在 scalacheck 中生成递归结构

Question

我正在尝试为名为@9​​87654321@ 的递归数据类型创建一个生成器。一行是命名Vals 的列表，其中Val 是原子Bin 或嵌套Row。

这是我的代码：

package com.dtci.data.anonymize.parquet

import java.nio.charset.StandardCharsets
import org.scalacheck.Gen

object TestApp extends App {

  sealed trait Val
  case class Bin(bytes: Array[Byte]) extends Val
  object Bin {
    def from_string(str: String): Bin = Bin(str.getBytes(StandardCharsets.UTF_8))
  }
  case class Row(flds: List[(String, Val)]) extends Val

  val gen_bin = Gen.alphaStr.map(Bin.from_string)
  val gen_field_name = Gen.alphaLowerStr
  val gen_field = Gen.zip(gen_field_name, gen_val)
  val gen_row = Gen.nonEmptyListOf(gen_field).map(Row.apply)
  def gen_val: Gen[Val] = Gen.oneOf(gen_bin, gen_row)

  gen_row.sample.get.flds.foreach( fld => println(s"${fld._1} --> ${fld._2}"))
}

它因以下堆栈跟踪而崩溃：

Exception in thread "main" java.lang.NullPointerException
    at org.scalacheck.Gen.$anonfun$flatMap$2(Gen.scala:84)
    at org.scalacheck.Gen$R.flatMap(Gen.scala:243)
    at org.scalacheck.Gen$R.flatMap$(Gen.scala:240)
    at org.scalacheck.Gen$R$$anon$3.flatMap(Gen.scala:228)
    at org.scalacheck.Gen.$anonfun$flatMap$1(Gen.scala:84)
    at org.scalacheck.Gen$Parameters.useInitialSeed(Gen.scala:318)
    at org.scalacheck.Gen$$anon$5.doApply(Gen.scala:255)
    at org.scalacheck.Gen$$anon$1.$anonfun$doApply$1(Gen.scala:110)
    at org.scalacheck.Gen$Parameters.useInitialSeed(Gen.scala:318)
    at org.scalacheck.Gen$$anon$1.doApply(Gen.scala:109)
    at org.scalacheck.Gen.$anonfun$map$1(Gen.scala:79)
    at org.scalacheck.Gen$Parameters.useInitialSeed(Gen.scala:318)
    at org.scalacheck.Gen$$anon$5.doApply(Gen.scala:255)
    at org.scalacheck.Gen.$anonfun$flatMap$2(Gen.scala:84)
    at org.scalacheck.Gen$R.flatMap(Gen.scala:243)
    at org.scalacheck.Gen$R.flatMap$(Gen.scala:240)
    at org.scalacheck.Gen$R$$anon$3.flatMap(Gen.scala:228)
    at org.scalacheck.Gen.$anonfun$flatMap$1(Gen.scala:84)
    at org.scalacheck.Gen$Parameters.useInitialSeed(Gen.scala:318)
    at org.scalacheck.Gen$$anon$5.doApply(Gen.scala:255)
    at org.scalacheck.Gen$$anon$1.$anonfun$doApply$1(Gen.scala:110)
    at org.scalacheck.Gen$Parameters.useInitialSeed(Gen.scala:318)
    at org.scalacheck.Gen$$anon$1.doApply(Gen.scala:109)
    at org.scalacheck.Gen$.$anonfun$sequence$2(Gen.scala:492)
    at scala.collection.LinearSeqOps.foldLeft(LinearSeq.scala:168)
    at scala.collection.LinearSeqOps.foldLeft$(LinearSeq.scala:164)
    at scala.collection.immutable.List.foldLeft(List.scala:79)
    at org.scalacheck.Gen$.$anonfun$sequence$1(Gen.scala:490)
    at org.scalacheck.Gen$Parameters.useInitialSeed(Gen.scala:318)
    at org.scalacheck.Gen$$anon$5.doApply(Gen.scala:255)
    at org.scalacheck.Gen.$anonfun$map$1(Gen.scala:79)
    at org.scalacheck.Gen$Parameters.useInitialSeed(Gen.scala:318)
    at org.scalacheck.Gen$$anon$5.doApply(Gen.scala:255)
    at org.scalacheck.Gen$$anon$1.$anonfun$doApply$1(Gen.scala:110)
    at org.scalacheck.Gen$Parameters.useInitialSeed(Gen.scala:318)
    at org.scalacheck.Gen$$anon$1.doApply(Gen.scala:109)
    at org.scalacheck.Gen.$anonfun$flatMap$2(Gen.scala:84)
    at org.scalacheck.Gen$R.flatMap(Gen.scala:243)
    at org.scalacheck.Gen$R.flatMap$(Gen.scala:240)
    at org.scalacheck.Gen$R$$anon$3.flatMap(Gen.scala:228)
    at org.scalacheck.Gen.$anonfun$flatMap$1(Gen.scala:84)
    at org.scalacheck.Gen$Parameters.useInitialSeed(Gen.scala:318)
    at org.scalacheck.Gen$$anon$5.doApply(Gen.scala:255)
    at org.scalacheck.Gen$.$anonfun$sized$1(Gen.scala:551)
    at org.scalacheck.Gen$Parameters.useInitialSeed(Gen.scala:318)
    at org.scalacheck.Gen$$anon$5.doApply(Gen.scala:255)
    at org.scalacheck.Gen$$anon$1.$anonfun$doApply$1(Gen.scala:110)
    at org.scalacheck.Gen$Parameters.useInitialSeed(Gen.scala:318)
    at org.scalacheck.Gen$$anon$1.doApply(Gen.scala:109)
    at org.scalacheck.Gen.$anonfun$map$1(Gen.scala:79)
    at org.scalacheck.Gen$Parameters.useInitialSeed(Gen.scala:318)
    at org.scalacheck.Gen$$anon$5.doApply(Gen.scala:255)
    at org.scalacheck.Gen.sample(Gen.scala:154)

