I want to count the months between two dates, and only compare with their months while ignoring the days, so 2012-01-31 and 2012-02-01 should have 1 month difference between them.
I do this with joda-time:
import org.joda.time.*;
PeriodType yearMonth = PeriodType.yearMonthDay().withDaysRemoved();
LocalDate dt1 = new LocalDate(2012, 1, 31);
LocalDate dt2 = new LocalDate(2012, 2, 1);
int months = new Period(dt1, dt2, yearMonth).getMonths();
System.out.println(months);
But I got printed output:
0
Although I used .withDaysRemoved(), but it's not working. Do I misuse it?
java.time
The Joda-Time project is now in maintenance mode. The team advises migration to the java.time classes.
LocalDate
As in Joda-Time, a LocalDate represents a date-only value without time-of-day and without a time zone.
LocalDate start = LocalDate.of ( 2012 , Month.JANUARY , 31 );
LocalDate stop = LocalDate.of ( 2012 , Month.MARCH , 1 );
Establish the place to hold our results.
long countMonths = 0;
YearMonth
The YearMonth class represents, well, a year and a month. This is the conduit to considering whole months, as you want, rather than number of days.
YearMonth ymStart = YearMonth.from ( start );
YearMonth ymStop = YearMonth.from ( stop );
This YearMonth class has methods for comparison such as isAfter, isBefore, and equals.
The ChronoUnit class can calculate elapsed months between a pair of Temporal objects. The YearMonth class fortunately implements Temporal and works with ChronoUnit.
if ( ymStart.isAfter ( ymStop ) ) {
// Handle error condition. Unexpected input values.
} else if ( ymStart.equals ( ymStop ) ) {
countMonths = 0;
} else { // Else different months.
countMonths = ChronoUnit.MONTHS.between ( ymStart , ymStop );
}
Dump to console.
System.out.println ( "start: " + start + " | stop: " + stop + " | countMonths: " + countMonths );
start: 2012-01-31 | stop: 2012-03-01 | countMonths: 2
Half-Open
The results above are based on the Half-Open approach to defining a span of time where the beginning is inclusive while the ending is exclusive. This approach is commonly used with date-time work, including throughout the java.time classes.
So, for example, lunch period starts at the first moment of the noon hour and runs up to, but not including, the first moment of 1 PM.
Another example: A week would start on a Monday and run up to, but not including, the following Monday. Or in this case, a pair of year-months from January and going up to, but not including, March; they yield a span of 2 months.
I believe using the Half-Open approach throughout your code leads to more clarity and less bugs. But if you or your users demand otherwise, add one to the results above.
About java.time
The java.time framework is built into Java 8 and later. These classes supplant the troublesome old date-time classes such as java.util.Date, .Calendar, & java.text.SimpleDateFormat.
The Joda-Time project, now in maintenance mode, advises migration to java.time.
To learn more, see the Oracle Tutorial. And search Stack Overflow for many examples and explanations.
Much of the java.time functionality is back-ported to Java 6 & 7 in ThreeTen-Backport and further adapted to Android in ThreeTenABP (see How to use…).
The ThreeTen-Extra project extends java.time with additional classes. This project is a proving ground for possible future additions to java.time. You may find some useful classes here such as Interval, YearWeek, YearQuarter, and more.
The month arithmetic of LocalDate will and must always use the day-of-month for its calculations. The fact that you have specified the period type such that the days shall not be used only means: The final result shall not show any days (but the day-of-month-component of LocalDate is internally still used).
However, if you use another type, namely YearMonth, then this type defines a different month arithmetic:
LocalDate dt1 = new LocalDate(2012, 1, 31);
LocalDate dt2 = new LocalDate(2012, 2, 1);
YearMonth ym1 = new YearMonth(dt1.getYear(), dt1.getMonthOfYear());
YearMonth ym2 = new YearMonth(dt2.getYear(), dt2.getMonthOfYear());
PeriodType yearMonth = PeriodType.yearMonthDay();
System.out.println(new Period(ym1, ym2, yearMonth).getMonths()); // 1
System.out.println(new Period(ym1, ym2, yearMonth.withDaysRemoved()).getMonths()); // 1
Here you can also see that - in context of YearMonth - suppressing the days in PeriodType is not relevant.
Related
I have a instance of a ZonedDatetime.
ZonedDateTime.now(ZoneId.of("America/New_York"))
I basically need a function that will take an instance of a ZonedDateTime and return the next 1 minute and 5 minute values.
So if the current time is:
2021-10-24T19:46:10.649817
The next minute will be 19:47:00 and the next 5 minute will be 19:50:00
The next 5 minute interval is always like:
1:00
1:05
1:10
1:15
1:20
1:25
...
