Add support for explicit string character code ranges

[josephjunker]

🚀 Feature Request

It would be nice to have a way to specify more fine-grained character code ranges for string generation.

Motivation

I want to generate alphanumeric strings, but just adding a generator for just that exact case feels overly specific. If I could pass a more fine-grained way of selecting subvalues then I could easily construct whatever specific arbitrary I needed.

Example

// generates a string consisting of single characters from the provided string
const hexGenerator = fc.string({ fromChars: "1234567890ABCDEF" });

// generates a string consisting of characters selected from the union of the below ranges
const alphanumericGenerator = fc.string({
    fromChars: fc.stringBuilders.charRange({
        // numerals
        encoding: "ascii",
        start: 48,
        end: 57
    }).concat(fc.stringBuilders.charRange({
        // uppercase letters
        encoding: "ascii",
        start: 65
        end: 90
    }).concat(fc.stringBuilders.charRange({
        // lowercase letters
        encoding: "ascii",
        start: 97,
        end: 122
    });

fromChars may have wrinkles with multi-codepoint characters. I'm not sure off the top of my head how to handle these, but I feel like this would still be useful even with edge cases around combining characters?

charRange is verbose but would have fewer edge cases. It feels like this sort of semigroup/builder-y interface could be extended to handle combining characters if one wanted. The interface is verbose so if this existed it might be nice to provide some common combinations out-of-the-box.

The two could be combined, I think. Maybe a fromAsciiChars function which returns these charRange objects? That would solve the edge cases with fromChars. It feels like it would be possible to generate multi-codepoint characters using this sort of combinator approach but I don't have an answer for exactly what that would look like.

---

Side note, I think fast-check is awesome, and if a feature like this was desired I would be happy to try to implement it myself. I didn't want to attempt a drive-by PR for something that may not fit the overall design goals of the library.

[gruhn]

Do you know about fc.stringMatching? With that you can generate strings based on a regex. In your example, not sure if you mean union or concatenation of ranges, but both is possible:

// concatenation:
fc.stringMatching(/^[0-9][a-z][A-Z]$/)

// union:
fc.stringMatching(/^[0-9a-zA-Z]$/)

[josephjunker]

I didn't realize there were more string functions than on the "primitives" docs page 🤦 stringMatching solves my problem. Thanks!