/src/mozilla-central/intl/icu/source/i18n/number_affixutils.cpp

Source (jump to first uncovered line)
// © 2017 and later: Unicode, Inc. and others.
// License & terms of use: http://www.unicode.org/copyright.html

#include "unicode/utypes.h"

#if !UCONFIG_NO_FORMATTING

#include "number_affixutils.h"
#include "unicode/utf16.h"
#include "unicode/uniset.h"

using namespace icu;
using namespace icu::number;
using namespace icu::number::impl;

TokenConsumer::~TokenConsumer() = default;
SymbolProvider::~SymbolProvider() = default;

int32_t AffixUtils::estimateLength(const UnicodeString &patternString, UErrorCode &status) {
    AffixPatternState state = STATE_BASE;
    int32_t offset = 0;
    int32_t length = 0;
    for (; offset < patternString.length();) {
        UChar32 cp = patternString.char32At(offset);

        switch (state) {
            case STATE_BASE:
                if (cp == u'\'') {
                    // First quote
                    state = STATE_FIRST_QUOTE;
                } else {
                    // Unquoted symbol
                    length++;
                }
                break;
            case STATE_FIRST_QUOTE:
                if (cp == u'\'') {
                    // Repeated quote
                    length++;
                    state = STATE_BASE;
                } else {
                    // Quoted code point
                    length++;
                    state = STATE_INSIDE_QUOTE;
                }
                break;
            case STATE_INSIDE_QUOTE:
                if (cp == u'\'') {
                    // End of quoted sequence
                    state = STATE_AFTER_QUOTE;
                } else {
                    // Quoted code point
                    length++;
                }
                break;
            case STATE_AFTER_QUOTE:
                if (cp == u'\'') {
                    // Double quote inside of quoted sequence
                    length++;
                    state = STATE_INSIDE_QUOTE;
                } else {
                    // Unquoted symbol
                    length++;
                }
                break;
            default:
                U_ASSERT(false);
        }

        offset += U16_LENGTH(cp);
    }

    switch (state) {
        case STATE_FIRST_QUOTE:
        case STATE_INSIDE_QUOTE:
            status = U_ILLEGAL_ARGUMENT_ERROR;
            break;
        default:
            break;
    }

    return length;
}

UnicodeString AffixUtils::escape(const UnicodeString &input) {
    AffixPatternState state = STATE_BASE;
    int32_t offset = 0;
    UnicodeString output;
    for (; offset < input.length();) {
        UChar32 cp = input.char32At(offset);

        switch (cp) {
            case u'\'':
                output.append(u"''", -1);
                break;

            case u'-':
            case u'+':
            case u'%':
            case u'‰':
            case u'¤':
                if (state == STATE_BASE) {
                    output.append(u'\'');
                    output.append(cp);
                    state = STATE_INSIDE_QUOTE;
                } else {
                    output.append(cp);
                }
                break;

            default:
                if (state == STATE_INSIDE_QUOTE) {
                    output.append(u'\'');
                    output.append(cp);
                    state = STATE_BASE;
                } else {
                    output.append(cp);
                }
                break;
        }
        offset += U16_LENGTH(cp);
    }

    if (state == STATE_INSIDE_QUOTE) {
        output.append(u'\'');
    }

    return output;
}

Field AffixUtils::getFieldForType(AffixPatternType type) {
    switch (type) {
        case TYPE_MINUS_SIGN:
            return Field::UNUM_SIGN_FIELD;
        case TYPE_PLUS_SIGN:
            return Field::UNUM_SIGN_FIELD;
        case TYPE_PERCENT:
            return Field::UNUM_PERCENT_FIELD;
        case TYPE_PERMILLE:
            return Field::UNUM_PERMILL_FIELD;
        case TYPE_CURRENCY_SINGLE:
            return Field::UNUM_CURRENCY_FIELD;
        case TYPE_CURRENCY_DOUBLE:
            return Field::UNUM_CURRENCY_FIELD;
        case TYPE_CURRENCY_TRIPLE:
            return Field::UNUM_CURRENCY_FIELD;
        case TYPE_CURRENCY_QUAD:
            return Field::UNUM_CURRENCY_FIELD;
        case TYPE_CURRENCY_QUINT:
            return Field::UNUM_CURRENCY_FIELD;
        case TYPE_CURRENCY_OVERFLOW:
            return Field::UNUM_CURRENCY_FIELD;
        default:
            U_ASSERT(false);
            return Field::UNUM_FIELD_COUNT; // suppress "control reaches end of non-void function"
    }
}