1:50
1:55
2:00
i.e. the next 5 minute interval is not based on exactly 5 minutes from now, but rather the next 5 minutes based on starting from the beginning of the hour. Same goes for the next 1 minute interval in the future.
def nextIntervals(zdt: ZonedDateTime): (ZonedDateTime, ZonedDateTime) = {
???
}
It is fairly simple to do so without hardcoding the values. Unfortunately I'm not familiar with scala so I'll give you some pseudo code, I believe you'll be able to easily translate it.
nextIntervals(zdt) {
timestamp = zdt.toUnixTimestamp();
return [
new ZonedDateTime(timestamp + (60 - timestamp % 60)),
new ZonedDateTime(timestamp + (300 - timestamp % 300))
]
}
The above code assumes that ZonedDateTime can be instantiated by giving it a unix timestamp, measured in seconds. And also that it can be converted to a unix timestamp.
The idea is pretty simple: the remainder of the modulus will be the time that has elapsed since the last required period (in your case 1 minute or 5 minutes). Take that away from the period itself and you have the time that's left until the next period. Add that to the current time and you have the exact datetime.
Edit:
Here's a working javascript example
function nextIntervals(date) {
let t = date.getTime();
return [
60e3,
300e3,
].map(i => new Date(t + i - t % i));
}
console.log(nextIntervals(new Date));
You can use the following functions to meet your requirements:
ZonedDateTime#plusMinutes
ZonedDateTime#minusMinutes
ZonedDateTime#truncatedTo
Demo:
import java.time.ZoneId;
import java.time.ZonedDateTime;
import java.time.temporal.ChronoUnit;
public class Main {
public static void main(String[] args) {
ZonedDateTime now = ZonedDateTime.now(ZoneId.of("America/New_York"));
ZonedDateTime nextMinute = now.plusMinutes(1).truncatedTo(ChronoUnit.MINUTES);
ZonedDateTime nextMultipleOfFiveMin = now.truncatedTo(ChronoUnit.MINUTES)
.minusMinutes(now.getMinute() % 5)
.plusMinutes(5);
System.out.println(now);
System.out.println(nextMinute);
System.out.println(nextMultipleOfFiveMin);
}
}
Output from a sample run:
2021-10-25T16:59:22.662943-04:00[America/New_York]
2021-10-25T17:00-04:00[America/New_York]
2021-10-25T17:00-04:00[America/New_York]
Output from another sample run after a while:
2021-10-25T17:05:09.596952-04:00[America/New_York]
2021-10-25T17:06-04:00[America/New_York]
2021-10-25T17:10-04:00[America/New_York]
ONLINE DEMO
Learn more about the modern Date-Time API from Trail: Date Time. Check this answer and this answer to learn how to use java.time API with JDBC.
Note: The java.util Date-Time API and their formatting API, SimpleDateFormat are outdated and error-prone. It is recommended to stop using them completely and switch to the modern Date-Time API*.
* If you are working for an Android project and your Android API level is still not compliant with Java-8, check Java 8+ APIs available through desugaring. Note that Android 8.0 Oreo already provides support for java.time.
We do need a little bit of hand-coded math to handle the 5-minute interval case. Excuse my Java syntax.
ZonedDateTime now = ZonedDateTime.now(ZoneId.of("America/New_York"));
System.out.println("Now: " + now);
// Truncate to the previous 5 minutes
ZonedDateTime zdt = now.truncatedTo(ChronoUnit.MINUTES);
zdt = zdt.withMinute(zdt.getMinute() / 5 * 5);
for (int i = 0; i <= 12; i++) {
zdt = zdt.plusMinutes(5);
System.out.println(zdt);
}
Example output:
Now: 2021-10-25T15:23:31.357567-04:00[America/New_York]
2021-10-25T15:25-04:00[America/New_York]
2021-10-25T15:30-04:00[America/New_York]
2021-10-25T15:35-04:00[America/New_York]
2021-10-25T15:40-04:00[America/New_York]
2021-10-25T15:45-04:00[America/New_York]
2021-10-25T15:50-04:00[America/New_York]
2021-10-25T15:55-04:00[America/New_York]
2021-10-25T16:00-04:00[America/New_York]
2021-10-25T16:05-04:00[America/New_York]
2021-10-25T16:10-04:00[America/New_York]
2021-10-25T16:15-04:00[America/New_York]
2021-10-25T16:20-04:00[America/New_York]
2021-10-25T16:25-04:00[America/New_York]
The trick to truncate to a whole multiple of 5 minutes is to divide by 5, obtain a whole number and discard any remainder, and multiply by 5 again.
