F# Avoid active pattern overwriting - f#

I have noticed I cannot create two active patterns with the same options, but I can have two with similar ones without any warning:
let (|A|B|C|) c =
if (c = 'a') then A
else if (c = 'b') then B
else C
let (|A|B|D|) c =
if (c = '1') then A
else if (c = '2') then B
else D
So when matching this way:
let check myvar =
match myvar with
| A -> printf "match A\n"
| n -> printf "match other %A\n" n
This happens:
check 'x' // match other 'x'
check 'a' // match other 'a' !!
check '1' // match A
I am a bit concerned of overwriting existing active pattern options inadvertently, for example in situations where the same word can appear in different patterns because different semantic contexts, like (|Direct|Indirect|) (route) and (|Alternating|Direct|) (current).
How can I avoid this situations?

I agree that shadowing of active patterns can be tricky - though it is the same problem that you get with discriminated union cases and record labels in F#. In case of types, you can always include the type name to resolve the ambiguity.
In case of active patterns, you can put them in modules - for example Pat1 and Pat2:
module Pat1 =
let (|A|B|C|) c =
if (c = 'a') then A
else if (c = 'b') then B
else C
module Pat2 =
let (|A|B|D|) c =
if (c = '1') then A
else if (c = '2') then B
else D
So, in your code, you can then use fully qualified name like Pat1.A or Pat2.A:
let check myvar =
match myvar with
| Pat1.A -> printf "match A\n"
| n -> printf "match other %A\n" n

I think your concerns apply to shadowing in general and not just active patterns. How often will you define two active patterns whose parameters and return values are the same and have overlapping case names? Generally, types mitigate potential shadowing problems. Along these lines, type annotations are your friend.

Related

Pattern matching against a string property

I'm de-serializing some mappings from JSON and later on I need to pattern match based on a string field of the de-serialized types like this:
let mappings = getWorkItemMappings
let result =
workItemMappings
|> Seq.find (fun (m: WorkItemMapping) -> m.Uuid = workTime.workItemUuid)
match mapping.Name with
Even if I complete the pattern match for all cases I still get Incomplete pattern matches on this expression.. Which is obvious to me due to the string type of the Name field.
Is there a way tell the compiler which values for the Name field are available?.
I think I could create a union type for the possible mapping types and try to de-serialize the JSON to this union type but I would like to if there's another option.
If you are pattern matching on a string value, the compiler has no static guarantee that it will only have certain values, because it is always possible to construct a string of a different value. The fact that it comes from JSON does not help - you may always have an invalid JSON.
The best option is to add a default case which throws a custom descriptive exception. Either one that you handle somewhere else (to indicate that the JSON file was invalid) or (if you check the validity elsewhere) something like this:
let parseFood f =
match f with
| "burger" -> 1
| "pizza" -> 2
| _ -> raise(invalidArg "f" $"Expected burger or pizza but got {f}")
Note that the F# compiler is very cautious. It does not even let you handle enum values using pattern matching, because under the cover, there are ways of creating invalid enum values! For example:
type Foo =
| A = 1
let f (a:Foo) =
match a with
| Foo.A -> 0
warning FS0104: Enums may take values outside known cases. For example, the value 'enum (0)' may indicate a case not covered by the pattern(s).
Very hard to understand what you're asking. Maybe this snippet can be of help. It demos how literal string constants can be used in pattern matching, and reused in functions. This gives some added safety and readability when adding and removing cases. If you prefer not to serialize a DU directly, then perhaps this is useful as part of the solution.
type MyDu =
| A
| B
| C
let [<Literal>] A' = "A"
let [<Literal>] B' = "B"
let [<Literal>] C' = "C"
let strToMyDuOption (s: string) =
match s with
| A' -> Some A
| B' -> Some B
| C'-> Some C
| _ -> None
let strToMyDu (s: string) =
match s with
| A' -> A
| B' -> B
| C'-> C
| s -> failwith $"MyDu case {s} is unknown."
let myDuToStr (x: MyDu) =
match x with
| A -> A'
| B -> B'
| C -> C'
// LINQPad
let dump x = x.Dump()
strToMyDuOption A' |> dump
strToMyDuOption "x" |> dump
myDuToStr A |> dump

How to avoid long pattern match based functions?

