ToSympyCode drops a lambda's body and throws on a set builder

[Rafael-SOWNet]

Two things MathS.ToSympyCode does with a binder, found while checking that the new forms in #976 export. Both are on master at 34f72d11 and neither is new — they reproduce with ordinary names.

1. A lambda exports with no body

MathS.ToSympyCode("lambda(x, x + 1)".ToEntity())

import sympy

x = sympy.Symbol('x')

expr = sympy.Lambda(x, )

sympy.Lambda(x, ) is a SyntaxError, so the generated file does not run. The body is simply not emitted.

2. A set builder throws

MathS.ToSympyCode("{ x : x > 0 }".ToEntity())

AngouriBugException: The given domain is not presented in those possible
 please report about it to the official repository

Reporting it, as instructed. A ConditionalSet reaches the exporter's domain switch with something the switch does not list; the natural export is sympy.ConditionalSet(x, x > 0, S.Complexes) or a Set comprehension.

Both are the export layer only — the entities themselves are fine, print correctly, and round-trip through the parser.

[Rafael-SOWNet]

Fixed in #1001 — and it was wider than these two. Both defects here are real and are fixed, but measuring the exporter by running what it emits rather than reading it turned up two more, and they affect every set export rather than the binders specifically.

What actually fails, on master at 5211ccd6

expr = FiniteSet(1, 2)                        # NameError: name 'FiniteSet' is not defined
expr = Interval(0, 1, ...)                    # NameError: name 'Interval' is not defined
expr = Union(FiniteSet(1, 2), FiniteSet(3))   # NameError: name 'Union' is not defined
expr = x in S.Reals                           # NameError: name 'S' is not defined
expr = sympy.Lambda(x, )                      # TypeError  (§1 above)

The preamble is import sympy and nothing else, so FiniteSet, Interval, Union, Intersection, Complement, ConditionSet and S were all unqualified and all NameError. The lambda and the set builder I filed are two symptoms of a file in which most exports did not run.

And a third fault, which is the one I would not have found by reading: Interval, Piecewise and the non-vector Matrix interpolated their children rather than exporting them — {Left}, not {Left.ToSymPy()}. A bare variable spells the same in Python as here, so it is invisible until the part is a function:

[sin(a); b]   was: Interval(sin(a), b, ...)     # Python has no sin

Plus one shape change: x in RR was x in S.Reals, and Python's in coerces its result to a bool, so even qualified it raises TypeError: did not evaluate to a bool: (-oo < x) & (x < oo). sympy.S.Reals.contains(x) answers with the condition, and still answers True/False when it can.

On §2's suggestion

You wrote the natural export is sympy.ConditionalSet(x, x > 0, S.Complexes). The class is ConditionSet, and the third argument is now sympy.S.UniversalSet rather than S.Complexes — Domain.Any is not the complex numbers, and S.UniversalSet is precisely the one SymPy prints ConditionSet without a third argument for, which is what "no restriction beyond the predicate" means on both sides.

Why nothing caught it

The tests assert substring containment. Assert.Contains("Piecewise((a, b), (c, d))") passes whether the parts were exported or interpolated, and passes on a program that does not run at all. Two existing cases were recording the broken output as correct. #1001 pins the whole emitted expression instead, and adds one case asserting every SymPy name carries its qualifier.

Measured by execution, with work/sympycheck extended from 24 cases to 45 to reach sets, binders, piecewise and matrices: 24 of 45 run on master, 43 with #1001, 45 with #1001 and #1000 (the last two are set builders whose preamble still declares #989's %1).