/src/icu/icu4c/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:
                UPRV_UNREACHABLE_EXIT;
        }

        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 {UFIELD_CATEGORY_NUMBER, UNUM_SIGN_FIELD};
        case TYPE_PLUS_SIGN:
            return {UFIELD_CATEGORY_NUMBER, UNUM_SIGN_FIELD};
        case TYPE_APPROXIMATELY_SIGN:
            return {UFIELD_CATEGORY_NUMBER, UNUM_APPROXIMATELY_SIGN_FIELD};
        case TYPE_PERCENT:
            return {UFIELD_CATEGORY_NUMBER, UNUM_PERCENT_FIELD};
        case TYPE_PERMILLE:
            return {UFIELD_CATEGORY_NUMBER, UNUM_PERMILL_FIELD};
        case TYPE_CURRENCY_SINGLE:
            return {UFIELD_CATEGORY_NUMBER, UNUM_CURRENCY_FIELD};
        case TYPE_CURRENCY_DOUBLE:
            return {UFIELD_CATEGORY_NUMBER, UNUM_CURRENCY_FIELD};
        case TYPE_CURRENCY_TRIPLE:
            return {UFIELD_CATEGORY_NUMBER, UNUM_CURRENCY_FIELD};
        case TYPE_CURRENCY_QUAD:
            return {UFIELD_CATEGORY_NUMBER, UNUM_CURRENCY_FIELD};
        case TYPE_CURRENCY_QUINT:
            return {UFIELD_CATEGORY_NUMBER, UNUM_CURRENCY_FIELD};
        case TYPE_CURRENCY_OVERFLOW:
            return {UFIELD_CATEGORY_NUMBER, UNUM_CURRENCY_FIELD};
        default:
            UPRV_UNREACHABLE_EXIT;
    }
}

int32_t
AffixUtils::unescape(const UnicodeString &affixPattern, FormattedStringBuilder &output, int32_t position,
                     const SymbolProvider &provider, Field field, 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,
                {UFIELD_CATEGORY_NUMBER, 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, field, 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) == Field(UFIELD_CATEGORY_NUMBER, 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_APPROXIMATELY_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:
                UPRV_UNREACHABLE_EXIT;
        }
    }
    // 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:
            UPRV_UNREACHABLE_EXIT;
    }
}

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: 2025-09-05 07:16

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	140k	TokenConsumer::~TokenConsumer() = default;
17	108k	SymbolProvider::~SymbolProvider() = default;
18
19	3.28k	int32_t AffixUtils::estimateLength(const UnicodeString &patternString, UErrorCode &status) {
20	3.28k	AffixPatternState state = STATE_BASE;
21	3.28k	int32_t offset = 0;
22	3.28k	int32_t length = 0;
23	3.76M	for (; offset < patternString.length();) {
24	3.75M	UChar32 cp = patternString.char32At(offset);
25
26	3.75M	switch (state) {
27	3.71M	case STATE_BASE:
28	3.71M	if (cp == u'\'') {
29		// First quote
30	189	state = STATE_FIRST_QUOTE;
31	3.71M	} else {
32		// Unquoted symbol
33	3.71M	length++;
34	3.71M	}
35	3.71M	break;
36	189	case STATE_FIRST_QUOTE:
37	189	if (cp == u'\'') {
38		// Repeated quote
39	124	length++;
40	124	state = STATE_BASE;
41	124	} else {
42		// Quoted code point
43	65	length++;
44	65	state = STATE_INSIDE_QUOTE;
45	65	}
46	189	break;
47	42.8k	case STATE_INSIDE_QUOTE:
48	42.8k	if (cp == u'\'') {
49		// End of quoted sequence
