1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/intl/icu/source/tools/tzcode/icuzdump.cpp Wed Dec 31 06:09:35 2014 +0100
1.3 @@ -0,0 +1,423 @@
1.4 +/*
1.5 +*******************************************************************************
1.6 +*
1.7 +* Copyright (C) 2007, International Business Machines
1.8 +* Corporation and others. All Rights Reserved.
1.9 +*
1.10 +*******************************************************************************
1.11 +* file name: icuzdump.cpp
1.12 +* encoding: US-ASCII
1.13 +* tab size: 8 (not used)
1.14 +* indentation:4
1.15 +*
1.16 +* created on: 2007-04-02
1.17 +* created by: Yoshito Umaoka
1.18 +*
1.19 +* This tool write out timezone transitions for ICU timezone. This tool
1.20 +* is used as a part of tzdata update process to check if ICU timezone
1.21 +* code works as well as the corresponding Olson stock localtime/zdump.
1.22 +*/
1.23 +
1.24 +#include <cstdlib>
1.25 +#include <cstring>
1.26 +#include <fstream>
1.27 +#include <sstream>
1.28 +#include <iostream>
1.29 +
1.30 +#include "unicode/utypes.h"
1.31 +#include "unicode/ustring.h"
1.32 +#include "unicode/timezone.h"
1.33 +#include "unicode/simpletz.h"
1.34 +#include "unicode/smpdtfmt.h"
1.35 +#include "unicode/decimfmt.h"
1.36 +#include "unicode/gregocal.h"
1.37 +#include "unicode/ustream.h"
1.38 +#include "unicode/putil.h"
1.39 +
1.40 +#include "uoptions.h"
1.41 +
1.42 +using namespace std;
1.43 +
1.44 +class DumpFormatter {
1.45 +public:
1.46 + DumpFormatter() {
1.47 + UErrorCode status = U_ZERO_ERROR;
1.48 + stz = new SimpleTimeZone(0, "");
1.49 + sdf = new SimpleDateFormat((UnicodeString)"yyyy-MM-dd EEE HH:mm:ss", Locale::getEnglish(), status);
1.50 + DecimalFormatSymbols *symbols = new DecimalFormatSymbols(Locale::getEnglish(), status);
1.51 + decf = new DecimalFormat("00", symbols, status);
1.52 + }
1.53 + ~DumpFormatter() {
1.54 + }
1.55 +
1.56 + UnicodeString& format(UDate time, int32_t offset, UBool isDst, UnicodeString& appendTo) {
1.57 + stz->setRawOffset(offset);
1.58 + sdf->setTimeZone(*stz);
1.59 + UnicodeString str = sdf->format(time, appendTo);
1.60 + if (offset < 0) {
1.61 + appendTo += "-";
1.62 + offset = -offset;
1.63 + } else {
1.64 + appendTo += "+";
1.65 + }
1.66 +
1.67 + int32_t hour, min, sec;
1.68 +
1.69 + offset /= 1000;
1.70 + sec = offset % 60;
1.71 + offset = (offset - sec) / 60;
1.72 + min = offset % 60;
1.73 + hour = offset / 60;
1.74 +
1.75 + decf->format(hour, appendTo);
1.76 + decf->format(min, appendTo);
1.77 + decf->format(sec, appendTo);
1.78 + appendTo += "[DST=";
1.79 + if (isDst) {
1.80 + appendTo += "1";
1.81 + } else {
1.82 + appendTo += "0";
1.83 + }
1.84 + appendTo += "]";
1.85 + return appendTo;
1.86 + }
1.87 +private:
1.88 + SimpleTimeZone* stz;
1.89 + SimpleDateFormat* sdf;
1.90 + DecimalFormat* decf;
1.91 +};
1.92 +
1.93 +class ICUZDump {
1.94 +public:
1.95 + ICUZDump() {
1.96 + formatter = new DumpFormatter();
1.97 + loyear = 1902;
1.98 + hiyear = 2050;
1.99 + tick = 1000;
1.100 + linesep = NULL;
1.101 + }
1.102 +
1.103 + ~ICUZDump() {
1.104 + }
1.105 +
