smalltalk - 从 OrderedCollection 中选择以下项目的最优雅方法是什么：

Question

给定这样的 OrderedCollection：

noise1
noise2
noise3
signal1
signal1
signal1
signal1
randomButInteresting
noise4
noise5

我想为新的 OrderedCollection 选择所有对象“signal1”以及这一系列“signal1”之后的对象，“randomButInteresting”。（一系列相同的信号在每个 Collection 中只出现一次。）

最优雅的方法是什么？

Answer 1

直接的方法类似于

| data result lastWasSignal |

data := #( #noise1 #noise2 #noise3 #signal1 #signal1 #signal1 #signal1 #randomButInteresting #noise4 #noise5 ).

lastWasSignal := false.
result := data select: [ :value |
    | isElementAppropriate |
    isElementAppropriate := value = #signal1 or: [ lastWasSignal ].
    lastWasSignal := value = #signal1.
    isElementAppropriate
].

result

是 O(n)。更聪明的方法是使用二分搜索找到仅出现一次的信号组的边界。

Answer 2

使用带有 PetitParser 的 Lukas 版本，但在结果中保留所有“signal1”：

" the parser that accepts the symbol #signal1 "
signal := PPPredicateObjectParser expect: #signal1.

" the parser that accepts many symbols #signal1 followed by something else "
pattern := signal plus , signal negate.

data := #(noise1 noise2 noise3 signal1 signal1 signal1 signal1 randomButInteresting noise4 noise5).

pattern flatten matchesSkipIn: data           -> an OrderedCollection(#(#signal1 #signal1 #signal1 #signal1 #randomButInteresting))

Answer 3

您可以使用PetitParser，因为您基本上匹配输入流上的特定模式。为清楚起见添加了一些注释的解析器定义如下：

" the parser that accepts the symbol #signal1 "
signal := PPPredicateObjectParser expect: #signal1.

" the parser that accepts the symbol #signal1 not followed by something else "
pattern := signal , signal negate.

" the parser that extract the second symbol "
parser := pattern map: [ :signal :random | random ].

当您在输入数据上运行它时，您会得到：

data := #(noise1 noise2 noise3 signal1 signal1 
          signal1 signal1 randomButInteresting
          noise4 noise5).
parser matchesIn: data -> #(randomButInteresting)

Answer 4

使用输入和输出流的不同解决方案（是的，我喜欢流:-)）：

data := #(noise1 noise2 noise3 signal1 signal1 signal1 signal1 randomButInteresting noise4 noise5).

"Let's create an OrderedCollection from an output stream"
OrderedCollection streamContents: [:output |
    |datast|
    "We are basically streaming on the input data, so let's use a stream:"
    datast := data readStream.
    "We ignore everything before #signal1"
    datast skipTo: #signal1.
    "We add the #signal1we just found"
    output nextPut: #signal1.
    "And we add all the subsequent #signal1"
    [datast peek = #signal1]
        whileTrue: [output nextPut: datast next].
    "Finally we add the element after the last #signal1"
    output nextPut: datast next
    ]