Sha256: 4156d3d8bb48d4aee6e8ffe5efcf8709b819024928b087bc049e6e61b01a8bc9
Contents?: true
Size: 1.15 KB
Versions: 327
Compression:
Stored size: 1.15 KB
Contents
defmodule Frequency do
@doc """
Count word frequency in parallel.
Returns a map of characters to frequencies.
"""
def frequency(texts, workers) do
groups = Enum.map(0..(workers-1), &stripe(&1, texts, workers))
Enum.map(groups, &Frequency.count_texts/1)
#:rpc.pmap({Frequency, :count_texts}, [], groups)
|> merge_freqs()
end
defp stripe(n, texts, workers) do
Enum.drop(texts, n) |> Enum.take_every(workers)
end
# Needs to be public because of how it's invoked by `:rpc.pmap/4`.
@doc false
def count_texts(texts) do
Enum.map(texts, &count_text/1)
|> merge_freqs()
end
defp count_text(string) do
String.downcase(string)
|> String.graphemes()
# At the time of writing Elixir doesn't yet have a way to determine if a
# character is a letter. So use a workaround with Regex.
String.replace(string, ~r/\P{L}+/u, "") # \P{L} = anything but a letter
|> String.downcase()
|> String.graphemes()
|> Enum.reduce(%{}, fn c, acc -> Map.update(acc, c, 1, &(&1+1)) end)
end
defp merge_freqs(map) do
Enum.reduce(map, %{}, fn d, acc ->
Map.merge(acc, d, fn _, a, b -> a+b end)
end)
end
end
Version data entries
327 entries across 327 versions & 1 rubygems