split string by char

Question

scala has a standard way of splitting a string in StringOps.split

it's behaviour somewhat surprised me though.

To demonstrate, using the quick convenience function

def sp(str: String) = str.split('.').toList

the following expressions all evaluate to true

(sp("") == List("")) //expected
(sp(".") == List()) //I would have expected List("", "")
(sp("a.b") == List("a", "b")) //expected
(sp(".b") == List("", "b")) //expected
(sp("a.") == List("a")) //I would have expected List("a", "")
(sp("..") == List()) // I would have expected List("", "", "")
(sp(".a.") == List("", "a")) // I would have expected List("", "a", "")

so I expected that split would return an array with (the number a separator occurrences) + 1 elements, but that's clearly not the case.

It is almost the above, but remove all trailing empty strings, but that's not true for splitting the empty string.

I'm failing to identify the pattern here. What rules does StringOps.split follow?

For bonus points, is there a good way (without too much copying/string appending) to get the split I'm expecting?

Answer 1

For curious you can find the code here.https://github.com/scala/scala/blob/v2.12.0-M1/src/library/scala/collection/immutable/StringLike.scala

See the split function with the character as an argument(line 206).

I think, the general pattern going on over here is, all the trailing empty splits results are getting ignored.

Except for the first one, for which "if no separator char is found then just send the whole string" logic is getting applied.

I am trying to find if there is any design documentation around these.

Also, if you use string instead of char for separator it will fall back to java regex split. As mentioned by @LRLucena, if you provide the limit parameter with a value more than size, you will get your trailing empty results. see http://docs.oracle.com/javase/7/docs/api/java/lang/String.html#split(java.lang.String,%20int)

Answer 2

You can use split with a regular expression. I´m not sure, but I guess that the second parameter is the largest size of the resulting array.

def sp(str: String) = str.split("\\.", str.length+1).toList