annotate service/mqtt_to_rdf/inference.py @ 1615:bcfa368e5498

change a Graph.__sub__ to Set.difference in verify() for a big speedup
author drewp@bigasterisk.com
date Mon, 06 Sep 2021 23:20:23 -0700
parents 97b2c3cfdb83
children 3a6ed545357f
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
1587
9a3a18c494f9 WIP new inferencer. no vars yet.
drewp@bigasterisk.com
parents:
diff changeset
1 """
9a3a18c494f9 WIP new inferencer. no vars yet.
drewp@bigasterisk.com
parents:
diff changeset
2 copied from reasoning 2021-08-29. probably same api. should
9a3a18c494f9 WIP new inferencer. no vars yet.
drewp@bigasterisk.com
parents:
diff changeset
3 be able to lib/ this out
9a3a18c494f9 WIP new inferencer. no vars yet.
drewp@bigasterisk.com
parents:
diff changeset
4 """
1588
0757fafbfdab WIP inferencer - partial var and function support
drewp@bigasterisk.com
parents: 1587
diff changeset
5 import itertools
1587
9a3a18c494f9 WIP new inferencer. no vars yet.
drewp@bigasterisk.com
parents:
diff changeset
6 import logging
1601
30463df12d89 infer() dumps stats
drewp@bigasterisk.com
parents: 1600
diff changeset
7 import time
1594
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
8 from collections import defaultdict
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
9 from dataclasses import dataclass, field
1607
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
10 from typing import Dict, Iterator, List, Set, Tuple, Union, cast
1588
0757fafbfdab WIP inferencer - partial var and function support
drewp@bigasterisk.com
parents: 1587
diff changeset
11
1587
9a3a18c494f9 WIP new inferencer. no vars yet.
drewp@bigasterisk.com
parents:
diff changeset
12 from prometheus_client import Summary
1607
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
13 from rdflib import BNode, Graph, Namespace, URIRef
1589
5c1055be3c36 WIP more debugging, working towards bnode-matching support
drewp@bigasterisk.com
parents: 1588
diff changeset
14 from rdflib.graph import ConjunctiveGraph, ReadOnlyGraphAggregate
1587
9a3a18c494f9 WIP new inferencer. no vars yet.
drewp@bigasterisk.com
parents:
diff changeset
15 from rdflib.term import Node, Variable
9a3a18c494f9 WIP new inferencer. no vars yet.
drewp@bigasterisk.com
parents:
diff changeset
16
1607
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
17 from candidate_binding import CandidateBinding
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
18 from inference_types import (BindableTerm, EvaluationFailed, ReadOnlyWorkingSet, Triple)
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
19 from lhs_evaluation import Evaluation
1605
449746d1598f WIP move evaluation to new file
drewp@bigasterisk.com
parents: 1603
diff changeset
20
1587
9a3a18c494f9 WIP new inferencer. no vars yet.
drewp@bigasterisk.com
parents:
diff changeset
21 log = logging.getLogger('infer')
1593
b0df43d5494c big rewrite- more classes, smaller methods, more typesafe, all current tests passing
drewp@bigasterisk.com
parents: 1592
diff changeset
22 INDENT = ' '
1587
9a3a18c494f9 WIP new inferencer. no vars yet.
drewp@bigasterisk.com
parents:
diff changeset
23
1601
30463df12d89 infer() dumps stats
drewp@bigasterisk.com
parents: 1600
diff changeset
24 INFER_CALLS = Summary('read_rules_calls', 'calls')
1587
9a3a18c494f9 WIP new inferencer. no vars yet.
drewp@bigasterisk.com
parents:
diff changeset
25
9a3a18c494f9 WIP new inferencer. no vars yet.
drewp@bigasterisk.com
parents:
diff changeset
26 ROOM = Namespace("http://projects.bigasterisk.com/room/")
9a3a18c494f9 WIP new inferencer. no vars yet.
drewp@bigasterisk.com
parents:
diff changeset
27 LOG = Namespace('http://www.w3.org/2000/10/swap/log#')
9a3a18c494f9 WIP new inferencer. no vars yet.
drewp@bigasterisk.com
parents:
diff changeset
28 MATH = Namespace('http://www.w3.org/2000/10/swap/math#')
9a3a18c494f9 WIP new inferencer. no vars yet.
drewp@bigasterisk.com
parents:
diff changeset
29
9a3a18c494f9 WIP new inferencer. no vars yet.
