Coverage Report

Created: 2023-02-22 06:51

/src/icu/source/common/simpleformatter.cpp
Line
Count
Source (jump to first uncovered line)
1
// © 2016 and later: Unicode, Inc. and others.
2
// License & terms of use: http://www.unicode.org/copyright.html
3
/*
4
******************************************************************************
5
* Copyright (C) 2014-2016, International Business Machines
6
* Corporation and others.  All Rights Reserved.
7
******************************************************************************
8
* simpleformatter.cpp
9
*/
10
11
#include "unicode/utypes.h"
12
#include "unicode/simpleformatter.h"
13
#include "unicode/unistr.h"
14
#include "uassert.h"
15
16
U_NAMESPACE_BEGIN
17
18
namespace {
19
20
/**
21
 * Argument numbers must be smaller than this limit.
22
 * Text segment lengths are offset by this much.
23
 * This is currently the only unused char value in compiled patterns,
24
 * except it is the maximum value of the first unit (max arg +1).
25
 */
26
const int32_t ARG_NUM_LIMIT = 0x100;
27
/**
28
 * Initial and maximum char/UChar value set for a text segment.
29
 * Segment length char values are from ARG_NUM_LIMIT+1 to this value here.
30
 * Normally 0xffff, but can be as small as ARG_NUM_LIMIT+1 for testing.
31
 */
32
const UChar SEGMENT_LENGTH_PLACEHOLDER_CHAR = 0xffff;
33
/**
34
 * Maximum length of a text segment. Longer segments are split into shorter ones.
35
 */
36
const int32_t MAX_SEGMENT_LENGTH = SEGMENT_LENGTH_PLACEHOLDER_CHAR - ARG_NUM_LIMIT;
37
38
enum {
39
    APOS = 0x27,
40
    DIGIT_ZERO = 0x30,
41
    DIGIT_ONE = 0x31,
42
    DIGIT_NINE = 0x39,
43
    OPEN_BRACE = 0x7b,
44
    CLOSE_BRACE = 0x7d
45
};
46
47
0
inline UBool isInvalidArray(const void *array, int32_t length) {
48
0
   return (length < 0 || (array == NULL && length != 0));
49
0
}
50
51
}  // namespace
52
53
0
SimpleFormatter &SimpleFormatter::operator=(const SimpleFormatter& other) {
54
0
    if (this == &other) {
55
0
        return *this;
56
0
    }
57
0
    compiledPattern = other.compiledPattern;
58
0
    return *this;
59
0
}
60
61
0
SimpleFormatter::~SimpleFormatter() {}
62
63
UBool SimpleFormatter::applyPatternMinMaxArguments(
64
        const UnicodeString &pattern,
65
        int32_t min, int32_t max,
66
0
        UErrorCode &errorCode) {
67
0
    if (U_FAILURE(errorCode)) {
68
0
        return FALSE;
69
0
    }
70
    // Parse consistent with MessagePattern, but
71
    // - support only simple numbered arguments
72
    // - build a simple binary structure into the result string
73
0
    const UChar *patternBuffer = pattern.getBuffer();
74
0
    int32_t patternLength = pattern.length();
75
    // Reserve the first char for the number of arguments.
76
0
    compiledPattern.setTo((UChar)0);
77
0
    int32_t textLength = 0;
78
0
    int32_t maxArg = -1;
79
0
    UBool inQuote = FALSE;
80
0
    for (int32_t i = 0; i < patternLength;) {
81
0
        UChar c = patternBuffer[i++];
82
0
        if (c == APOS) {
83
0
            if (i < patternLength && (c = patternBuffer[i]) == APOS) {
84
                // double apostrophe, skip the second one
85
0
                ++i;
86
0
            } else if (inQuote) {
87
                // skip the quote-ending apostrophe
88
0
                inQuote = FALSE;
89
0
                continue;
90
0
            } else if (c == OPEN_BRACE || c == CLOSE_BRACE) {
91
                // Skip the quote-starting apostrophe, find the end of the quoted literal text.
92
0
                ++i;
93
0
                inQuote = TRUE;
94
0
            } else {
95
                // The apostrophe is part of literal text.
96
0
                c = APOS;
97
0
            }
98
0
        } else if (!inQuote && c == OPEN_BRACE) {
99
0
            if (textLength > 0) {
100
0
                compiledPattern.setCharAt(compiledPattern.length() - textLength - 1,
101
0
                                          (UChar)(ARG_NUM_LIMIT + textLength));
102
0
                textLength = 0;
103
0
            }
104
0
            int32_t argNumber;
105
0
            if ((i + 1) < patternLength &&
106
0
                    0 <= (argNumber = patternBuffer[i] - DIGIT_ZERO) && argNumber <= 9 &&
107
0
                    patternBuffer[i + 1] == CLOSE_BRACE) {
108
0
                i += 2;
109
0
            } else {
110
                // Multi-digit argument number (no leading zero) or syntax error.
