Generating the pretty bit at the end of a URL

Question

I'm having a dip into F# and am attempting not to write it like it's C#.

One area that bothers me is using members of System.String. These often need to be chained together, but the only way to do so that I've seen so far is pretty much the same as C#.

For example, I've written a function to generate the pretty bit at the end of a URL (similar to SE):

module String =
 open System.Text.RegularExpressions
 ///Makes a string suitable for use as part of a URL: 
 ///Removes syntax, replaces spaces with dashes and lowers case, 
 ///then trims first word after 30 chars
 let prettify (x:string) = 
 let parsed = Regex.Replace(x.Trim(),"[^0-9a-zA-Z ]","").Replace(" ","-").ToLower()
 match String.length parsed with
 | m when m > 30 -> 
 match parsed.IndexOf("-",30-1) with
 | p when p < 0 -> parsed
 | p -> parsed.[..p]
 | _ -> parsed

Is there a way to refactor this to be "better F#"? I'm particularly concerned about how to better handle the methods on System.String and Regex.Replace. (The match part is a little convoluted to handle the edge case of the last word being the one that takes it over 30 chars.)

• Jon Egerton: I posted my latest iteration of this method. There have been a few, but all had the horrible String functions line.
• Mauricio Scheffer: IMHO before focusing on syntax/style you should focus on conceptual properties. For example, that function is partial and impure. Most likely you don't want that in such a function.
• Jon Egerton: @MauricioScheffer: Thanks for feeding back to me. Could you give me a quick pointer on both of those aspects please?
• Mauricio Scheffer: It's partial because it doesn't terminate for certain inputs (e.g. an exception). It's impure because its output depends on a variable that is not part of its input (see programmers.stackexchange.com/questions/184574/… ). As an exercise, try to figure out which inputs cause it to throw, and what is the "implicit" input variable that makes it impure.

Answer 1

F# does have a String module, but it's almost empty. If you want to make string manipulations look more idiomatic, you'll have write functions that wrap System.String methods yourself (quick search didn't find anything existing):

let trim (input:string) =
 input.Trim()
let regexReplace (pattern:string) (replacement:string) (input:string) =
 Regex.Replace(input, pattern, replacement)
let replace (pattern:string) (replacement:string) (input:string) =
 input.Replace(pattern, replacement)
let toLower (input:string) =
 input.ToLower()
...
let parsed =
 x
 |> trim
 |> regexReplace "[^0-9a-zA-Z ]" ""
 |> replace " " "-"
 |> toLower

---

match String.length parsed with
 | m when m > 30 -> ...
 | _ -> parsed

You don't need to try to differentiate from C# so hard. if would be more appropriate here:

if String.length parsed > 30 then ...
else parsed

---

match parsed.IndexOf("-",30-1) with
 | p when p < 0 -> parsed
 | p -> parsed.[..p]

You can rewrite this to avoid having to reference p in both branches:

match parsed.IndexOf("-",30-1) with
 | p when p >= 0 -> parsed.[..p]
 | _ -> parsed

Also, use more descriptive names than p, like index or maybe pos.

---

Also, this code usually returns a string that's over 30 characters and ends with -. Is that what you intended?

• Jon Egerton: Thanks for that. I'd found the string module and used length from it, but as you say, the other functions in it weren't very helpful here. I'd considered those wrappers, as an option if I need to do a lot of such manipulation.
• Jon Egerton: btw, its intended to be >30 often as I want the last bit to be a full word. Will adjust for the trailing -