Regex string replace considerably fast than regular replace in kotlin

Question

I have been trying to replace characters in string like below

data = data.replace(Regex("[a-z:]", RegexOption.IGNORE_CASE), "")
        .replace(Regex("/", RegexOption.IGNORE_CASE), ".")
        .replace(Regex(",", RegexOption.IGNORE_CASE), "")
        .replace(Regex("'", RegexOption.IGNORE_CASE), "")
        .replace(Regex("é",RegexOption.IGNORE_CASE),"")
        .replace(Regex("ê",RegexOption.IGNORE_CASE),"")
        .replace(Regex("ö",RegexOption.IGNORE_CASE),"")
        .replace(Regex("Ä",RegexOption.IGNORE_CASE),"")
        .replace(Regex("ä",RegexOption.IGNORE_CASE),"")
        .replace(Regex("ä |",RegexOption.IGNORE_CASE),"")

And

data = data.replace(Regex("[a-z:]", RegexOption.IGNORE_CASE), "")
        .replace("/", ".")
        .replace(",", "")
        .replace("'", "")
        .replace("é","")
        .replace("ê","")
        .replace("ö","")
        .replace("Ä","")
        .replace("ä","")

And I measured time required for both of this code and surprisingly code with regex turned out at least 20 times faster than normal replace.

As long as I have been reading about regex, they say regex is expensive operation, am I missing something?

Is the space + `|` a typo? See `.replace(Regex("ä |",RegexOption.IGNORE_CASE),"")`. What is the outcome if you fix it (`"ä |"` => `"ä"`)? — Wiktor Stribiżew, Jan 09 '19 at 07:54
*"And I measured time required for both..."* Did you do so [properly](https://stackoverflow.com/questions/504103/how-do-i-write-a-correct-micro-benchmark-in-java)? — T.J. Crowder, Jan 09 '19 at 07:56
Separately: Why the multiple calls to `replace`? The above requires just two, not nine. — T.J. Crowder, Jan 09 '19 at 07:57
@T.J.Crowder that's not the point, I was going through existing codebase and found this code. Was trying to improve it when I came across this issue. — Amit Bhandari, Jan 09 '19 at 07:59
@WiktorStribiżew Not a typo, checked without it as well. Result is same. — Amit Bhandari, Jan 09 '19 at 07:59
@T.J.Crowder I did simple time comparison using currentTimeMillis. It came 3 ms on average for regex and >40ms on average for normal replace — Amit Bhandari, Jan 09 '19 at 08:00
Ok, but note that `.replace(Regex("ä |",RegexOption.IGNORE_CASE),"")` and `.replace("ä","")` are not equal in what they do as the first one removes `ä` + space and the second only removes `ä`. — Wiktor Stribiżew, Jan 09 '19 at 08:01
@AmitBhandari - Please read the answers to [the question I linked](https://stackoverflow.com/questions/504103/how-do-i-write-a-correct-micro-benchmark-in-java). You can't do perf analysis on the JVM like that. — T.J. Crowder, Jan 09 '19 at 08:03
For sure `ä |` fails to match because there would be no `ä`s after the previous call to replace. — revo, Jan 09 '19 at 08:10
Offtopic, but you can write one single regex for all the empty string replacements. This way you can eliminate all those redundant method chainings. — Adam, Jan 09 '19 at 08:58

Regex string replace considerably fast than regular replace in kotlin

0 Answers0