drewp@bigasterisk.com
parents:
diff changeset
30
1594
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
31 @dataclass
1593
b0df43d5494c big rewrite- more classes, smaller methods, more typesafe, all current tests passing
drewp@bigasterisk.com
parents: 1592
diff changeset
32 class Lhs:
1594
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
33 graph: Graph
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
34
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
35 def __post_init__(self):
1608
f928eb06a4f6 cleaning up inner loop
drewp@bigasterisk.com
parents: 1607
diff changeset
36 # do precomputation in here that's not specific to the workingSet
f928eb06a4f6 cleaning up inner loop
drewp@bigasterisk.com
parents: 1607
diff changeset
37 self.staticRuleStmts = Graph()
1611
a794a150a89b a little inner-loop performace. same number of iterations, but less time on logging and some stmt filtering
drewp@bigasterisk.com
parents: 1610
diff changeset
38 self.nonStaticRuleStmts = Graph()
a794a150a89b a little inner-loop performace. same number of iterations, but less time on logging and some stmt filtering
drewp@bigasterisk.com
parents: 1610
diff changeset
39
1608
f928eb06a4f6 cleaning up inner loop
drewp@bigasterisk.com
parents: 1607
diff changeset
40 self.lhsBindables: Set[BindableTerm] = set()
f928eb06a4f6 cleaning up inner loop
drewp@bigasterisk.com
parents: 1607
diff changeset
41 self.lhsBnodes: Set[BNode] = set()
1594
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
42 for ruleStmt in self.graph:
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
43 varsAndBnodesInStmt = [term for term in ruleStmt if isinstance(term, (Variable, BNode))]
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
44 self.lhsBindables.update(varsAndBnodesInStmt)
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
45 self.lhsBnodes.update(x for x in varsAndBnodesInStmt if isinstance(x, BNode))
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
46 if not varsAndBnodesInStmt:
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
47 self.staticRuleStmts.add(ruleStmt)
1611
a794a150a89b a little inner-loop performace. same number of iterations, but less time on logging and some stmt filtering
drewp@bigasterisk.com
parents: 1610
diff changeset
48 else:
a794a150a89b a little inner-loop performace. same number of iterations, but less time on logging and some stmt filtering
drewp@bigasterisk.com
parents: 1610
diff changeset
49 self.nonStaticRuleStmts.add(ruleStmt)
1593
b0df43d5494c big rewrite- more classes, smaller methods, more typesafe, all current tests passing
drewp@bigasterisk.com
parents: 1592
diff changeset
50
1615
bcfa368e5498 change a Graph.__sub__ to Set.difference in verify() for a big speedup
drewp@bigasterisk.com
parents: 1614
diff changeset
51 self.nonStaticRuleStmtsSet = set(self.nonStaticRuleStmts)
bcfa368e5498 change a Graph.__sub__ to Set.difference in verify() for a big speedup
drewp@bigasterisk.com
parents: 1614
diff changeset
52
1602
e3c44ac6d3c5 do findEvals once at setRules time
drewp@bigasterisk.com
parents: 1601
diff changeset
53 self.evaluations = list(Evaluation.findEvals(self.graph))
e3c44ac6d3c5 do findEvals once at setRules time
drewp@bigasterisk.com
parents: 1601
diff changeset
54
1609
34f2817320cc new tests for a smaller part of the inner loop
drewp@bigasterisk.com
parents: 1608
diff changeset
55 def __repr__(self):
34f2817320cc new tests for a smaller part of the inner loop
drewp@bigasterisk.com
parents: 1608
diff changeset
56 return f"Lhs({graphDump(self.graph)})"
34f2817320cc new tests for a smaller part of the inner loop
drewp@bigasterisk.com
parents: 1608
diff changeset
57
1607
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
58 def findCandidateBindings(self, workingSet: ReadOnlyWorkingSet, stats) -> Iterator['BoundLhs']:
1593
b0df43d5494c big rewrite- more classes, smaller methods, more typesafe, all current tests passing
drewp@bigasterisk.com
parents: 1592
diff changeset
59 """bindings that fit the LHS of a rule, using statements from workingSet and functions
b0df43d5494c big rewrite- more classes, smaller methods, more typesafe, all current tests passing
drewp@bigasterisk.com
parents: 1592
diff changeset
60 from LHS"""
1597
387a9cb66517 logging adjustments
drewp@bigasterisk.com
parents: 1596
diff changeset
61 log.debug(f'{INDENT*3} nodesToBind: {self.lhsBindables}')
1607
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
62 stats['findCandidateBindingsCalls'] += 1
1593
b0df43d5494c big rewrite- more classes, smaller methods, more typesafe, all current tests passing
drewp@bigasterisk.com
parents: 1592
diff changeset
63
1600
89a50242cb5e cleanup internal names, imports
drewp@bigasterisk.com
parents: 1599
diff changeset
64 if not self._allStaticStatementsMatch(workingSet):
1607
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
65 stats['findCandidateBindingEarlyExits'] += 1
1593
b0df43d5494c big rewrite- more classes, smaller methods, more typesafe, all current tests passing
drewp@bigasterisk.com
parents: 1592
diff changeset
66 return
b0df43d5494c big rewrite- more classes, smaller methods, more typesafe, all current tests passing
