1 | // © 2016 and later: Unicode, Inc. and others. |
2 | // License & terms of use: http://www.unicode.org/copyright.html |
3 | /* |
4 | ************************************************************************** |
5 | * Copyright (c) 2002-2010, International Business Machines Corporation * |
6 | * and others. All Rights Reserved. * |
7 | ************************************************************************** |
8 | * Date Name Description * |
9 | * 01/28/2002 aliu Creation. * |
10 | ************************************************************************** |
11 | */ |
12 | #ifndef TRIDPARS_H |
13 | #define TRIDPARS_H |
14 | |
15 | #include "unicode/utypes.h" |
16 | |
17 | #if !UCONFIG_NO_TRANSLITERATION |
18 | |
19 | #include "unicode/uobject.h" |
20 | #include "unicode/unistr.h" |
21 | |
22 | U_NAMESPACE_BEGIN |
23 | |
24 | class Transliterator; |
25 | class UnicodeSet; |
26 | class UVector; |
27 | |
28 | /** |
29 | * Parsing component for transliterator IDs. This class contains only |
30 | * static members; it cannot be instantiated. Methods in this class |
31 | * parse various ID formats, including the following: |
32 | * |
33 | * A basic ID, which contains source, target, and variant, but no |
34 | * filter and no explicit inverse. Examples include |
35 | * "Latin-Greek/UNGEGN" and "Null". |
36 | * |
37 | * A single ID, which is a basic ID plus optional filter and optional |
38 | * explicit inverse. Examples include "[a-zA-Z] Latin-Greek" and |
39 | * "Lower (Upper)". |
40 | * |
41 | * A compound ID, which is a sequence of one or more single IDs, |
42 | * separated by semicolons, with optional forward and reverse global |
43 | * filters. The global filters are UnicodeSet patterns prepended or |
44 | * appended to the IDs, separated by semicolons. An appended filter |
45 | * must be enclosed in parentheses and applies in the reverse |
46 | * direction. |
47 | * |
48 | * @author Alan Liu |
49 | */ |
50 | class TransliteratorIDParser /* not : public UObject because all methods are static */ { |
51 | |
52 | public: |
53 | |
54 | /** |
55 | * A structure containing the parsed data of a filtered ID, that |
56 | * is, a basic ID optionally with a filter. |
57 | * |
58 | * 'source' and 'target' will always be non-null. The 'variant' |
59 | * will be non-null only if a non-empty variant was parsed. |
60 | * |
61 | * 'sawSource' is true if there was an explicit source in the |
62 | * parsed id. If there was no explicit source, then an implied |
63 | * source of ANY is returned and 'sawSource' is set to false. |
64 | * |
65 | * 'filter' is the parsed filter pattern, or null if there was no |
66 | * filter. |
67 | */ |
68 | class Specs : public UMemory { |
69 | public: |
70 | UnicodeString source; // not null |
71 | UnicodeString target; // not null |
72 | UnicodeString variant; // may be null |
73 | UnicodeString filter; // may be null |
74 | UBool sawSource; |
75 | Specs(const UnicodeString& s, const UnicodeString& t, |
76 | const UnicodeString& v, UBool sawS, |
77 | const UnicodeString& f); |
78 | |
79 | private: |
80 | |
81 | Specs(const Specs &other); // forbid copying of this class |
82 | Specs &operator=(const Specs &other); // forbid copying of this class |
83 | }; |
84 | |
85 | /** |
86 | * A structure containing the canonicalized data of a filtered ID, |
87 | * that is, a basic ID optionally with a filter. |
88 | * |
89 | * 'canonID' is always non-null. It may be the empty string "". |
90 | * It is the id that should be assigned to the created |
91 | * transliterator. It _cannot_ be instantiated directly. |
92 | * |
93 | * 'basicID' is always non-null and non-empty. It is always of |
94 | * the form S-T or S-T/V. It is designed to be fed to low-level |
95 | * instantiation code that only understands these two formats. |
96 | * |
97 | * 'filter' may be null, if there is none, or non-null and |
98 | * non-empty. |
99 | */ |
100 | class SingleID : public UMemory { |
