExponentMatcher.match(segment, result, exponentSign, status);
120 result.quantity.bogus = wasBogus;
121
122 if (segment.getOffset() != digitsOffset) {
123 // At least one exponent digit was matched.
124 result.flags |= FLAG_HAS_EXPONENT;
125 } else {
126 // No exponent digits were matched; un-match the exponent separator.
127 segment.setOffset(exponentStart);
128 }
129 return digitsReturnValue;
130
131 } else if (overlap1 == segment.length()) {
132 // Partial exponent separator match
133 return true;
134 }
135
136 // No match
137 return false;
138 }
139
140 bool ScientificMatcher::smokeTest(const StringSegment& segment) const {
141 return segment.startsWith(fExponentSeparatorString);
142 }
143
144 UnicodeString ScientificMatcher::toString() const {
145 return u"<Scientific>";
146 }
147
148
149 #endif /* #if !UCONFIG_NO_FORMATTING */