1 /*
2 *******************************************************************************
3 *
4 * Copyright (C) 2007, International Business Machines
5 * Corporation and others. All Rights Reserved.
6 *
7 *******************************************************************************
8 * file name: icuzdump.cpp
9 * encoding: US-ASCII
10 * tab size: 8 (not used)
11 * indentation:4
12 *
13 * created on: 2007-04-02
14 * created by: Yoshito Umaoka
15 *
16 * This tool write out timezone transitions for ICU timezone. This tool
17 * is used as a part of tzdata update process to check if ICU timezone
18 * code works as well as the corresponding Olson stock localtime/zdump.
19 */
20
21 #include <cstdlib>
22 #include <cstring>
23 #include <fstream>
24 #include <sstream>
25 #include <iostream>
26
27 #include "unicode/utypes.h"
28 #include "unicode/ustring.h"
29 #include "unicode/timezone.h"
30 #include "unicode/simpletz.h"
31 #include "unicode/smpdtfmt.h"
32 #include "unicode/decimfmt.h"
33 #include "unicode/gregocal.h"
34 #include "unicode/ustream.h"
35 #include "unicode/putil.h"
36
37 #include "uoptions.h"
38
39 using namespace std;
40
41 class DumpFormatter {
42 public:
DumpFormatter()43 DumpFormatter() {
44 UErrorCode status = U_ZERO_ERROR;
45 stz = new SimpleTimeZone(0, "");
46 sdf = new SimpleDateFormat((UnicodeString)"yyyy-MM-dd EEE HH:mm:ss", Locale::getEnglish(), status);
47 DecimalFormatSymbols *symbols = new DecimalFormatSymbols(Locale::getEnglish(), status);
48 decf = new DecimalFormat("00", symbols, status);
49 }
~DumpFormatter()50 ~DumpFormatter() {
51 }
52
format(UDate time,int32_t offset,UBool isDst,UnicodeString & appendTo)53 UnicodeString& format(UDate time, int32_t offset, UBool isDst, UnicodeString& appendTo) {
54 stz->setRawOffset(offset);
55 sdf->setTimeZone(*stz);
56 UnicodeString str = sdf->format(time, appendTo);
57 if (offset < 0) {
58 appendTo += "-";
59 offset = -offset;
60 } else {
61 appendTo += "+";
62 }
63
64 int32_t hour, min, sec;
65
66 offset /= 1000;
67 sec = offset % 60;
68 offset = (offset - sec) / 60;
69 min = offset % 60;
70 hour = offset / 60;
71
72 decf->format(hour, appendTo);
73 decf->format(min, appendTo);
74 decf->format(sec, appendTo);
75 appendTo += "[DST=";
76 if (isDst) {
77 appendTo += "1";
78 } else {
79 appendTo += "0";
80 }
81 appendTo += "]";
82 return appendTo;
83 }
84 private:
85 SimpleTimeZone* stz;
86 SimpleDateFormat* sdf;
87 DecimalFormat* decf;
88 };
89
90 class ICUZDump {
91 public:
ICUZDump()92 ICUZDump() {
93 formatter = new DumpFormatter();
94 loyear = 1902;
95 hiyear = 2050;
96 tick = 1000;
97 linesep = NULL;
98 }
99
~ICUZDump()100 ~ICUZDump() {
101 }
102
setLowYear(int32_t lo)103 void setLowYear(int32_t lo) {
104 loyear = lo;
105 }
106
setHighYear(int32_t hi)107 void setHighYear(int32_t hi) {
108 hiyear = hi;
109 }
110
setTick(int32_t t)111 void setTick(int32_t t) {
112 tick = t;
113 }
114
setTimeZone(TimeZone * tz)115 void setTimeZone(TimeZone* tz) {
116 timezone = tz;
117 }
118
setDumpFormatter(DumpFormatter * fmt)119 void setDumpFormatter(DumpFormatter* fmt) {
120 formatter = fmt;
121 }
122
setLineSeparator(const char * sep)123 void setLineSeparator(const char* sep) {
124 linesep = sep;
125 }
126
dump(ostream & out)127 void dump(ostream& out) {
128 UErrorCode status = U_ZERO_ERROR;
129 UDate SEARCH_INCREMENT = 12 * 60 * 60 * 1000; // half day
130 UDate t, cutlo, cuthi;
131 int32_t rawOffset, dstOffset;
132 UnicodeString str;
133
134 getCutOverTimes(cutlo, cuthi);
135 t = cutlo;
136 timezone->getOffset(t, FALSE, rawOffset, dstOffset, status);
137 while (t < cuthi) {
138 int32_t newRawOffset, newDstOffset;
139 UDate newt = t + SEARCH_INCREMENT;
140