int32_t
AffixUtils::unescape(const UnicodeString &affixPattern, NumberStringBuilder &output, int32_t position,
                     const SymbolProvider &provider, UErrorCode &status) {
    int32_t length = 0;
    AffixTag tag;
    while (hasNext(tag, affixPattern)) {
        tag = nextToken(tag, affixPattern, status);
        if (U_FAILURE(status)) { return length; }
        if (tag.type == TYPE_CURRENCY_OVERFLOW) {
            // Don't go to the provider for this special case
            length += output.insertCodePoint(position + length, 0xFFFD, UNUM_CURRENCY_FIELD, status);
        } else if (tag.type < 0) {
            length += output.insert(
                    position + length, provider.getSymbol(tag.type), getFieldForType(tag.type), status);
        } else {
            length += output.insertCodePoint(position + length, tag.codePoint, UNUM_FIELD_COUNT, status);
        }
    }
    return length;
}

int32_t AffixUtils::unescapedCodePointCount(const UnicodeString &affixPattern,
                                            const SymbolProvider &provider, UErrorCode &status) {
    int32_t length = 0;
    AffixTag tag;
    while (hasNext(tag, affixPattern)) {
        tag = nextToken(tag, affixPattern, status);
        if (U_FAILURE(status)) { return length; }
        if (tag.type == TYPE_CURRENCY_OVERFLOW) {
            length += 1;
        } else if (tag.type < 0) {
            length += provider.getSymbol(tag.type).length();
        } else {
            length += U16_LENGTH(tag.codePoint);
        }
    }
    return length;
}

bool
AffixUtils::containsType(const UnicodeString &affixPattern, AffixPatternType type, UErrorCode &status) {
    if (affixPattern.length() == 0) {
        return false;
    }
    AffixTag tag;
    while (hasNext(tag, affixPattern)) {
        tag = nextToken(tag, affixPattern, status);
        if (U_FAILURE(status)) { return false; }
        if (tag.type == type) {
            return true;
        }
    }
    return false;
}

bool AffixUtils::hasCurrencySymbols(const UnicodeString &affixPattern, UErrorCode &status) {
    if (affixPattern.length() == 0) {
        return false;
    }
    AffixTag tag;
    while (hasNext(tag, affixPattern)) {
        tag = nextToken(tag, affixPattern, status);
        if (U_FAILURE(status)) { return false; }
        if (tag.type < 0 && getFieldForType(tag.type) == UNUM_CURRENCY_FIELD) {
            return true;
        }
    }
    return false;
}

UnicodeString AffixUtils::replaceType(const UnicodeString &affixPattern, AffixPatternType type,
                                      char16_t replacementChar, UErrorCode &status) {
    UnicodeString output(affixPattern); // copy
    if (affixPattern.length() == 0) {
        return output;
    };
    AffixTag tag;
    while (hasNext(tag, affixPattern)) {
        tag = nextToken(tag, affixPattern, status);
        if (U_FAILURE(status)) { return output; }
        if (tag.type == type) {
            output.replace(tag.offset - 1, 1, replacementChar);
        }
    }
    return output;
}

bool AffixUtils::containsOnlySymbolsAndIgnorables(const UnicodeString& affixPattern,
                                                  const UnicodeSet& ignorables, UErrorCode& status) {
    if (affixPattern.length() == 0) {
        return true;
    };
    AffixTag tag;
    while (hasNext(tag, affixPattern)) {
        tag = nextToken(tag, affixPattern, status);
        if (U_FAILURE(status)) { return false; }
        if (tag.type == TYPE_CODEPOINT && !ignorables.contains(tag.codePoint)) {
            return false;
        }
    }
    return true;
}

void AffixUtils::iterateWithConsumer(const UnicodeString& affixPattern, TokenConsumer& consumer,
                                     UErrorCode& status) {
    if (affixPattern.length() == 0) {
        return;
    };
    AffixTag tag;
    while (hasNext(tag, affixPattern)) {
        tag = nextToken(tag, affixPattern, status);
        if (U_FAILURE(status)) { return; }
        consumer.consumeToken(tag.type, tag.codePoint, status);
        if (U_FAILURE(status)) { return; }
    }
}