When using union types with quite a few constructors I almost always find myself implementing lots of logic in single function, i.e. handling all cases in one function. Sometimes I would like to extract logic for single case to separate function, but one cannot have a function accepting only one "constructor" as parameter.
Example:
Assume that we have typical "expression" type :
type Formula =
| Operator of OperatorKind * Formula * Formula
| Number of double
| Function of string * Formula list
[...]
Then, we would like to calculate expression :
let rec calculate efValues formula =
match formula with
| Number n -> [...]
| Operator (kind, lFormula, rFormula) -> [...]
| [...]
Such function would be very long and growing with every new Formula constructor.
How can I avoid that and clean up such code? Are long pattern matching constructs inevitable?
You can define the Operator case of the Formula union using an explicit tuple:
type Formula =
| Operator of (string * Formula * Formula)
| Number of double
If you do this, the compiler will let you pattern match using both Operator(name, left, right) and using a single argument Operator args, so you can write something like:
let evalOp (name, l, r) = 0.0
let eval f =
match f with
| Number n -> 0.0
| Operator args -> evalOp args
I would find this a bit confusing, so it might be better to be more explicit in the type definition and use a named tuple (which is equivalent to the above):
type OperatorInfo = string * Formula * Formula
and Formula =
| Operator of OperatorInfo
| Number of double
Or perhaps be even more explicit and use a record:
type OperatorInfo =
{ Name : string
Left : Formula
Right : Formula }
and Formula =
| Operator of OperatorInfo
| Number of double
Then you can pattern match using one of the following:
| Operator args -> (...)
| Operator { Name = n; Left = l; Right = r } -> (...)
I would say you typically want to handle all the cases in a single function. That's the main selling point of unions - they force you to handle all the cases in one way or another. That said, I can see where you're coming from.
If I had a big union and only cared about a single case, I would handle it like this, wrapping the result in an option:
let doSomethingForOneCase (form: Formula) =
match form with
| Formula (op, l, r) ->
let result = (...)
Some result
| _ -> None
And then handle None in whatever way is appropriate at the call site.
Note that this is in line with the signature required by partial active patterns, so if you decide that you need to use this function as a case in another match expression, you can easily wrap it up in an active pattern to get the nice syntax.

change properties of an element in a set and put it back in the set

Here's the problem,
Suppose we have a datatype, and a set of the datatype:
data Color=red(int n, str name)|black(int n, str name);
set[Color] coloredBalls={red(1,"a"), red(2,"b")};
I would like to iterate through the elements of the set and change the elements as follows:
for(Color aColor <- coloredBalls)
{
if(aColor.n == 2)
aColor.str="bb"
//I would like to change the node: red(2,"b") to red(2,"bb") in the set coloredBalls
//but am unable to express it in Rascal, unless I recourse to using a "Map", which I
//think obfuscates the code quite a bit. Maybe using switch/visit with case does the trick?
}
There are many ways to build the new set you want to have (you can't actually change the existing set of course).
You can use visit:
coloredBalls = visit (coloredBalls) {
case red(a,b) => red(a, b + "b")
}
Or use a comprehension with an is:
coloredBalls = { x is red ? x[name=name+"b"] : x | x <- coloredBalls};
Or with pattern match on the generator side:
coloredBalls = {red(a, b + "b") | red(a,b) <- coloredBalls} + { y | y <- coloredBalls, !(y is red)};
Or with a pattern match on the insert side of the comprehension:
coloredBalls = {red(a,b) := c ? red(a,b+"b") : c | c <- coloredBalls};

Am I properly using function composition?

