string.contains()

[ben-albrecht]

Chapel domains support a .contains() method for querying if a value is in the domain (as of #5041).

Should Chapel strings be consistent with the domain interface and support string.contains() for querying if a substring is in the string?

The string type currently supports string.find() and string.count(), however these are slightly more verbose to use in a conditional than string.contains():

if myString.find('waldo') > 0 { }
if myString.count('waldo') > 0 { }
if myString.contains('waldo') { }

[rsh3khar]

@ben-albrecht

Any updates if this feature is to be implemented or not? If yes, I would like to work on this.

[ben-albrecht]

@RajShekhar7 - it is not implemented, but this requires design discussion before pursuing implementation. Check out contributor info to understand that process better.

[rsh3khar]

@ben-albrecht - I've read the Contributor Info, but help me understand, should I open a new issue for design discussions in this case if I want to implement it?

[Aniket21mathur]

@RajShekhar7 I think you can give your design inputs on this thread itself.

Thanks !

[ben-albrecht]

I've read the Contributor Info, but help me understand, should I open a new issue for design discussions in this case if I want to implement it?

This thread is fine. The big thing is that we need other core developers to agree to the design before pursuing it.

[Yudhishthira1406]

@ben-albrecht, I would like to work on this.

This thread is fine. The big thing is that we need other core developers to agree to the design before pursuing it.

I want to discuss how to implement this as I have 1 or 2 approaches in mind, if all others agree.

[ben-albrecht]

I think the implementation for this is fairly straightforward, e.g.

proc string.contains(s: string) {
  return this.count(s) > 0;
}

The big thing is that we need other core developers to agree to the design before pursuing it. To move this forward, someone should make a post to chapel developers discourse requesting input, as instructed here: https://github.com/chapel-lang/chapel/blob/master/doc/rst/developer/bestPractices/ContributorInfo.rst#design

[e-kayrakli]

I find the idea very appealing. And in general, find the behavior as Ben sketched out above to be the right behavior. I would like to know/discuss:

• Examples from other languages where such a function exits, if any.
• Would the name be confusing? "abcd".contains("bd"); I think this should return false, but there is at least one case (string.strip) where we pass a string to represent list of characters, rather than an actual string. I personally don't think that would cause confusion here. An alternative would be to name it containsSubstring or something, but I like wordiness much less than a slight possibility of confusion.
• Edge cases: does any string contain an empty string? Does empty string contain empty string? For that, I am curious what Ben's implementation sketch above would do, and whether that behavior is confusing in the context of string.contains

[Yudhishthira1406]

@e-kayrakli I did a bit of google search and found out that.

• Python has __contains__() and Java has contains() methods.
• An empty string is a universal substring. I cross-checked it with the python's __contains()__ function.

[ben-albrecht]

Closing since a more recent feature request has captured the info from here and has additional info with a future checked in: #19119