michael@0: /*
michael@0: *******************************************************************************
michael@0: *
michael@0: *   Copyright (C) 2007, International Business Machines
michael@0: *   Corporation and others.  All Rights Reserved.
michael@0: *
michael@0: *******************************************************************************
michael@0: *   file name:  icuzdump.cpp
michael@0: *   encoding:   US-ASCII
michael@0: *   tab size:   8 (not used)
michael@0: *   indentation:4
michael@0: *
michael@0: *   created on: 2007-04-02
michael@0: *   created by: Yoshito Umaoka
michael@0: *
michael@0: *   This tool write out timezone transitions for ICU timezone.  This tool
michael@0: *   is used as a part of tzdata update process to check if ICU timezone
michael@0: *   code works as well as the corresponding Olson stock localtime/zdump.
michael@0: */
michael@0: 
michael@0: #include <cstdlib>
michael@0: #include <cstring>
michael@0: #include <fstream>
michael@0: #include <sstream>
michael@0: #include <iostream>
michael@0: 
michael@0: #include "unicode/utypes.h"
michael@0: #include "unicode/ustring.h"
michael@0: #include "unicode/timezone.h"
michael@0: #include "unicode/simpletz.h"
michael@0: #include "unicode/smpdtfmt.h"
michael@0: #include "unicode/decimfmt.h"
michael@0: #include "unicode/gregocal.h"
michael@0: #include "unicode/ustream.h"
michael@0: #include "unicode/putil.h"
michael@0: 
michael@0: #include "uoptions.h"
michael@0: 
michael@0: using namespace std;
michael@0: 
michael@0: class DumpFormatter {
michael@0: public:
michael@0:     DumpFormatter() {
michael@0:         UErrorCode status = U_ZERO_ERROR;
michael@0:         stz = new SimpleTimeZone(0, "");
michael@0:         sdf = new SimpleDateFormat((UnicodeString)"yyyy-MM-dd EEE HH:mm:ss", Locale::getEnglish(), status);
michael@0:         DecimalFormatSymbols *symbols = new DecimalFormatSymbols(Locale::getEnglish(), status);
michael@0:         decf = new DecimalFormat("00", symbols, status);
michael@0:     }
michael@0:     ~DumpFormatter() {
michael@0:     }
michael@0: 
michael@0:     UnicodeString& format(UDate time, int32_t offset, UBool isDst, UnicodeString& appendTo) {
michael@0:         stz->setRawOffset(offset);
michael@0:         sdf->setTimeZone(*stz);
michael@0:         UnicodeString str = sdf->format(time, appendTo);
michael@0:         if (offset < 0) {
michael@0:             appendTo += "-";
michael@0:             offset = -offset;
michael@0:         } else {
michael@0:             appendTo += "+";
michael@0:         }
michael@0: 
michael@0:         int32_t hour, min, sec;
michael@0: 
michael@0:         offset /= 1000;
michael@0:         sec = offset % 60;
michael@0:         offset = (offset - sec) / 60;
michael@0:         min = offset % 60;
michael@0:         hour = offset / 60;
michael@0: 
michael@0:         decf->format(hour, appendTo);
michael@0:         decf->format(min, appendTo);
michael@0:         decf->format(sec, appendTo);
michael@0:         appendTo += "[DST=";
michael@0:         if (isDst) {
michael@0:             appendTo += "1";
michael@0:         } else {
michael@0:             appendTo += "0";
michael@0:         }
michael@0:         appendTo += "]";
michael@0:         return appendTo;
michael@0:     }
michael@0: private:
michael@0:     SimpleTimeZone*     stz;
michael@0:     SimpleDateFormat*   sdf;
michael@0:     DecimalFormat*      decf;
michael@0: };
michael@0: 
michael@0: class ICUZDump {
michael@0: public:
michael@0:     ICUZDump() {
michael@0:         formatter = new DumpFormatter();
michael@0:         loyear = 1902;
michael@0:         hiyear = 2050;
michael@0:         tick = 1000;
michael@0:         linesep = NULL;
michael@0:     }
michael@0: 
michael@0:     ~ICUZDump() {
michael@0:     }
michael@0: 
michael@0:     void setLowYear(int32_t lo) {
michael@0:         loyear = lo;
michael@0:     }
michael@0: 
michael@0:     void setHighYear(int32_t hi) {
michael@0:         hiyear = hi;
michael@0:     }
michael@0: 
michael@0:     void setTick(int32_t t) {
michael@0:         tick = t;
michael@0:     }
michael@0: 
michael@0:     void setTimeZone(TimeZone* tz) {
michael@0:         timezone = tz;