In an effort to understand the capabilities of functional programming I put together a few basic functions that you can compose together to build complex regular expressions. Now after some testing I have found this works but you can write some horrible code in any language that will work. Is this the kind of code you would find a professional F# programmer writing or am I abusing the feature?
Note: test is specifically what I am referring to.
type State = { input:string; index:int; succeeded:bool }
type Matcher = State -> State
let term (cs:char Set) =
fun s ->
if s.succeeded && s.index < s.input.Length && cs.Contains s.input.[s.index] then
{ input = s.input; index = s.index + 1; succeeded = true }
else
{ input = s.input; index = s.index; succeeded = false }
let quantify (term, min, max) =
let rec inner (s:State, count) =
if s.succeeded && s.index < s.input.Length && count <= max then
inner (term { input = s.input; index = s.index + 1; succeeded = true }, count + 1)
elif count >= min && count <= max then
{ input = s.input; index = s.index - 1; succeeded = true }
else
s
fun s -> inner (s, 0)
let disjunction leftTerm rightTerm =
fun s ->
let left = leftTerm s
if not left.succeeded then
let right = rightTerm s
if not right.succeeded then
{ input = s.input; index = s.index; succeeded = false }
else
right
else
left
let matcher input terms =
let r = terms { input = input; index = 0; succeeded = true }
if r.succeeded then r.input.Substring (0, r.index) else null
let test = // (abc|xyz)a{2,3}bc
disjunction // (abc|xyz)
(term (set "a") >> term (set "b") >> term (set "c"))
(term (set "x") >> term (set "y") >> term (set "z"))
>> quantify (term (set "a"), 2, 3) // (a{2,3})
>> term (set "b") // b
>> term (set "c") // c
let main () : unit =
printfn "%s" (matcher "xyzaabc" test)
System.Console.ReadKey true |> ignore
main()
The code looks pretty good to me.
I'm not sure if this was your intention or a coincidence, but you're implementing something quite similar to "parser combinators", which is a topic of many academic papers :-). I think that Monadic Parser Combinators is quite readable (it has examples in Haskell, but you should be able to translate them to F#).
Regarding the function composition operator. I'm generally not a big fan of using the operator too much, because it often obfuscates the code. However, in your example it makes a good sense because you can easily imagine that >> means "this group should be followed by that group", which is easy to interpret.
The only minor change that I would do is to choose some nice custom operator for the disjunction operation and define a few more primitive operations, so that you can write for example this:
// Test against several terms in sequence
let sequence terms = (fun state -> terms |> Seq.fold (>>) state)
// Test for a substring
let substring s = sequence [ for c in s -> term (set [c]) ]
let test = // (abc|xyz)a{2,3}bc
( substring "abc" <|> substring "xyz" )
>> quantify 2 3 (term (set "a")) // (a{2,3})
>> substring "bc" // bc
This is more higher-level description, so it removes some of the >> operators in favor of functions that are more descriptive (and encapsulate >>). I also changed quantify to take multiple arguments instead of a tripple (which is a minor change)
If you want to play with this further, then you can take a look at the article and try to write F# computation expression builder that would allow you to use parser { .. } syntax.
This is generally good style but you're missing some tricks and still have quite a bit of redundancy. Maybe more like this:
let valid (s: State) = s.succeeded && s.index < s.input.Length
...
let disjunction leftTerm rightTerm s =
let left = leftTerm s
if left.succeeded then left else
let right = rightTerm s
if right.succeeded then right else
{ s with succeeded = false }
...
let test =
let f s = set s |> term
let (++) s t = f s >> f t
disjunction ("a" ++ "b" ++ "c") ("x" ++ "y" ++ "z")
>> quantify (f "a", 2, 3)
>> "b" ++ "c"
You might prefer to accumulate a value representing a computation rather than closures because it makes debugging much easier.

Randomly choose an instance from union in F#

In F#, given
type MyType = A | B | C | D | E | F | G
How do I randomly define an instance of MyType?
This ought to work:
let randInst<'t>() =
let cases = Reflection.FSharpType.GetUnionCases(typeof<'t>)
let index = System.Random().Next(cases.Length)
let case = cases.[index]
Reflection.FSharpValue.MakeUnion(case, [||]) :?> 't
This code assumes that the union cases are all nullary and that the type you're using is actually a union type, but it would be easy to explicitly check those assumptions and throw meaningful exceptions if desired.
Select a random number, then pattern match that number with different branches returning a different instant?

Resources