Refactor

[korenmiklos]

1. All command rewrites moved to rewrites.jl. Helper functions remained in codegen.jl.
2. For simpler commands, all setup code (bitmask, creating sdf, checking for syntax) is prewritten. You can just add $setup to the beginning of the quote block.
3. $teardown is a function that passes through the argument given to it, while executing some side effects necessary to finish the command. For example, removing _n if it was added to the df.

So, the typical use case is:

function rewrite(::Val{:regress}, command::Command)
    gc = generate_command(command; options=[:variables, :ifable])
    (; df, local_copy, sdf, gdf, setup, teardown, arguments) = gc
    quote
        $setup
        reg($sdf, @formula $(arguments[1]) ~ $(sum(arguments[2:end]))) |> $teardown
    end |> esc
end

[korenmiklos]

julia> @regress df a _n @if _n < _N
[ Info: The variance-covariance matrix is not invertible. F-statistic not computed
                            FixedEffectModel
=========================================================================
Number of obs:                    9   Converged:                     true
dof (model):                      1   dof (residuals):                  6
R²:                           1.000   R² adjusted:                  1.000
F-statistic:                    NaN   P-value:                        NaN
=========================================================================
             Estimate  Std. Error  t-stat  Pr(>|t|)  Lower 95%  Upper 95%
─────────────────────────────────────────────────────────────────────────
_n                1.0         0.0     Inf    <1e-99        1.0        1.0
(Intercept)       0.0         0.0    NaN     NaN           0.0        0.0
=========================================================================

julia> df
10×1 DataFrame
 Row │ a
     │ Int64
─────┼───────
   1 │     1
   2 │     2
   3 │     3
   4 │     4
   5 │     5
   6 │     6
   7 │     7
   8 │     8
   9 │     9
  10 │    10