141 timezone->getOffset(newt, FALSE, newRawOffset, newDstOffset, status);
142
143 UBool bSameOffset = (rawOffset + dstOffset) == (newRawOffset + newDstOffset);
144 UBool bSameDst = ((dstOffset != 0) && (newDstOffset != 0)) || ((dstOffset == 0) && (newDstOffset == 0));
145
146 if (!bSameOffset || !bSameDst) {
147 // find the boundary
148 UDate lot = t;
149 UDate hit = newt;
150 while (true) {
151 int32_t diff = (int32_t)(hit - lot);
152 if (diff <= tick) {
153 break;
154 }
155 UDate medt = lot + ((diff / 2) / tick) * tick;
156 int32_t medRawOffset, medDstOffset;
157 timezone->getOffset(medt, FALSE, medRawOffset, medDstOffset, status);
158
159 bSameOffset = (rawOffset + dstOffset) == (medRawOffset + medDstOffset);
160 bSameDst = ((dstOffset != 0) && (medDstOffset != 0)) || ((dstOffset == 0) && (medDstOffset == 0));
161
162 if (!bSameOffset || !bSameDst) {
163 hit = medt;
164 } else {
165 lot = medt;
166 }
167 }
168 // write out the boundary
169 str.remove();
170 formatter->format(lot, rawOffset + dstOffset, (dstOffset == 0 ? FALSE : TRUE), str);
171 out << str << " > ";
172 str.remove();
173 formatter->format(hit, newRawOffset + newDstOffset, (newDstOffset == 0 ? FALSE : TRUE), str);
174 out << str;
175 if (linesep != NULL) {
176 out << linesep;
177 } else {
178 out << endl;
179 }
180
181 rawOffset = newRawOffset;
182 dstOffset = newDstOffset;
183 }
184 t = newt;
185 }
186 }
187
188 private:
getCutOverTimes(UDate & lo,UDate & hi)189 void getCutOverTimes(UDate& lo, UDate& hi) {
190 UErrorCode status = U_ZERO_ERROR;
191 GregorianCalendar* gcal = new GregorianCalendar(timezone, Locale::getEnglish(), status);
192 gcal->clear();
193 gcal->set(loyear, 0, 1, 0, 0, 0);
194 lo = gcal->getTime(status);
195 gcal->set(hiyear, 0, 1, 0, 0, 0);
196 hi = gcal->getTime(status);
197 }
198
dumpZone(ostream & out,const char * linesep,UnicodeString tzid,int32_t low,int32_t high)199 void dumpZone(ostream& out, const char* linesep, UnicodeString tzid, int32_t low, int32_t high) {
200 }
201
202 TimeZone* timezone;
203 int32_t loyear;
204 int32_t hiyear;
205 int32_t tick;
206
207 DumpFormatter* formatter;
208 const char* linesep;
209 };
210
211 class ZoneIterator {
212 public:
ZoneIterator(UBool bAll=FALSE)213 ZoneIterator(UBool bAll = FALSE) {
214 if (bAll) {
215 zenum = TimeZone::createEnumeration();
216 }
217 else {
218 zenum = NULL;
219 zids = NULL;
220 idx = 0;
221 numids = 1;
222 }
223 }
224
ZoneIterator(const char ** ids,int32_t num)225 ZoneIterator(const char** ids, int32_t num) {
226 zenum = NULL;
227 zids = ids;
228 idx = 0;
229 numids = num;
230 }
231
~ZoneIterator()232 ~ZoneIterator() {
233 if (zenum != NULL) {
234 delete zenum;
235 }
236 }
237
next()238 TimeZone* next() {
239 TimeZone* tz = NULL;
240 if (zenum != NULL) {
241 UErrorCode status = U_ZERO_ERROR;
242 const UnicodeString* zid = zenum->snext(status);
243 if (zid != NULL) {
244 tz = TimeZone::createTimeZone(*zid);
245 }
246 }
247 else {
248 if (idx < numids) {
249 if (zids != NULL) {
250 tz = TimeZone::createTimeZone((const UnicodeString&)zids[idx]);
251 }
252 else {
253 tz = TimeZone::createDefault();
254 }
255 idx++;
256 }
257 }
258 return tz;
259 }
260
261 private:
262 const char** zids;
263 StringEnumeration* zenum;
264 int32_t idx;
265 int32_t numids;
266 };
267
268 enum {
269 kOptHelpH = 0,
270 kOptHelpQuestionMark,
271 kOptAllZones,
272 kOptCutover,
273 kOptDestDir,
274 kOptLineSep
275 };
276
277 static UOption options[]={
278 UOPTION_HELP_H,
279 UOPTION_HELP_QUESTION_MARK,
280 UOPTION_DEF("allzones", 'a', UOPT_NO_ARG),
281 UOPTION_DEF("cutover", 'c', UOPT_REQUIRES_ARG),
282 UOPTION_DEF("destdir", 'd', UOPT_REQUIRES_ARG),
283 UOPTION_DEF("linesep", 'l', UOPT_REQUIRES_ARG)
