adjust some python tests to work with PyPy

Part of #467.

* tests/python/bdddict.py, tests/python/ltl2tgba.py,
tests/python/ltlparse.py, tests/python/ltlsimple.py,
tests/python/sccinfo.py, tests/python/simstate.py,
tests/python/split.py, tests/python/tra2tba.py: Adjust to
deal with a non-refcounted Python implementation.

tests/python/bdddict.py

@@ -1,5 +1,5 @@
 # -*- mode: python; coding: utf-8 -*-
-# Copyright (C) 2019 Laboratoire de Recherche et Développement de l'Epita
+# Copyright (C) 2019, 2021 Laboratoire de Recherche et Développement de l'Epita
 # (LRDE).
 #
 # This file is part of Spot, a model checking library.
@@ -20,6 +20,17 @@
 # Make sure we can leep track of BDD association in Python using bdd_dict, as
 # discussed in issue #372.
 
+# CPython use reference counting, so that automata are destructed
+# when we expect them to be.   However other implementations like
+# PyPy may call destructors latter, causing different output.
+from platform import python_implementation
+if python_implementation() == 'CPython':
+    def gcollect():
+        pass
+else:
+    import gc
+    def gcollect():
+        gc.collect()
 
 import spot
 
@@ -88,18 +99,22 @@ debug("h2")
 check_ok()
 
 del aut
+gcollect()
 debug("-aut")
 check_ok()
 
 del word
+gcollect()
 debug("-word")
 check_ok()
 
 del h
+gcollect()
 debug("-h")
 check_ok()
 
 del h2
+gcollect()
 debug("-h2")
 check_nok()
 
@@ -110,9 +125,11 @@ var = bdict.register_anonymous_variables(1, h3)
 debug("h3")
 assert var == 2
 del h2
+gcollect()
 debug("-h2")
 check_ok()
 del h3
+gcollect()
 debug("-h3")
 check_nok()
 
@@ -121,4 +138,5 @@ debug("h2")
 bdict.unregister_variable(bdict.varnum("b"), h2)
 bdict.unregister_variable(bdict.varnum("a"), h2)
 debug("-b,-a")
+gcollect()
 check_nok()

tests/python/ltl2tgba.py

@@ -1,6 +1,6 @@
 # -*- mode: python; coding: utf-8 -*-
-# Copyright (C) 2009, 2010, 2012, 2014, 2015, 2016 Laboratoire de Recherche et
+# Copyright (C) 2009, 2010, 2012, 2014-2016, 2021 Laboratoire de Recherche
-# Développement de l'Epita (LRDE).
+# et Développement de l'Epita (LRDE).
 # Copyright (C) 2003, 2004 Laboratoire d'Informatique de Paris 6 (LIP6),
 # département Systèmes Répartis Coopératifs (SRC), Université Pierre
 # et Marie Curie.
@@ -28,7 +28,6 @@ import sys
 import getopt
 import spot
 
-
 def usage(prog):
     sys.stderr.write("""Usage: %s [OPTIONS...] formula
 
@@ -131,4 +130,11 @@ else:
 
 del pf
 del dict
-assert spot.fnode_instances_check()
+
+
+# In CPython, reference counting will cause the above del()
+# to actually destruct the above formulas.  However that is
+# not necessary in other implementations.
+from platform import python_implementation
+if python_implementation() == 'CPython':
+    assert spot.fnode_instances_check()

tests/python/ltlparse.py

@@ -1,5 +1,5 @@
 # -*- mode: python; coding: utf-8 -*-
-# Copyright (C) 2009-2012, 2014-2017, 2019 Laboratoire de Recherche et
+# Copyright (C) 2009-2012, 2014-2017, 2019, 2021 Laboratoire de Recherche et
 # Développement de l'Epita (LRDE).
 # Copyright (C) 2003, 2004 Laboratoire d'Informatique de Paris 6 (LIP6),
 # département Systèmes Répartis Coopératifs (SRC), Université Pierre
@@ -199,6 +199,12 @@ for (x, msg) in [('a->', "missing right operand for \"implication operator\""),
     assert msg in err
     del f9
 
+# force GC before fnode_instances_check(), unless it's CPython
+from platform import python_implementation
+if python_implementation() != 'CPython':
+    import gc
+    gc.collect()
+
 assert spot.fnode_instances_check()
 
 f = spot.formula_F(2, 4, spot.formula_ap("a"))

tests/python/ltlsimple.py

@@ -1,6 +1,6 @@
 # -*- mode: python; coding: utf-8 -*-
-# Copyright (C) 2009, 2010, 2012, 2015, 2018 Laboratoire de Recherche et
+# Copyright (C) 2009, 2010, 2012, 2015, 2018, 2021 Laboratoire de
-# Développement de l'Epita (LRDE).
+# Recherche et Développement de l'Epita (LRDE).
 # Copyright (C) 2003, 2004 Laboratoire d'Informatique de Paris 6 (LIP6),
 # département Systemes Répartis Coopératifs (SRC), Université Pierre
 # et Marie Curie.
@@ -23,6 +23,12 @@
 import spot
 import sys
 