AffixTag AffixUtils::nextToken(AffixTag tag, const UnicodeString &patternString, UErrorCode &status) {
    int32_t offset = tag.offset;
    int32_t state = tag.state;
    for (; offset < patternString.length();) {
        UChar32 cp = patternString.char32At(offset);
        int32_t count = U16_LENGTH(cp);

        switch (state) {
            case STATE_BASE:
                switch (cp) {
                    case u'\'':
                        state = STATE_FIRST_QUOTE;
                        offset += count;
                        // continue to the next code point
                        break;
                    case u'-':
                        return makeTag(offset + count, TYPE_MINUS_SIGN, STATE_BASE, 0);
                    case u'+':
                        return makeTag(offset + count, TYPE_PLUS_SIGN, STATE_BASE, 0);
                    case u'%':
                        return makeTag(offset + count, TYPE_PERCENT, STATE_BASE, 0);
                    case u'‰':
                        return makeTag(offset + count, TYPE_PERMILLE, STATE_BASE, 0);
                    case u'¤':
                        state = STATE_FIRST_CURR;
                        offset += count;
                        // continue to the next code point
                        break;
                    default:
                        return makeTag(offset + count, TYPE_CODEPOINT, STATE_BASE, cp);
                }
                break;
            case STATE_FIRST_QUOTE:
                if (cp == u'\'') {
                    return makeTag(offset + count, TYPE_CODEPOINT, STATE_BASE, cp);
                } else {
                    return makeTag(offset + count, TYPE_CODEPOINT, STATE_INSIDE_QUOTE, cp);
                }
            case STATE_INSIDE_QUOTE:
                if (cp == u'\'') {
                    state = STATE_AFTER_QUOTE;
                    offset += count;
                    // continue to the next code point
                    break;
                } else {
                    return makeTag(offset + count, TYPE_CODEPOINT, STATE_INSIDE_QUOTE, cp);
                }
            case STATE_AFTER_QUOTE:
                if (cp == u'\'') {
                    return makeTag(offset + count, TYPE_CODEPOINT, STATE_INSIDE_QUOTE, cp);
                } else {
                    state = STATE_BASE;
                    // re-evaluate this code point
                    break;
                }
            case STATE_FIRST_CURR:
                if (cp == u'¤') {
                    state = STATE_SECOND_CURR;
                    offset += count;
                    // continue to the next code point
                    break;
                } else {
                    return makeTag(offset, TYPE_CURRENCY_SINGLE, STATE_BASE, 0);
                }
            case STATE_SECOND_CURR:
                if (cp == u'¤') {
                    state = STATE_THIRD_CURR;
                    offset += count;
                    // continue to the next code point
                    break;
                } else {
                    return makeTag(offset, TYPE_CURRENCY_DOUBLE, STATE_BASE, 0);
                }
            case STATE_THIRD_CURR:
                if (cp == u'¤') {
                    state = STATE_FOURTH_CURR;
                    offset += count;
                    // continue to the next code point
                    break;
                } else {
                    return makeTag(offset, TYPE_CURRENCY_TRIPLE, STATE_BASE, 0);
                }
            case STATE_FOURTH_CURR:
                if (cp == u'¤') {
                    state = STATE_FIFTH_CURR;
                    offset += count;
                    // continue to the next code point
                    break;
                } else {
                    return makeTag(offset, TYPE_CURRENCY_QUAD, STATE_BASE, 0);
                }
            case STATE_FIFTH_CURR:
                if (cp == u'¤') {
                    state = STATE_OVERFLOW_CURR;
                    offset += count;
                    // continue to the next code point
                    break;
                } else {
                    return makeTag(offset, TYPE_CURRENCY_QUINT, STATE_BASE, 0);
                }
            case STATE_OVERFLOW_CURR:
                if (cp == u'¤') {
                    offset += count;
                    // continue to the next code point and loop back to this state
                    break;
                } else {
                    return makeTag(offset, TYPE_CURRENCY_OVERFLOW, STATE_BASE, 0);
                }
            default:
                U_ASSERT(false);
        }
    }
    // End of string
    switch (state) {
        case STATE_BASE:
            // No more tokens in string.
            return {-1};
        case STATE_FIRST_QUOTE:
        case STATE_INSIDE_QUOTE:
            // For consistent behavior with the JDK and ICU 58, set an error here.
            status = U_ILLEGAL_ARGUMENT_ERROR;
            return {-1};
        case STATE_AFTER_QUOTE:
            // No more tokens in string.
            return {-1};
        case STATE_FIRST_CURR:
            return makeTag(offset, TYPE_CURRENCY_SINGLE, STATE_BASE, 0);
        case STATE_SECOND_CURR:
            return makeTag(offset, TYPE_CURRENCY_DOUBLE, STATE_BASE, 0);
        case STATE_THIRD_CURR:
            return makeTag(offset, TYPE_CURRENCY_TRIPLE, STATE_BASE, 0);
        case STATE_FOURTH_CURR:
            return makeTag(offset, TYPE_CURRENCY_QUAD, STATE_BASE, 0);
        case STATE_FIFTH_CURR:
            return makeTag(offset, TYPE_CURRENCY_QUINT, STATE_BASE, 0);
        case STATE_OVERFLOW_CURR:
            return makeTag(offset, TYPE_CURRENCY_OVERFLOW, STATE_BASE, 0);
        default:
            U_ASSERT(false);
            return {-1}; // suppress "control reaches end of non-void function"
    }
}