michael@0:     }
michael@0: 
michael@0:     void setDumpFormatter(DumpFormatter* fmt) {
michael@0:         formatter = fmt;
michael@0:     }
michael@0: 
michael@0:     void setLineSeparator(const char* sep) {
michael@0:         linesep = sep;
michael@0:     }
michael@0: 
michael@0:     void dump(ostream& out) {
michael@0:         UErrorCode status = U_ZERO_ERROR;
michael@0:         UDate SEARCH_INCREMENT = 12 * 60 * 60 * 1000; // half day
michael@0:         UDate t, cutlo, cuthi;
michael@0:         int32_t rawOffset, dstOffset;
michael@0:         UnicodeString str;
michael@0: 
michael@0:         getCutOverTimes(cutlo, cuthi);
michael@0:         t = cutlo;
michael@0:         timezone->getOffset(t, FALSE, rawOffset, dstOffset, status);
michael@0:         while (t < cuthi) {
michael@0:             int32_t newRawOffset, newDstOffset;
michael@0:             UDate newt = t + SEARCH_INCREMENT;
michael@0: 
michael@0:             timezone->getOffset(newt, FALSE, newRawOffset, newDstOffset, status);
michael@0: 
michael@0:             UBool bSameOffset = (rawOffset + dstOffset) == (newRawOffset + newDstOffset);
michael@0:             UBool bSameDst = ((dstOffset != 0) && (newDstOffset != 0)) || ((dstOffset == 0) && (newDstOffset == 0));
michael@0: 
michael@0:             if (!bSameOffset || !bSameDst) {
michael@0:                 // find the boundary
michael@0:                 UDate lot = t;
michael@0:                 UDate hit = newt;
michael@0:                 while (true) {
michael@0:                     int32_t diff = (int32_t)(hit - lot);
michael@0:                     if (diff <= tick) {
michael@0:                         break;
michael@0:                     }
michael@0:                     UDate medt = lot + ((diff / 2) / tick) * tick;
michael@0:                     int32_t medRawOffset, medDstOffset;
michael@0:                     timezone->getOffset(medt, FALSE, medRawOffset, medDstOffset, status);
michael@0: 
michael@0:                     bSameOffset = (rawOffset + dstOffset) == (medRawOffset + medDstOffset);
michael@0:                     bSameDst = ((dstOffset != 0) && (medDstOffset != 0)) || ((dstOffset == 0) && (medDstOffset == 0));
michael@0: 
michael@0:                     if (!bSameOffset || !bSameDst) {
michael@0:                         hit = medt;
michael@0:                     } else {
michael@0:                         lot = medt;
michael@0:                     }
michael@0:                 }
michael@0:                 // write out the boundary
michael@0:                 str.remove();
michael@0:                 formatter->format(lot, rawOffset + dstOffset, (dstOffset == 0 ? FALSE : TRUE), str);
michael@0:                 out << str << " > ";
michael@0:                 str.remove();
michael@0:                 formatter->format(hit, newRawOffset + newDstOffset, (newDstOffset == 0 ? FALSE : TRUE), str);
michael@0:                 out << str;
michael@0:                 if (linesep != NULL) {
michael@0:                     out << linesep;
michael@0:                 } else {
michael@0:                     out << endl;
michael@0:                 }
michael@0: 
michael@0:                 rawOffset = newRawOffset;
michael@0:                 dstOffset = newDstOffset;
michael@0:             }
michael@0:             t = newt;
michael@0:         }
michael@0:     }
michael@0: 
michael@0: private:
michael@0:     void getCutOverTimes(UDate& lo, UDate& hi) {
michael@0:         UErrorCode status = U_ZERO_ERROR;
michael@0:         GregorianCalendar* gcal = new GregorianCalendar(timezone, Locale::getEnglish(), status);
michael@0:         gcal->clear();
michael@0:         gcal->set(loyear, 0, 1, 0, 0, 0);
michael@0:         lo = gcal->getTime(status);
michael@0:         gcal->set(hiyear, 0, 1, 0, 0, 0);
michael@0:         hi = gcal->getTime(status);
michael@0:     }
michael@0: 
michael@0:     void dumpZone(ostream& out, const char* linesep, UnicodeString tzid, int32_t low, int32_t high) {
michael@0:     }
michael@0: 
michael@0:     TimeZone*   timezone;
michael@0:     int32_t     loyear;
michael@0:     int32_t     hiyear;
michael@0:     int32_t     tick;
michael@0: 
michael@0:     DumpFormatter*  formatter;
