1/*
2*******************************************************************************
3*
4*   Copyright (C) 2007-2014, 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
39using namespace std;
40
41class DumpFormatter {
42public:
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    }
50    ~DumpFormatter() {
51    }
52
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    }
84private:
85    SimpleTimeZone*     stz;
86    SimpleDateFormat*   sdf;
87    DecimalFormat*      decf;
88};
89
90class ICUZDump {
91public:
92    ICUZDump() {
93        formatter = new DumpFormatter();
94        loyear = 1902;
95        hiyear = 2050;
96        tick = 1000;
97        linesep = NULL;
98    }
99
100    ~ICUZDump() {
101    }
102
103    void setLowYear(int32_t lo) {
104        loyear = lo;
105    }
106
107    void setHighYear(int32_t hi) {
108        hiyear = hi;
109    }
110
111    void setTick(int32_t t) {
112        tick = t;
113    }
114
115    void setTimeZone(TimeZone* tz) {
116        timezone = tz;
117    }
118
119    void setDumpFormatter(DumpFormatter* fmt) {
120        formatter = fmt;
121    }
122
123    void setLineSeparator(const char* sep) {
124        linesep = sep;
125    }
126
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
188private:
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
199    TimeZone*   timezone;
200    int32_t     loyear;
201    int32_t     hiyear;
202    int32_t     tick;
203
204    DumpFormatter*  formatter;
205    const char*  linesep;
206};
207
208class ZoneIterator {
209public:
210    ZoneIterator(UBool bAll = FALSE) {
211        if (bAll) {
212            zenum = TimeZone::createEnumeration();
213        }
214        else {
215            zenum = NULL;
216            zids = NULL;
217            idx = 0;
218            numids = 1;
219        }
220    }
221
222    ZoneIterator(const char** ids, int32_t num) {
223        zenum = NULL;
224        zids = ids;
225        idx = 0;
226        numids = num;
227    }
228
229    ~ZoneIterator() {
230        if (zenum != NULL) {
231            delete zenum;
232        }
233    }
234
235    TimeZone* next() {
236        TimeZone* tz = NULL;
237        if (zenum != NULL) {
238            UErrorCode status = U_ZERO_ERROR;
239            const UnicodeString* zid = zenum->snext(status);
240            if (zid != NULL) {
241                tz = TimeZone::createTimeZone(*zid);
242            }
243        }
244        else {
245            if (idx < numids) {
246                if (zids != NULL) {
247                    tz = TimeZone::createTimeZone((const UnicodeString&)zids[idx]);
248                }
249                else {
250                    tz = TimeZone::createDefault();
251                }
252                idx++;
253            }
254        }
255        return tz;
256    }
257
258private:
259    const char** zids;
260    StringEnumeration* zenum;
261    int32_t idx;
262    int32_t numids;
263};
264
265enum {
266  kOptHelpH = 0,
267  kOptHelpQuestionMark,
268  kOptAllZones,
269  kOptCutover,
270  kOptDestDir,
271  kOptLineSep
272};
273
274static UOption options[]={
275    UOPTION_HELP_H,
276    UOPTION_HELP_QUESTION_MARK,
277    UOPTION_DEF("allzones", 'a', UOPT_NO_ARG),
278    UOPTION_DEF("cutover", 'c', UOPT_REQUIRES_ARG),
279    UOPTION_DEF("destdir", 'd', UOPT_REQUIRES_ARG),
280    UOPTION_DEF("linesep", 'l', UOPT_REQUIRES_ARG)
281};
282
283extern int
284main(int argc, char *argv[]) {
285    int32_t low = 1902;
286    int32_t high = 2038;
287    UBool bAll = FALSE;
288    const char *dir = NULL;
289    const char *linesep = NULL;
290
291    U_MAIN_INIT_ARGS(argc, argv);
292    argc = u_parseArgs(argc, argv, sizeof(options)/sizeof(options[0]), options);
293
294    if (argc < 0) {
295        cerr << "Illegal command line argument(s)" << endl << endl;
296    }
297
298    if (argc < 0 || options[kOptHelpH].doesOccur || options[kOptHelpQuestionMark].doesOccur) {
299        cerr
300            << "Usage: icuzdump [-options] [zoneid1 zoneid2 ...]" << endl