50	316	state = STATE_AFTER_QUOTE;
51	42.5k	} else {
52		// Quoted code point
53	42.5k	length++;
54	42.5k	}
55	42.8k	break;
56	4.70k	case STATE_AFTER_QUOTE:
57	4.70k	if (cp == u'\'') {
58		// Double quote inside of quoted sequence
59	251	length++;
60	251	state = STATE_INSIDE_QUOTE;
61	4.44k	} else {
62		// Unquoted symbol
63	4.44k	length++;
64	4.44k	}
65	4.70k	break;
66	0	default:
67	0	UPRV_UNREACHABLE_EXIT;
68	3.75M	}
69
70	3.75M	offset += U16_LENGTH(cp);
71	3.75M	}
72
73	3.28k	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	3.28k	default:
79	3.28k	break;
80	3.28k	}
81
82	3.28k	return length;
83	3.28k	}
84
85	2.08M	UnicodeString AffixUtils::escape(const UnicodeString &input) {
86	2.08M	AffixPatternState state = STATE_BASE;
87	2.08M	int32_t offset = 0;
88	2.08M	UnicodeString output;
89	2.08M	for (; offset < input.length();) {
90	0	UChar32 cp = input.char32At(offset);
91
92	0	switch (cp) {
93	0	case u'\'':
94	0	output.append(u"''", -1);
95	0	break;
96
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
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
124	2.08M	if (state == STATE_INSIDE_QUOTE) {
125	0	output.append(u'\'');
126	0	}
127
128	2.08M	return output;
129	2.08M	}
130
131	262k	Field AffixUtils::getFieldForType(AffixPatternType type) {
132	262k	switch (type) {
133	106k	case TYPE_MINUS_SIGN:
134	106k	return {UFIELD_CATEGORY_NUMBER, UNUM_SIGN_FIELD};
135	107k	case TYPE_PLUS_SIGN:
136	107k	return {UFIELD_CATEGORY_NUMBER, UNUM_SIGN_FIELD};
137	462	case TYPE_APPROXIMATELY_SIGN:
138	462	return {UFIELD_CATEGORY_NUMBER, UNUM_APPROXIMATELY_SIGN_FIELD};
139	19.7k	case TYPE_PERCENT:
140	19.7k	return {UFIELD_CATEGORY_NUMBER, UNUM_PERCENT_FIELD};
141	1.80k	case TYPE_PERMILLE:
142	1.80k	return {UFIELD_CATEGORY_NUMBER, UNUM_PERMILL_FIELD};
143	25.4k	case TYPE_CURRENCY_SINGLE:
144	25.4k	return {UFIELD_CATEGORY_NUMBER, UNUM_CURRENCY_FIELD};
145	446	case TYPE_CURRENCY_DOUBLE:
146	446	return {UFIELD_CATEGORY_NUMBER, UNUM_CURRENCY_FIELD};
147	461	case TYPE_CURRENCY_TRIPLE:
148	461	return {UFIELD_CATEGORY_NUMBER, UNUM_CURRENCY_FIELD};
149	58	case TYPE_CURRENCY_QUAD:
150	58	return {UFIELD_CATEGORY_NUMBER, UNUM_CURRENCY_FIELD};
151	30	case TYPE_CURRENCY_QUINT:
152	30	return {UFIELD_CATEGORY_NUMBER, UNUM_CURRENCY_FIELD};
153	0	case TYPE_CURRENCY_OVERFLOW:
154	0	return {UFIELD_CATEGORY_NUMBER, UNUM_CURRENCY_FIELD};
155	0	default:
156	0	UPRV_UNREACHABLE_EXIT;
157	262k	}
158	262k	}
159
160		int32_t
161		AffixUtils::unescape(const UnicodeString &affixPattern, FormattedStringBuilder &output, int32_t position,
162	809k	const SymbolProvider &provider, Field field, UErrorCode &status) {
163	809k	int32_t length = 0;
164	809k	AffixTag tag;
165	1.01M	while (hasNext(tag, affixPattern)) {
166	206k	tag = nextToken(tag, affixPattern, status);
167	206k	if (U_FAILURE(status)) { return length; }
168	206k	if (tag.type == TYPE_CURRENCY_OVERFLOW) {
169		// Don't go to the provider for this special case
170	0	length += output.insertCodePoint(
171	0	position + length,
172	0	0xFFFD,
173	0	{UFIELD_CATEGORY_NUMBER, UNUM_CURRENCY_FIELD},