michael@0:     const char*  linesep;
michael@0: };
michael@0: 
michael@0: class ZoneIterator {
michael@0: public:
michael@0:     ZoneIterator(UBool bAll = FALSE) {
michael@0:         if (bAll) {
michael@0:             zenum = TimeZone::createEnumeration();
michael@0:         }
michael@0:         else {
michael@0:             zenum = NULL;
michael@0:             zids = NULL;
michael@0:             idx = 0;
michael@0:             numids = 1;
michael@0:         }
michael@0:     }
michael@0: 
michael@0:     ZoneIterator(const char** ids, int32_t num) {
michael@0:         zenum = NULL;
michael@0:         zids = ids;
michael@0:         idx = 0;
michael@0:         numids = num;
michael@0:     }
michael@0: 
michael@0:     ~ZoneIterator() {
michael@0:         if (zenum != NULL) {
michael@0:             delete zenum;
michael@0:         }
michael@0:     }
michael@0: 
michael@0:     TimeZone* next() {
michael@0:         TimeZone* tz = NULL;
michael@0:         if (zenum != NULL) {
michael@0:             UErrorCode status = U_ZERO_ERROR;
michael@0:             const UnicodeString* zid = zenum->snext(status);
michael@0:             if (zid != NULL) {
michael@0:                 tz = TimeZone::createTimeZone(*zid);
michael@0:             }
michael@0:         }
michael@0:         else {
michael@0:             if (idx < numids) {
michael@0:                 if (zids != NULL) {
michael@0:                     tz = TimeZone::createTimeZone((const UnicodeString&)zids[idx]);
michael@0:                 }
michael@0:                 else {
michael@0:                     tz = TimeZone::createDefault();
michael@0:                 }
michael@0:                 idx++;
michael@0:             }
michael@0:         }
michael@0:         return tz;
michael@0:     }
michael@0: 
michael@0: private:
michael@0:     const char** zids;
michael@0:     StringEnumeration* zenum;
michael@0:     int32_t idx;
michael@0:     int32_t numids;
michael@0: };
michael@0: 
michael@0: enum { 
michael@0:   kOptHelpH = 0,
michael@0:   kOptHelpQuestionMark,
michael@0:   kOptAllZones,
michael@0:   kOptCutover,
michael@0:   kOptDestDir,
michael@0:   kOptLineSep
michael@0: };
michael@0: 
michael@0: static UOption options[]={
michael@0:     UOPTION_HELP_H,
michael@0:     UOPTION_HELP_QUESTION_MARK,
michael@0:     UOPTION_DEF("allzones", 'a', UOPT_NO_ARG),
michael@0:     UOPTION_DEF("cutover", 'c', UOPT_REQUIRES_ARG),
michael@0:     UOPTION_DEF("destdir", 'd', UOPT_REQUIRES_ARG),
michael@0:     UOPTION_DEF("linesep", 'l', UOPT_REQUIRES_ARG)
michael@0: };
michael@0: 
michael@0: extern int
michael@0: main(int argc, char *argv[]) {
michael@0:     int32_t low = 1902;
michael@0:     int32_t high = 2038;
michael@0:     UBool bAll = FALSE;
michael@0:     const char *dir = NULL;
michael@0:     const char *linesep = NULL;
michael@0: 
michael@0:     U_MAIN_INIT_ARGS(argc, argv);
michael@0:     argc = u_parseArgs(argc, argv, sizeof(options)/sizeof(options[0]), options);
michael@0: 
michael@0:     if (argc < 0) {
michael@0:         cerr << "Illegal command line argument(s)" << endl << endl;
michael@0:     }
michael@0: 
michael@0:     if (argc < 0 || options[kOptHelpH].doesOccur || options[kOptHelpQuestionMark].doesOccur) {
michael@0:         cerr
michael@0:             << "Usage: icuzdump [-options] [zoneid1 zoneid2 ...]" << endl
michael@0:             << endl
michael@0:             << "\tDump all offset transitions for the specified zones." << endl
michael@0:             << endl
michael@0:             << "Options:" << endl
michael@0:             << "\t-a       : Dump all available zones." << endl
michael@0:             << "\t-d <dir> : When specified, write transitions in a file under" << endl
michael@0:             << "\t           the directory for each zone." << endl
michael@0:             << "\t-l <sep> : New line code type used in file outputs. CR or LF (default)"
michael@0:             << "\t           or CRLF." << endl
michael@0:             << "\t-c [<low_year>,]<high_year>" << endl
michael@0:             << "\t         : When specified, dump transitions starting <low_year>" << endl
michael@0:             << "\t           (inclusive) up to <high_year> (exclusive).  The default" << endl