drewp@bigasterisk.com
parents: 1592
diff changeset
67
1607
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
68 for binding in self._possibleBindings(workingSet, stats):
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
69 log.debug('')
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
70 log.debug(f'{INDENT*4}*trying {binding.binding}')
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
71
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
72 if not binding.verify(workingSet):
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
73 log.debug(f'{INDENT*4} this binding did not verify')
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
74 stats['permCountFailingVerify'] += 1
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
75 continue
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
76
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
77 stats['permCountSucceeding'] += 1
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
78 yield binding
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
79
1608
f928eb06a4f6 cleaning up inner loop
drewp@bigasterisk.com
parents: 1607
diff changeset
80 def _allStaticStatementsMatch(self, workingSet: ReadOnlyWorkingSet) -> bool:
1613
03ed8c9abd5b forget GRAPH_ID optimization in this case
drewp@bigasterisk.com
parents: 1612
diff changeset
81 # bug: see TestSelfFulfillingRule.test3 for a case where this rule's
03ed8c9abd5b forget GRAPH_ID optimization in this case
drewp@bigasterisk.com
parents: 1612
diff changeset
82 # static stmt is matched by a non-static stmt in the rule itself
1608
f928eb06a4f6 cleaning up inner loop
drewp@bigasterisk.com
parents: 1607
diff changeset
83 for ruleStmt in self.staticRuleStmts:
f928eb06a4f6 cleaning up inner loop
drewp@bigasterisk.com
parents: 1607
diff changeset
84 if ruleStmt not in workingSet:
f928eb06a4f6 cleaning up inner loop
drewp@bigasterisk.com
parents: 1607
diff changeset
85 log.debug(f'{INDENT*3} {ruleStmt} not in working set- skip rule')
f928eb06a4f6 cleaning up inner loop
drewp@bigasterisk.com
parents: 1607
diff changeset
86 return False
f928eb06a4f6 cleaning up inner loop
drewp@bigasterisk.com
parents: 1607
diff changeset
87 return True
f928eb06a4f6 cleaning up inner loop
drewp@bigasterisk.com
parents: 1607
diff changeset
88
1607
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
89 def _possibleBindings(self, workingSet, stats) -> Iterator['BoundLhs']:
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
90 """this yields at least the working bindings, and possibly others"""
1600
89a50242cb5e cleanup internal names, imports
drewp@bigasterisk.com
parents: 1599
diff changeset
91 candidateTermMatches: Dict[BindableTerm, Set[Node]] = self._allCandidateTermMatches(workingSet)
1614
97b2c3cfdb83 refactor: inline an odometer algorithm in place of itertools.product
drewp@bigasterisk.com
parents: 1613
diff changeset
92 for bindRow in self._product(candidateTermMatches):
97b2c3cfdb83 refactor: inline an odometer algorithm in place of itertools.product
drewp@bigasterisk.com
parents: 1613
diff changeset
93 try:
97b2c3cfdb83 refactor: inline an odometer algorithm in place of itertools.product
drewp@bigasterisk.com
parents: 1613
diff changeset
94 yield BoundLhs(self, bindRow)
97b2c3cfdb83 refactor: inline an odometer algorithm in place of itertools.product
drewp@bigasterisk.com
parents: 1613
diff changeset
95 except EvaluationFailed:
97b2c3cfdb83 refactor: inline an odometer algorithm in place of itertools.product
drewp@bigasterisk.com
parents: 1613
diff changeset
96 stats['permCountFailingEval'] += 1
1593
b0df43d5494c big rewrite- more classes, smaller methods, more typesafe, all current tests passing
drewp@bigasterisk.com
parents: 1592
diff changeset
97
1614
97b2c3cfdb83 refactor: inline an odometer algorithm in place of itertools.product
drewp@bigasterisk.com
parents: 1613
diff changeset
98 def _product(self, candidateTermMatches: Dict[BindableTerm, Set[Node]]) -> Iterator[CandidateBinding]:
1600
89a50242cb5e cleanup internal names, imports
drewp@bigasterisk.com
parents: 1599
diff changeset
99 orderedVars, orderedValueSets = _organize(candidateTermMatches)
89a50242cb5e cleanup internal names, imports
drewp@bigasterisk.com
parents: 1599
diff changeset
100 self._logCandidates(orderedVars, orderedValueSets)
1597
387a9cb66517 logging adjustments
drewp@bigasterisk.com
parents: 1596
diff changeset
101 log.debug(f'{INDENT*3} trying all permutations:')
1614
97b2c3cfdb83 refactor: inline an odometer algorithm in place of itertools.product
drewp@bigasterisk.com
parents: 1613
diff changeset
102 if not orderedValueSets:
97b2c3cfdb83 refactor: inline an odometer algorithm in place of itertools.product
drewp@bigasterisk.com
parents: 1613
diff changeset
103 yield CandidateBinding({})
97b2c3cfdb83 refactor: inline an odometer algorithm in place of itertools.product
drewp@bigasterisk.com
parents: 1613
diff changeset
104 return
97b2c3cfdb83 refactor: inline an odometer algorithm in place of itertools.product
drewp@bigasterisk.com
parents: 1613
diff changeset
105 if not orderedValueSets or not all(orderedValueSets):