111
                // MessagePattern permits PatternProps.skipWhiteSpace(pattern, index)
112
                // around the number, but this class does not.
113
0
                argNumber = -1;
114
0
                if (i < patternLength && DIGIT_ONE <= (c = patternBuffer[i++]) && c <= DIGIT_NINE) {
115
0
                    argNumber = c - DIGIT_ZERO;
116
0
                    while (i < patternLength &&
117
0
                            DIGIT_ZERO <= (c = patternBuffer[i++]) && c <= DIGIT_NINE) {
118
0
                        argNumber = argNumber * 10 + (c - DIGIT_ZERO);
119
0
                        if (argNumber >= ARG_NUM_LIMIT) {
120
0
                            break;
121
0
                        }
122
0
                    }
123
0
                }
124
0
                if (argNumber < 0 || c != CLOSE_BRACE) {
125
0
                    errorCode = U_ILLEGAL_ARGUMENT_ERROR;
126
0
                    return FALSE;
127
0
                }
128
0
            }
129
0
            if (argNumber > maxArg) {
130
0
                maxArg = argNumber;
131
0
            }
132
0
            compiledPattern.append((UChar)argNumber);
133
0
            continue;
134
0
        }  // else: c is part of literal text
135
        // Append c and track the literal-text segment length.
136
0
        if (textLength == 0) {
137
            // Reserve a char for the length of a new text segment, preset the maximum length.
138
0
            compiledPattern.append(SEGMENT_LENGTH_PLACEHOLDER_CHAR);
139
0
        }
140
0
        compiledPattern.append(c);
141
0
        if (++textLength == MAX_SEGMENT_LENGTH) {
142
0
            textLength = 0;
143
0
        }
144
0
    }
145
0
    if (textLength > 0) {
146
0
        compiledPattern.setCharAt(compiledPattern.length() - textLength - 1,
147
0
                                  (UChar)(ARG_NUM_LIMIT + textLength));
148
0
    }
149
0
    int32_t argCount = maxArg + 1;
150
0
    if (argCount < min || max < argCount) {
151
0
        errorCode = U_ILLEGAL_ARGUMENT_ERROR;
152
0
        return FALSE;
153
0
    }
154
0
    compiledPattern.setCharAt(0, (UChar)argCount);
155
0
    return TRUE;
156
0
}
157
158
UnicodeString& SimpleFormatter::format(
159
        const UnicodeString &value0,
160
0
        UnicodeString &appendTo, UErrorCode &errorCode) const {
161
0
    const UnicodeString *values[] = { &value0 };
162
0
    return formatAndAppend(values, 1, appendTo, NULL, 0, errorCode);
163
0
}
164
165
UnicodeString& SimpleFormatter::format(
166
        const UnicodeString &value0,
167
        const UnicodeString &value1,
168
0
        UnicodeString &appendTo, UErrorCode &errorCode) const {
169
0
    const UnicodeString *values[] = { &value0, &value1 };
170
0
    return formatAndAppend(values, 2, appendTo, NULL, 0, errorCode);
171
0
}
172
173
UnicodeString& SimpleFormatter::format(
174
        const UnicodeString &value0,
175
        const UnicodeString &value1,
176
        const UnicodeString &value2,
177
0
        UnicodeString &appendTo, UErrorCode &errorCode) const {
178
0
    const UnicodeString *values[] = { &value0, &value1, &value2 };
179
0
    return formatAndAppend(values, 3, appendTo, NULL, 0, errorCode);
180
0
}
181
182
UnicodeString& SimpleFormatter::formatAndAppend(
183
        const UnicodeString *const *values, int32_t valuesLength,
184
        UnicodeString &appendTo,
185
0
        int32_t *offsets, int32_t offsetsLength, UErrorCode &errorCode) const {
186
0
    if (U_FAILURE(errorCode)) {
187
0
        return appendTo;
188
0
    }
189
0
    if (isInvalidArray(values, valuesLength) || isInvalidArray(offsets, offsetsLength) ||
190
0
            valuesLength < getArgumentLimit()) {
191
0
        errorCode = U_ILLEGAL_ARGUMENT_ERROR;
192
0
        return appendTo;
193
0
    }
194
0
    return format(compiledPattern.getBuffer(), compiledPattern.length(), values,
195
0
                  appendTo, NULL, TRUE,
196
0
                  offsets, offsetsLength, errorCode);
197
0
}
198
199
UnicodeString &SimpleFormatter::formatAndReplace(
200
        const UnicodeString *const *values, int32_t valuesLength,
201
        UnicodeString &result,
202
0
        int32_t *offsets, int32_t offsetsLength, UErrorCode &errorCode) const {
203
0
    if (U_FAILURE(errorCode)) {
204
0
        return result;
205
0
    }
206
0
    if (isInvalidArray(values, valuesLength) || isInvalidArray(offsets, offsetsLength)) {
207
0
        errorCode = U_ILLEGAL_ARGUMENT_ERROR;
208
0
        return result;
209
0
    }
210
0
    const UChar *cp = compiledPattern.getBuffer();
211
0
    int32_t cpLength = compiledPattern.length();
212
0
    if (valuesLength < getArgumentLimit(cp, cpLength)) {
213
0
        errorCode = U_ILLEGAL_ARGUMENT_ERROR;
214
0
        return result;
215
0
    }
216
217
    // If the pattern starts with an argument whose value is the same object
218
    // as the result, then we keep the result contents and append to it.
