Parsing a textfile and splitting it efficiently

I would like to parse a text file which looks more or less like this:

TYPE1=123
TYPE2="SOMETEXT"
TYPE3="SOMETEXT_BUT
ON_MULTIPLE_
LINES"
TYPE4=456

If a value spans multiple lines it always has quotation marks. If it only spans one line it has either quotation marks or not (unfortunately it does not matter if it's a number or string - there can also be a string without quotation marks - the format is not very consistent)

And I'm currently figuring out how I could split them by type and parse it efficiently. I could do a readlines and split by "=". That would work for everything except TYPE3 in the above example because it spans through multiple lines.

So I'm thinking about reading the whole file into a String and then doing some regex, e.g. (.*)=("([^"]*)"|.*\n) which would result in the first capturing group always beeing the type and the last capturing group the value. I just fear that for larger files this might be to slow and cause issues.

Is there a better/more efficient way to solve this parsing problem?

I came up with this straight-forward read-through of the lines. Not sure it's more efficient than loading the whole file and using Regex, but it could be useful for huge files since it only reads one line at a time.

fun readCustomPropertiesFile(file: File): Map<String, String> {
    val map = mutableMapOf<String, String>()
    var entry = ""
    var entryComplete = true
    file.forEachLine { line ->
        if (entryComplete && '=' !in line){
            println("Line is invalid: $line")
            return@forEachLine
        }
        entry = if (entryComplete) line else "$entry\n$line"
        val (key, value) = entry.split('=', limit = 2)
        val startQuote = value.startsWith('"')
        val endQuote = value.endsWith('"')
        entryComplete = !startQuote || startQuote == endQuote
        if (entryComplete) {
            map[key] = if (startQuote && endQuote) value.substring(1, value.length - 1) else value
        }
    }
    return map
}

Your format is so near to .properties, with = and in a bit different form multi-line properties, that I would adapt the data, and use Properties. Using UTF-8 would need a small adaption, and you are done: production quality.

I think your idea of reading the file into a string and then applying a regex would work fine. Some points:

You don't need (and probably don't want) the \n in the pattern.

• You probably want only 123 as the value of TYPE1, not 123\n.
• . doesn't match \n, so .* stops matching if/when it hits a \n anyway.
• And if it happens that the file ends without a newline, the pattern-with-\n will fail to match, but the pattern-without will succeed.

If it's possible that a string value can contain an =, then (.*)= isn't going to work. E.g., if a line is TYPE2="SOME=TEXT", then the (.*) will match TYPE2="SOME, which you presumably don't want. You can fix this by using ([^=]*)= or maybe (\w*)=, depending on the particulars of the format.