284 };
285
286 extern int
main(int argc,char * argv[])287 main(int argc, char *argv[]) {
288 int32_t low = 1902;
289 int32_t high = 2038;
290 UBool bAll = FALSE;
291 const char *dir = NULL;
292 const char *linesep = NULL;
293
294 U_MAIN_INIT_ARGS(argc, argv);
295 argc = u_parseArgs(argc, argv, sizeof(options)/sizeof(options[0]), options);
296
297 if (argc < 0) {
298 cerr << "Illegal command line argument(s)" << endl << endl;
299 }
300
301 if (argc < 0 || options[kOptHelpH].doesOccur || options[kOptHelpQuestionMark].doesOccur) {
302 cerr
303 << "Usage: icuzdump [-options] [zoneid1 zoneid2 ...]" << endl
304 << endl
305 << "\tDump all offset transitions for the specified zones." << endl
306 << endl
307 << "Options:" << endl
308 << "\t-a : Dump all available zones." << endl
309 << "\t-d <dir> : When specified, write transitions in a file under" << endl
310 << "\t the directory for each zone." << endl
311 << "\t-l <sep> : New line code type used in file outputs. CR or LF (default)"
312 << "\t or CRLF." << endl
313 << "\t-c [<low_year>,]<high_year>" << endl
314 << "\t : When specified, dump transitions starting <low_year>" << endl
315 << "\t (inclusive) up to <high_year> (exclusive). The default" << endl
316 << "\t values are 1902(low) and 2038(high)." << endl;
317 return argc < 0 ? U_ILLEGAL_ARGUMENT_ERROR : U_ZERO_ERROR;
318 }
319
320 bAll = options[kOptAllZones].doesOccur;
321
322 if (options[kOptDestDir].doesOccur) {
323 dir = options[kOptDestDir].value;
324 }
325
326 if (options[kOptLineSep].doesOccur) {
327 if (strcmp(options[kOptLineSep].value, "CR") == 0) {
328 linesep = "\r";
329 } else if (strcmp(options[kOptLineSep].value, "CRLF") == 0) {
330 linesep = "\r\n";
331 } else if (strcmp(options[kOptLineSep].value, "LF") == 0) {
332 linesep = "\n";
333 }
334 }
335
336 if (options[kOptCutover].doesOccur) {
337 char* comma = (char*)strchr(options[kOptCutover].value, ',');
338 if (comma == NULL) {
339 high = atoi(options[kOptCutover].value);
340 } else {
341 *comma = 0;
342 low = atoi(options[kOptCutover].value);
343 high = atoi(comma + 1);
344 }
345 }
346
347 ICUZDump dumper;
348 dumper.setLowYear(low);
349 dumper.setHighYear(high);
350 if (dir != NULL && linesep != NULL) {
351 // use the specified line separator only for file output
352 dumper.setLineSeparator((const char*)linesep);
353 }
354
355 ZoneIterator* zit;
356 if (bAll) {
357 zit = new ZoneIterator(TRUE);
358 } else {
359 if (argc <= 1) {
360 zit = new ZoneIterator();
361 } else {
362 zit = new ZoneIterator((const char**)&argv[1], argc - 1);
363 }
364 }
365
366 UnicodeString id;
367 if (dir != NULL) {
368 // file output
369 ostringstream path;
370 ios::openmode mode = ios::out;
371 if (linesep != NULL) {
372 mode |= ios::binary;
373 }
374 for (;;) {
375 TimeZone* tz = zit->next();
376 if (tz == NULL) {
377 break;
378 }
379 dumper.setTimeZone(tz);
380 tz->getID(id);
381
382 // target file path
383 path.str("");
384 path << dir << U_FILE_SEP_CHAR;
385 id = id.findAndReplace("/", "-");
386 path << id;
387
388 ofstream* fout = new ofstream(path.str().c_str(), mode);
389 if (fout->fail()) {
390 cerr << "Cannot open file " << path << endl;
391 delete fout;
392 delete tz;
393 break;
394 }
395
396 dumper.dump(*fout);
397 fout->close();
398 delete fout;
399 delete tz;
400 }
401
402 } else {
403 // stdout
404 UBool bFirst = TRUE;
405 for (;;) {
406 TimeZone* tz = zit->next();
407 if (tz == NULL) {
408 break;
409 }
410 dumper.setTimeZone(tz);
411 tz->getID(id);
412 if (bFirst) {
413 bFirst = FALSE;
414 } else {
415 cout << endl;
416 }
417 cout << "ZONE: " << id << endl;
418 dumper.dump(cout);
419 delete tz;
420 }
421 }
422 delete zit;
423 }
424