97b2c3cfdb83 refactor: inline an odometer algorithm in place of itertools.product
drewp@bigasterisk.com
parents: 1613
diff changeset
106 # some var or bnode has no options at all
97b2c3cfdb83 refactor: inline an odometer algorithm in place of itertools.product
drewp@bigasterisk.com
parents: 1613
diff changeset
107 return
97b2c3cfdb83 refactor: inline an odometer algorithm in place of itertools.product
drewp@bigasterisk.com
parents: 1613
diff changeset
108 rings: List[Iterator[Node]] = [itertools.cycle(valSet) for valSet in orderedValueSets]
97b2c3cfdb83 refactor: inline an odometer algorithm in place of itertools.product
drewp@bigasterisk.com
parents: 1613
diff changeset
109 currentSet: List[Node] = [next(ring) for ring in rings]
97b2c3cfdb83 refactor: inline an odometer algorithm in place of itertools.product
drewp@bigasterisk.com
parents: 1613
diff changeset
110 starts = [valSet[-1] for valSet in orderedValueSets]
97b2c3cfdb83 refactor: inline an odometer algorithm in place of itertools.product
drewp@bigasterisk.com
parents: 1613
diff changeset
111 while True:
97b2c3cfdb83 refactor: inline an odometer algorithm in place of itertools.product
drewp@bigasterisk.com
parents: 1613
diff changeset
112 for col, curr in enumerate(currentSet):
97b2c3cfdb83 refactor: inline an odometer algorithm in place of itertools.product
drewp@bigasterisk.com
parents: 1613
diff changeset
113 currentSet[col] = next(rings[col])
97b2c3cfdb83 refactor: inline an odometer algorithm in place of itertools.product
drewp@bigasterisk.com
parents: 1613
diff changeset
114 log.debug(repr(currentSet))
97b2c3cfdb83 refactor: inline an odometer algorithm in place of itertools.product
drewp@bigasterisk.com
parents: 1613
diff changeset
115 yield CandidateBinding(dict(zip(orderedVars, currentSet)))
97b2c3cfdb83 refactor: inline an odometer algorithm in place of itertools.product
drewp@bigasterisk.com
parents: 1613
diff changeset
116 if curr is not starts[col]:
97b2c3cfdb83 refactor: inline an odometer algorithm in place of itertools.product
drewp@bigasterisk.com
parents: 1613
diff changeset
117 break
97b2c3cfdb83 refactor: inline an odometer algorithm in place of itertools.product
drewp@bigasterisk.com
parents: 1613
diff changeset
118 if col == len(orderedValueSets) - 1:
97b2c3cfdb83 refactor: inline an odometer algorithm in place of itertools.product
drewp@bigasterisk.com
parents: 1613
diff changeset
119 return
1592
d7b66234064b pure reordering of funcs to make the next diffs smaller
drewp@bigasterisk.com
parents: 1591
diff changeset
120
1600
89a50242cb5e cleanup internal names, imports
drewp@bigasterisk.com
parents: 1599
diff changeset
121 def _allCandidateTermMatches(self, workingSet: ReadOnlyWorkingSet) -> Dict[BindableTerm, Set[Node]]:
1593
b0df43d5494c big rewrite- more classes, smaller methods, more typesafe, all current tests passing
drewp@bigasterisk.com
parents: 1592
diff changeset
122 """the total set of terms each variable could possibly match"""
1592
d7b66234064b pure reordering of funcs to make the next diffs smaller
drewp@bigasterisk.com
parents: 1591
diff changeset
123
1593
b0df43d5494c big rewrite- more classes, smaller methods, more typesafe, all current tests passing
drewp@bigasterisk.com
parents: 1592
diff changeset
124 candidateTermMatches: Dict[BindableTerm, Set[Node]] = defaultdict(set)
1594
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
125 for lhsStmt in self.graph:
1597
387a9cb66517 logging adjustments
drewp@bigasterisk.com
parents: 1596
diff changeset
126 log.debug(f'{INDENT*4} possibles for this lhs stmt: {lhsStmt}')
1608
f928eb06a4f6 cleaning up inner loop
drewp@bigasterisk.com
parents: 1607
diff changeset
127 for i, trueStmt in enumerate(workingSet):
1597
387a9cb66517 logging adjustments
drewp@bigasterisk.com
parents: 1596
diff changeset
128 # log.debug(f'{INDENT*5} consider this true stmt ({i}): {trueStmt}')
1592
d7b66234064b pure reordering of funcs to make the next diffs smaller
drewp@bigasterisk.com
parents: 1591
diff changeset
129
1594
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
130 for v, vals in self._bindingsFromStatement(lhsStmt, trueStmt):
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
131 candidateTermMatches[v].update(vals)
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
132
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
133 for trueStmt in itertools.chain(workingSet, self.graph):
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
134 for b in self.lhsBnodes:
1593
b0df43d5494c big rewrite- more classes, smaller methods, more typesafe, all current tests passing
drewp@bigasterisk.com
parents: 1592
diff changeset
135 for t in [trueStmt[0], trueStmt[2]]:
b0df43d5494c big rewrite- more classes, smaller methods, more typesafe, all current tests passing
drewp@bigasterisk.com
parents: 1592
diff changeset
136 if isinstance(t, (URIRef, BNode)):
b0df43d5494c big rewrite- more classes, smaller methods, more typesafe, all current tests passing
drewp@bigasterisk.com
parents: 1592
diff changeset
137 candidateTermMatches[b].add(t)
b0df43d5494c big rewrite- more classes, smaller methods, more typesafe, all current tests passing
drewp@bigasterisk.com
parents: 1592
