Concatenate query list field once

[preciz]

Assited-by: Codex CLI:GPT 5.6 Sol

Concats field string once.
It's faster and uses less memory.

Bench:

Mix.install([:benchee])

defmodule Before do
  def encode({field, list}) do
    field
    |> encode_pair(list)
    |> IO.iodata_to_binary()
  end

  defp encode_pair(parent_field, list) when is_list(list) do
    mapper = fn value ->
      [?&, encode_pair(parent_field <> "[]", value)]
    end

    list
    |> Enum.flat_map(mapper)
    |> prune()
  end

  defp encode_pair(field, value), do: [field, ?= | to_string(value)]

  defp prune([?& | rest]), do: rest
  defp prune([]), do: []
end

defmodule After do
  def encode({parent_field, list}) do
    parent_field
    |> encode_pair(list)
    |> IO.iodata_to_binary()
  end

  defp encode_pair(parent_field, list) when is_list(list) do
    field = parent_field <> "[]"
    Enum.map_intersperse(list, ?&, fn value -> encode_pair(field, value) end)
  end

  defp encode_pair(field, value), do: [field, ?= | to_string(value)]
end

Benchee.run(
  %{
    "before" => &Before.encode/1,
    "after" => &After.encode/1
  },
  inputs: %{
    "5 tags" => {"tags", ~w(elixir phoenix plug http web)},
    "20 category IDs" => {"category_ids", Enum.to_list(1..20)},
    "100 record IDs" => {"record_ids", Enum.to_list(1..100)}
  },
  pre_check: :all_same,
  warmup: 1,
  time: 2,
  memory_time: 1
)

Results:

Operating System: Linux
CPU Information: AMD Ryzen 7 8845HS w
Number of Available Cores: 16
Available memory: 54.72 GB
Elixir 1.20.3
Erlang 29.0.5
JIT enabled: true

Benchmark suite executing with the following configuration:
warmup: 1 s
time: 2 s
memory time: 1 s
reduction time: 0 ns
parallel: 1
inputs: 100 record IDs, 20 category IDs, 5 tags
Estimated total run time: 24 s
Excluding outliers: false

Benchmarking after with input 100 record IDs ...
Benchmarking after with input 20 category IDs ...
Benchmarking after with input 5 tags ...
Benchmarking before with input 100 record IDs ...
Benchmarking before with input 20 category IDs ...
Benchmarking before with input 5 tags ...
Calculating statistics...
Formatting results...

##### With input 100 record IDs #####
Name             ips        average  deviation         median         99th %
after       149.27 K        6.70 μs    ±86.56%        6.40 μs       11.46 μs
before       75.19 K       13.30 μs    ±90.31%       10.03 μs       92.55 μs

Comparison: 
after       149.27 K
before       75.19 K - 1.99x slower +6.60 μs

Memory usage statistics:

Name      Memory usage
after          8.73 KB
before        18.02 KB - 2.07x memory usage +9.30 KB

**All measurements for memory usage were the same**

##### With input 20 category IDs #####
Name             ips        average  deviation         median         99th %
after       664.63 K        1.50 μs   ±597.09%        1.33 μs        2.60 μs
before      323.32 K        3.09 μs   ±482.35%        2.13 μs        7.21 μs

Comparison: 
after       664.63 K
before      323.32 K - 2.06x slower +1.59 μs

Memory usage statistics:

Name      Memory usage
after          1.85 KB
before         3.65 KB - 1.97x memory usage +1.80 KB

**All measurements for memory usage were the same**

##### With input 5 tags #####
Name             ips        average  deviation         median         99th %
after         3.39 M      295.24 ns   ±830.22%         251 ns         621 ns
before        1.43 M      700.15 ns  ±1525.31%         461 ns        1342 ns

Comparison: 
after         3.39 M
before        1.43 M - 2.37x slower +404.92 ns

Memory usage statistics:

Name      Memory usage
after            472 B
before           872 B - 1.85x memory usage +400 B

**All measurements for memory usage were the same**

[josevalim]

💚 💙 💜 💛 ❤️