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/char16_t 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 char16_t 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 | inline UBool isInvalidArray(const void *array, int32_t length) { |
48 | return (length < 0 || (array == nullptr && length != 0)); |
49 | } |
50 | |
51 | } // namespace |
52 | |
53 | SimpleFormatter &SimpleFormatter::operator=(const SimpleFormatter& other) { |
54 | if (this == &other) { |
55 | return *this; |
56 | } |
57 | compiledPattern = other.compiledPattern; |
58 | return *this; |
59 | } |
60 | |
61 | SimpleFormatter::~SimpleFormatter() {} |
62 | |
63 | UBool SimpleFormatter::applyPatternMinMaxArguments( |
64 | const UnicodeString &pattern, |
65 | int32_t min, int32_t max, |
66 | UErrorCode &errorCode) { |
67 | if (U_FAILURE(errorCode)) { |
68 | return false; |
69 | } |
70 | // Parse consistent with MessagePattern, but |
71 | // - support only simple numbered arguments |
72 | // - build a simple binary structure into the result string |
73 | const char16_t *patternBuffer = pattern.getBuffer(); |
74 | int32_t patternLength = pattern.length(); |
75 | // Reserve the first char for the number of arguments. |
76 | compiledPattern.setTo((char16_t)0); |
77 | int32_t textLength = 0; |
78 | int32_t maxArg = -1; |
79 | UBool inQuote = false; |
80 | for (int32_t i = 0; i < patternLength;) { |
81 | char16_t c = patternBuffer[i++]; |
82 | if (c == APOS) { |
83 | if (i < patternLength && (c = patternBuffer[i]) == APOS) { |
84 | // double apostrophe, skip the second one |
85 | ++i; |
86 | } else if (inQuote) { |
87 | // skip the quote-ending apostrophe |
88 | inQuote = false; |
89 | continue; |
90 | } else if (c == OPEN_BRACE || c == CLOSE_BRACE) { |
91 | // Skip the quote-starting apostrophe, find the end of the quoted literal text. |
92 | ++i; |
93 | inQuote = true; |
94 | } else { |
95 | // The apostrophe is part of literal text. |
96 | c = APOS; |
97 | } |
98 | } else if (!inQuote && c == OPEN_BRACE) { |
99 | if (textLength > 0) { |
100 | compiledPattern.setCharAt(compiledPattern.length() - textLength - 1, |
101 | (char16_t)(ARG_NUM_LIMIT + textLength)); |
102 | textLength = 0; |
103 | } |
104 | int32_t argNumber; |
105 | if ((i + 1) < patternLength && |
106 | 0 <= (argNumber = patternBuffer[i] - DIGIT_ZERO) && argNumber <= 9 && |
107 | patternBuffer[i + 1] == CLOSE_BRACE) { |
108 | i += 2; |
109 | } 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 | argNumber = -1; |
114 | if (i < patternLength && DIGIT_ONE <= (c = patternBuffer[i++]) && c <= DIGIT_NINE) { |
115 | argNumber = c - DIGIT_ZERO; |
116 | while (i < patternLength && |
117 | DIGIT_ZERO <= (c = patternBuffer[i++]) && c <= DIGIT_NINE) { |
118 | argNumber = argNumber * 10 + (c - DIGIT_ZERO); |
119 | if (argNumber >= ARG_NUM_LIMIT) { |
120 | break; |
121 | } |
122 | } |
123 | } |
124 | if (argNumber < 0 || c != CLOSE_BRACE) { |
125 | errorCode = U_ILLEGAL_ARGUMENT_ERROR; |
126 | return false; |
127 | } |
128 | } |
129 | if (argNumber > maxArg) { |
130 | maxArg = argNumber; |
131 | } |
132 | compiledPattern.append((char16_t)argNumber); |
133 | continue; |
134 | } // else: c is part of literal text |