diff changeset
138 return candidateTermMatches
b0df43d5494c big rewrite- more classes, smaller methods, more typesafe, all current tests passing
drewp@bigasterisk.com
parents: 1592
diff changeset
139
1594
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
140 def _bindingsFromStatement(self, stmt1: Triple, stmt2: Triple) -> Iterator[Tuple[Variable, Set[Node]]]:
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
141 """if these stmts match otherwise, what BNode or Variable mappings do we learn?
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
142
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
143 e.g. stmt1=(?x B ?y) and stmt2=(A B C), then we yield (?x, {A}) and (?y, {C})
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
144 or stmt1=(_:x B C) and stmt2=(A B C), then we yield (_:x, {A})
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
145 or stmt1=(?x B C) and stmt2=(A B D), then we yield nothing
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
146 """
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
147 bindingsFromStatement = {}
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
148 for term1, term2 in zip(stmt1, stmt2):
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
149 if isinstance(term1, (BNode, Variable)):
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
150 bindingsFromStatement.setdefault(term1, set()).add(term2)
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
151 elif term1 != term2:
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
152 break
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
153 else:
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
154 for v, vals in bindingsFromStatement.items():
1597
387a9cb66517 logging adjustments
drewp@bigasterisk.com
parents: 1596
diff changeset
155 log.debug(f'{INDENT*5} {v=} {vals=}')
1594
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
156 yield v, vals
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
157
1600
89a50242cb5e cleanup internal names, imports
drewp@bigasterisk.com
parents: 1599
diff changeset
158 def _logCandidates(self, orderedVars, orderedValueSets):
1593
b0df43d5494c big rewrite- more classes, smaller methods, more typesafe, all current tests passing
drewp@bigasterisk.com
parents: 1592
diff changeset
159 if not log.isEnabledFor(logging.DEBUG):
b0df43d5494c big rewrite- more classes, smaller methods, more typesafe, all current tests passing
drewp@bigasterisk.com
parents: 1592
diff changeset
160 return
1597
387a9cb66517 logging adjustments
drewp@bigasterisk.com
parents: 1596
diff changeset
161 log.debug(f'{INDENT*3} resulting candidate terms:')
1593
b0df43d5494c big rewrite- more classes, smaller methods, more typesafe, all current tests passing
drewp@bigasterisk.com
parents: 1592
diff changeset
162 for v, vals in zip(orderedVars, orderedValueSets):
1597
387a9cb66517 logging adjustments
drewp@bigasterisk.com
parents: 1596
diff changeset
163 log.debug(f'{INDENT*4} {v!r} could be:')
1593
b0df43d5494c big rewrite- more classes, smaller methods, more typesafe, all current tests passing
drewp@bigasterisk.com
parents: 1592
diff changeset
164 for val in vals:
1597
387a9cb66517 logging adjustments
drewp@bigasterisk.com
parents: 1596
diff changeset
165 log.debug(f'{INDENT*5}{val!r}')
1592
d7b66234064b pure reordering of funcs to make the next diffs smaller
drewp@bigasterisk.com
parents: 1591
diff changeset
166
d7b66234064b pure reordering of funcs to make the next diffs smaller
drewp@bigasterisk.com
parents: 1591
diff changeset
167
1607
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
168 @dataclass
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
169 class BoundLhs:
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
170 lhs: Lhs
1610
6fc48ef4c696 mysteriously lost an important line
drewp@bigasterisk.com
parents: 1609
diff changeset
171 binding: CandidateBinding
1607
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
172
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
173 def __post_init__(self):
1613
03ed8c9abd5b forget GRAPH_ID optimization in this case
drewp@bigasterisk.com
parents: 1612
diff changeset
174 self.usedByFuncs = Graph()
1607
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
175 self._applyFunctions()
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
176
1608
f928eb06a4f6 cleaning up inner loop
drewp@bigasterisk.com
parents: 1607
diff changeset
177 self.graphWithoutEvals = self.lhs.graph - self.usedByFuncs
1607
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
178
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
179 def _applyFunctions(self):
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
180 """may grow the binding with some results"""
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
181 while True:
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
182 delta = self._applyFunctionsIteration()
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
183 if delta == 0:
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
184 break
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
185
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
