bug in StringRandomizer for large String

[AVE-cesar]

I've found a little bug when I use your wonderful tool on a large String.
I am using version 3.7.0.

This is an extract of my class:

@Size(max=2147483647)
public String getName() {
	return name;
}

The impacted line is:

    int length = minLength + random.nextInt(maxLength - minLength + 1);

the calculated length is: -2147483648 but the length MUST be positive !

This is the stack trace I've got:

Exception in thread "main" io.github.benas.randombeans.api.ObjectGenerationException: Unable to generate a random instance of type class a.b.Person
	at io.github.benas.randombeans.EnhancedRandomImpl.doPopulateBean(EnhancedRandomImpl.java:124)
	at io.github.benas.randombeans.EnhancedRandomImpl.nextObject(EnhancedRandomImpl.java:77)
	at ave.bertrand.MainClass.main(MainClass.java:18)
Caused by: java.lang.IllegalArgumentException: bound must be positive
	at java.util.Random.nextInt(Random.java:388)
	at io.github.benas.randombeans.randomizers.text.StringRandomizer.getRandomValue(StringRandomizer.java:256)
	at io.github.benas.randombeans.randomizers.text.StringRandomizer.getRandomValue(StringRandomizer.java:36)
	at io.github.benas.randombeans.FieldPopulator.populateField(FieldPopulator.java:78)
	at io.github.benas.randombeans.EnhancedRandomImpl.populateField(EnhancedRandomImpl.java:158)
	at io.github.benas.randombeans.EnhancedRandomImpl.populateFields(EnhancedRandomImpl.java:149)
	at io.github.benas.randombeans.EnhancedRandomImpl.doPopulateBean(EnhancedRandomImpl.java:120)
	... 2 more

[fmbenhassine]

Hi @AVE-cesar

Thank you for reporting this issue!
Indeed, random beans fails miserably on this value.

I will fix this issue and get back to you asap.

Kr
Mahmoud

[fmbenhassine]

Hi,

2147483647 is actually the value of Integer.MAX_VALUE, so @Size(max=2147483647) is equivalent to @Size. This is perfectly valid value to be specified as a user, and the issue happens because Integer.MAX_VALUE + 1 == Integer.MIN_VALUE (explained in details on SO here).

The +1 in the formula was introduced in 587e600 to fix the case where the generated String can never have the max length.

I fixed the issue and now the random String length is correctly calculated. However, when the random String length is too high, a OOM might happen and random beans can do nothing to prevent it (see #289).

The fix has been deployed to maven central in 3.8.0-SNAPSHOT. I'm closing this issue for now. @AVE-cesar It would be great if you give the fix a try (see here on how to use the snapshot version). Please add a comment here if you find any issue.

Kr,
Mahmoud