219
    // Otherwise we replace its contents.
220
0
    int32_t firstArg = -1;
221
    // If any non-initial argument value is the same object as the result,
222
    // then we first copy its contents and use that instead while formatting.
223
0
    UnicodeString resultCopy;
224
0
    if (getArgumentLimit(cp, cpLength) > 0) {
225
0
        for (int32_t i = 1; i < cpLength;) {
226
0
            int32_t n = cp[i++];
227
0
            if (n < ARG_NUM_LIMIT) {
228
0
                if (values[n] == &result) {
229
0
                    if (i == 2) {
230
0
                        firstArg = n;
231
0
                    } else if (resultCopy.isEmpty() && !result.isEmpty()) {
232
0
                        resultCopy = result;
233
0
                    }
234
0
                }
235
0
            } else {
236
0
                i += n - ARG_NUM_LIMIT;
237
0
            }
238
0
        }
239
0
    }
240
0
    if (firstArg < 0) {
241
0
        result.remove();
242
0
    }
243
0
    return format(cp, cpLength, values,
244
0
                  result, &resultCopy, FALSE,
245
0
                  offsets, offsetsLength, errorCode);
246
0
}
247
248
UnicodeString SimpleFormatter::getTextWithNoArguments(
249
        const UChar *compiledPattern,
250
        int32_t compiledPatternLength,
251
        int32_t* offsets,
252
0
        int32_t offsetsLength) {
253
0
    for (int32_t i = 0; i < offsetsLength; i++) {
254
0
        offsets[i] = -1;
255
0
    }
256
0
    int32_t capacity = compiledPatternLength - 1 -
257
0
            getArgumentLimit(compiledPattern, compiledPatternLength);
258
0
    UnicodeString sb(capacity, 0, 0);  // Java: StringBuilder
259
0
    for (int32_t i = 1; i < compiledPatternLength;) {
260
0
        int32_t n = compiledPattern[i++];
261
0
        if (n > ARG_NUM_LIMIT) {
262
0
            n -= ARG_NUM_LIMIT;
263
0
            sb.append(compiledPattern + i, n);
264
0
            i += n;
265
0
        } else if (n < offsetsLength) {
266
            // TODO(ICU-20406): This does not distinguish between "{0}{1}" and "{1}{0}".
267
            // Consider removing this function and replacing it with an iterator interface.
268
0
            offsets[n] = sb.length();
269
0
        }
270
0
    }
271
0
    return sb;
272
0
}
273
274
UnicodeString &SimpleFormatter::format(
275
        const UChar *compiledPattern, int32_t compiledPatternLength,
276
        const UnicodeString *const *values,
277
        UnicodeString &result, const UnicodeString *resultCopy, UBool forbidResultAsValue,
278
        int32_t *offsets, int32_t offsetsLength,
279
0
        UErrorCode &errorCode) {
280
0
    if (U_FAILURE(errorCode)) {
281
0
        return result;
282
0
    }
283
0
    for (int32_t i = 0; i < offsetsLength; i++) {
284
0
        offsets[i] = -1;
285
0
    }
286
0
    for (int32_t i = 1; i < compiledPatternLength;) {
287
0
        int32_t n = compiledPattern[i++];
288
0
        if (n < ARG_NUM_LIMIT) {
289
0
            const UnicodeString *value = values[n];
290
0
            if (value == NULL) {
291
0
                errorCode = U_ILLEGAL_ARGUMENT_ERROR;
292
0
                return result;
293
0
            }
294
0
            if (value == &result) {
295
0
                if (forbidResultAsValue) {
296
0
                    errorCode = U_ILLEGAL_ARGUMENT_ERROR;
297
0
                    return result;
298
0
                }
299
0
                if (i == 2) {
300
                    // We are appending to result which is also the first value object.
301
0
                    if (n < offsetsLength) {
302
0
                        offsets[n] = 0;
303
0
                    }
304
0
                } else {
305
0
                    if (n < offsetsLength) {
306
0
                        offsets[n] = result.length();
307
0
                    }
308
0
                    result.append(*resultCopy);
309
0
                }
310
0
            } else {
311
0
                if (n < offsetsLength) {
312
0
                    offsets[n] = result.length();
313
0
                }
314
0
                result.append(*value);
315
0
            }
316
0
        } else {
317
0
            int32_t length = n - ARG_NUM_LIMIT;
318
0
            result.append(compiledPattern + i, length);
319
0
            i += length;
320
0
        }
321
0
    }
322
0
    return result;
323
0
}
324
325
U_NAMESPACE_END