我的代码出了什么问题，我自己诊断它的最佳方法是什么？

作为说明，我已经看到关于 Gen.oneOf 严格并且需要 Gen.lzy 用于递归结构的评论。但是，如果在我的代码中，我将 gen_val 的定义包装在 Gen.lzy(...) 中，那么我会得到堆栈溢出，而不是当前的空指针异常。

Answer 1

首先，小心使用object Main extends App。我发现它的字段初始化语义不如具有逐行语义的普通旧 main 明显：

object Main {
  def main(args: Array[String]): Unit = {...}
}

这可能是NullPointerException 的问题。

通常，可以通过仔细检查字段初始化顺序并将部分（或全部）val 标记为lazy 来解决此问题。

StackOverflowError 的出现是因为生成的数据结构太深。

通常，当您处理任何类型的递归时，请始终考虑 基本情况 何时应该停止递归以及 步骤 最终会遇到 基本情况。

在您的特定情况下，我们可以使用Gen.sized 和Gen.resize，它们负责生成元素的“大”程度（查看文档和谷歌了解更多信息）。

package com.dtci.data.anonymize.parquet

import java.nio.charset.StandardCharsets
import org.scalacheck.Gen

object Main extends App {

  sealed trait Val
  case class Bin(bytes: Array[Byte]) extends Val
  object Bin {
    def from_string(str: String): Bin = Bin(str.getBytes(StandardCharsets.UTF_8))
  }
  case class Row(flds: List[(String, Val)]) extends Val

  val gen_bin = Gen.alphaStr.map(Bin.from_string)
  val gen_field_name = Gen.alphaLowerStr
  val gen_field = Gen.zip(gen_field_name, gen_val)
  val gen_row = Gen.sized(size => Gen.resize(size / 2, Gen.nonEmptyListOf(gen_field).map(Row.apply)))

  def gen_val: Gen[Val] = Gen.sized { size =>
    if (size <= 0) {
      gen_bin
    } else {
      Gen.oneOf(gen_bin, gen_row)
    }
  }

  gen_row.sample.get.flds.foreach(fld => println(s"${fld._1} --> ${fld._2}"))
}