具有可选字符串属性的一组对象,改进一个接受String参数并在Set中找到最佳匹配的函数

问题描述 投票:0回答:2

我们正在尝试优化我们一直在使用的一些“加权”匹配算法,并决定咨询互联网以获得更多想法

我们有一个Struct MyStruct,有5个可选属性(在swift中,这只意味着属性可以是nil):

prop1: String?
prop2: String?
prop3: String?
prop4: String?
prop5: String?

然后我们有一组MyStruct(保证没有2个实例具有相同的确切属性),

structArray: Set<MyStruct>

我们有一个函数,它接受这个数组,以及字典中的1-5个属性,返回1个最匹配的单个实例。如果任何属性不匹配,则实例立即退出竞争

func findBestMatch(forSet set:Set<MyStruct>, andArgs argDict:[String:String]) -> MyStruct? {
  //Will use this to store any matches, as well as an associated match score
  var bestMatch: MyStruct?
  var topScore = 0
  for element in set {
    var score = 0
    if let p1 = argDict["p1"] {
      if p1 == element.prop1 {
        score += 16 //p1 match has highest weight
      } else {
        continue
      }
    }

    if let p2 = argDict["p2"] {
      if p2 == element.prop2 {
        score += 8 //p2 match has second-highest weight
      } else {
        continue
      }
    }

    //etc for the other 3 properties

    if score > topScore {
      topScore = score
      bestMatch = element 
    }
  }
  return bestMatch
}

例:

exInstance1
  prop1 = "no good"
  prop2 = nil
  prop3 = "goodbye

exInstance2
  prop1 = "hello"
  prop2 = "noproblem"
  prop3 = "goodbye"

exInstance3
  prop1 = nil
  prop2 = nil
  prop3 = "goodbye"

exampleSet: Set<MyStruct> = [exInstance1, exInstance2, exInstance3]

matchingProperties: [String:String] = {
  "p1": "hello",
  "p3": "goodbye"
}


findBestMatch(forSet: exampleSet, andArgs: matchingProperties)

在prop3上exInstance1只有1个匹配,但因为prop1根本不匹配,所以exInstance没有得分

exInstance2匹配两个属性,得分为20

exInstance3匹配一个属性,得分为4

选择并返回exInstance2


问题:有更好的方法吗?如果没有,有什么办法可以改进这个算法吗?

swift algorithm pattern-matching string-matching
2个回答
1
投票

如果你将for循环外的字典访问分解出来,我只会看到一点点改进,例如:

func findBestMatch(forSet set:Set<MyStruct>, andArgs argDict:[String:String]) -> MyStruct? {
    //Will use this to store any matches, as well as an associated match score
    var bestMatch: MyStruct?
    var topScore = 0
    let p1 = argDict["p1"]
    let p2 = argDict["p2"]  // and so on
    for element in set {
        var score = 0
        if let p1 = p1 {
            if p1 == element.prop1 {
                score += 16 //p1 match has highest weight
            } else {
                continue
            }
        }

        //etc for the other properties

        if score > topScore {
            topScore = score
            bestMatch = element
        }
    }
    return bestMatch
}

1
投票

在这种情况下,可以进行以下优化:

  1. 不要使用字典。相反,直接传递参数。
  2. 在循环的每次迭代中,对每个属性使用可选绑定。这可以避免。
  3. 如果对象的所有字段都不是可选的并且等于传递给函数的那些字段,那么它是最好的。

基于以上所述,我提出以下解决方案:

import Foundation

struct MyStruct {
    var prop1: String? = nil
    var prop2: String? = nil
    var prop3: String? = nil
}

extension MyStruct: Hashable {}

func findBestMatch(for set: Set<MyStruct>,
                   oProperty1: String? = nil,
                   oProperty2: String? = nil,
                   oProperty3: String? = nil) -> MyStruct?
{
    let mask000 = 0b00000000
    let mask001 = 0b00000001
    let mask010 = 0b00000010
    let mask011 = 0b00000011
    let mask100 = 0b00000100
    let mask101 = 0b00000101
    let mask110 = 0b00000110
    let mask111 = 0b00000111

    var mask = mask000

    if let _ = oProperty1 {
        mask |= mask001
    }
    if let _ = oProperty2 {
        mask |= mask010
    }
    if let _ = oProperty3 {
        mask |= mask100
    }

    if mask == mask000 {
        return nil
    } else if mask == mask001 {
        let prop3 = oProperty3!
        return set.first(where: { $0.prop3 == prop3 })
    } else if mask == mask010 {
        let prop2 = oProperty2!
        return set.first(where: { $0.prop2 == prop2 })
    } else if mask == mask011 {
        let prop2 = oProperty2!
        let prop3 = oProperty3!
        return set.first(where: { $0.prop2 == prop2 && $0.prop3 == prop3 })
    } else if mask == mask100 {
        let prop1 = oProperty1!
        return set.first(where: { $0.prop1 == prop1 })
    } else if mask == mask101 {
        let prop1 = oProperty1!
        let prop3 = oProperty3!
        return set.first(where: { $0.prop1 == prop1 && $0.prop3 == prop3 })
    } else if mask == mask110 {
        let prop1 = oProperty1!
        let prop2 = oProperty2!
        return set.first(where: { $0.prop1 == prop1 && $0.prop2 == prop2 })
    } else if mask == mask111 {
        let prop1 = oProperty1!
        let prop2 = oProperty2!
        let prop3 = oProperty3!
        return set.first(where: { $0.prop1 == prop1 && $0.prop2 == prop2 && $0.prop3 == prop3 })
    }
    return nil
}

let exInstance1 = MyStruct(prop1: "no good", prop2: nil, prop3: "goodbye")
let exInstance2 = MyStruct(prop1: "hello", prop2: "noproblem", prop3: "goodbye")
let exInstance3 = MyStruct(prop1: nil, prop2: nil, prop3: "goodbye")

let exampleSet: Set<MyStruct> = [
    exInstance1,
    exInstance2,
    exInstance3,
]

if let object = findBestMatch(for: exampleSet, oProperty1: "hello", oProperty2: nil, oProperty3: "goodbye") {
    print(object) // print MyStruct(prop1: Optional("hello"), prop2: Optional("noproblem"), prop3: Optional("goodbye"))
} else {
    print("not found")
}

我们将逐步分析您的示例:

  1. mask = 000
  2. oProperty1 != nil。 => Qazxswpoi
  3. mask = mask | 100 = 000 | 100 = 100。 => Qazxswpoi
  4. oProperty2 == nil。 => Qazxswpoi
  5. mask = 100。 => oProperty3 != nil
© www.soinside.com 2019 - 2024. All rights reserved.