为什么 Java 时间戳解析在毫秒部分以 0 为前缀？

Question

final String OLD_FORMAT = "mm:ss.SS";        
final String TARGET_FORMAT = "HH:mm:ss,SSS";           

String timeInputStr="00:17.20";  // input is always in mm.ss.SS m--> minutes, ss-> seconds , and SSS is decimal fraction of seconds always , not actual milliseconds 
String timeOutputStr="";
Date d=new Date();
DateFormat  formatter= new SimpleDateFormat(OLD_FORMAT); 
DateFormat nformatter= new SimpleDateFormat(TARGET_FORMAT);          
try{   
        d = formatter.parse(timeInputStr);
}
catch (ParseException e){
         System.out.println("Can't Parse date "+d + " from: " +lrcTime );
}

timeInputStr=nformatter.format(d);
System.out.println( "For Input String: " + lrcTime +  " -> Parsed date "+ formatter.format(d) +  "-> will print as to  " + timeInputStr);
return timeOutputStr;   

It's giving me the following output:

For Input String: 00:17.20 -> Parsed date 00:17.20-> will print as to  00:00:17,020

But I want to parse such string as 00:00:17,200

What am I missing?

Answer 1

There is an ambiguity that is improved in the newer date time classes:

The old SimpleDateFormat

• S = millisecond

  SS with 20 means 20 ms, hence SSS will be 020 ms.

The newer DateTimeFormatter

• S = fraction of second.

  Will give SSS = 200 ms

The rationale behind the new interpretation is that with microseconds and nanoseconds a 2 ms as ,2 does not really fit with things like ,SSSSSS - as you remarked.

Answer 2

The formatter interprets the .20 as 20 milliseconds, not .2 seconds (which would be 200ms). To resolve this issue, you can simply add a zero to your String.

d = formatter.parse(lrcTime + "0");