icuSources/tools/tzcode/icuzdump.cpp

   1 // © 2016 and later: Unicode, Inc. and others.
   2 // License & terms of use: http://www.unicode.org/copyright.html
   3 /*
   4 *******************************************************************************
   5 *
   6 *   Copyright (C) 2007-2016, International Business Machines
   7 *   Corporation and others.  All Rights Reserved.
   8 *
   9 *******************************************************************************
  10 *   file name:  icuzdump.cpp
  11 *   encoding:   UTF-8
  12 *   tab size:   8 (not used)
  13 *   indentation:4
  14 *
  15 *   created on: 2007-04-02
  16 *   created by: Yoshito Umaoka
  17 *
  18 *   This tool write out timezone transitions for ICU timezone.  This tool
  19 *   is used as a part of tzdata update process to check if ICU timezone
  20 *   code works as well as the corresponding Olson stock localtime/zdump.
  21 */
  22
  23 #include <cstdlib>
  24 #include <cstring>
  25 #include <fstream>
  26 #include <sstream>
  27 #include <iostream>
  28
  29 #include "unicode/utypes.h"
  30 #include "unicode/ustring.h"
  31 #include "unicode/timezone.h"
  32 #include "unicode/simpletz.h"
  33 #include "unicode/smpdtfmt.h"
  34 #include "unicode/decimfmt.h"
  35 #include "unicode/gregocal.h"
  36 #include "unicode/ustream.h"
  37 #include "unicode/putil.h"
  38
  39 #include "cmemory.h"
  40 #include "uoptions.h"
  41
  42 using namespace std;
  43
  44 class DumpFormatter {
  45 public:
  46     DumpFormatter() {
  47         UErrorCode status = U_ZERO_ERROR;
  48         stz = new SimpleTimeZone(0, "");
  49         sdf = new SimpleDateFormat((UnicodeString)"yyyy-MM-dd EEE HH:mm:ss", Locale::getEnglish(), status);
  50         DecimalFormatSymbols *symbols = new DecimalFormatSymbols(Locale::getEnglish(), status);
  51         decf = new DecimalFormat("00", symbols, status);
  52     }
  53     ~DumpFormatter() {
  54     }
  55
  56     UnicodeString& format(UDate time, int32_t offset, UBool isDst, UnicodeString& appendTo) {
  57         stz->setRawOffset(offset);
  58         sdf->setTimeZone(*stz);
  59         UnicodeString str = sdf->format(time, appendTo);
  60         if (offset < 0) {
  61             appendTo += "-";
  62             offset = -offset;
  63         } else {
  64             appendTo += "+";
  65         }
  66
  67         int32_t hour, min, sec;
  68
  69         offset /= 1000;
  70         sec = offset % 60;
  71         offset = (offset - sec) / 60;
  72         min = offset % 60;
  73         hour = offset / 60;
  74
  75         decf->format(hour, appendTo);
  76         decf->format(min, appendTo);
  77         decf->format(sec, appendTo);
  78         appendTo += "[DST=";
  79         if (isDst) {
  80             appendTo += "1";
  81         } else {
  82             appendTo += "0";
  83         }
  84         appendTo += "]";
  85         return appendTo;
  86     }
  87 private:
  88     SimpleTimeZone*     stz;
  89     SimpleDateFormat*   sdf;
  90     DecimalFormat*      decf;
  91 };
  92
  93 class ICUZDump {
  94 public:
  95     ICUZDump() {
  96         formatter = new DumpFormatter();
  97         loyear = 1902;
  98         hiyear = 2050;
  99         tick = 1000;
 100         linesep = NULL;
 101     }
 102
 103     ~ICUZDump() {
 104     }
 105
 106     void setLowYear(int32_t lo) {
 107         loyear = lo;
 108     }
 109
 110     void setHighYear(int32_t hi) {
 111         hiyear = hi;
 112     }
 113
 114     void setTick(int32_t t) {
 115         tick = t;
 116     }
 117
 118     void setTimeZone(TimeZone* tz) {
 119         timezone = tz;
 120     }
 121
 122     void setDumpFormatter(DumpFormatter* fmt) {
 123         formatter = fmt;
 124     }
 125
 126     void setLineSeparator(const char* sep) {