1.106 + void setLowYear(int32_t lo) {
1.107 + loyear = lo;
1.108 + }
1.109 +
1.110 + void setHighYear(int32_t hi) {
1.111 + hiyear = hi;
1.112 + }
1.113 +
1.114 + void setTick(int32_t t) {
1.115 + tick = t;
1.116 + }
1.117 +
1.118 + void setTimeZone(TimeZone* tz) {
1.119 + timezone = tz;
1.120 + }
1.121 +
1.122 + void setDumpFormatter(DumpFormatter* fmt) {
1.123 + formatter = fmt;
1.124 + }
1.125 +
1.126 + void setLineSeparator(const char* sep) {
1.127 + linesep = sep;
1.128 + }
1.129 +
1.130 + void dump(ostream& out) {
1.131 + UErrorCode status = U_ZERO_ERROR;
1.132 + UDate SEARCH_INCREMENT = 12 * 60 * 60 * 1000; // half day
1.133 + UDate t, cutlo, cuthi;
1.134 + int32_t rawOffset, dstOffset;
1.135 + UnicodeString str;
1.136 +
1.137 + getCutOverTimes(cutlo, cuthi);
1.138 + t = cutlo;
1.139 + timezone->getOffset(t, FALSE, rawOffset, dstOffset, status);
1.140 + while (t < cuthi) {
1.141 + int32_t newRawOffset, newDstOffset;
1.142 + UDate newt = t + SEARCH_INCREMENT;
1.143 +
1.144 + timezone->getOffset(newt, FALSE, newRawOffset, newDstOffset, status);
1.145 +
1.146 + UBool bSameOffset = (rawOffset + dstOffset) == (newRawOffset + newDstOffset);
1.147 + UBool bSameDst = ((dstOffset != 0) && (newDstOffset != 0)) || ((dstOffset == 0) && (newDstOffset == 0));
1.148 +
1.149 + if (!bSameOffset || !bSameDst) {
1.150 + // find the boundary
1.151 + UDate lot = t;
1.152 + UDate hit = newt;
1.153 + while (true) {
1.154 + int32_t diff = (int32_t)(hit - lot);
1.155 + if (diff <= tick) {
1.156 + break;
1.157 + }
1.158 + UDate medt = lot + ((diff / 2) / tick) * tick;
1.159 + int32_t medRawOffset, medDstOffset;
1.160 + timezone->getOffset(medt, FALSE, medRawOffset, medDstOffset, status);
1.161 +
1.162 + bSameOffset = (rawOffset + dstOffset) == (medRawOffset + medDstOffset);
1.163 + bSameDst = ((dstOffset != 0) && (medDstOffset != 0)) || ((dstOffset == 0) && (medDstOffset == 0));
1.164 +
1.165 + if (!bSameOffset || !bSameDst) {
1.166 + hit = medt;
1.167 + } else {
1.168 + lot = medt;
1.169 + }
1.170 + }
1.171 + // write out the boundary
1.172 + str.remove();
1.173 + formatter->format(lot, rawOffset + dstOffset, (dstOffset == 0 ? FALSE : TRUE), str);
1.174 + out << str << " > ";
1.175 + str.remove();
1.176 + formatter->format(hit, newRawOffset + newDstOffset, (newDstOffset == 0 ? FALSE : TRUE), str);
1.177 + out << str;
1.178 + if (linesep != NULL) {
1.179 + out << linesep;
1.180 + } else {
1.181 + out << endl;
1.182 + }
1.183 +
1.184 + rawOffset = newRawOffset;
1.185 + dstOffset = newDstOffset;
1.186 + }
1.187 + t = newt;
1.188 + }
1.189 + }
1.190 +
1.191 +private:
1.192 + void getCutOverTimes(UDate& lo, UDate& hi) {
1.193 + UErrorCode status = U_ZERO_ERROR;
1.194 + GregorianCalendar* gcal = new GregorianCalendar(timezone, Locale::getEnglish(), status);
1.195 + gcal->clear();
1.196 + gcal->set(loyear, 0, 1, 0, 0, 0);
1.197 + lo = gcal->getTime(status);
1.198 + gcal->set(hiyear, 0, 1, 0, 0, 0);
1.199 + hi = gcal->getTime(status);
1.200 + }
1.201 +
1.202 + void dumpZone(ostream& out, const char* linesep, UnicodeString tzid, int32_t low, int32_t high) {