bool AffixUtils::hasNext(const AffixTag &tag, const UnicodeString &string) {
    // First check for the {-1} and default initializer syntax.
    if (tag.offset < 0) {
        return false;
    } else if (tag.offset == 0) {
        return string.length() > 0;
    }
    // The rest of the fields are safe to use now.
    // Special case: the last character in string is an end quote.
    if (tag.state == STATE_INSIDE_QUOTE && tag.offset == string.length() - 1 &&
        string.charAt(tag.offset) == u'\'') {
        return false;
    } else if (tag.state != STATE_BASE) {
        return true;
    } else {
        return tag.offset < string.length();
    }
}

#endif /* #if !UCONFIG_NO_FORMATTING */

Coverage Report

Created: 2018-09-25 14:53

Line	Count	Source (jump to first uncovered line)
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
8		#include "number_affixutils.h"
9		#include "unicode/utf16.h"
10		#include "unicode/uniset.h"
11
12		using namespace icu;
13		using namespace icu::number;
14		using namespace icu::number::impl;
15
16	0	TokenConsumer::~TokenConsumer() = default;
17	0	SymbolProvider::~SymbolProvider() = default;
18
19	0	int32_t AffixUtils::estimateLength(const UnicodeString &patternString, UErrorCode &status) {
20	0	AffixPatternState state = STATE_BASE;
21	0	int32_t offset = 0;
22	0	int32_t length = 0;
23	0	for (; offset < patternString.length();) {
24	0	UChar32 cp = patternString.char32At(offset);
25	0
26	0	switch (state) {
27	0	case STATE_BASE:
28	0	if (cp == u'\'') {
29	0	// First quote
30	0	state = STATE_FIRST_QUOTE;
31	0	} else {
32	0	// Unquoted symbol
33	0	length++;
34	0	}
35	0	break;
36	0	case STATE_FIRST_QUOTE:
37	0	if (cp == u'\'') {
38	0	// Repeated quote
39	0	length++;
40	0	state = STATE_BASE;
41	0	} else {
42	0	// Quoted code point
43	0	length++;
44	0	state = STATE_INSIDE_QUOTE;
45	0	}
46	0	break;
47	0	case STATE_INSIDE_QUOTE:
48	0	if (cp == u'\'') {
49	0	// End of quoted sequence
50	0	state = STATE_AFTER_QUOTE;
51	0	} else {
52	0	// Quoted code point
53	0	length++;
54	0	}
55	0	break;
56	0	case STATE_AFTER_QUOTE:
57	0	if (cp == u'\'') {
58	0	// Double quote inside of quoted sequence
59	0	length++;
60	0	state = STATE_INSIDE_QUOTE;
61	0	} else {
62	0	// Unquoted symbol
63	0	length++;
64	0	}
65	0	break;
66	0	default:
67	0	U_ASSERT(false);
68	0	}
69	0
70	0	offset += U16_LENGTH(cp);
71	0	}
72	0
73	0	switch (state) {
74	0	case STATE_FIRST_QUOTE:
75	0	case STATE_INSIDE_QUOTE:
76	0	status = U_ILLEGAL_ARGUMENT_ERROR;
77	0	break;
78	0	default:
79	0	break;
80	0	}
81	0
82	0	return length;
83	0	}
84
85	0	UnicodeString AffixUtils::escape(const UnicodeString &input) {
86	0	AffixPatternState state = STATE_BASE;
87	0	int32_t offset = 0;
88	0	UnicodeString output;