The 1-minute interval is similar, only a bit simpler: we don’t need to do any math ourselves, java.time takes care of it all.
I have a 6 digit value from which i have to get the date in scala. For eg if the value is - 119003 then the output should be
1=20 century
19=2019 year
003= january 3
The output should be 2019/01/03
I have tried ti split the value first and then get the date. But i am not sure how to proceed as i am new to scala
I think you'll have to do the century calculations manually. After that you can let the java.time library do all the rest.
import java.time.LocalDate
import java.time.format.DateTimeFormatter
val in = "119003"
val cent = in.head.asDigit + 19
val res = LocalDate.parse(cent+in.tail, DateTimeFormatter.ofPattern("yyyyDDD"))
.format(DateTimeFormatter.ofPattern("yyyy/MM/dd"))
//res: String = 2019/01/03
The Date class of Java 1.0 used 1900-based years, so 119 would mean 2019, for example. This use was deprecated already in Java 1.1 more than 20 years ago, so it’s surprising to see it survive into Scala.
When you say 6 digit value, I take it to be a number (not a string).
The answer by jwvh is correct. My variant would be like (sorry about the Java code, please translate yourself):
int value = 119003;
int year1900based = value / 1000;
int dayOfYear = value % 1000;
LocalDate date = LocalDate.ofYearDay(year1900based + 1900, dayOfYear);
System.out.println(date);
2019-01-03
If you’ve got a string, I would slice it into two parts only, 119 and 003 (not three parts as in your comment). Parse each into an int and proceed as above.
If you need 2019/01/03 format in your output, use a DateTimeFormatter for that. Inside your program, do keep the LocalDate, not a String.
I trying to add few years to current time. My code looks like:
// ten yeas ago
int backYears = 10;
Instant instant = ChronoUnit.YEARS.addTo(Instant.now(), -backYears);
But I got an exception:
java.time.temporal.UnsupportedTemporalTypeException: Unsupported unit: Years
at java.time.Instant.plus(Instant.java:862)
When I opened the method Instant.plus I see the following:
#Override
public Instant plus(long amountToAdd, TemporalUnit unit) {
if (unit instanceof ChronoUnit) {
switch ((ChronoUnit) unit) {
case NANOS: return plusNanos(amountToAdd);
case MICROS: return plus(amountToAdd / 1000_000, (amountToAdd % 1000_000) * 1000);
case MILLIS: return plusMillis(amountToAdd);
case SECONDS: return plusSeconds(amountToAdd);
case MINUTES: return plusSeconds(Math.multiplyExact(amountToAdd, SECONDS_PER_MINUTE));
case HOURS: return plusSeconds(Math.multiplyExact(amountToAdd, SECONDS_PER_HOUR));
case HALF_DAYS: return plusSeconds(Math.multiplyExact(amountToAdd, SECONDS_PER_DAY / 2));
case DAYS: return plusSeconds(Math.multiplyExact(amountToAdd, SECONDS_PER_DAY));
}
throw new UnsupportedTemporalTypeException("Unsupported unit: " + unit);
}
return unit.addTo(this, amountToAdd);
}
As you can see MONTHS and YEARS are unsupported. But why?
With an old java.util.Calendar I can do that easily:
Calendar c = Calendar.getInstance();
c.setTime(date);
c.add(Calendar.YEAR, amount);
return c.getTime();
The only one reason what I guess is that we don't know how many days in a month and year because of leap day 29 Feb.
But to be honest we also have a leap second.
Thus I think that this is a bug and all ChronoUnits should be supported.
The only one question is: do we need to take in account leap second and leap day.
As for my needs it's okay just to assume that month has 30 days and year 365.
I don't need to make something like Calendar.roll() but this can satisfy me too.
Let’s try something out. I am taking an instant as ZonedDateTime and subtracting 10 years in different time zones.
OffsetDateTime origin = OffsetDateTime.of(2018, 3, 1, 0, 0, 0, 0, ZoneOffset.UTC);
Instant originInstant = origin.toInstant();
Instant tenYearsBackKyiv = origin.atZoneSameInstant(ZoneId.of("Europe/Kiev"))
.minusYears(10)
.toInstant();
long hoursSubtractedKyiv = ChronoUnit.HOURS.between(tenYearsBackKyiv, originInstant);
System.out.println("Hours subtracted in Київ: " + hoursSubtractedKyiv);
Instant tenYearsBackSaoPaulo = origin.atZoneSameInstant(ZoneId.of("America/Sao_Paulo"))
.minusYears(10)
.toInstant();
long hoursSubtractedSaoPaulo = ChronoUnit.HOURS.between(tenYearsBackSaoPaulo, originInstant);
System.out.println("Hours subtracted in São Paulo: " + hoursSubtractedSaoPaulo);
The output is:
Hours subtracted in Київ: 87648
Hours subtracted in São Paulo: 87672
As you can see, 24 hours more (1 day more) is subtracted in São Paulo compared to Київ (Kyiv, Kiev). You may already have figured out that it’s because there we pass from 1 March to 29 February three times in leap years, in Київ only twice.
