Posexional row as struct

[GPrimola]

This is a proposal for #11.

Obs 1: There's no breaking changes.
Obs 2: I apologize about the size of this PR. 😅

Major Changes

• Option [:as_struct] to PosexionalRow.__using__/1 macro. Not breaking.
• New field Posexional.Field.Field that parses field type text. It behaves like Ecto.Schema.field/3. Not breaking.

Examples

PosexionalRow.__using__/1 with option [:as_struct] and PosexionalRow.field/4 with simple data types:

Suppose we have the following row definition to be mapped to a struct:

defmodule Row1 do
  use PosexionalRow, [:as_struct]

  guesser &__MODULE__.guesser/1

  fixed_value "row1"
  field :id, :id, 3
  field :binary_id, :binary_id, 3
  field :integer, :integer, 2
  field :float, :float, 6
  field :boolean, :boolean, 5
  field :string, :string, 4
  field :binary, :binary, 4

  def guesser("row1" <> _), do: true
  def guesser(_), do: false
end

defmodule StructRowFile do
  use PosexionalFile

  row Row1
end

and a positional file (example.txt - not included) with the content row1123ABA00432.99falseJohn Doe. Then

StructRowFile.read("row1123ABA00432.99falseJohn Doe")
=>
[
  %Row1{
    binary: "Doe",
    binary_id: "ABA",
    boolean: false,
    float: 432.99,
    id: 123,
    integer: 0,
    string: "John"
  }
]

PosexionalRow.__using__/1 with option [:as_struct] and PosexionalRow.field/4 with complex data types:

defmodule Row2 do
  use PosexionalRow, [:as_struct]

  guesser &__MODULE__.guesser/1

  field :custom, :custom, 4, parser: &String.to_atom/1
  field :array, :array, 4, separator: ""
  field :list, :list, 7, separator: "|"
  field :decimal, :decimal, 8
  field :date, :date, 8, format_in: "{0M}{0D}{YYYY}"
  field :time, :time, 8, format_in: "%H%M%S%f"
  field :time2, :time, 6, format_in: "%H%M%S"
  field :datetime, :datetime, 14, format_in: "%d%m%Y%H%M%S", timezone: "America/Sao_Paulo"
  field :naive_dt, :naive_datetime, 23, format_in: "%d/%m/%Y@%H:%M:%S.%L"
  field :utc_dt, :utc_datetime, 23, format_in: "%Y-%m-%d %H:%M:%S.%L"

  def guesser("afoo" <> _), do: true
  def guesser(_), do: false
end

defmodule StructRowFile do
  use PosexionalFile

  row Row2
end

StructRowFile.read("afoo1234cat|dog123.45  01312021235959990745001209198807450031/12/2021@23:59:59.9992022-01-01 00:00:00.000")
=>
[
%Row2{
     array: ["1", "2", "3", "4"],
     custom: :afoo,
     date: ~D[2021-01-31],
     datetime: %DateTime{
       year: 1988,
       month: 9,
       day: 12,
       hour: 7,
       minute: 45,
       second: 00,
       utc_offset: -10800,
       time_zone: "America/Sao_Paulo",
       std_offset: 0,
       zone_abbr: "-03"
     },
     decimal: 123.45,
     list: ["cat", "dog"],
     naive_dt: ~N[2021-12-31 23:59:59.999],
     time: ~T[23:59:59.99],
     time2: ~T[07:45:00],
     utc_dt: ~U[2022-01-01 00:00:00Z]
   }
]

If you want more information about the behaviour, you can check the test/posexional/field/field_test.exs file where there are examples of bad input and how it's been handled.

Please, feel free to comment, propose changes, code refactor, variable, module, function and option naming, etc. I'll be glad to tackle everything!

Congrats about the lib, though! It's been very helpful! 😄

[neslinesli93]

Amazing work, for real ❤️

Unfortunately, our CI service is purposely disabled for public projects, so it can't be run on this PR 😢
I've tried locally to run credo and dialyzer and I got a bunch of errors. You can try mix check and see that credo complains about single pipes, and some other stuff.

Apart from that, I'm not a huge fan of having a Posexional.Field.Field module, mostly for the repetition of words and because imho it's not that self-explanatory. Maybe something like TypedField would be better? I like the reference to Ecto, but the repetition kills me :D

[GPrimola]

Hey @neslinesli93,

Thank you for your review! I've already tackled everything you pointed out!
And agree, Posexional.Field.Field was a bad name! Thanks for the suggestion! ;)

If you have anything more to add, please feel free to do it! :)

[chess4ever]

Hi @GPrimola ! You have done a really great work here! Thank you for you contribution!
We would like to merge your PR a soon as possible since it is valuable.

I have run dialyzer locally (in the container) and there are still some issues (I count 9 of those).

Let us know if you need help to solve them! We are glad to help!

Thank you again!

[chess4ever]

@GPrimola is it ready to be merged?

[GPrimola]

Hey @chess4ever, thanks!

Actually I missed letting dialyzer run fully.
Now I think everything should be fine!

Thank you!