1.203 + }
1.204 +
1.205 + TimeZone* timezone;
1.206 + int32_t loyear;
1.207 + int32_t hiyear;
1.208 + int32_t tick;
1.209 +
1.210 + DumpFormatter* formatter;
1.211 + const char* linesep;
1.212 +};
1.213 +
1.214 +class ZoneIterator {
1.215 +public:
1.216 + ZoneIterator(UBool bAll = FALSE) {
1.217 + if (bAll) {
1.218 + zenum = TimeZone::createEnumeration();
1.219 + }
1.220 + else {
1.221 + zenum = NULL;
1.222 + zids = NULL;
1.223 + idx = 0;
1.224 + numids = 1;
1.225 + }
1.226 + }
1.227 +
1.228 + ZoneIterator(const char** ids, int32_t num) {
1.229 + zenum = NULL;
1.230 + zids = ids;
1.231 + idx = 0;
1.232 + numids = num;
1.233 + }
1.234 +
1.235 + ~ZoneIterator() {
1.236 + if (zenum != NULL) {
1.237 + delete zenum;
1.238 + }
1.239 + }
1.240 +
1.241 + TimeZone* next() {
1.242 + TimeZone* tz = NULL;
1.243 + if (zenum != NULL) {
1.244 + UErrorCode status = U_ZERO_ERROR;
1.245 + const UnicodeString* zid = zenum->snext(status);
1.246 + if (zid != NULL) {
1.247 + tz = TimeZone::createTimeZone(*zid);
1.248 + }
1.249 + }
1.250 + else {
1.251 + if (idx < numids) {
1.252 + if (zids != NULL) {
1.253 + tz = TimeZone::createTimeZone((const UnicodeString&)zids[idx]);
1.254 + }
1.255 + else {
1.256 + tz = TimeZone::createDefault();
1.257 + }
1.258 + idx++;
1.259 + }
1.260 + }
1.261 + return tz;
1.262 + }
1.263 +
1.264 +private:
1.265 + const char** zids;
1.266 + StringEnumeration* zenum;
1.267 + int32_t idx;
1.268 + int32_t numids;
1.269 +};
1.270 +
1.271 +enum {
1.272 + kOptHelpH = 0,
1.273 + kOptHelpQuestionMark,
1.274 + kOptAllZones,
1.275 + kOptCutover,
1.276 + kOptDestDir,
1.277 + kOptLineSep
1.278 +};
1.279 +
1.280 +static UOption options[]={
1.281 + UOPTION_HELP_H,
1.282 + UOPTION_HELP_QUESTION_MARK,
1.283 + UOPTION_DEF("allzones", 'a', UOPT_NO_ARG),
1.284 + UOPTION_DEF("cutover", 'c', UOPT_REQUIRES_ARG),
1.285 + UOPTION_DEF("destdir", 'd', UOPT_REQUIRES_ARG),
1.286 + UOPTION_DEF("linesep", 'l', UOPT_REQUIRES_ARG)
1.287 +};
1.288 +
1.289 +extern int
1.290 +main(int argc, char *argv[]) {
1.291 + int32_t low = 1902;
1.292 + int32_t high = 2038;
1.293 + UBool bAll = FALSE;
1.294 + const char *dir = NULL;
1.295 + const char *linesep = NULL;
1.296 +
1.297 + U_MAIN_INIT_ARGS(argc, argv);
1.298 + argc = u_parseArgs(argc, argv, sizeof(options)/sizeof(options[0]), options);
1.299 +
1.300 + if (argc < 0) {
1.301 + cerr << "Illegal command line argument(s)" << endl << endl;
1.302 + }
1.303 +
1.304 + if (argc < 0 || options[kOptHelpH].doesOccur || options[kOptHelpQuestionMark].doesOccur) {
1.305 + cerr
1.306 + << "Usage: icuzdump [-options] [zoneid1 zoneid2 ...]" << endl
1.307 + << endl
1.308 + << "\tDump all offset transitions for the specified zones." << endl
1.309 + << endl
1.310 + << "Options:" << endl
1.311 + << "\t-a : Dump all available zones." << endl
1.312 + << "\t-d <dir> : When specified, write transitions in a file under" << endl
1.313 + << "\t the directory for each zone." << endl
1.314 + << "\t-l <sep> : New line code type used in file outputs. CR or LF (default)"
1.315 + << "\t or CRLF." << endl