101 | public: |
102 | UnicodeString canonID; |
103 | UnicodeString basicID; |
104 | UnicodeString filter; |
105 | SingleID(const UnicodeString& c, const UnicodeString& b, |
106 | const UnicodeString& f); |
107 | SingleID(const UnicodeString& c, const UnicodeString& b); |
108 | Transliterator* createInstance(); |
109 | |
110 | private: |
111 | |
112 | SingleID(const SingleID &other); // forbid copying of this class |
113 | SingleID &operator=(const SingleID &other); // forbid copying of this class |
114 | }; |
115 | |
116 | /** |
117 | * Parse a filter ID, that is, an ID of the general form |
118 | * "[f1] s1-t1/v1", with the filters optional, and the variants optional. |
119 | * @param id the id to be parsed |
120 | * @param pos INPUT-OUTPUT parameter. On input, the position of |
121 | * the first character to parse. On output, the position after |
122 | * the last character parsed. |
123 | * @return a SingleID object or null if the parse fails |
124 | */ |
125 | static SingleID* parseFilterID(const UnicodeString& id, int32_t& pos); |
126 | |
127 | /** |
128 | * Parse a single ID, that is, an ID of the general form |
129 | * "[f1] s1-t1/v1 ([f2] s2-t3/v2)", with the parenthesized element |
130 | * optional, the filters optional, and the variants optional. |
131 | * @param id the id to be parsed |
132 | * @param pos INPUT-OUTPUT parameter. On input, the position of |
133 | * the first character to parse. On output, the position after |
134 | * the last character parsed. |
135 | * @param dir the direction. If the direction is REVERSE then the |
136 | * SingleID is constructed for the reverse direction. |
137 | * @return a SingleID object or null |
138 | */ |
139 | static SingleID* parseSingleID(const UnicodeString& id, int32_t& pos, |
140 | int32_t dir, UErrorCode& status); |
141 | |
142 | /** |
143 | * Parse a global filter of the form "[f]" or "([f])", depending |
144 | * on 'withParens'. |
145 | * @param id the pattern the parse |
146 | * @param pos INPUT-OUTPUT parameter. On input, the position of |
147 | * the first character to parse. On output, the position after |
148 | * the last character parsed. |
149 | * @param dir the direction. |
150 | * @param withParens INPUT-OUTPUT parameter. On entry, if |
151 | * withParens[0] is 0, then parens are disallowed. If it is 1, |
152 | * then parens are required. If it is -1, then parens are |
153 | * optional, and the return result will be set to 0 or 1. |
154 | * @param canonID OUTPUT parameter. The pattern for the filter |
155 | * added to the canonID, either at the end, if dir is FORWARD, or |
156 | * at the start, if dir is REVERSE. The pattern will be enclosed |
157 | * in parentheses if appropriate, and will be suffixed with an |
158 | * ID_DELIM character. May be null. |
159 | * @return a UnicodeSet object or null. A non-null results |
160 | * indicates a successful parse, regardless of whether the filter |
161 | * applies to the given direction. The caller should discard it |
162 | * if withParens != (dir == REVERSE). |
163 | */ |
164 | static UnicodeSet* parseGlobalFilter(const UnicodeString& id, int32_t& pos, |
165 | int32_t dir, |
166 | int32_t& withParens, |
167 | UnicodeString* canonID); |
168 | |
169 | /** |
170 | * Parse a compound ID, consisting of an optional forward global |
171 | * filter, a separator, one or more single IDs delimited by |
172 | * separators, an an optional reverse global filter. The |
173 | * separator is a semicolon. The global filters are UnicodeSet |
174 | * patterns. The reverse global filter must be enclosed in |
175 | * parentheses. |
176 | * @param id the pattern the parse |
177 | * @param dir the direction. |
178 | * @param canonID OUTPUT parameter that receives the canonical ID, |
179 | * consisting of canonical IDs for all elements, as returned by |
180 | * parseSingleID(), separated by semicolons. Previous contents |
181 | * are discarded. |
182 | * @param list OUTPUT parameter that receives a list of SingleID |
