Copyright | (C) 2012-16 Edward Kmett |
---|---|

License | BSD-style (see the file LICENSE) |

Maintainer | Edward Kmett <ekmett@gmail.com> |

Stability | provisional |

Portability | Rank2Types |

Safe Haskell | None |

Language | Haskell98 |

This module provides combinators for breadth-first searching within arbitrary traversals.

- data Level i a
- levels :: ATraversal s t a b -> IndexedTraversal Int s t (Level () a) (Level () b)
- ilevels :: AnIndexedTraversal i s t a b -> IndexedTraversal Int s t (Level i a) (Level j b)

# Documentation

This data type represents a path-compressed copy of one level of a source data structure. We can safely use path-compression because we know the depth of the tree.

Path compression is performed by viewing a `Level`

as a PATRICIA trie of the
paths into the structure to leaves at a given depth, similar in many ways
to a `IntMap`

, but unlike a regular PATRICIA trie we do not need
to store the mask bits merely the depth of the fork.

One invariant of this structure is that underneath a `Two`

node you will not
find any `Zero`

nodes, so `Zero`

can only occur at the root.

TraversableWithIndex i (Level i) Source | |

FoldableWithIndex i (Level i) Source | |

FunctorWithIndex i (Level i) Source | |

Functor (Level i) Source | |

Foldable (Level i) Source | |

Traversable (Level i) Source | |

(Eq i, Eq a) => Eq (Level i a) Source | |

(Ord i, Ord a) => Ord (Level i a) Source | |

(Read i, Read a) => Read (Level i a) Source | |

(Show i, Show a) => Show (Level i a) Source | |

levels :: ATraversal s t a b -> IndexedTraversal Int s t (Level () a) (Level () b) Source

This provides a breadth-first `Traversal`

of the individual `levels`

of any other `Traversal`

via iterative deepening depth-first search. The levels are returned to you in a compressed format.

This can permit us to extract the `levels`

directly:

`>>>`

[Zero,Zero,One () 'h',Two 0 (One () 'e') (One () 'w'),Two 0 (One () 'l') (One () 'o'),Two 0 (One () 'l') (One () 'r'),Two 0 (One () 'o') (One () 'l'),One () 'd']`["hello","world"]^..levels (traverse.traverse)`

But we can also traverse them in turn:

`>>>`

"hewlolrold"`["hello","world"]^..levels (traverse.traverse).traverse`

We can use this to traverse to a fixed depth in the tree of (`<*>`

) used in the `Traversal`

:

`>>>`

["HEllo","World"]`["hello","world"] & taking 4 (levels (traverse.traverse)).traverse %~ toUpper`

Or we can use it to traverse the first `n`

elements in found in that `Traversal`

regardless of the depth
at which they were found.

`>>>`

["HELlo","World"]`["hello","world"] & taking 4 (levels (traverse.traverse).traverse) %~ toUpper`

The resulting `Traversal`

of the `levels`

which is indexed by the depth of each `Level`

.

`>>>`

[(2,'d'),(3,'o'),(3,'c'),(4,'g'),(4,'a'),(5,'t')]`["dog","cat"]^@..levels (traverse.traverse) <. traverse`

*Note:* Internally this is implemented by using an illegal `Applicative`

, as it extracts information
in an order that violates the `Applicative`

laws.

ilevels :: AnIndexedTraversal i s t a b -> IndexedTraversal Int s t (Level i a) (Level j b) Source

This provides a breadth-first `Traversal`

of the individual levels of any other `Traversal`

via iterative deepening depth-first search. The levels are returned to you in a compressed format.

This is similar to `levels`

, but retains the index of the original `IndexedTraversal`

, so you can
access it when traversing the levels later on.

`>>>`

[((0,0),'d'),((0,1),'o'),((1,0),'c'),((0,2),'g'),((1,1),'a'),((1,2),'t')]`["dog","cat"]^@..ilevels (traversed<.>traversed).itraversed`

The resulting `Traversal`

of the levels which is indexed by the depth of each `Level`

.

`>>>`

[((2,(0,0)),'d'),((3,(0,1)),'o'),((3,(1,0)),'c'),((4,(0,2)),'g'),((4,(1,1)),'a'),((5,(1,2)),'t')]`["dog","cat"]^@..ilevels (traversed<.>traversed)<.>itraversed`

*Note:* Internally this is implemented by using an illegal `Applicative`

, as it extracts information
in an order that violates the `Applicative`

laws.