301            << endl
302            << "\tDump all offset transitions for the specified zones." << endl
303            << endl
304            << "Options:" << endl
305            << "\t-a       : Dump all available zones." << endl
306            << "\t-d <dir> : When specified, write transitions in a file under" << endl
307            << "\t           the directory for each zone." << endl
308            << "\t-l <sep> : New line code type used in file outputs. CR or LF (default)"
309            << "\t           or CRLF." << endl
310            << "\t-c [<low_year>,]<high_year>" << endl
311            << "\t         : When specified, dump transitions starting <low_year>" << endl
312            << "\t           (inclusive) up to <high_year> (exclusive).  The default" << endl
313            << "\t           values are 1902(low) and 2038(high)." << endl;
314        return argc < 0 ? U_ILLEGAL_ARGUMENT_ERROR : U_ZERO_ERROR;
315    }
316
317    bAll = options[kOptAllZones].doesOccur;
318
319    if (options[kOptDestDir].doesOccur) {
320        dir = options[kOptDestDir].value;
321    }
322
323    if (options[kOptLineSep].doesOccur) {
324        if (strcmp(options[kOptLineSep].value, "CR") == 0) {
325            linesep = "\r";
326        } else if (strcmp(options[kOptLineSep].value, "CRLF") == 0) {
327            linesep = "\r\n";
328        } else if (strcmp(options[kOptLineSep].value, "LF") == 0) {
329            linesep = "\n";
330        }
331    }
332
333    if (options[kOptCutover].doesOccur) {
334        char* comma = (char*)strchr(options[kOptCutover].value, ',');
335        if (comma == NULL) {
336            high = atoi(options[kOptCutover].value);
337        } else {
338            *comma = 0;
339            low = atoi(options[kOptCutover].value);
340            high = atoi(comma + 1);
341        }
342    }
343
344    ICUZDump dumper;
345    dumper.setLowYear(low);
346    dumper.setHighYear(high);
347    if (dir != NULL && linesep != NULL) {
348        // use the specified line separator only for file output
349        dumper.setLineSeparator((const char*)linesep);
350    }
351
352    ZoneIterator* zit;
353    if (bAll) {
354        zit = new ZoneIterator(TRUE);
355    } else {
356        if (argc <= 1) {
357            zit = new ZoneIterator();
358        } else {
359            zit = new ZoneIterator((const char**)&argv[1], argc - 1);
360        }
361    }
362
363    UnicodeString id;
364    if (dir != NULL) {
365        // file output
366        ostringstream path;
367        ios::openmode mode = ios::out;
368        if (linesep != NULL) {
369            mode |= ios::binary;
370        }
371        for (;;) {
372            TimeZone* tz = zit->next();
373            if (tz == NULL) {
374                break;
375            }
376            dumper.setTimeZone(tz);
377            tz->getID(id);
378
379            // target file path
380            path.str("");
381            path << dir << U_FILE_SEP_CHAR;
382            id = id.findAndReplace("/", "-");
383            path << id;
384
385            ofstream* fout = new ofstream(path.str().c_str(), mode);
386            if (fout->fail()) {
387                cerr << "Cannot open file " << path << endl;
388                delete fout;
389                delete tz;
390                break;
391            }
392
393            dumper.dump(*fout);
394            fout->close();
395            delete fout;
396            delete tz;
397        }
398
399    } else {
400        // stdout
401        UBool bFirst = TRUE;
402        for (;;) {
403            TimeZone* tz = zit->next();
404            if (tz == NULL) {
405                break;
406            }
407            dumper.setTimeZone(tz);
408            tz->getID(id);
409            if (bFirst) {
410                bFirst = FALSE;
411            } else {
412                cout << endl;
413            }
414            cout << "ZONE: " << id << endl;
415            dumper.dump(cout);
416            delete tz;
417        }
418    }
419    delete zit;
420}
421