Verify if String matches a format String

Question

In Java, how can you determine if a String matches a format string (ie: song%03d.mp3)?

In other words, how would you implement the following function?

/**
* @return true if formatted equals String.format(format, something), false otherwise.
**/
boolean matches(String formatted, String format);

Examples:

matches("hello world!", "hello %s!"); // true
matches("song001.mp3", "song%03d.mp3"); // true
matches("potato", "song%03d.mp3"); // false

Maybe there's a way to convert a format string into a regex?

Clarification

The format String is a parameter. I don't know it in advance. song%03d.mp3 is just an example. It could be any other format string.

If it helps, I can assume that the format string will only have one parameter.

Solution 1

I don't know of a library that does that. Here is an example how to convert a format pattern into a regex. Notice that Pattern.quote is important to handle accidental regexes in the format string.

// copied from java.util.Formatter
// %[argument_index$][flags][width][.precision][t]conversion
private static final String formatSpecifier
    = "%(\\d+\\$)?([-#+ 0,(\\<]*)?(\\d+)?(\\.\\d+)?([tT])?([a-zA-Z%])";

private static final Pattern formatToken = Pattern.compile(formatSpecifier);

public Pattern convert(final String format) {
    final StringBuilder regex = new StringBuilder();
    final Matcher matcher = formatToken.matcher(format);
    int lastIndex = 0;
    regex.append('^');
    while (matcher.find()) {
        regex.append(Pattern.quote(format.substring(lastIndex, matcher.start())));
        regex.append(convertToken(matcher.group(1), matcher.group(2), matcher.group(3), 
                                  matcher.group(4), matcher.group(5), matcher.group(6)));
        lastIndex = matcher.end();
    }
    regex.append(Pattern.quote(format.substring(lastIndex, format.length())));
    regex.append('$');
    return Pattern.compile(regex.toString());
}

Of course, implementing convertToken will be a challenge. Here is something to start with:

private static String convertToken(String index, String flags, String width, String precision, String temporal, String conversion) {
    if (conversion.equals("s")) {
        return "[\\w\\d]*";
    } else if (conversion.equals("d")) {
        return "[\\d]{" + width + "}";
    }
    throw new IllegalArgumentException("%" + index + flags + width + precision + temporal + conversion);
}

Solution 2

You can use Java regular expressions - please see http://www.vogella.de/articles/JavaRegularExpressions/article.html

Thanks...

Solution 3

Since you do not know the format in advance, you will have to write a method that converts a format string into a regexp. Not trivial, but possible. Here is a simple example for the 2 testcases you have given:

public static String getRegexpFromFormatString(String format)
{
    String toReturn = format;

    // escape some special regexp chars
    toReturn = toReturn.replaceAll("\\.", "\\\\.");
    toReturn = toReturn.replaceAll("\\!", "\\\\!");

    if (toReturn.indexOf("%") >= 0)
    {
        toReturn = toReturn.replaceAll("%s", "[\\\\w]+"); //accepts 0-9 A-Z a-z _

        while (toReturn.matches(".*%([0-9]+)[d]{1}.*"))
        {
            String digitStr = toReturn.replaceFirst(".*%([0-9]+)[d]{1}.*", "$1");
            int numDigits = Integer.parseInt(digitStr);
            toReturn = toReturn.replaceFirst("(.*)(%[0-9]+[d]{1})(.*)", "$1[0-9]{" + numDigits + "}$3");
        }
    }

    return "^" + toReturn + "$";
}

and some test code:

public static void main(String[] args) throws Exception
{
    String formats[] = {"hello %s!", "song%03d.mp3", "song%03d.mp3"};
    for (int i=0; i<formats.length; i++)
    {
        System.out.println("Format in [" + i + "]: " + formats[i]);
        System.out.println("Regexp out[" + i + "]: " + getRegexp(formats[i]));
    }

    String[] words = {"hello world!", "song001.mp3", "potato"};
    for (int i=0; i<formats.length; i++)
    {
        System.out.println("Word [" + i + "]: " + words[i] +
            " : matches=" + words[i].matches(getRegexpFromFormatString(formats[i])));
    }
}

Solution 4

There is not a simple way to do this. A straight-forward way would be to write some code that converts format strings (or a simpler subset of them) to regular expressions and then match those using the standard regular expression classes.

A better way is probably to rethink/refactor your code. Why do you want this?

View more solutions

Comments

[hpique]

The format String is a parameter. I don't know it in advance. song%03d.mp3 was just an example.

[hpique]

The format String is a parameter. I don't know it in advance. song%03d.mp3 was just an example.

[hpique]

The format String is a parameter. I don't know it in advance. song%03d.mp3 was just an example.

[hpique]

And how do you convert a generic format string into a pattern?

[hpique]

I'd love to use regular expressions, but what I'm given is a format string.

[Yhn]

Hence my comment about replacing format codes like %03d with their regular expression equivalent :). The page you linked completely defines the possible codes and prefixes, you'd need to write a function that searches those codes and replaces them.a %d would be replaced with \d+; %03d could become \d{3}\d? (to ensure a minumum of 3, but possibly "infinite" digits.

[hpique]

That's what I would like to avoid. I didn't write the whole code.

[dtech]

There is definitly no native way to do this. So you either need to rethink your input/code, write your own converter or find a converter that does this. E.g. why exactly does you need to use a format string?

[hpique]

We choose format strings because they're pretty much the same across all platforms, unlike regex.

[Mario Duarte]

Well, if you're giving it as a parameter for a java application why don't you just use Java regexps?

[Vishwas Mehra]

@hgpc ok I've modified my answer appropriately. It's more than I would usually do for a SO answer but I was intrigued. :) You would have to perfect/complete this for production use but it is an idea for how to approach this if necessary.

[rascio]

But you have to write a regex...this is what i don't understand...how is created the regex? you need something that creates the regex automatically? or you need something that checks if a string contains a regex?

[hpique]

Because the Java app is one of the many clients that receive this input.

[hpique]

+1 This is more or less what I'm doing right now. Thanks for posting code.

[dtech]

Perl5 compatible regular expressions are implemented in nearly all programming languages. But if you have to do this the only thing you can do is write a converter. It's not that hard. Also note that you're using the wrong format for the job. Format strings are only intended for data -> string(s). Regular expressions are broader. What you're doing now is basically re-inventing an impractical regex notation.

[Cephalopod]

If you want to be a hero, you can publish your code as open source.

[hpique]

I'm no stranger to publishing open-source code, but this is too specific to publish.