186 def _applyFunctionsIteration(self):
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
187 before = len(self.binding.binding)
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
188 delta = 0
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
189 for ev in self.lhs.evaluations:
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
190 newBindings, usedGraph = ev.resultBindings(self.binding)
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
191 self.usedByFuncs += usedGraph
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
192 self.binding.addNewBindings(newBindings)
1608
f928eb06a4f6 cleaning up inner loop
drewp@bigasterisk.com
parents: 1607
diff changeset
193 delta = len(self.binding.binding) - before
f928eb06a4f6 cleaning up inner loop
drewp@bigasterisk.com
parents: 1607
diff changeset
194 log.debug(f'{INDENT*4} eval rules made {delta} new bindings')
1607
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
195 return delta
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
196
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
197 def verify(self, workingSet: ReadOnlyWorkingSet) -> bool:
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
198 """Can this bound lhs be true all at once in workingSet?"""
1615
bcfa368e5498 change a Graph.__sub__ to Set.difference in verify() for a big speedup
drewp@bigasterisk.com
parents: 1614
diff changeset
199 rem = cast(Set[Triple], self.lhs.nonStaticRuleStmtsSet.difference(self.usedByFuncs))
bcfa368e5498 change a Graph.__sub__ to Set.difference in verify() for a big speedup
drewp@bigasterisk.com
parents: 1614
diff changeset
200 boundLhs = self.binding.apply(rem)
1607
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
201
1611
a794a150a89b a little inner-loop performace. same number of iterations, but less time on logging and some stmt filtering
drewp@bigasterisk.com
parents: 1610
diff changeset
202 if log.isEnabledFor(logging.DEBUG):
a794a150a89b a little inner-loop performace. same number of iterations, but less time on logging and some stmt filtering
drewp@bigasterisk.com
parents: 1610
diff changeset
203 boundLhs = list(boundLhs)
a794a150a89b a little inner-loop performace. same number of iterations, but less time on logging and some stmt filtering
drewp@bigasterisk.com
parents: 1610
diff changeset
204 self._logVerifyBanner(boundLhs, workingSet)
1607
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
205
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
206 for stmt in boundLhs:
1611
a794a150a89b a little inner-loop performace. same number of iterations, but less time on logging and some stmt filtering
drewp@bigasterisk.com
parents: 1610
diff changeset
207 log.debug(f'{INDENT*4} check for %s', stmt)
1607
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
208
1611
a794a150a89b a little inner-loop performace. same number of iterations, but less time on logging and some stmt filtering
drewp@bigasterisk.com
parents: 1610
diff changeset
209 if stmt not in workingSet:
1607
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
210 log.debug(f'{INDENT*5} stmt not known to be true')
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
211 return False
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
212 return True
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
213
1611
a794a150a89b a little inner-loop performace. same number of iterations, but less time on logging and some stmt filtering
drewp@bigasterisk.com
parents: 1610
diff changeset
214 def _logVerifyBanner(self, boundLhs, workingSet: ReadOnlyWorkingSet):
1607
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
215 log.debug(f'{INDENT*4}/ verify all bindings against this boundLhs:')
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
216 for stmt in sorted(boundLhs):
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
217 log.debug(f'{INDENT*4}|{INDENT} {stmt}')
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
218
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
219 # log.debug(f'{INDENT*4}| and against this workingSet:')
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
220 # for stmt in sorted(workingSet):
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
221 # log.debug(f'{INDENT*4}|{INDENT} {stmt}')
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
222
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
223 log.debug(f'{INDENT*4}\\')
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
224
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
225
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
226 @dataclass
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
227 class Rule:
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
228 lhsGraph: Graph
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
229 rhsGraph: Graph
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
230
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
231 def __post_init__(self):
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
232 self.lhs = Lhs(self.lhsGraph)
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
233
1612
272f78d4671a mark skipped tests. move applyRule into Rule. minor cleanups.
