erlang Elixir -如何改进代码和风格

q3qa4bjr 于 2022-12-08 发布在 Erlang

关注(0)|答案(1)|浏览(138)

The goal was a script which reads the file, line by line, containing file paths (Windows and Linux). It strips the path leaving only a file name with an extension. Then replaces any special characters in the file name with an "_" - underscore and at the end reduces the consecutive underscores with just one. Like st__a___ck becomes st_a_ck. I got it working but I believe there may be a better/nicer looking way of doing this. I'm a very beginner and still learning to think the Elixir/functional way. What I want is to see different ways of doing this, ways of improving and elixifying a bit.
The test sample:

c:\program files\mydir\mydir2\my&@Doc.doc 
c:\program files\mydir\mydir2\myD$oc2.doc\ 
c:\\program files\\mydir\\mydir2\\myD;'oc2.doc
c:\\program files\\mydir\mydir2\\my[Doc2.doc\\
/home/python/projects/files.py
/home/python/projects/files.py/
//home//python//projects//files.py
//home//python//projects//files.py//
c:\program files\mydir\mydir2\my!D#oc.doc 
c:\program files\mydir\mydir2\myDoc2.doc\ 
c:\\program files\\mydir\\mydir2\\my';Doc2.doc
c:\\program files\\mydir\mydir2\\myD&$%oc2.doc\\
/home/python/projects/f_)*iles.py
/home/python/projects/files.py/
//home//python//projects//fi=-les.py
//home//python//projects//fil !%es.py//
/home/python/projects/f_)* iles.py
/home/python/projects/fi les.py/
//home//python//projects//fii___kiii=- les.py 
//home//python//projects//ff###f!%#illfffl! %es.py//

The code:

defmodule Paths do

     def read_file(filename) do
         File.stream!(filename)
         |> Enum.map( &(String.replace(&1,"\\","/")) )
         |> Enum.map( &(String.trim(&1,"\n")) )
         |> Enum.map( &(String.trim(&1,"/")) )
         |> Enum.map( &(String.split(&1,"/")) )
         |> Enum.map( &(List.last(&1)) )
         |> Enum.map( &(String.split(&1,".")) )
         |> Enum.map( &(remove_special)/1 )
         |> Enum.map( &(print_name_and_suffix)/1 )

     end
     defp print_name_and_suffix(str) do
         [h|t] = str
         IO.puts "Name: #{h}\t suffix: #{t}\t: #{h}.#{t}"
     end
     defp remove_special(str) do
         [h|t] = str
         h = String.replace(h, ~r/[\W]/, "_")
         h = String.replace(h, ~r/_+/, "_")
         [h]++t
     end

end

Paths.read_file("test.txt")

Any insights much appreciated.
EDIT: I refactored the code a little. Which version is more Elixir style like?

defmodule Paths do

     def read_file(filename) do
         File.stream!(filename)
         |> Enum.map( &(format_path)/1 )
         |> Enum.map( &(remove_special)/1 )
         |> Enum.map( &(print_name_and_suffix)/1 )

     end

     defp format_path(path) do
             path
             |> String.replace("\\","/")
             |> String.trim("\n")
             |> String.trim("/")
             |> String.trim("\\")
     end

     defp print_name_and_suffix(str) do
         [h|t] = str
         IO.puts "Name: #{h}\t suffix: #{t}\t: #{h}#{t}"
     end

     defp remove_special(str) do
         ext = Path.extname(str)
         filename = Path.basename(str)
             |> String.trim(ext)
             |> String.replace(~r/[\W]/, "_")
             |> String.replace( ~r/_+/, "_")

         [filename]++ext
     end

end

Paths.read_file("test.txt")

erlang

来源：https://stackoverflow.com/questions/53877753/elixir-how-to-improve-the-code-and-style

1条答案

按热度按时间

balp4ylt1#

I would point to the generic problems with the code in the first place.

File.stream!/3 produces a Stream explicitly designed to be processed lazily (so we don't keep the whole content of the file in memory). Passing it to Enum.map/2 makes zero sense. Use Stream.map/2 to keep processing the file lazily or use Flow.map/2 to parallelize the mapping operations and use all available cores (you keep the laziness too!).
Formatting matters. We use 2 spaces for the indent. Use Elixir Formatter (or mix task formatter ) to format your code.
Decompose directly in function head wherever possible (instead of defp print_name_and_suffix(str), do: [h|t] = str ... do directly defp print_name_and_suffix([h|t]) .
Minimize the number of calls to replacement in strings since each requires the separate string pass to substitute characters.
Use different function clauses with pattern matching to simplify life.
Try to use binary pattern matching and recursion wherever applicable.

That said, the most [opinionated] Elixirish approach would be:

defmodule Paths do
  def read_file(filename) do
    filename
    |> File.stream!()
    # Uncomment next line and replace all Steam calls with Flow 
    # to embrace multi core parallelism
    # |> Flow.from_enumerable()  
    |> Stream.map(&right_trim/1)
    |> Stream.map(&strip_path/1)
    |> Stream.map(&split_and_cleanup/1)
    |> Stream.map(&name_and_suffix/1)
    |> Enum.to_list()
  end

  defp right_trim(str), do: Regex.replace(~r/\W+\z/, str, "")

  defp strip_path(input, acc \\ "")
  defp strip_path("", acc), do: acc
  defp strip_path(<<"\\", rest :: binary>>, acc), do: strip_path(rest, "")
  defp strip_path(<<"/", rest :: binary>>, acc), do: strip_path(rest, "")
  defp strip_path(<<chr :: binary-size(1), rest :: binary>>, acc),
    do: strip_path(rest, acc <> chr)

  defp split_and_cleanup(str) do
    str
    |> String.split(".")
    |> Enum.map(&String.replace(&1, ~r/[_\W]+/, "_"))
  end

  defp name_and_suffix([file, ext]) do
    IO.puts "Name: #{file}\t suffix: .#{ext}\t: #{file}.#{ext}"
  end
end

Paths.read_file("/tmp/test.txt")

Please pay attention mostly to strip_path/2 function, it does recursively parse the input string, returning the part after the last slash, forward or backward. I could use String.split/2 or any internal function from String module but I explicitly had it implemented with a most functional approach.

赞(0）回复(0）举报 2022-12-08

我来回答

erlang Elixir -如何改进代码和风格

1条答案

相关问题

热门标签

最新问答