1.316 + << "\t-c [<low_year>,]<high_year>" << endl
1.317 + << "\t : When specified, dump transitions starting <low_year>" << endl
1.318 + << "\t (inclusive) up to <high_year> (exclusive). The default" << endl
1.319 + << "\t values are 1902(low) and 2038(high)." << endl;
1.320 + return argc < 0 ? U_ILLEGAL_ARGUMENT_ERROR : U_ZERO_ERROR;
1.321 + }
1.322 +
1.323 + bAll = options[kOptAllZones].doesOccur;
1.324 +
1.325 + if (options[kOptDestDir].doesOccur) {
1.326 + dir = options[kOptDestDir].value;
1.327 + }
1.328 +
1.329 + if (options[kOptLineSep].doesOccur) {
1.330 + if (strcmp(options[kOptLineSep].value, "CR") == 0) {
1.331 + linesep = "\r";
1.332 + } else if (strcmp(options[kOptLineSep].value, "CRLF") == 0) {
1.333 + linesep = "\r\n";
1.334 + } else if (strcmp(options[kOptLineSep].value, "LF") == 0) {
1.335 + linesep = "\n";
1.336 + }
1.337 + }
1.338 +
1.339 + if (options[kOptCutover].doesOccur) {
1.340 + char* comma = (char*)strchr(options[kOptCutover].value, ',');
1.341 + if (comma == NULL) {
1.342 + high = atoi(options[kOptCutover].value);
1.343 + } else {
1.344 + *comma = 0;
1.345 + low = atoi(options[kOptCutover].value);
1.346 + high = atoi(comma + 1);
1.347 + }
1.348 + }
1.349 +
1.350 + ICUZDump dumper;
1.351 + dumper.setLowYear(low);
1.352 + dumper.setHighYear(high);
1.353 + if (dir != NULL && linesep != NULL) {
1.354 + // use the specified line separator only for file output
1.355 + dumper.setLineSeparator((const char*)linesep);
1.356 + }
1.357 +
1.358 + ZoneIterator* zit;
1.359 + if (bAll) {
1.360 + zit = new ZoneIterator(TRUE);
1.361 + } else {
1.362 + if (argc <= 1) {
1.363 + zit = new ZoneIterator();
1.364 + } else {
1.365 + zit = new ZoneIterator((const char**)&argv[1], argc - 1);
1.366 + }
1.367 + }
1.368 +
1.369 + UnicodeString id;
1.370 + if (dir != NULL) {
1.371 + // file output
1.372 + ostringstream path;
1.373 + ios::openmode mode = ios::out;
1.374 + if (linesep != NULL) {
1.375 + mode |= ios::binary;
1.376 + }
1.377 + for (;;) {
1.378 + TimeZone* tz = zit->next();
1.379 + if (tz == NULL) {
1.380 + break;
1.381 + }
1.382 + dumper.setTimeZone(tz);
1.383 + tz->getID(id);
1.384 +
1.385 + // target file path
1.386 + path.str("");
1.387 + path << dir << U_FILE_SEP_CHAR;
1.388 + id = id.findAndReplace("/", "-");
1.389 + path << id;
1.390 +
1.391 + ofstream* fout = new ofstream(path.str().c_str(), mode);
1.392 + if (fout->fail()) {
1.393 + cerr << "Cannot open file " << path << endl;
1.394 + delete fout;
1.395 + delete tz;
1.396 + break;
1.397 + }
1.398 +
1.399 + dumper.dump(*fout);
1.400 + fout->close();
1.401 + delete fout;
1.402 + delete tz;
1.403 + }
1.404 +
1.405 + } else {
1.406 + // stdout
1.407 + UBool bFirst = TRUE;
1.408 + for (;;) {
1.409 + TimeZone* tz = zit->next();
1.410 + if (tz == NULL) {
1.411 + break;
1.412 + }
1.413 + dumper.setTimeZone(tz);
1.414 + tz->getID(id);
1.415 + if (bFirst) {
1.416 + bFirst = FALSE;
1.417 + } else {
1.418 + cout << endl;
1.419 + }
1.420 + cout << "ZONE: " << id << endl;
1.421 + dumper.dump(cout);
1.422 + delete tz;
1.423 + }
1.424 + }
1.425 + delete zit;
1.426 +}
