You don’t need the Symbolics.jl package at all for this. Once you have Julia code as a string from SymPy, you can just parse it and evaluate it into a function.
Unfortunately, it looks like sympy.julia_code produces invalid Julia code, because it thinks it is a good idea to produce a “vectorized” expression but it uses invalid syntax (it generates strings like q2.^5.*q3, where the second . is ambiguous). (I guess, coming from Python, they mistakenly think that it’s a good idea to “vectorize” everything, rather than putting it into a scalar function and vectorizing at call-time if needed.) Fortunately, this is easily fixed by just replacing the dotted operators with non-dotted versions:
p1_1_int_julia_fixed = replace(p1_1_int_julia, r"\.([/*+^])" => s"\1")
which you can then parse and put into a function, e.g.
@eval myfunc(q1, q2, q3, xi, r, L0) = $(Meta.parse(p1_1_int_julia_fixed))