# 推导式 [![badge](https://img.shields.io/endpoint.svg?url=https%3A%2F%2Fgezf7g7pd5.execute-api.ap-northeast-1.amazonaws.com%2Fdefault%2Fsource_up_to_date%3Fowner%3Derg-lang%26repos%3Derg%26ref%3Dmain%26path%3Ddoc/EN/syntax/28_comprehension.md%26commit_hash%3De959b3e54bfa8cee4929743b0193a129e7525c61)](https://gezf7g7pd5.execute-api.ap-northeast-1.amazonaws.com/default/source_up_to_date?owner=erg-lang&repos=erg&ref=main&path=doc/EN/syntax/28_comprehension.md&commit_hash=e959b3e54bfa8cee4929743b0193a129e7525c61) Array和`[expr | (name <- iterable)+ (predicate)*]`, set和`{expr | (name <- iterable)+ (predicate)*}`, 你可以创建一个字典`{key: value | (name <- iterable)+ (predicate)*}`. 由`|`分隔的子句的第一部分称为布局子句(位置子句)，第二部分称为绑定子句(绑定子句)，第三部分称为保护子句(条件子句) 保护子句可以省略，但绑定子句不能省略，保护子句不能在绑定子句之前理解示例 ```python # 布局子句是 i # 绑定子句是 i <- [0, 1, 2] assert [i | i <- [0, 1, 2]] == [0, 1, 2] # 布局子句是 i / 2 # 绑定子句是 i <- 0..2 assert [i/2 | i <- 0..2] == [0.0, 0.5, 1.0] # 布局子句是 (i, j) # 绑定子句 i <- 0..2, j <- 0..2 # 保护子句是 (i + j) % 2 == 0 assert [(i, j) | i <- 0..2; j <- 0..2; (i + j) % 2 == 0] == [(0, 0), (0, 2), (1, 1), (2, 0), (2, 2)] assert {i % 2 | i <- 0..9} == {0, 1} assert {k: v | k <- ["a", "b"]; v <- [1, 2]} == {"a": 1, "b": 2} ``` Erg推导式受到Haskell的启发，但有一些不同对于Haskell列表推导，变量的顺序会对结果产生影响，但在Erg中这并不重要 ``` haskell -- Haskell [(i, j) | i <- [1..3], j <- [3..5]] == [(1,3),(1,4),(1,5),(2 ,3),(2,4),(2,5),(3,3),(3,4),(3,5)] [(i, j) | j <- [3..5], i <- [1..3]] == [(1,3),(2,3),(3,3),(1 ,4),(2,4),(3,4),(1,5),(2,5),(3,5)] ``` ```python # Erg assert [(i, j) | i <- 1..<3; j <- 3..<5] == [(i, j) | j <- 3..<5; i <- 1.. <3] ``` 该规范与Python的规范相同 ```python # Python assert [(i, j) for i in range(1, 3) for j in range(3, 5)] == [(i, j) for j in range(3, 5) for i in range(1, 3)] ``` ## 筛子类型与推导类似的是筛类型。筛子类型是以`{Name: Type | Predicate}`创建的(枚举类型) refinement类型的情况下，只能指定一个Name，不能指定布局(但是如果是tuple类型可以处理多个值)，Predicate可以在编译时计算，即，只能指定一个常量表达式 ```python Nat = {I: Int | I >= 0} # 如果谓词表达式只有and，可以替换为: # Nat2D = {(I, J): (Int, Int) | I >= 0; J >= 0} Nat2D = {(I, J): (Int, Int) | I >= 0 and J >= 0} ```