 127         linesep = sep;
 128     }
 129
 130     void dump(ostream& out) {
 131         UErrorCode status = U_ZERO_ERROR;
 132         UDate SEARCH_INCREMENT = 12 * 60 * 60 * 1000; // half day
 133         UDate t, cutlo, cuthi;
 134         int32_t rawOffset, dstOffset;
 135         UnicodeString str;
 136
 137         getCutOverTimes(cutlo, cuthi);
 138         t = cutlo;
 139         timezone->getOffset(t, FALSE, rawOffset, dstOffset, status);
 140         while (t < cuthi) {
 141             int32_t newRawOffset, newDstOffset;
 142             UDate newt = t + SEARCH_INCREMENT;
 143
 144             timezone->getOffset(newt, FALSE, newRawOffset, newDstOffset, status);
 145
 146             UBool bSameOffset = (rawOffset + dstOffset) == (newRawOffset + newDstOffset);
 147             UBool bSameDst = ((dstOffset != 0) && (newDstOffset != 0)) || ((dstOffset == 0) && (newDstOffset == 0));
 148
 149             if (!bSameOffset || !bSameDst) {
 150                 // find the boundary
 151                 UDate lot = t;
 152                 UDate hit = newt;
 153                 while (true) {
 154                     int32_t diff = (int32_t)(hit - lot);
 155                     if (diff <= tick) {
 156                         break;
 157                     }
 158                     UDate medt = lot + ((diff / 2) / tick) * tick;
 159                     int32_t medRawOffset, medDstOffset;
 160                     timezone->getOffset(medt, FALSE, medRawOffset, medDstOffset, status);
 161
 162                     bSameOffset = (rawOffset + dstOffset) == (medRawOffset + medDstOffset);
 163                     bSameDst = ((dstOffset != 0) && (medDstOffset != 0)) || ((dstOffset == 0) && (medDstOffset == 0));
 164
 165                     if (!bSameOffset || !bSameDst) {
 166                         hit = medt;
 167                     } else {
 168                         lot = medt;
 169                     }
 170                 }
 171                 // write out the boundary
 172                 str.remove();
 173                 formatter->format(lot, rawOffset + dstOffset, (dstOffset == 0 ? FALSE : TRUE), str);
 174                 out << str << " > ";
 175                 str.remove();
 176                 formatter->format(hit, newRawOffset + newDstOffset, (newDstOffset == 0 ? FALSE : TRUE), str);
 177                 out << str;
 178                 if (linesep != NULL) {
 179                     out << linesep;
 180                 } else {
 181                     out << endl;
 182                 }
 183
 184                 rawOffset = newRawOffset;
 185                 dstOffset = newDstOffset;
 186             }
 187             t = newt;
 188         }
 189     }
 190
 191 private:
 192     void getCutOverTimes(UDate& lo, UDate& hi) {
 193         UErrorCode status = U_ZERO_ERROR;
 194         GregorianCalendar* gcal = new GregorianCalendar(timezone, Locale::getEnglish(), status);
 195         gcal->clear();
 196         gcal->set(loyear, 0, 1, 0, 0, 0);
 197         lo = gcal->getTime(status);
 198         gcal->set(hiyear, 0, 1, 0, 0, 0);
 199         hi = gcal->getTime(status);
 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:
 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
 225     ZoneIterator(const char** ids, int32_t num) {
 226         zenum = NULL;
 227         zids = ids;
 228         idx = 0;
 229         numids = num;
 230     }
 231
 232     ~ZoneIterator() {
 233         if (zenum != NULL) {
 234             delete zenum;
 235         }
 236     }
 237
 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
 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, UPRV_LENGTHOF(options), 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.str() << 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 }