89	0	for (; offset < input.length();) {
90	0	UChar32 cp = input.char32At(offset);
91	0
92	0	switch (cp) {
93	0	case u'\'':
94	0	output.append(u"''", -1);
95	0	break;
96	0
97	0	case u'-':
98	0	case u'+':
99	0	case u'%':
100	0	case u'‰':
101	0	case u'¤':
102	0	if (state == STATE_BASE) {
103	0	output.append(u'\'');
104	0	output.append(cp);
105	0	state = STATE_INSIDE_QUOTE;
106	0	} else {
107	0	output.append(cp);
108	0	}
109	0	break;
110	0
111	0	default:
112	0	if (state == STATE_INSIDE_QUOTE) {
113	0	output.append(u'\'');
114	0	output.append(cp);
115	0	state = STATE_BASE;
116	0	} else {
117	0	output.append(cp);
118	0	}
119	0	break;
120	0	}
121	0	offset += U16_LENGTH(cp);
122	0	}
123	0
124	0	if (state == STATE_INSIDE_QUOTE) {
125	0	output.append(u'\'');
126	0	}
127	0
128	0	return output;
129	0	}
130
131		Field AffixUtils::getFieldForType(AffixPatternType type) {
132		switch (type) {
133		case TYPE_MINUS_SIGN:
134		return Field::UNUM_SIGN_FIELD;
135		case TYPE_PLUS_SIGN:
136		return Field::UNUM_SIGN_FIELD;
137		case TYPE_PERCENT:
138		return Field::UNUM_PERCENT_FIELD;
139		case TYPE_PERMILLE:
140		return Field::UNUM_PERMILL_FIELD;
141		case TYPE_CURRENCY_SINGLE:
142		return Field::UNUM_CURRENCY_FIELD;
143		case TYPE_CURRENCY_DOUBLE:
144		return Field::UNUM_CURRENCY_FIELD;
145		case TYPE_CURRENCY_TRIPLE:
146		return Field::UNUM_CURRENCY_FIELD;
147		case TYPE_CURRENCY_QUAD:
148		return Field::UNUM_CURRENCY_FIELD;
149		case TYPE_CURRENCY_QUINT:
150		return Field::UNUM_CURRENCY_FIELD;
151		case TYPE_CURRENCY_OVERFLOW:
152		return Field::UNUM_CURRENCY_FIELD;
153		default:
154		U_ASSERT(false);
155		return Field::UNUM_FIELD_COUNT; // suppress "control reaches end of non-void function"
156		}
157		}
158
159		int32_t
160		AffixUtils::unescape(const UnicodeString &affixPattern, NumberStringBuilder &output, int32_t position,
161	0	const SymbolProvider &provider, UErrorCode &status) {
162	0	int32_t length = 0;
163	0	AffixTag tag;
164	0	while (hasNext(tag, affixPattern)) {
165	0	tag = nextToken(tag, affixPattern, status);
166	0	if (U_FAILURE(status)) { return length; }
167	0	if (tag.type == TYPE_CURRENCY_OVERFLOW) {
168	0	// Don't go to the provider for this special case
169	0	length += output.insertCodePoint(position + length, 0xFFFD, UNUM_CURRENCY_FIELD, status);
170	0	} else if (tag.type < 0) {
171	0	length += output.insert(
172	0	position + length, provider.getSymbol(tag.type), getFieldForType(tag.type), status);
173	0	} else {
174	0	length += output.insertCodePoint(position + length, tag.codePoint, UNUM_FIELD_COUNT, status);
175	0	}
176	0	}
177	0	return length;
178	0	}
179
180		int32_t AffixUtils::unescapedCodePointCount(const UnicodeString &affixPattern,
181	0	const SymbolProvider &provider, UErrorCode &status) {