michael@0:             << "\t           values are 1902(low) and 2038(high)." << endl;
michael@0:         return argc < 0 ? U_ILLEGAL_ARGUMENT_ERROR : U_ZERO_ERROR;
michael@0:     }
michael@0: 
michael@0:     bAll = options[kOptAllZones].doesOccur;
michael@0: 
michael@0:     if (options[kOptDestDir].doesOccur) {
michael@0:         dir = options[kOptDestDir].value;
michael@0:     }
michael@0: 
michael@0:     if (options[kOptLineSep].doesOccur) {
michael@0:         if (strcmp(options[kOptLineSep].value, "CR") == 0) {
michael@0:             linesep = "\r";
michael@0:         } else if (strcmp(options[kOptLineSep].value, "CRLF") == 0) {
michael@0:             linesep = "\r\n";
michael@0:         } else if (strcmp(options[kOptLineSep].value, "LF") == 0) {
michael@0:             linesep = "\n";
michael@0:         }
michael@0:     }
michael@0: 
michael@0:     if (options[kOptCutover].doesOccur) {
michael@0:         char* comma = (char*)strchr(options[kOptCutover].value, ',');
michael@0:         if (comma == NULL) {
michael@0:             high = atoi(options[kOptCutover].value);
michael@0:         } else {
michael@0:             *comma = 0;
michael@0:             low = atoi(options[kOptCutover].value);
michael@0:             high = atoi(comma + 1);
michael@0:         }
michael@0:     }
michael@0: 
michael@0:     ICUZDump dumper;
michael@0:     dumper.setLowYear(low);
michael@0:     dumper.setHighYear(high);
michael@0:     if (dir != NULL && linesep != NULL) {
michael@0:         // use the specified line separator only for file output
michael@0:         dumper.setLineSeparator((const char*)linesep);
michael@0:     }
michael@0: 
michael@0:     ZoneIterator* zit;
michael@0:     if (bAll) {
michael@0:         zit = new ZoneIterator(TRUE);
michael@0:     } else {
michael@0:         if (argc <= 1) {
michael@0:             zit = new ZoneIterator();
michael@0:         } else {
michael@0:             zit = new ZoneIterator((const char**)&argv[1], argc - 1);
michael@0:         }
michael@0:     }
michael@0: 
michael@0:     UnicodeString id;
michael@0:     if (dir != NULL) {
michael@0:         // file output
michael@0:         ostringstream path;
michael@0:         ios::openmode mode = ios::out;
michael@0:         if (linesep != NULL) {
michael@0:             mode |= ios::binary;
michael@0:         }
michael@0:         for (;;) {
michael@0:             TimeZone* tz = zit->next();
michael@0:             if (tz == NULL) {
michael@0:                 break;
michael@0:             }
michael@0:             dumper.setTimeZone(tz);
michael@0:             tz->getID(id);
michael@0: 
michael@0:             // target file path
michael@0:             path.str("");
michael@0:             path << dir << U_FILE_SEP_CHAR;
michael@0:             id = id.findAndReplace("/", "-");
michael@0:             path << id;
michael@0: 
michael@0:             ofstream* fout = new ofstream(path.str().c_str(), mode);
michael@0:             if (fout->fail()) {
michael@0:                 cerr << "Cannot open file " << path << endl;
michael@0:                 delete fout;
michael@0:                 delete tz;
michael@0:                 break;
michael@0:             }
michael@0: 
michael@0:             dumper.dump(*fout);
michael@0:             fout->close();
michael@0:             delete fout;
michael@0:             delete tz;
michael@0:         }
michael@0: 
michael@0:     } else {
michael@0:         // stdout
michael@0:         UBool bFirst = TRUE;
michael@0:         for (;;) {
michael@0:             TimeZone* tz = zit->next();
michael@0:             if (tz == NULL) {
michael@0:                 break;
michael@0:             }
michael@0:             dumper.setTimeZone(tz);
michael@0:             tz->getID(id);
michael@0:             if (bFirst) {
michael@0:                 bFirst = FALSE;
michael@0:             } else {
michael@0:                 cout << endl;
michael@0:             }
michael@0:             cout << "ZONE: " << id << endl;
michael@0:             dumper.dump(cout);
michael@0:             delete tz;
michael@0:         }
michael@0:     }
michael@0:     delete zit;
michael@0: }