An Iterator for Sliding Windows. Probably originally based on something from Word2Vec.jl

SlidingWindowIter.jl

type SlidingWindow
    ws::Any #Normally a word stream
    lsize::Int64
    rsize::Int64
end


Base.iteratorsize(::Type{SlidingWindow}) = Base.SizeUnknown()
window_length(window) = window.lsize + 1 + window.rsize

function Base.start(window::SlidingWindow)
    ws_state = start(window.ws)
    words = String["" for ii in 1:window_length(window)]
    
    for ii in 0:window_length(window)-2
	    _,(words,ws_state) = next(window, (words, ws_state))
    end
    (words, ws_state)
end

function Base.done(window::SlidingWindow, state)
    (words, ws_state) = state
    done(window.ws, ws_state) #CHECKME: Is this getting all the words I want at the end?
end

function Base.next(window::SlidingWindow, state)
    (words, ws_state) = state        
    next_word, ws_state = next(window.ws, ws_state)
    push!(words, next_word)    
    words = words[2:end] #PREMOPT: could use a circular bufffer
    
    (filter(w->length(w)>0, words), (words,ws_state))
end

function sliding_window(words; lsize=5, rsize=5)
    SlidingWindow(words, lsize, rsize)
end

SlidingWindowTest.jl

using FactCheck

facts("Should get a sliding window") do 
	ww=words_of(data)
	windows = collect(sliding_window(ww,lsize=1,rsize=1))
	@fact windows[1] --> String["the","king","and"]
	@fact windows[2] --> String["king","and", "his"]

	bigger_windows = collect(sliding_window(ww,lsize=2,rsize=2))
	@fact bigger_windows[end] --> String["find","the","land","of","gold"]
end