182	0	int32_t length = 0;
183	0	AffixTag tag;
184	0	while (hasNext(tag, affixPattern)) {
185	0	tag = nextToken(tag, affixPattern, status);
186	0	if (U_FAILURE(status)) { return length; }
187	0	if (tag.type == TYPE_CURRENCY_OVERFLOW) {
188	0	length += 1;
189	0	} else if (tag.type < 0) {
190	0	length += provider.getSymbol(tag.type).length();
191	0	} else {
192	0	length += U16_LENGTH(tag.codePoint);
193	0	}
194	0	}
195	0	return length;
196	0	}
197
198		bool
199	0	AffixUtils::containsType(const UnicodeString &affixPattern, AffixPatternType type, UErrorCode &status) {
200	0	if (affixPattern.length() == 0) {
201	0	return false;
202	0	}
203	0	AffixTag tag;
204	0	while (hasNext(tag, affixPattern)) {
205	0	tag = nextToken(tag, affixPattern, status);
206	0	if (U_FAILURE(status)) { return false; }
207	0	if (tag.type == type) {
208	0	return true;
209	0	}
210	0	}
211	0	return false;
212	0	}
213
214	0	bool AffixUtils::hasCurrencySymbols(const UnicodeString &affixPattern, UErrorCode &status) {
215	0	if (affixPattern.length() == 0) {
216	0	return false;
217	0	}
218	0	AffixTag tag;
219	0	while (hasNext(tag, affixPattern)) {
220	0	tag = nextToken(tag, affixPattern, status);
221	0	if (U_FAILURE(status)) { return false; }
222	0	if (tag.type < 0 && getFieldForType(tag.type) == UNUM_CURRENCY_FIELD) {
223	0	return true;
224	0	}
225	0	}
226	0	return false;
227	0	}
228
229		UnicodeString AffixUtils::replaceType(const UnicodeString &affixPattern, AffixPatternType type,
230	0	char16_t replacementChar, UErrorCode &status) {
231	0	UnicodeString output(affixPattern); // copy
232	0	if (affixPattern.length() == 0) {
233	0	return output;
234	0	};
235	0	AffixTag tag;
236	0	while (hasNext(tag, affixPattern)) {
237	0	tag = nextToken(tag, affixPattern, status);
238	0	if (U_FAILURE(status)) { return output; }
239	0	if (tag.type == type) {
240	0	output.replace(tag.offset - 1, 1, replacementChar);
241	0	}
242	0	}
243	0	return output;
244	0	}
245
246		bool AffixUtils::containsOnlySymbolsAndIgnorables(const UnicodeString& affixPattern,
247	0	const UnicodeSet& ignorables, UErrorCode& status) {
248	0	if (affixPattern.length() == 0) {
249	0	return true;
250	0	};
251	0	AffixTag tag;
252	0	while (hasNext(tag, affixPattern)) {
253	0	tag = nextToken(tag, affixPattern, status);
254	0	if (U_FAILURE(status)) { return false; }
255	0	if (tag.type == TYPE_CODEPOINT && !ignorables.contains(tag.codePoint)) {
256	0	return false;
257	0	}
258	0	}
259	0	return true;
260	0	}
261
262		void AffixUtils::iterateWithConsumer(const UnicodeString& affixPattern, TokenConsumer& consumer,
263	0	UErrorCode& status) {
264	0	if (affixPattern.length() == 0) {
265	0	return;
266	0	};
267	0	AffixTag tag;
268	0	while (hasNext(tag, affixPattern)) {
