Parsing french dates in Java [duplicate]

Answer 1

In French, abbreviated month names have a period.

See this page at Yale University Library, Abbreviations of the Names of the Months. Lists a few dozen languages.

“mars” is the full name for March (four letters). That name is so short as to not require abbreviating. No abbreviation, so no period. Same for “mai” (May), “juin” (June), and août (August).

Also, as you may have noticed, the first letter is lowercase in French but uppercase in English.

Joda-Time

I tried this in Joda-Time 2.4 in Java 8 on Mac OS X Mountain Lion. [Jump down for java.time, Joda-Time’s replacement]

LocalDate localDate = DateTimeFormat.forPattern( "dd MMM yyyy" ).withLocale( java.util.Locale.FRENCH ).parseLocalDate( "10 juil 2014" );

Same Problem: Missing Period

Both juillet and juil. successfully parse as French, but juil fails and throws an exception. The month abbreviation is expected to have a period terminator.

Workaround: Insert Period

Let's use substring and lastIndexOf to tear apart the string, add a period, and rebuild the string.

Test if the string contains: " janv ", " févr ", " avr ", " juil ", " sept ", " oct ", " nov ", " déc ". Note the use of spaces of both sides in case you get a string with the full month name rather than abbreviation.

String inputRaw = "10 juil 2014";
int indexOfSecondSpace = inputRaw.lastIndexOf( " " );
String input = inputRaw.substring( 0, indexOfSecondSpace ) + "." + inputRaw.substring( indexOfSecondSpace );
DateTimeFormatter formatter = DateTimeFormat.forPattern( "dd MMM yyyy" ).withLocale( java.util.Locale.FRENCH );
LocalDate localDate = formatter.parseLocalDate( input );

System.out.println( inputRaw + " → " + input + " → " + localDate );

When run.

10 juil 2014 → 10 juil. 2014 → 2014-07-10

Or call replace to do a replacement of:

• " janv " → " janv. "
• " févr " → " févr. "
• " avr " → " avr. "
• " juil " → " juil. "
• " sept " → " sept. "
• " oct " → " oct. "
• " nov " → " nov. "
• " déc " → " déc. "

Sanity-Check

In the real world, I would add some sanity-checks to ensure the input matches our expectations such as having two spaces in middle and none on the beginning or end.

java.time

Java 8 and later comes with the java.time framework built-in. These new classes supplant the old java.util.Date/.Calendar and related classes that have proven to be poorly designed, confusing, and troublesome. The new java.time classes are inspired by Joda-Time, defined by JSR 310, extended by the ThreeTen-Extra project, explained in the Oracle Tutorial, and backported to Java 6 & 7 as well as backported to Android.

The java.time classes include the handy Month enum. The getDisplayName generates localized name of month.

Similarly the DateTimeFormatter class also generates localized text. Call the ofLocalized… methods.

System.out.println ( "US | Québec | France" );
for ( Month month : Month.values () ) {
    TextStyle style = TextStyle.SHORT;
    String us = month.getDisplayName ( style , Locale.US );
    String quebec = month.getDisplayName ( style , Locale.CANADA_FRENCH );
    String france = month.getDisplayName ( style , Locale.FRANCE );
    System.out.println ( us + " | " + quebec + " | " + france );
}

We get the same behavior in java.time as seen in Joda-Time: In French the abbreviated months have a period. And month names are entirely lowercase.

US | Québec | France
Jan | janv. | janv.
Feb | févr. | févr.
Mar | mars | mars
Apr | avr. | avr.
May | mai | mai
Jun | juin | juin
Jul | juil. | juil.
Aug | août | août
Sep | sept. | sept.
Oct | oct. | oct.
Nov | nov. | nov.
Dec | déc. | déc.

Answer 2

Building on @tobias_k's comment, here is code that will find any month in a date string where the French short month abbreviation is expected to end with a period but doesn't, and replace it with the correct abbreviation including the period.

import java.util.Locale;
import java.util.regex.Matcher;
import java.util.regex.Pattern;
import java.text.DateFormatSymbols;

    public String fixFrenchMonths(String date) {
        for (String mois : DateFormatSymbols
                    .getInstance(Locale.FRENCH).getShortMonths()) {
            if (mois.endsWith(".")) {
                Pattern sansDot = Pattern.compile("(" +
                    Pattern.quote(mois.substring(0, mois.length()-1)) +
                    "(?!\\.))");
                Matcher matcher = sansDot.matcher(date);
                if (matcher.find()) {
                    date = matcher.replaceFirst(mois);
                }
            }
        }
        return date;
    }

Note: "mois" is French for "month", and "sansDot" means "withoutDot". That may be a trifle too clever, perhaps. It uses a zero-width negative lookahead to make sure it doesn't replace an abbreviation that already contains a dot. It also uses Pattern.quote on the data from DateFormatSymbols. This is probably overkill, since we don't expect that to include any characters that are regex metacharacters (except the dot itself, which we strip off), but it's probably better safe than sorry when passing data from some place we don't control into Pattern.compile.