38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239 | @dataclass
class SnakeLogSolver(Solver):
"""
A solver that uses snakelog.
Snakelog is a lightweight Datalog engine that uses SQLite as a backend,
for more details see [this blog post](https://www.philipzucker.com/snakelog-post/).
While Snakelog is only supports a limited subset of Datalog, it has the advantage of being
**fast** and requiring no additional dependencies. It is well suited for simple traversal-style
logic programming problems, such as the one below.
>>> from typedlogic.integrations.frameworks.pydantic import FactBaseModel
>>> from typedlogic import Implies, And, Variable
>>> class AncestorOf(FactBaseModel):
... ancestor: str
... descendant: str
>>> class ParentOf(FactBaseModel):
... parent: str
... child: str
>>> solver = SnakeLogSolver(strict=True)
>>> from typedlogic import SentenceGroup, PredicateDefinition
>>> solver.add_predicate_definition(PredicateDefinition(predicate="AncestorOf", arguments={'ancestor': str, 'descendant': str}))
>>> solver.add_predicate_definition(PredicateDefinition(predicate="ParentOf", arguments={'parent': str, 'child': str}))
>>> solver.add_fact(ParentOf(parent='p1', child='p1a'))
>>> solver.add_fact(ParentOf(parent='p1a', child='p1aa'))
>>> X = Variable("X")
>>> Y = Variable("Y")
>>> Z = Variable("Z")
>>> solver.add_sentence(Implies(Term(ParentOf.__name__, X, Y), Term(AncestorOf.__name__, X, Y)))
>>> solver.add_sentence(Implies(And(Term(AncestorOf.__name__, X, Z),
... Term(AncestorOf.__name__, Z, Y)),
... Term(AncestorOf.__name__, X, Y)))
>>> model = solver.model()
>>> facts = [str(f) for f in model.ground_terms]
>>> for f in sorted(facts):
... print(f)
AncestorOf(p1, p1a)
AncestorOf(p1, p1aa)
AncestorOf(p1a, p1aa)
ParentOf(p1, p1a)
ParentOf(p1a, p1aa)
This solver does not implement the open-world assumption.
>>> from typedlogic.profiles import OpenWorld
>>> solver.profile.impl(OpenWorld)
False
"""
_wrapped_solver: Optional[snakelog.BaseSolver] = None
predicate_map: Optional[Dict[str, PredicateDefinition]] = None
sentences: List[Sentence] = field(default_factory=list)
profile: ClassVar[Profile] = MixedProfile(ClassicDatalog(), UnsortedLogic(), ExcludedProfile(PropositionalLogic()))
methods_supported: ClassVar[List[Method]] = [
Method(name="litelog", impl_class=litelog.Solver, is_default=True),
Method(name="souffle", impl_class=SouffleSolver),
]
@property
def wrapped_solver(self) -> snakelog.BaseSolver:
if self._wrapped_solver is None:
impl_class = self.method.impl_class
if impl_class is None:
raise ValueError("No implementation class defined")
self._wrapped_solver = impl_class()
return self._wrapped_solver
def check(self) -> Solution:
return Solution(satisfiable=None)
def models(self) -> Iterator[Model]:
s = self.wrapped_solver
s.run()
facts = []
if not self.predicate_map:
raise ValueError("Predicates have not been defined")
for p, pd in self.predicate_map.items():
tbl = self.to_predicate(p)
try:
res = s.con.execute(f"SELECT * FROM {tbl}")
for fact in res.fetchall():
bindings = dict(zip(pd.arguments.keys(), fact[0:], strict=False))
fact = Term(p, bindings)
facts.append(fact)
except sqlite3.OperationalError:
# TODO: better way to detect zero implications
pass
m = Model(source_object=s, ground_terms=facts)
yield m
def prove(self, sentence: Sentence) -> Optional[bool]:
return super().prove(sentence)
def add_fact(self, fact: FactMixin) -> None:
p = self.to_predicate(fact_predicate(fact))