drewp@bigasterisk.com
parents: 1611
diff changeset
234 def applyRule(self, workingSet: Graph, implied: Graph, stats: Dict):
272f78d4671a mark skipped tests. move applyRule into Rule. minor cleanups.
drewp@bigasterisk.com
parents: 1611
diff changeset
235 for bound in self.lhs.findCandidateBindings(ReadOnlyGraphAggregate([workingSet]), stats):
272f78d4671a mark skipped tests. move applyRule into Rule. minor cleanups.
drewp@bigasterisk.com
parents: 1611
diff changeset
236 log.debug(f'{INDENT*3} rule has a working binding:')
272f78d4671a mark skipped tests. move applyRule into Rule. minor cleanups.
drewp@bigasterisk.com
parents: 1611
diff changeset
237
272f78d4671a mark skipped tests. move applyRule into Rule. minor cleanups.
drewp@bigasterisk.com
parents: 1611
diff changeset
238 for lhsBoundStmt in bound.binding.apply(bound.graphWithoutEvals):
272f78d4671a mark skipped tests. move applyRule into Rule. minor cleanups.
drewp@bigasterisk.com
parents: 1611
diff changeset
239 log.debug(f'{INDENT*5} adding {lhsBoundStmt=}')
272f78d4671a mark skipped tests. move applyRule into Rule. minor cleanups.
drewp@bigasterisk.com
parents: 1611
diff changeset
240 workingSet.add(lhsBoundStmt)
272f78d4671a mark skipped tests. move applyRule into Rule. minor cleanups.
drewp@bigasterisk.com
parents: 1611
diff changeset
241 for newStmt in bound.binding.apply(self.rhsGraph):
272f78d4671a mark skipped tests. move applyRule into Rule. minor cleanups.
drewp@bigasterisk.com
parents: 1611
diff changeset
242 log.debug(f'{INDENT*5} adding {newStmt=}')
272f78d4671a mark skipped tests. move applyRule into Rule. minor cleanups.
drewp@bigasterisk.com
parents: 1611
diff changeset
243 workingSet.add(newStmt)
272f78d4671a mark skipped tests. move applyRule into Rule. minor cleanups.
drewp@bigasterisk.com
parents: 1611
diff changeset
244 implied.add(newStmt)
272f78d4671a mark skipped tests. move applyRule into Rule. minor cleanups.
drewp@bigasterisk.com
parents: 1611
diff changeset
245
1607
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
246
1587
9a3a18c494f9 WIP new inferencer. no vars yet.
drewp@bigasterisk.com
parents:
diff changeset
247 class Inference:
9a3a18c494f9 WIP new inferencer. no vars yet.
drewp@bigasterisk.com
parents:
diff changeset
248
9a3a18c494f9 WIP new inferencer. no vars yet.
drewp@bigasterisk.com
parents:
diff changeset
249 def __init__(self) -> None:
1607
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
250 self.rules = []
1587
9a3a18c494f9 WIP new inferencer. no vars yet.
drewp@bigasterisk.com
parents:
diff changeset
251
9a3a18c494f9 WIP new inferencer. no vars yet.
drewp@bigasterisk.com
parents:
diff changeset
252 def setRules(self, g: ConjunctiveGraph):
1607
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
253 self.rules: List[Rule] = []
1599
abbf0eb0e640 fix a bug with a slightly moer complicated set of rules
drewp@bigasterisk.com
parents: 1598
diff changeset
254 for stmt in g:
abbf0eb0e640 fix a bug with a slightly moer complicated set of rules
drewp@bigasterisk.com
parents: 1598
diff changeset
255 if stmt[1] == LOG['implies']:
1607
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
256 self.rules.append(Rule(stmt[0], stmt[2]))
1612
272f78d4671a mark skipped tests. move applyRule into Rule. minor cleanups.
drewp@bigasterisk.com
parents: 1611
diff changeset
257 # other stmts should go to a default working set?