The old and now outdated Calendar class always had a time zone in it, so knew in which time zone to subtract years (another thing is it was happy to give you a result even in situations where it was unclear which result you wanted). The modern classes ZonedDateTime, OffsetDateTime and LocalDateTime can do the same. So use them. An Instant conceptually doesn’t have a time zone, so refuses to do operations that depend on time zone (I know it’s implemented using UTC, but we should regard this as an irrelevant implementation detail, not as a part of the specification of the interface to the class).
Neither the old nor the modern classes take leap seoncds into account, and you are right, only therefore can an Instant add and subtract days, hours and minutes.
I can easilly calculate time period between the first day of month and current time:
/**
* Returns the time range between the first day of month and current time in milliseconds.
*
* #param zoneId time zone ID.
* #return a {#code long} array, where at index: 0 - the first day of month midnight time; 1 - current time.
*/
public static long[] monthDateRange(ZoneId zoneId) {
long[] toReturn = new long[2];
ZonedDateTime nowZdt = LocalDateTime.now().atZone(zoneId);
ZonedDateTime startZdt = nowZdt.withDayOfMonth(1);
toReturn[0] = startZdt.toInstant().toEpochMilli();
toReturn[1] = nowZdt.toInstant().toEpochMilli();
return toReturn;
}
But how to start counting at the first day (midnight) of current week?
tl;dr
ZonedDateTime
.now( ZoneId.of( "Asia/Kolkata" ) ) // Current moment in a particular time zone.
.toLocalDate() // Extract date-only value, losing the time-of-day and time zone components.
.with( TemporalAdjusters.previousOrSame( DayOfWeek.SUNDAY ) ) // Move to another day-of-week, or same date if this is the desired day-of-week.
.atStartOfDay( ZoneId.of( "Asia/Kolkata" ) ) // Determine the first moment of the day. Do *not* assume this time-of-day is 00:00:00 as anomalies such as Daylight Saving Time (DST) may mean otherwise such as 01:00:00.
.toInstant() // Adjust into UTC, same moment, same point on the timeline, but viewed through the lens of UTC time zone.
.toEpochMilli() // Extract a count-from-epoch in milliseconds. I do *not* recommend tracking date-time this way, but the Question requires this number.
Details
The Answer by Gruodis is good, but here's an alternative that is a bit more direct and flexible.
Get current moment as a ZonedDateTime.
ZoneId z = ZoneId.of( "Pacific/Auckland" ) ;
ZonedDateTime now = ZonedDateTime.now( z ) ;
TemporalAdjuster
The TemporalAdjuster interface lets you manipulate a date-time value to get a fresh date-time value. The TemporalAdjusters class (note plural s) provides several handy implementations. Use the DayOfWeek enum to specify what day you consider to be the first day of the week.
DayOfWeek dowStartOfWeek = DayOfWeek.MONDAY ;
LocalDate weekStartDate = now.toLocalDate().with( TemporalAdjusters.previousOrSame( DayOfWeek.MONDAY ) ) ;
ZonedDateTime start = weekStartDate.atStartOfDay( z ) ; // Determine first moment of the day. Note: *not* always 00:00:00.
See this code run live at IdeOne.com.
2017-08-21T00:00+12:00[Pacific/Auckland]
2017-08-21T08:44:46.439+12:00[Pacific/Auckland]
Span of time
To report your span of time, pou could indeed extract a count-from-epoch of whole seconds, if required.
long epochSeconds = start.toEpochSecond() ;
Or extract milliseconds via Instant.
long epochMillis = start.toInstant().toEpochMilli() ;
But keep in mind that both those numbers truncate any further fractional second, as the java.time types resolve to nanoseconds.
Besides truncation, there are other reasons to avoid tracking date-time as a count-from-epoch. Since such values are meaningless to the human eye, debugging is much more difficult and faulty data may escape your notice. Also, you may assume the epoch is 1970-01-01T00:00:00Z, but there are at least another couple dozen epochs is use by common software systems. Yet another problem is ambiguity over the granularity of the count, where some systems use whole seconds, others use milliseconds, others use microseconds, others nanoseconds, and still others use other resolutions.