269	0	tag = nextToken(tag, affixPattern, status);
270	0	if (U_FAILURE(status)) { return; }
271	0	consumer.consumeToken(tag.type, tag.codePoint, status);
272	0	if (U_FAILURE(status)) { return; }
273	0	}
274	0	}
275
276	0	AffixTag AffixUtils::nextToken(AffixTag tag, const UnicodeString &patternString, UErrorCode &status) {
277	0	int32_t offset = tag.offset;
278	0	int32_t state = tag.state;
279	0	for (; offset < patternString.length();) {
280	0	UChar32 cp = patternString.char32At(offset);
281	0	int32_t count = U16_LENGTH(cp);
282	0
283	0	switch (state) {
284	0	case STATE_BASE:
285	0	switch (cp) {
286	0	case u'\'':
287	0	state = STATE_FIRST_QUOTE;
288	0	offset += count;
289	0	// continue to the next code point
290	0	break;
291	0	case u'-':
292	0	return makeTag(offset + count, TYPE_MINUS_SIGN, STATE_BASE, 0);
293	0	case u'+':
294	0	return makeTag(offset + count, TYPE_PLUS_SIGN, STATE_BASE, 0);
295	0	case u'%':
296	0	return makeTag(offset + count, TYPE_PERCENT, STATE_BASE, 0);
297	0	case u'‰':
298	0	return makeTag(offset + count, TYPE_PERMILLE, STATE_BASE, 0);
299	0	case u'¤':
300	0	state = STATE_FIRST_CURR;
301	0	offset += count;
302	0	// continue to the next code point
303	0	break;
304	0	default:
305	0	return makeTag(offset + count, TYPE_CODEPOINT, STATE_BASE, cp);
306	0	}
307	0	break;
308	0	case STATE_FIRST_QUOTE:
309	0	if (cp == u'\'') {
310	0	return makeTag(offset + count, TYPE_CODEPOINT, STATE_BASE, cp);
311	0	} else {
312	0	return makeTag(offset + count, TYPE_CODEPOINT, STATE_INSIDE_QUOTE, cp);
313	0	}
314	0	case STATE_INSIDE_QUOTE:
315	0	if (cp == u'\'') {
316	0	state = STATE_AFTER_QUOTE;
317	0	offset += count;
318	0	// continue to the next code point
319	0	break;
320	0	} else {
321	0	return makeTag(offset + count, TYPE_CODEPOINT, STATE_INSIDE_QUOTE, cp);
322	0	}
323	0	case STATE_AFTER_QUOTE:
324	0	if (cp == u'\'') {
325	0	return makeTag(offset + count, TYPE_CODEPOINT, STATE_INSIDE_QUOTE, cp);
326	0	} else {
327	0	state = STATE_BASE;
328	0	// re-evaluate this code point
329	0	break;
330	0	}
331	0	case STATE_FIRST_CURR:
332	0	if (cp == u'¤') {
333	0	state = STATE_SECOND_CURR;
334	0	offset += count;
335	0	// continue to the next code point
336	0	break;
337	0	} else {
338	0	return makeTag(offset, TYPE_CURRENCY_SINGLE, STATE_BASE, 0);
339	0	}
340	0	case STATE_SECOND_CURR:
341	0	if (cp == u'¤') {
342	0	state = STATE_THIRD_CURR;
343	0	offset += count;
344	0	// continue to the next code point
345	0	break;
346	0	} else {
347	0	return makeTag(offset, TYPE_CURRENCY_DOUBLE, STATE_BASE, 0);
348	0	}
349	0	case STATE_THIRD_CURR:
350	0	if (cp == u'¤') {
351	0	state = STATE_FOURTH_CURR;
352	0	offset += count;
353	0	// continue to the next code point
354	0	break;
355	0	} else {