183 | * objects representing the parsed IDs. Previous contents are |
184 | * discarded. |
185 | * @param globalFilter OUTPUT parameter that receives a pointer to |
186 | * a newly created global filter for this ID in this direction, or |
187 | * null if there is none. |
188 | * @return true if the parse succeeds, that is, if the entire |
189 | * id is consumed without syntax error. |
190 | */ |
191 | static UBool parseCompoundID(const UnicodeString& id, int32_t dir, |
192 | UnicodeString& canonID, |
193 | UVector& list, |
194 | UnicodeSet*& globalFilter); |
195 | |
196 | /** |
197 | * Convert the elements of the 'list' vector, which are SingleID |
198 | * objects, into actual Transliterator objects. In the course of |
199 | * this, some (or all) entries may be removed. If all entries |
200 | * are removed, the Null transliterator will be added. |
201 | * |
202 | * Delete entries with empty basicIDs; these are generated by |
203 | * elements like "(A)" in the forward direction, or "A()" in |
204 | * the reverse. THIS MAY RESULT IN AN EMPTY VECTOR. Convert |
205 | * SingleID entries to actual transliterators. |
206 | * |
207 | * @param list vector of SingleID objects. On exit, vector |
208 | * of one or more Transliterators. |
209 | * @param ec Output param to receive a success or an error code. |
210 | * @return new value of insertIndex. The index will shift if |
211 | * there are empty items, like "(Lower)", with indices less than |
212 | * insertIndex. |
213 | */ |
214 | static void instantiateList(UVector& list, |
215 | UErrorCode& ec); |
216 | |
217 | /** |
218 | * Parse an ID into pieces. Take IDs of the form T, T/V, S-T, |
219 | * S-T/V, or S/V-T. If the source is missing, return a source of |
220 | * ANY. |
221 | * @param id the id string, in any of several forms |
222 | * @param source the given source. |
223 | * @param target the given target. |
224 | * @param variant the given variant |
225 | * @param isSourcePresent If TRUE then the source is present. |
226 | * If the source is not present, ANY will be |
227 | * given as the source, and isSourcePresent will be null |
228 | * @return an array of 4 strings: source, target, variant, and |
229 | * isSourcePresent. If the source is not present, ANY will be |
230 | * given as the source, and isSourcePresent will be null. Otherwise |
231 | * isSourcePresent will be non-null. The target may be empty if the |
232 | * id is not well-formed. The variant may be empty. |
233 | */ |
234 | static void IDtoSTV(const UnicodeString& id, |
235 | UnicodeString& source, |
236 | UnicodeString& target, |
237 | UnicodeString& variant, |
238 | UBool& isSourcePresent); |
239 | |
240 | /** |
241 | * Given source, target, and variant strings, concatenate them into a |
242 | * full ID. If the source is empty, then "Any" will be used for the |
243 | * source, so the ID will always be of the form s-t/v or s-t. |
244 | */ |
245 | static void STVtoID(const UnicodeString& source, |
246 | const UnicodeString& target, |
247 | const UnicodeString& variant, |
248 | UnicodeString& id); |
249 | |
250 | /** |
251 | * Register two targets as being inverses of one another. For |
252 | * example, calling registerSpecialInverse("NFC", "NFD", true) causes |
253 | * Transliterator to form the following inverse relationships: |
254 | * |
255 | * <pre>NFC => NFD |
256 | * Any-NFC => Any-NFD |
257 | * NFD => NFC |
258 | * Any-NFD => Any-NFC</pre> |
259 | * |
260 | * (Without the special inverse registration, the inverse of NFC |
261 | * would be NFC-Any.) Note that NFD is shorthand for Any-NFD, but |
262 | * that the presence or absence of "Any-" is preserved. |
263 | * |
264 | * <p>The relationship is symmetrical; registering (a, b) is |
265 | * equivalent to registering (b, a). |
266 | * |
267 | * <p>The relevant IDs must still be registered separately as |
268 | * factories or classes. |
269 | * |
