I'm getting an unexpected behaviour with `String.split(regex)` when there are no matches. Does `spli...

# getting-started

[v79]

I'm getting an unexpected behaviour with String.split(regex) when there are no matches. Does split() return "the rest of the string" when a regex match fails? Code in 🧵

[v79]

fun main() {
    val regex = Regex("-{3} #")
    val willMatch = """
    ---
    apple: banana
    --- #matches
    Yay
    """
    val willNotMatch = """
    ---
    Not a valid match
    """
    println("Find the number of matches")
    println(regex.findAll(willMatch).count()) // expect 1, got 1
    println(regex.findAll(willNotMatch).count()) // expect 0, got 0

    println("Split and build a map from the matches. Why does size != count from above?")
    println(willMatch.split(regex).map {it.trim()}.associate{it.substringBefore("\n").trim() to it.substringAfter("\n")}.size) // expect 1, got 2
    println(willNotMatch.split(regex).map {it.trim()}.associate{it.substringBefore("\n") to it.substringAfter("\n")}.size) // expect 0, got 1

}

[v79]

Playground here: https://pl.kotl.in/-C7TfpD5T

[ephemient]

split, with a regex or not, always returns at least 1 segment

[hho]

Just as the Java API docs state:

If the expression does not match any part of the input then the resulting array has just one element, namely this string.

https://docs.oracle.com/javase/8/docs/api/java/lang/String.html#split-java.lang.String-int-

[ephemient]

Java's string split actually changed behavior between 7 and 8

[ephemient]

Kotlin's string split is always consistent though

[ephemient]

simpler example to demonstrate:

"a-b".split('-') == listOf("a", "b")

there is 1 match and 2 results

[v79]

Thanks for the confirmation. I guess I'll need to check for matches, then only do the split if the match succeeds.

[ephemient]

(the weird edge cases come from empty strings or empty matches)

[ephemient]

why?

[v79]

I'm extracting text that may be split by named blocks ("--- #blockName"). I need the names and the text that follows. But I don't want to split when there is no block name.

[ephemient]

so? no matter what, it sounds like you don't want the content before the first match, and so .split().drop(1) always gets you what you want whether there is a match or not

[ephemient]

or possibly split isn't the right tool anyway, if you can potentially have both named and unnamed blocks

[ephemient]

e.g. https://pl.kotl.in/ZAzlNnaNn can't be done by just split()