Interval
So instead of returning mere long integer numbers, I suggest returning an object. A pair of Instant objects work, which is what is used by the Interval class in the ThreeTen-Extra project. That class has several very handy methods I expect the calling code may find useful such as contains, encloses, abuts, overlaps, span, isEmpty, and more.
org.threeten.extra.Interval interval = Interval.of( start.toInstant() , now.toInstant() ) ;
You can apply a time zone to view either the beginning or ending through the lens of a region’s own wall-clock time.
ZonedDateTime zdtStart = interval.getStart().atZone( z ); // Or `getEnd()`.
The solution:
/**
* Returns the time range between the first day of current week midnight and current time in milliseconds.
*
* #param zoneId time zone ID.
* #return a {#code long} array, where at index: 0 - the first day of current week midnight time; 1 - current time.
*/
public static long[] monthDateRange(ZoneId zoneId) {
long[] toReturn = new long[2];
//ZonedDateTime nowZdt = LocalDateTime.now().atZone(zoneId);
ZonedDateTime nowZdt = ZonedDateTime.now(zoneId);//As suggested by Basil Bourque (tested).
//ZonedDateTime startZdt = nowZdt.with(ChronoField.DAY_OF_WEEK, 1);
ZonedDateTime startZdt = nowZdt.with(TemporalAdjusters.previousOrSame(DayOfWeek.MONDAY));//As suggested by Basil Bourque (tested).
startZdt = startZdt.toLocalDate ().atStartOfDay(zoneId);
toReturn[0] = startZdt.toInstant().toEpochMilli();
toReturn[1] = nowZdt.toInstant().toEpochMilli();
return toReturn;
}
See this code run live at IdeOne.com.
If I've got a time_t value from gettimeofday() or compatible in a Unix environment (e.g., Linux, BSD), is there a compact algorithm available that would be able to tell me the corresponding week number within the month?
Ideally the return value would work in similar to the way %W behaves in strftime() , except giving the week within the month rather than the week within the year.
I think Java has a W formatting token that does something more or less like what I'm asking.
[Everything below written after answers were posted by David Nehme, Branan, and Sparr.]
I realized that to return this result in a similar way to %W, we want to count the number of Mondays that have occurred in the month so far. If that number is zero, then 0 should be returned.
Thanks to David Nehme and Branan in particular for their solutions which started things on the right track. The bit of code returning [using Branan's variable names] ((ts->mday - 1) / 7) tells the number of complete weeks that have occurred before the current day.
However, if we're counting the number of Mondays that have occurred so far, then we want to count the number of integral weeks, including today, then consider if the fractional week left over also contains any Mondays.
To figure out whether the fractional week left after taking out the whole weeks contains a Monday, we need to consider ts->mday % 7 and compare it to the day of the week, ts->wday. This is easy to see if you write out the combinations, but if we insure the day is not Sunday (wday > 0), then anytime ts->wday <= (ts->mday % 7) we need to increment the count of Mondays by 1. This comes from considering the number of days since the start of the month, and whether, based on the current day of the week within the the first fractional week, the fractional week contains a Monday.
So I would rewrite Branan's return statement as follows:
return (ts->tm_mday / 7) + ((ts->tm_wday > 0) && (ts->tm_wday <= (ts->tm_mday % 7)));
If you define the first week to be days 1-7 of the month, the second week days 8-14, ... then the following code will work.
int week_of_month( const time_t *my_time)
{
struct tm *timeinfo;
timeinfo =localtime(my_time);
return 1 + (timeinfo->tm_mday-1) / 7;
}
Assuming your first week is week 1:
int getWeekOfMonth()
{
time_t my_time;
struct tm *ts;
my_time = time(NULL);
ts = localtime(&my_time);
return ((ts->tm_mday -1) / 7) + 1;
}
For 0-index, drop the +1 in the return statement.
Consider this pseudo-code, since I am writing it in mostly C syntax but pretending I can borrow functionality from other languages (string->int assignment, string->time conversion). Adapt or expand for your language of choice.
int week_num_in_month(time_t timestamp) {
int first_weekday_of_month, day_of_month;
day_of_month = strftime(timestamp,"%d");
first_weekday_of_month = strftime(timefstr(strftime(timestamp,"%d/%m/01")),"%w");
return (day_of_month + first_weekday_of_month - 1 ) / 7 + 1;
}
Obviously I am assuming that you want to handle weeks of the month the way the standard time functions handle weeks of the year, as opposed to just days 1-7, 8-13, etc.