135 | // Append c and track the literal-text segment length. |
136 | if (textLength == 0) { |
137 | // Reserve a char for the length of a new text segment, preset the maximum length. |
138 | compiledPattern.append(SEGMENT_LENGTH_PLACEHOLDER_CHAR); |
139 | } |
140 | compiledPattern.append(c); |
141 | if (++textLength == MAX_SEGMENT_LENGTH) { |
142 | textLength = 0; |
143 | } |
144 | } |
145 | if (textLength > 0) { |
146 | compiledPattern.setCharAt(compiledPattern.length() - textLength - 1, |
147 | (char16_t)(ARG_NUM_LIMIT + textLength)); |
148 | } |
149 | int32_t argCount = maxArg + 1; |
150 | if (argCount < min || max < argCount) { |
151 | errorCode = U_ILLEGAL_ARGUMENT_ERROR; |
152 | return false; |
153 | } |
154 | compiledPattern.setCharAt(0, (char16_t)argCount); |
155 | return true; |
156 | } |
157 | |
158 | UnicodeString& SimpleFormatter::format( |
159 | const UnicodeString &value0, |
160 | UnicodeString &appendTo, UErrorCode &errorCode) const { |
161 | const UnicodeString *values[] = { &value0 }; |
162 | return formatAndAppend(values, 1, appendTo, nullptr, 0, errorCode); |
163 | } |
164 | |
165 | UnicodeString& SimpleFormatter::format( |
166 | const UnicodeString &value0, |
167 | const UnicodeString &value1, |
168 | UnicodeString &appendTo, UErrorCode &errorCode) const { |
169 | const UnicodeString *values[] = { &value0, &value1 }; |
170 | return formatAndAppend(values, 2, appendTo, nullptr, 0, errorCode); |
171 | } |
172 | |
173 | UnicodeString& SimpleFormatter::format( |
174 | const UnicodeString &value0, |
175 | const UnicodeString &value1, |
176 | const UnicodeString &value2, |
177 | UnicodeString &appendTo, UErrorCode &errorCode) const { |
178 | const UnicodeString *values[] = { &value0, &value1, &value2 }; |
179 | return formatAndAppend(values, 3, appendTo, nullptr, 0, errorCode); |
180 | } |
181 | |
182 | UnicodeString& SimpleFormatter::formatAndAppend( |
183 | const UnicodeString *const *values, int32_t valuesLength, |
184 | UnicodeString &appendTo, |
185 | int32_t *offsets, int32_t offsetsLength, UErrorCode &errorCode) const { |
186 | if (U_FAILURE(errorCode)) { |
187 | return appendTo; |
188 | } |
189 | if (isInvalidArray(values, valuesLength) || isInvalidArray(offsets, offsetsLength) || |
190 | valuesLength < getArgumentLimit()) { |
191 | errorCode = U_ILLEGAL_ARGUMENT_ERROR; |
192 | return appendTo; |
193 | } |
194 | return format(compiledPattern.getBuffer(), compiledPattern.length(), values, |
195 | appendTo, nullptr, true, |
196 | offsets, offsetsLength, errorCode); |
197 | } |
198 | |
199 | UnicodeString &SimpleFormatter::formatAndReplace( |
200 | const UnicodeString *const *values, int32_t valuesLength, |
201 | UnicodeString &result, |
202 | int32_t *offsets, int32_t offsetsLength, UErrorCode &errorCode) const { |
203 | if (U_FAILURE(errorCode)) { |
204 | return result; |
205 | } |
206 | if (isInvalidArray(values, valuesLength) || isInvalidArray(offsets, offsetsLength)) { |
207 | errorCode = U_ILLEGAL_ARGUMENT_ERROR; |
208 | return result; |
209 | } |
210 | const char16_t *cp = compiledPattern.getBuffer(); |
211 | int32_t cpLength = compiledPattern.length(); |