174	0	status);
175	206k	} else if (tag.type < 0) {
176	200k	length += output.insert(
177	200k	position + length, provider.getSymbol(tag.type), getFieldForType(tag.type), status);
178	200k	} else {
179	6.39k	length += output.insertCodePoint(position + length, tag.codePoint, field, status);
180	6.39k	}
181	206k	}
182	809k	return length;
183	809k	}
184
185		int32_t AffixUtils::unescapedCodePointCount(const UnicodeString &affixPattern,
186	0	const SymbolProvider &provider, UErrorCode &status) {
187	0	int32_t length = 0;
188	0	AffixTag tag;
189	0	while (hasNext(tag, affixPattern)) {
190	0	tag = nextToken(tag, affixPattern, status);
191	0	if (U_FAILURE(status)) { return length; }
192	0	if (tag.type == TYPE_CURRENCY_OVERFLOW) {
193	0	length += 1;
194	0	} else if (tag.type < 0) {
195	0	length += provider.getSymbol(tag.type).length();
196	0	} else {
197	0	length += U16_LENGTH(tag.codePoint);
198	0	}
199	0	}
200	0	return length;
201	0	}
202
203		bool
204	36.5k	AffixUtils::containsType(const UnicodeString &affixPattern, AffixPatternType type, UErrorCode &status) {
205	36.5k	if (affixPattern.length() == 0) {
206	14.7k	return false;
207	14.7k	}
208	21.8k	AffixTag tag;
209	3.42M	while (hasNext(tag, affixPattern)) {
210	3.40M	tag = nextToken(tag, affixPattern, status);
211	3.40M	if (U_FAILURE(status)) { return false; }
212	3.40M	if (tag.type == type) {
213	1.32k	return true;
214	1.32k	}
215	3.40M	}
216	20.4k	return false;
217	21.8k	}
218
219	2.02M	bool AffixUtils::hasCurrencySymbols(const UnicodeString &affixPattern, UErrorCode &status) {
220	2.02M	if (affixPattern.length() == 0) {
221	1.95M	return false;
222	1.95M	}
223	68.8k	AffixTag tag;
224	60.7M	while (hasNext(tag, affixPattern)) {
225	60.7M	tag = nextToken(tag, affixPattern, status);
226	60.7M	if (U_FAILURE(status)) { return false; }
227	60.7M	if (tag.type < 0 && getFieldForType(tag.type) == Field(UFIELD_CATEGORY_NUMBER, UNUM_CURRENCY_FIELD)) {
228	26.4k	return true;
229	26.4k	}
230	60.7M	}
231	42.4k	return false;
232	68.8k	}
233
234		UnicodeString AffixUtils::replaceType(const UnicodeString &affixPattern, AffixPatternType type,
235	0	char16_t replacementChar, UErrorCode &status) {
236	0	UnicodeString output(affixPattern); // copy
237	0	if (affixPattern.length() == 0) {
238	0	return output;
239	0	}
240	0	AffixTag tag;
241	0	while (hasNext(tag, affixPattern)) {
242	0	tag = nextToken(tag, affixPattern, status);
243	0	if (U_FAILURE(status)) { return output; }
244	0	if (tag.type == type) {
245	0	output.replace(tag.offset - 1, 1, replacementChar);
246	0	}
247	0	}
248	0	return output;
249	0	}
250
251		bool AffixUtils::containsOnlySymbolsAndIgnorables(const UnicodeString& affixPattern,
252	0	const UnicodeSet& ignorables, UErrorCode& status) {
253	0	if (affixPattern.length() == 0) {
254	0	return true;
255	0	}
256	0	AffixTag tag;
257	0	while (hasNext(tag, affixPattern)) {
258	0	tag = nextToken(tag, affixPattern, status);
259	0	if (U_FAILURE(status)) { return false; }
260	0	if (tag.type == TYPE_CODEPOINT && !ignorables.contains(tag.codePoint)) {
261	0	return false;
262	0	}
263	0	}