+# Some of the tests here assume timely destructor calls, as they occur
+# in the the reference-counted CPython implementation.  Other
+# implementation such as PyPy, should skip those tests.
+from platform import python_implementation
+is_cpython = python_implementation() == 'CPython'
+
 # ----------------------------------------------------------------------
 a = spot.formula.ap('a')
 b = spot.formula.ap('b')
@@ -71,7 +77,8 @@ f5 = spot.formula.Xor(F, c)
 
 del a, b, c, T, F, f1, f2, f4, f5
 
-assert spot.fnode_instances_check()
+if is_cpython:
+    assert spot.fnode_instances_check()
 
 # ----------------------------------------------------------------------
 assert str([str(x) for x in spot.formula('a &b & c')]) == "['a', 'b', 'c']"

tests/python/sccinfo.py

@@ -59,7 +59,7 @@ l = sorted(list(l0) + list(l1) + list(l2) + list(l3))
 assert l == [0, 1, 2, 3, 4]
 
 i = si.initial()
-todo = {i}
+todo = [i]
 seen = {i}
 trans = []
 transi = []
@@ -72,13 +72,13 @@ while todo:
     for s in si.succ(e):
         if s not in seen:
             seen.add(s)
-            todo.add(s)
+            todo.append(s)
 assert seen == {0, 1, 2, 3}
 assert trans == [(0, 0), (0, 1), (0, 2), (0, 3),
                  (2, 0), (2, 1), (2, 2), (2, 4),
-                 (1, 1), (4, 1), (4, 4), (3, 3)]
+                 (3, 3), (4, 1), (4, 4), (1, 1)]
 assert transi == [(0, 0, 1), (0, 2, 3), (2, 0, 6),
-                  (2, 2, 8), (1, 1, 5), (4, 4, 12), (3, 3, 10)]
+                  (2, 2, 8), (3, 3, 10), (4, 4, 12), (1, 1, 5)]
 
 assert not spot.is_weak_automaton(a, si)
 

tests/python/simstate.py

@@ -20,6 +20,18 @@
 import spot
 from sys import exit
 
+# CPython use reference counting, so that automata are destructed
+# when we expect them to be.   However other implementations like
+# PyPy may call destructors latter, causing different output.
+from platform import python_implementation
+if python_implementation() == 'CPython':
+    def gcollect():
+        pass
+else:
+    import gc
+    def gcollect():
+        gc.collect()
+
 aut = spot.automaton("""
 HOA: v1
 States: 2
@@ -63,6 +75,10 @@ State: 0 "[0,1]" {0}
 [t] 0
 --END--"""
 
+del aut
+del aut2
+gcollect()
+
 aut = spot.translate('GF((p0 -> Gp0) R p1)')
 
 daut = spot.tgba_determinize(aut, True)
@@ -92,6 +108,10 @@ State: 2 "{₀[0]₀}{₁[1]₁}"
 [0&1] 1 {2}
 --END--"""
 
+del aut
+del daut
+gcollect()
+
 aut = spot.automaton("""
 HOA: v1
 States: 2
@@ -175,6 +195,7 @@ assert b.num_states() == 3
 b.set_init_state(1)
 b.purge_unreachable_states()
 b.copy_state_names_from(a)
+
 assert b.to_str() == """HOA: v1
 States: 1
 Start: 0

tests/python/split.py

@@ -1,5 +1,5 @@
 # -*- mode: python; coding: utf-8 -*-
-# Copyright (C) 2018-2020 Laboratoire de Recherche et
+# Copyright (C) 2018-2021 Laboratoire de Recherche et
 # Développement de l'Epita
 #
 # This file is part of Spot, a model checking library.
@@ -19,6 +19,19 @@
 
 import spot
 
+# CPython use reference counting, so that automata are destructed
+# when we expect them to be.   However other implementations like
+# PyPy may call destructors latter, causing different output.
+from platform import python_implementation
+if python_implementation() == 'CPython':
+    def gcollect():
+        pass
+else:
+    import gc
+    def gcollect():
+        gc.collect()
+
+
 
 def incl(a, b):
     return not b.intersects(spot.dualize(spot.tgba_determinize(a)))
@@ -50,6 +63,10 @@ def str_diff(expect, obtained):
 aut, s = do_split('(FG !a) <-> (GF b)', ['a'], ['b'])
 assert equiv(aut, spot.unsplit_2step(s))
 
+del aut
+del s
+gcollect()
+
 aut, s = do_split('GFa && GFb', ['a'], ['b'])
 assert equiv(aut, spot.unsplit_2step(s))
 assert str_diff("""HOA: v1
@@ -73,6 +90,10 @@ State: 2
 [1] 0 {1}
 --END--""", s.to_str() )
 