212 | if (valuesLength < getArgumentLimit(cp, cpLength)) { |
213 | errorCode = U_ILLEGAL_ARGUMENT_ERROR; |
214 | return result; |
215 | } |
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 | 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 | UnicodeString resultCopy; |
224 | if (getArgumentLimit(cp, cpLength) > 0) { |
225 | for (int32_t i = 1; i < cpLength;) { |
226 | int32_t n = cp[i++]; |
227 | if (n < ARG_NUM_LIMIT) { |
228 | if (values[n] == &result) { |
229 | if (i == 2) { |
230 | firstArg = n; |
231 | } else if (resultCopy.isEmpty() && !result.isEmpty()) { |
232 | resultCopy = result; |
233 | } |
234 | } |
235 | } else { |
236 | i += n - ARG_NUM_LIMIT; |
237 | } |
238 | } |
239 | } |
240 | if (firstArg < 0) { |
241 | result.remove(); |
242 | } |
243 | return format(cp, cpLength, values, |
244 | result, &resultCopy, false, |
245 | offsets, offsetsLength, errorCode); |
246 | } |
247 | |
248 | UnicodeString SimpleFormatter::getTextWithNoArguments( |
249 | const char16_t *compiledPattern, |
250 | int32_t compiledPatternLength, |
251 | int32_t* offsets, |
252 | int32_t offsetsLength) { |
253 | for (int32_t i = 0; i < offsetsLength; i++) { |
254 | offsets[i] = -1; |
255 | } |
256 | int32_t capacity = compiledPatternLength - 1 - |
257 | getArgumentLimit(compiledPattern, compiledPatternLength); |
258 | UnicodeString sb(capacity, 0, 0); // Java: StringBuilder |
259 | for (int32_t i = 1; i < compiledPatternLength;) { |
260 | int32_t n = compiledPattern[i++]; |
261 | if (n > ARG_NUM_LIMIT) { |
262 | n -= ARG_NUM_LIMIT; |
263 | sb.append(compiledPattern + i, n); |
264 | i += n; |
265 | } 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 | offsets[n] = sb.length(); |
269 | } |
270 | } |
271 | return sb; |
272 | } |
273 | |
274 | UnicodeString &SimpleFormatter::format( |
275 | const char16_t *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 | UErrorCode &errorCode) { |
280 | if (U_FAILURE(errorCode)) { |
281 | return result; |
282 | } |
283 | for (int32_t i = 0; i < offsetsLength; i++) { |
284 | offsets[i] = -1; |
285 | } |
286 | for (int32_t i = 1; i < compiledPatternLength;) { |
287 | int32_t n = compiledPattern[i++]; |
288 | if (n < ARG_NUM_LIMIT) { |
289 | const UnicodeString *value = values[n]; |
290 | if (value == nullptr) { |
291 | errorCode = U_ILLEGAL_ARGUMENT_ERROR; |
292 | return result; |
293 | } |
294 | if (value == &result) { |
295 | if (forbidResultAsValue) { |
296 | errorCode = U_ILLEGAL_ARGUMENT_ERROR; |
297 | return result; |
298 | } |
299 | if (i == 2) { |
300 | // We are appending to result which is also the first value object. |
301 | if (n < offsetsLength) { |
302 | offsets[n] = 0; |
303 | } |
304 | } else { |
305 | if (n < offsetsLength) { |
306 | offsets[n] = result.length(); |
307 | } |
308 | result.append(*resultCopy); |
309 | } |
310 | } else { |
311 | if (n < offsetsLength) { |
312 | offsets[n] = result.length(); |
313 | } |
314 | result.append(*value); |
315 | } |
316 | } else { |
317 | int32_t length = n - ARG_NUM_LIMIT; |
318 | result.append(compiledPattern + i, length); |
319 | i += length; |
320 | } |
321 | } |
322 | return result; |
323 | } |
324 | |
325 | U_NAMESPACE_END |
326 | |