module Utils

import TOML: parsefile
import DelimitedFiles: readdlm
import DSP

function load_init_params(filename::String)
  meta = parsefile(filename)["metadata"]
  params = NamedTuple(Dict(Symbol(p["symbol"]) => p["initial_value"] for p in meta["parameters"]))
  lower_bounds = [p["min"] for p in meta["parameters"]]
  upper_bounds = [p["max"] for p in meta["parameters"]]
  return (meta=meta, params=params, lower_bounds=lower_bounds, upper_bounds=upper_bounds)
end

function select_columns(QIE_array, qmin, qmax, bin, err_mul)
  #""Select q / I(q) / err(q) data	-	Change I(q) value from mm^-1 to nm^-1'

  Q = Float64[]
  IDATA = Float64[]
  ERR = Float64[]

  for i in axes(QIE_array, 1)
    # keep only the values of q within bounds
    if qmin <= QIE_array[i, 1] <= qmax
      # skip all but the (k*bin)-th values
      if i % bin == 0
        if QIE_array[i, 2] > 0
          push!(Q, QIE_array[i, 1])
          push!(IDATA, 1e-6 * QIE_array[i, 2])
          if err_mul == 0
            push!(ERR, 0.01 * 1e-6 * QIE_array[i, 2])
          elseif QIE_array[i, 3] * err_mul > 0.01 * QIE_array[i, 2]
            push!(ERR, 1e-6 * QIE_array[i, 3] * err_mul)
          else
            push!(ERR, 0.01 * 1e-6 * QIE_array[i, 2])
          end
        end
      end
    end
  end

  return (Q, IDATA, ERR)
end

"""
    compute_logscale_weights(q)

Compute the weights W, which are proportional to the distance in the log space
"""
function compute_logscale_weights(q)
  diff_log_q = diff(log10.(q))
  left_diff_log_q = view(diff_log_q, [firstindex(diff_log_q); eachindex(diff_log_q)])
  right_diff_log_q = view(diff_log_q, [eachindex(diff_log_q); lastindex(diff_log_q)])

  return left_diff_log_q + right_diff_log_q
end

function load_config(filename::String)
  meta, params_init, lower_bounds, upper_bounds = load_init_params(filename)
  f = open(meta["data_file"], "r")
  qie_data = readdlm(f, '\t')
  close(f)
  return (meta=meta,
    params_init=params_init,
    lower_bounds=lower_bounds,
    upper_bounds=upper_bounds,
    qie_deta=qie_data)
end

function lowpass_filter(i; σ=1)
  half_w = 10
  x = range(-5, 5; length=2half_w + 1)
  i_padded = view(i, [fill(firstindex(i), half_w); eachindex(i); fill(lastindex(i), half_w)])
  k = exp.(-0.5 * ((x / σ) .^ 2)) / (2 * σ * sqrt(2π))
  k ./= sum(k)
  @show sum(k)
  r = DSP.conv(i_padded, k)[2half_w+1:end-2half_w]
  @show size(r)
  return r
end

function chi2(I_data, I_model, err)
  return sum((I_data .- I_model ./ err) .^ 2)
end

end # module Utils