270 | * <p>Only the targets are specified. Special inverses always |
271 | * have the form Any-Target1 <=> Any-Target2. The target should |
272 | * have canonical casing (the casing desired to be produced when |
273 | * an inverse is formed) and should contain no whitespace or other |
274 | * extraneous characters. |
275 | * |
276 | * @param target the target against which to register the inverse |
277 | * @param inverseTarget the inverse of target, that is |
278 | * Any-target.getInverse() => Any-inverseTarget |
279 | * @param bidirectional if true, register the reverse relation |
280 | * as well, that is, Any-inverseTarget.getInverse() => Any-target |
281 | */ |
282 | static void registerSpecialInverse(const UnicodeString& target, |
283 | const UnicodeString& inverseTarget, |
284 | UBool bidirectional, |
285 | UErrorCode &status); |
286 | |
287 | /** |
288 | * Free static memory. |
289 | */ |
290 | static void cleanup(); |
291 | |
292 | private: |
293 | //---------------------------------------------------------------- |
294 | // Private implementation |
295 | //---------------------------------------------------------------- |
296 | |
297 | // forbid instantiation |
298 | TransliteratorIDParser(); |
299 | |
300 | /** |
301 | * Parse an ID into component pieces. Take IDs of the form T, |
302 | * T/V, S-T, S-T/V, or S/V-T. If the source is missing, return a |
303 | * source of ANY. |
304 | * @param id the id string, in any of several forms |
305 | * @param pos INPUT-OUTPUT parameter. On input, pos[0] is the |
306 | * offset of the first character to parse in id. On output, |
307 | * pos[0] is the offset after the last parsed character. If the |
308 | * parse failed, pos[0] will be unchanged. |
309 | * @param allowFilter if true, a UnicodeSet pattern is allowed |
310 | * at any location between specs or delimiters, and is returned |
311 | * as the fifth string in the array. |
312 | * @return a Specs object, or null if the parse failed. If |
313 | * neither source nor target was seen in the parsed id, then the |
314 | * parse fails. If allowFilter is true, then the parsed filter |
315 | * pattern is returned in the Specs object, otherwise the returned |
316 | * filter reference is null. If the parse fails for any reason |
317 | * null is returned. |
318 | */ |
319 | static Specs* parseFilterID(const UnicodeString& id, int32_t& pos, |
320 | UBool allowFilter); |
321 | |
322 | /** |
323 | * Givens a Specs object, convert it to a SingleID object. The |
324 | * Spec object is a more unprocessed parse result. The SingleID |
325 | * object contains information about canonical and basic IDs. |
326 | * @param specs the given Specs object. |
327 | * @param dir either FORWARD or REVERSE. |
328 | * @return a SingleID; never returns null. Returned object always |
329 | * has 'filter' field of null. |
330 | */ |
331 | static SingleID* specsToID(const Specs* specs, int32_t dir); |
332 | |
333 | /** |
334 | * Given a Specs object, return a SingleID representing the |
335 | * special inverse of that ID. If there is no special inverse |
336 | * then return null. |
337 | * @param specs the given Specs. |
338 | * @return a SingleID or null. Returned object always has |
339 | * 'filter' field of null. |
340 | */ |
341 | static SingleID* specsToSpecialInverse(const Specs& specs, UErrorCode &status); |
342 | |
343 | /** |
344 | * Glue method to get around access problems in C++. |
345 | * @param id the id string for the transliterator, in any of several forms |
346 | * @param canonID the given canonical ID |
347 | */ |
348 | static Transliterator* createBasicInstance(const UnicodeString& id, |
349 | const UnicodeString* canonID); |
350 | |
351 | /** |
352 | * Initialize static memory. |
353 | */ |
354 | static void U_CALLCONV init(UErrorCode &status); |
355 | |
356 | friend class SingleID; |
357 | }; |
358 | |
359 | U_NAMESPACE_END |
360 | |
361 | #endif /* #if !UCONFIG_NO_TRANSLITERATION */ |
362 | |
363 | #endif |
364 | |