atom = litelog.Atom(p, list(fact_arg_values(fact)))
self.wrapped_solver.add(atom)
self.sentences.append(fact)
def add_sentence(self, sentence: Sentence) -> None:
try:
for sentence in to_horn_rules(sentence):
for snakelog_expr in self.to_clauses(sentence):
self.wrapped_solver.add(snakelog_expr)
self.sentences.append(sentence)
except NotInProfileError as e:
logger.info(f"SKIPPING: {sentence} // {e}")
if self.strict:
raise e
def _string_type(self) -> str:
# TODO: remove after the following is fixed
# https://github.com/philzook58/snakelog/issues/4
if self.method_name == "souffle":
return "symbol"
return "TEXT"
def add_predicate_definition(self, predicate_definition: PredicateDefinition) -> None:
if not self.predicate_map:
self.predicate_map = {}
s = self.wrapped_solver
string_type = self._string_type()
arg_types = [string_type for _ in predicate_definition.arguments.keys()]
sig = [self.to_predicate(predicate_definition.predicate)] + arg_types
s.Relation(*sig)
self.predicate_map[predicate_definition.predicate] = predicate_definition
def to_predicate(self, predicate: str) -> str:
return predicate.lower()
def to_clauses(self, sentence: Sentence) -> List[Union[litelog.Clause, litelog.Atom]]:
if isinstance(sentence, tlog.Forall):
return self.to_clauses(sentence.sentence)
if isinstance(sentence, tlog.Implied):
return self.to_clauses(tlog.Implies(sentence.operands[1], sentence.operands[0]))
if isinstance(sentence, tlog.Iff):
return self.to_clauses(
tlog.And(tlog.Implies(sentence.left, sentence.right), tlog.Implies(sentence.right, sentence.left))
)
if isinstance(sentence, tlog.And):
sentences = []
for s in sentence.operands:
sentences.extend(self.to_clauses(s))
return sentences
return [self.to_clause(sentence)]
def to_clause(self, sentence: Sentence) -> Union[litelog.Clause, litelog.Atom]:
if isinstance(sentence, tlog.Forall):
return self.to_clause(sentence.sentence)
if isinstance(sentence, tlog.Implies):
head = self.to_atom(sentence.consequent)
body = self.to_body(sentence.antecedent)
return litelog.Clause(head, body)
if isinstance(sentence, tlog.Term):
# unit clause
return self.to_atom(sentence)
raise NotInProfileError(f"Unknown clause type {type(sentence)} :: {sentence}")
def to_atom(self, sentence: Sentence) -> litelog.Atom:
if isinstance(sentence, tlog.Term):
def _render_arg(arg):
if arg is None:
return None
if isinstance(arg, tlog.Variable):
# TODO: this should be the norm after normalization
arg = arg.name
return Var(arg.upper())
else:
if isinstance(arg, PRIMITIVE_TYPES):
return arg
else:
return str(arg)
return litelog.Atom(
self.to_predicate(sentence.predicate), [_render_arg(a) for a in sentence.bindings.values()]
)
if isinstance(sentence, typedlogic.pybridge.FactMixin):
def _render_arg(arg):
if arg is None:
return None
return Var(arg.upper())
p = self.to_predicate(fact_predicate(sentence))
return litelog.Atom(p, [_render_arg(a) for a in fact_arg_values(sentence)])
raise NotInProfileError(f"Unknown atom type {type(sentence)} :: {sentence}")
def to_body(self, sentence: Sentence) -> litelog.Body:
if isinstance(sentence, tlog.And):
atoms = [self.to_atom(s) for s in sentence.operands]
return litelog.Body(atoms)
if isinstance(sentence, (tlog.Term, typedlogic.pybridge.FactMixin)):
return litelog.Body([self.to_atom(sentence)])
raise NotInProfileError(f"Unknown body type {type(sentence)} :: {sentence}")
def dump(self) -> str:
return str(self.wrapped_solver)
|