add SymPy plugin

[ranocha]

I hope this follows the recommended style. When it works, it would be really great to get a new release of Latexify.jl.

[gustaphe]

Latexify is not intended as a collection of plugins for different types, the ones that are here are holdovers from before the recipe system was introduced.

Depending on if the SymPy devs want to support this or not, I would suggest either making a PR pulling in Latexify and creating a recipe in SymPy, or creating a SymPyLatexify glue package.

(I'm not a maintainer, just someone who has done this before)

[ranocha]

Hmm, I see. Could you please comment on #185 in that case? The problem is that people might want to avoid pulling in all dependencies of Latexify.jl and the official recipe system using a macro doesn't work with Requires.jl.

[ranocha]

And I don't think glue packages are a good solution. If I want to use two packages together, I don't want to depend on a third one.

[ChrisRackauckas]

SymPy just needs to add the recipe to their package and then everything is 👍

[gustaphe]

I was just working on a comment, but it took a while because I can't figure out Requires.
As the author of UnitfulLatexify I know your pain. If/when JuliaLang/Pkg.jl#1285 is solved I'm celebrating. OTOH, pulling in a package isn't that hard or expensive in julia, I've learned to not worry too much about it.

[ranocha]

Okay, let's say I prepare a PR to SymPy.jl using @latexrecipe. How should this look like?

Trying to mimic the SymEngine plugin by converting a SymPy.Sym to a String or an Expr doesn't work:

julia> using SymPy, Latexify

julia> α = SymPy.symbols("α")
α

julia> @latexrecipe function _(x::Sym)
           string(x)
       end

julia> latexify(- 3/(8*α) + 1/(8*α^2))
ERROR: MethodError: no method matching get(::Char, ::Symbol, ::Symbol)
Closest candidates are:
  get(::PyCall.PyDict{K, V}, ::Any, ::Any) where {K, V} at /home/hendrik/.julia/packages/PyCall/BD546/src/conversions.jl:496
  get(::IOContext, ::Any, ::Any) at show.jl:343
  get(::NamedTuple, ::Union{Integer, Symbol}, ::Any) at namedtuple.jl:309
  ...
Stacktrace:
 [1] process_latexify(args::Sym; kwargs::Base.Pairs{Symbol, Union{}, Tuple{}, NamedTuple{(), Tuple{}}})
   @ Latexify ~/.julia/packages/Latexify/afimT/src/latexify_function.jl:21
 [2] process_latexify(args::Sym)
   @ Latexify ~/.julia/packages/Latexify/afimT/src/latexify_function.jl:18
 [3] latexify(args::Sym; kwargs::Base.Pairs{Symbol, Union{}, Tuple{}, NamedTuple{(), Tuple{}}})
   @ Latexify ~/.julia/packages/Latexify/afimT/src/latexify_function.jl:3
 [4] latexify(args::Sym)
   @ Latexify ~/.julia/packages/Latexify/afimT/src/latexify_function.jl:2
 [5] top-level scope
   @ REPL[4]:1

julia> @latexrecipe function _(x::Sym)
           Meta.parse(string(x))
       end

julia> latexify(- 3/(8*α) + 1/(8*α^2))
ERROR: MethodError: no method matching iterate(::Expr)
Closest candidates are:
  iterate(::Union{LinRange, StepRangeLen}) at range.jl:821
  iterate(::Union{LinRange, StepRangeLen}, ::Integer) at range.jl:821
  iterate(::T) where T<:Union{Base.KeySet{<:Any, <:Dict}, Base.ValueIterator{<:Dict}} at dict.jl:695
  ...
Stacktrace:
 [1] indexed_iterate(I::Expr, i::Int64)
   @ Base ./tuple.jl:92
 [2] process_latexify(args::Sym; kwargs::Base.Pairs{Symbol, Union{}, Tuple{}, NamedTuple{(), Tuple{}}})
   @ Latexify ~/.julia/packages/Latexify/afimT/src/latexify_function.jl:18
 [3] process_latexify
   @ ~/.julia/packages/Latexify/afimT/src/latexify_function.jl:18 [inlined]
 [4] latexify(args::Sym; kwargs::Base.Pairs{Symbol, Union{}, Tuple{}, NamedTuple{(), Tuple{}}})
   @ Latexify ~/.julia/packages/Latexify/afimT/src/latexify_function.jl:3
 [5] latexify(args::Sym)
   @ Latexify ~/.julia/packages/Latexify/afimT/src/latexify_function.jl:2
 [6] top-level scope
   @ REPL[6]:1

Do I really have to use some internal knowledge about undocumented implementation details of @latexrecipe to fix it using something like

julia> @latexrecipe function _(x::Sym)
           (string(x), ), kwargs
       end

julia> latexify(- 3/(8*α) + 1/(8*α^2))
L"$\frac{-3}{8 \cdot \alpha} + \frac{1}{8 \cdot \alpha^{2}}$"

Or are there other suggestions?

[ranocha]

Nevermind, I fixed this specific problem in #188.

[ranocha]

I converted this PR into a draft. As requested above, I converted the compatibility code for Latexify and SymPy into a PR to SymPy.jl. Additionally, I created a new PR #187 with a bug fix I developed while working on this PR. When these two PRs get merged, this PR can be closed.

[ranocha]

Closed in favor of #187 and JuliaPy/SymPy.jl#439