356	0	return makeTag(offset, TYPE_CURRENCY_TRIPLE, STATE_BASE, 0);
357	0	}
358	0	case STATE_FOURTH_CURR:
359	0	if (cp == u'¤') {
360	0	state = STATE_FIFTH_CURR;
361	0	offset += count;
362	0	// continue to the next code point
363	0	break;
364	0	} else {
365	0	return makeTag(offset, TYPE_CURRENCY_QUAD, STATE_BASE, 0);
366	0	}
367	0	case STATE_FIFTH_CURR:
368	0	if (cp == u'¤') {
369	0	state = STATE_OVERFLOW_CURR;
370	0	offset += count;
371	0	// continue to the next code point
372	0	break;
373	0	} else {
374	0	return makeTag(offset, TYPE_CURRENCY_QUINT, STATE_BASE, 0);
375	0	}
376	0	case STATE_OVERFLOW_CURR:
377	0	if (cp == u'¤') {
378	0	offset += count;
379	0	// continue to the next code point and loop back to this state
380	0	break;
381	0	} else {
382	0	return makeTag(offset, TYPE_CURRENCY_OVERFLOW, STATE_BASE, 0);
383	0	}
384	0	default:
385	0	U_ASSERT(false);
386	0	}
387	0	}
388	0	// End of string
389	0	switch (state) {
390	0	case STATE_BASE:
391	0	// No more tokens in string.
392	0	return {-1};
393	0	case STATE_FIRST_QUOTE:
394	0	case STATE_INSIDE_QUOTE:
395	0	// For consistent behavior with the JDK and ICU 58, set an error here.
396	0	status = U_ILLEGAL_ARGUMENT_ERROR;
397	0	return {-1};
398	0	case STATE_AFTER_QUOTE:
399	0	// No more tokens in string.
400	0	return {-1};
401	0	case STATE_FIRST_CURR:
402	0	return makeTag(offset, TYPE_CURRENCY_SINGLE, STATE_BASE, 0);
403	0	case STATE_SECOND_CURR:
404	0	return makeTag(offset, TYPE_CURRENCY_DOUBLE, STATE_BASE, 0);
405	0	case STATE_THIRD_CURR:
406	0	return makeTag(offset, TYPE_CURRENCY_TRIPLE, STATE_BASE, 0);
407	0	case STATE_FOURTH_CURR:
408	0	return makeTag(offset, TYPE_CURRENCY_QUAD, STATE_BASE, 0);
409	0	case STATE_FIFTH_CURR:
410	0	return makeTag(offset, TYPE_CURRENCY_QUINT, STATE_BASE, 0);
411	0	case STATE_OVERFLOW_CURR:
412	0	return makeTag(offset, TYPE_CURRENCY_OVERFLOW, STATE_BASE, 0);
413	0	default:
414	0	U_ASSERT(false);
415	0	return {-1}; // suppress "control reaches end of non-void function"
416	0	}
417	0	}
418
419	0	bool AffixUtils::hasNext(const AffixTag &tag, const UnicodeString &string) {
420	0	// First check for the {-1} and default initializer syntax.
421	0	if (tag.offset < 0) {
422	0	return false;
423	0	} else if (tag.offset == 0) {
424	0	return string.length() > 0;
425	0	}
426	0	// The rest of the fields are safe to use now.
427	0	// Special case: the last character in string is an end quote.
428	0	if (tag.state == STATE_INSIDE_QUOTE && tag.offset == string.length() - 1 &&
429	0	string.charAt(tag.offset) == u'\'') {
430	0	return false;
431	0	} else if (tag.state != STATE_BASE) {
432	0	return true;
433	0	} else {
434	0	return tag.offset < string.length();
435	0	}
436	0	}
437
438		#endif /* #if !UCONFIG_NO_FORMATTING */