+del aut
+del s
+gcollect()
+
 aut, s = do_split('! ((G (req -> (F ack))) && (G (go -> (F grant))))',
                   ['go', 'req'], ['ack'])
 assert equiv(aut, spot.unsplit_2step(s))
@@ -113,6 +134,10 @@ assert equiv(aut, spot.unsplit_2step(s))
 # [!3] 2
 # --END--"""
 
+del aut
+del s
+gcollect()
+
 aut, s = do_split('((G (((! g_0) || (! g_1)) && ((r_0 && (X r_1)) -> (F (g_0 \
     && g_1))))) && (G (r_0 -> F g_0))) && (G (r_1 -> F g_1))',
                   ['r_0', 'r_1'], ['g_0', 'g_1'])

tests/python/tra2tba.py

@@ -1,6 +1,6 @@
 # -*- mode: python; coding: utf-8 -*-
-# Copyright (C) 2016-2018, 2020  Laboratoire de Recherche et Développement
+# Copyright (C) 2016-2018, 2020-2021  Laboratoire de Recherche
-# de l'Epita
+# et Développement de l'Epita
 #
 # This file is part of Spot, a model checking library.
 #
@@ -19,6 +19,12 @@
 
 import spot
 
+# CPython use reference counting, so that automata are destructed
+# when we expect them to be.   However other implementations like
+# PyPy may call destructors latter, causing different output.
+from platform import python_implementation
+is_cpython = python_implementation() == 'CPython'
+
 # Test 1.
 aut = spot.automaton("""
 HOA: v1
@@ -365,7 +371,10 @@ State: 7
 --END--"""
 
 res = spot.remove_fin(aut)
-assert(res.to_str('hoa') == exp)
+if is_cpython:
+    assert(res.to_str('hoa') == exp)
+else:
+    assert(res.equivalent_to(spot.automaton(exp)))
 
 # Test 9.
 aut = spot.automaton("""
@@ -401,7 +410,10 @@ State: 1
 --END--"""
 
 res = spot.remove_fin(aut)
-assert(res.to_str('hoa') == exp)
+if is_cpython:
+    assert(res.to_str('hoa') == exp)
+else:
+    assert(res.equivalent_to(spot.automaton(exp)))
 
 # Test 10.
 aut = spot.automaton("""
@@ -440,7 +452,10 @@ State: 2 {0}
 --END--"""
 
 res = spot.remove_fin(aut)
-assert(res.to_str('hoa') == exp)
+if is_cpython:
+    assert(res.to_str('hoa') == exp)
+else:
+    assert(res.equivalent_to(spot.automaton(exp)))
 
 # Test 11.
 aut = spot.automaton("""
@@ -477,7 +492,10 @@ State: 1
 --END--"""
 
 res = spot.remove_fin(aut)
-assert(res.to_str('hoa') == exp)
+if is_cpython:
+    assert(res.to_str('hoa') == exp)
+else:
+    assert(res.equivalent_to(spot.automaton(exp)))
 
 # Different order for rabin_to_buchi_if_realizable() due to merge_edges() not
 # being called.  This is on purpose: the edge order should match exactly the
@@ -499,7 +517,10 @@ State: 1
 [!0] 0 {0}
 --END--"""
 res = spot.rabin_to_buchi_if_realizable(aut)
-assert(res.to_str('hoa') == exp2)
+if is_cpython:
+    assert(res.to_str('hoa') == exp2)
+else:
+    assert(res.equivalent_to(spot.automaton(exp2)))
 
 # Test 12.
 aut = spot.automaton("""
@@ -543,7 +564,10 @@ State: 3 {0}
 --END--"""
 
 res = spot.remove_fin(aut)
-assert(res.to_str('hoa') == exp)
+if is_cpython:
+    assert(res.to_str('hoa') == exp)
+else:
+    assert(res.equivalent_to(spot.automaton(exp)))
 
 # Test 13.
 aut = spot.automaton("""
@@ -590,7 +614,10 @@ State: 1
 --END--"""
 
 res = spot.remove_fin(aut)
-assert(res.to_str('hoa') == exp)
+if is_cpython:
+    assert(res.to_str('hoa') == exp)
+else:
+    assert(res.equivalent_to(spot.automaton(exp)))
 
 # rabin_to_buchi_if_realizable() does not call merge_edges() on purpose: the
 # edge order should match exactly the original automaton.
@@ -616,7 +643,10 @@ State: 1
 --END--"""
 
 res = spot.rabin_to_buchi_if_realizable(aut)
-assert(res.to_str('hoa') == exp2)
+if is_cpython:
+    assert(res.to_str('hoa') == exp2)
+else:
+    assert(res.equivalent_to(spot.automaton(exp2)))
 
 # Test 14.
 aut = spot.automaton("""
@@ -650,5 +680,8 @@ State: 1
 --END--"""
 
 res = spot.remove_fin(aut)
-assert(res.to_str('hoa') == exp)
+if is_cpython:
+    assert(res.to_str('hoa') == exp)
+else:
+    assert(res.equivalent_to(spot.automaton(exp)))
 assert spot.rabin_to_buchi_if_realizable(aut) is None