264	0	return true;
265	0	}
266
267		void AffixUtils::iterateWithConsumer(const UnicodeString& affixPattern, TokenConsumer& consumer,
268	140k	UErrorCode& status) {
269	140k	if (affixPattern.length() == 0) {
270	0	return;
271	0	}
272	140k	AffixTag tag;
273	60.1M	while (hasNext(tag, affixPattern)) {
274	59.9M	tag = nextToken(tag, affixPattern, status);
275	59.9M	if (U_FAILURE(status)) { return; }
276	59.9M	consumer.consumeToken(tag.type, tag.codePoint, status);
277	59.9M	if (U_FAILURE(status)) { return; }
278	59.9M	}
279	140k	}
280
281	124M	AffixTag AffixUtils::nextToken(AffixTag tag, const UnicodeString &patternString, UErrorCode &status) {
282	124M	int32_t offset = tag.offset;
283	124M	int32_t state = tag.state;
284	125M	for (; offset < patternString.length();) {
285	125M	UChar32 cp = patternString.char32At(offset);
286	125M	int32_t count = U16_LENGTH(cp);
287
288	125M	switch (state) {
289	120M	case STATE_BASE:
290	120M	switch (cp) {
291	414k	case u'\'':
292	414k	state = STATE_FIRST_QUOTE;
293	414k	offset += count;
294		// continue to the next code point
295	414k	break;
296	204k	case u'-':
297	204k	return makeTag(offset + count, TYPE_MINUS_SIGN, STATE_BASE, 0);
298	116k	case u'+':
299	116k	return makeTag(offset + count, TYPE_PLUS_SIGN, STATE_BASE, 0);
300	462	case u'~':
301	462	return makeTag(offset + count, TYPE_APPROXIMATELY_SIGN, STATE_BASE, 0);
302	39.8k	case u'%':
303	39.8k	return makeTag(offset + count, TYPE_PERCENT, STATE_BASE, 0);
304	3.45k	case u'‰':
305	3.45k	return makeTag(offset + count, TYPE_PERMILLE, STATE_BASE, 0);
306	60.1k	case u'¤':
307	60.1k	state = STATE_FIRST_CURR;
308	60.1k	offset += count;
309		// continue to the next code point
310	60.1k	break;
311	119M	default:
312	119M	return makeTag(offset + count, TYPE_CODEPOINT, STATE_BASE, cp);
313	120M	}
314	474k	break;
315	474k	case STATE_FIRST_QUOTE:
316	414k	if (cp == u'\'') {
317	350k	return makeTag(offset + count, TYPE_CODEPOINT, STATE_BASE, cp);
318	350k	} else {
319	63.5k	return makeTag(offset + count, TYPE_CODEPOINT, STATE_INSIDE_QUOTE, cp);
320	63.5k	}
321	3.91M	case STATE_INSIDE_QUOTE:
322	3.91M	if (cp == u'\'') {
323	418k	state = STATE_AFTER_QUOTE;
324	418k	offset += count;
325		// continue to the next code point
326	418k	break;
327	3.50M	} else {
328	3.50M	return makeTag(offset + count, TYPE_CODEPOINT, STATE_INSIDE_QUOTE, cp);
329	3.50M	}
330	418k	case STATE_AFTER_QUOTE:
331	418k	if (cp == u'\'') {
332	355k	return makeTag(offset + count, TYPE_CODEPOINT, STATE_INSIDE_QUOTE, cp);
333	355k	} else {
334	62.7k	state = STATE_BASE;
335		// re-evaluate this code point
336	62.7k	break;
337	62.7k	}
338	25.8k	case STATE_FIRST_CURR:
339	25.8k	if (cp == u'¤') {
340	4.57k	state = STATE_SECOND_CURR;
341	4.57k	offset += count;
342		// continue to the next code point
343	4.57k	break;
344	21.3k	} else {
345	21.3k	return makeTag(offset, TYPE_CURRENCY_SINGLE, STATE_BASE, 0);
346	21.3k	}
347	4.05k	case STATE_SECOND_CURR:
348	4.05k	if (cp == u'¤') {
349	2.65k	state = STATE_THIRD_CURR;
350	2.65k	offset += count;
351		// continue to the next code point
352	2.65k	break;
