我将哪些参数传递给 Scala 中的“过滤器”function

Question

I am having trouble understanding this Scala code involving a lambda expression:

object ScalaList {
  case class Student(rollNo : Int, name : String, marks : Int)
  val students=   List(Student(1,"A",10),Student(2,"B",14))
  var toppers = students.filter(s=>s.marks>10)    //> toppers  : List[ScalaList.Student] = List(Student(2,B,14))
}

It works, but the filter function takes an argument 's' that I didn't define. What is 's'. Also, if this were pandas, I think I would need to supply an axis=1 parameter. Why don't I need it in Scala

Answer 1

Perhaps it would help to compare Python implementation

>>> from pydantic import BaseModel
>>>
>>> class Student(BaseModel):
...     rollNo: int
...     name: str
...     marks: int
...
>>> students = [
...     Student(rollNo=1, name="A", marks=10),
...     Student(rollNo=2, name="B", marks=14)
... ]
>>>
>>> list(filter(lambda s: s.marks > 10, students))
[Student(rollNo=2, name='B', marks=14)]

with similar Scala implementation

scala> case class Student(
     |     rollNo: Int,
     |     name: String,
     |     marks: Int
     | )
     | val students = List(
     |     Student(1, "A", 10),
     |     Student(2, "B", 14)
     | )
     | students.filter(s => s.marks > 10)
class Student
val students: List[Student] = List(Student(1,A,10), Student(2,B,14))
val res0: List[Student] = List(Student(2,B,14))

where we see Scala lambda

s => s.marks > 10

is similar to Python lambda

lambda s: s.marks > 10

which should make it clear that declared lambda parameter s becomes a local variable inside the body of the lambda.

Consider Scala API documentation for List#filter as well as Scala for Python Developers section of Scala 3 Book.

Answer 2

Note that the argument to the filter method is not s , but rather the entirety of s => s.marks > 10 : this is the syntax to define an anonymous function that takes some s and returns the result of the comparison of its marks field with the literal 10 . In this context, s is just the made up name of a parameter of the function, just like it would be if it was defined as a method as in the following example:

def marksGreaterThan10(s: Student): Boolean = s.marks > 10

One thing that could make someone a bit confused about the syntax is that the compiler itself is able to infer the types based on your input, making it very compact and readable, but perhaps a bit opaque for newcomers. Note that without the appropriate context (ie a list of Student s is being filtered, therefore the input type of the function must be a Student ), the compiler would complain about the missing parameter type and you would need to provide one, as in the following example (where I assign the anonymous function to a variable, so that I can use it in some other part of the program):

val marksGreaterThan10 = (s: Student) => s.marks > 10

Note that both a method and a function can be passed to a higher-order function (a function that can take a function as a parameter or return one as an output) like filter .

final case class Student(marks: Int)

val marksGreaterThan10Function = (s: Student) => s.marks > 10

def marksGreaterThan10Method(s: Student): Boolean = s.marks > 10

val students = Seq(Student(1), Student(11))

// Both of these returns `Seq(Student(11))`
students.filter(marksGreaterThan10Function)
students.filter(marksGreaterThan10Method)

You can play around with the code above here on Scastie.

If you are familiar with the Java lambda expression syntax, -> is more or less equivalent to => and you would have written something like:

students.stream().filter(s -> s.marks > 10)