353	2.65k	} else {
354	1.40k	return makeTag(offset, TYPE_CURRENCY_DOUBLE, STATE_BASE, 0);
355	1.40k	}
356	1.82k	case STATE_THIRD_CURR:
357	1.82k	if (cp == u'¤') {
358	760	state = STATE_FOURTH_CURR;
359	760	offset += count;
360		// continue to the next code point
361	760	break;
362	1.06k	} else {
363	1.06k	return makeTag(offset, TYPE_CURRENCY_TRIPLE, STATE_BASE, 0);
364	1.06k	}
365	678	case STATE_FOURTH_CURR:
366	678	if (cp == u'¤') {
367	432	state = STATE_FIFTH_CURR;
368	432	offset += count;
369		// continue to the next code point
370	432	break;
371	432	} else {
372	246	return makeTag(offset, TYPE_CURRENCY_QUAD, STATE_BASE, 0);
373	246	}
374	419	case STATE_FIFTH_CURR:
375	419	if (cp == u'¤') {
376	0	state = STATE_OVERFLOW_CURR;
377	0	offset += count;
378		// continue to the next code point
379	0	break;
380	419	} else {
381	419	return makeTag(offset, TYPE_CURRENCY_QUINT, STATE_BASE, 0);
382	419	}
383	0	case STATE_OVERFLOW_CURR:
384	0	if (cp == u'¤') {
385	0	offset += count;
386		// continue to the next code point and loop back to this state
387	0	break;
388	0	} else {
389	0	return makeTag(offset, TYPE_CURRENCY_OVERFLOW, STATE_BASE, 0);
390	0	}
391	0	default:
392	0	UPRV_UNREACHABLE_EXIT;
393	125M	}
394	125M	}
395		// End of string
396	35.7k	switch (state) {
397	0	case STATE_BASE:
398		// No more tokens in string.
399	0	return {-1};
400	0	case STATE_FIRST_QUOTE:
401	0	case STATE_INSIDE_QUOTE:
402		// For consistent behavior with the JDK and ICU 58, set an error here.
403	0	status = U_ILLEGAL_ARGUMENT_ERROR;
404	0	return {-1};
405	0	case STATE_AFTER_QUOTE:
406		// No more tokens in string.
407	0	return {-1};
408	34.2k	case STATE_FIRST_CURR:
409	34.2k	return makeTag(offset, TYPE_CURRENCY_SINGLE, STATE_BASE, 0);
410	518	case STATE_SECOND_CURR:
411	518	return makeTag(offset, TYPE_CURRENCY_DOUBLE, STATE_BASE, 0);
412	824	case STATE_THIRD_CURR:
413	824	return makeTag(offset, TYPE_CURRENCY_TRIPLE, STATE_BASE, 0);
414	82	case STATE_FOURTH_CURR:
415	82	return makeTag(offset, TYPE_CURRENCY_QUAD, STATE_BASE, 0);
416	13	case STATE_FIFTH_CURR:
417	13	return makeTag(offset, TYPE_CURRENCY_QUINT, STATE_BASE, 0);
418	0	case STATE_OVERFLOW_CURR:
419	0	return makeTag(offset, TYPE_CURRENCY_OVERFLOW, STATE_BASE, 0);
420	0	default:
421	0	UPRV_UNREACHABLE_EXIT;
422	35.7k	}
423	35.7k	}
424
425	125M	bool AffixUtils::hasNext(const AffixTag &tag, const UnicodeString &string) {
426		// First check for the {-1} and default initializer syntax.
427	125M	if (tag.offset < 0) {
428	0	return false;
429	125M	} else if (tag.offset == 0) {
430	1.04M	return string.length() > 0;
431	1.04M	}
432		// The rest of the fields are safe to use now.
433		// Special case: the last character in string is an end quote.
434	124M	if (tag.state == STATE_INSIDE_QUOTE && tag.offset == string.length() - 1 &&
435	124M	string.charAt(tag.offset) == u'\'') {
436	730	return false;
437	124M	} else if (tag.state != STATE_BASE) {
438	3.91M	return true;
439	120M	} else {
440	120M	return tag.offset < string.length();
441	120M	}
442	124M	}
443
444		#endif /* #if !UCONFIG_NO_FORMATTING */