1587
9a3a18c494f9 WIP new inferencer. no vars yet.
drewp@bigasterisk.com
parents:
diff changeset
258
1601
30463df12d89 infer() dumps stats
drewp@bigasterisk.com
parents: 1600
diff changeset
259 @INFER_CALLS.time()
1587
9a3a18c494f9 WIP new inferencer. no vars yet.
drewp@bigasterisk.com
parents:
diff changeset
260 def infer(self, graph: Graph):
9a3a18c494f9 WIP new inferencer. no vars yet.
drewp@bigasterisk.com
parents:
diff changeset
261 """
9a3a18c494f9 WIP new inferencer. no vars yet.
drewp@bigasterisk.com
parents:
diff changeset
262 returns new graph of inferred statements.
9a3a18c494f9 WIP new inferencer. no vars yet.
drewp@bigasterisk.com
parents:
diff changeset
263 """
1597
387a9cb66517 logging adjustments
drewp@bigasterisk.com
parents: 1596
diff changeset
264 log.info(f'{INDENT*0} Begin inference of graph len={graph.__len__()} with rules len={len(self.rules)}:')
1601
30463df12d89 infer() dumps stats
drewp@bigasterisk.com
parents: 1600
diff changeset
265 startTime = time.time()
1607
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
266 stats: Dict[str, Union[int, float]] = defaultdict(lambda: 0)
1589
5c1055be3c36 WIP more debugging, working towards bnode-matching support
drewp@bigasterisk.com
parents: 1588
diff changeset
267 # everything that is true: the input graph, plus every rule conclusion we can make
1593
b0df43d5494c big rewrite- more classes, smaller methods, more typesafe, all current tests passing
drewp@bigasterisk.com
parents: 1592
diff changeset
268 workingSet = Graph()
b0df43d5494c big rewrite- more classes, smaller methods, more typesafe, all current tests passing
drewp@bigasterisk.com
parents: 1592
diff changeset
269 workingSet += graph
1587
9a3a18c494f9 WIP new inferencer. no vars yet.
drewp@bigasterisk.com
parents:
diff changeset
270
1594
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
271 # just the statements that came from RHS's of rules that fired.
1587
9a3a18c494f9 WIP new inferencer. no vars yet.
drewp@bigasterisk.com
parents:
diff changeset
272 implied = ConjunctiveGraph()
9a3a18c494f9 WIP new inferencer. no vars yet.
drewp@bigasterisk.com
parents:
diff changeset
273
9a3a18c494f9 WIP new inferencer. no vars yet.
drewp@bigasterisk.com
parents:
diff changeset
274 bailout_iterations = 100
9a3a18c494f9 WIP new inferencer. no vars yet.
drewp@bigasterisk.com
parents:
diff changeset
275 delta = 1
1607
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
276 stats['initWorkingSet'] = cast(int, workingSet.__len__())
1587
9a3a18c494f9 WIP new inferencer. no vars yet.
drewp@bigasterisk.com
parents:
diff changeset
277 while delta > 0 and bailout_iterations > 0:
1597
387a9cb66517 logging adjustments
drewp@bigasterisk.com
parents: 1596
diff changeset
278 log.info(f'{INDENT*1}*iteration ({bailout_iterations} left)')
1587
9a3a18c494f9 WIP new inferencer. no vars yet.
drewp@bigasterisk.com
parents:
diff changeset
279 bailout_iterations -= 1
9a3a18c494f9 WIP new inferencer. no vars yet.
drewp@bigasterisk.com
parents:
diff changeset
280 delta = -len(implied)
1607
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
281 self._iterateAllRules(workingSet, implied, stats)
1587
9a3a18c494f9 WIP new inferencer. no vars yet.
drewp@bigasterisk.com
parents:
diff changeset
282 delta += len(implied)
1607
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
283 stats['iterations'] += 1
1597
387a9cb66517 logging adjustments
drewp@bigasterisk.com
parents: 1596
diff changeset
284 log.info(f'{INDENT*2} this inference iteration added {delta} more implied stmts')
1607
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
285 stats['timeSpent'] = round(time.time() - startTime, 3)
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
286 stats['impliedStmts'] = len(implied)
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
287 log.info(f'{INDENT*0} Inference done {dict(stats)}. Implied:')
1587
9a3a18c494f9 WIP new inferencer. no vars yet.
drewp@bigasterisk.com
parents:
diff changeset
288 for st in implied:
1597
387a9cb66517 logging adjustments
drewp@bigasterisk.com
parents: 1596
diff changeset
289 log.info(f'{INDENT*1} {st}')
1587
9a3a18c494f9 WIP new inferencer. no vars yet.
drewp@bigasterisk.com
parents:
diff changeset
290 return implied
9a3a18c494f9 WIP new inferencer. no vars yet.
drewp@bigasterisk.com
parents:
diff changeset
291
1607
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
292 def _iterateAllRules(self, workingSet: Graph, implied: Graph, stats):
1612
272f78d4671a mark skipped tests. move applyRule into Rule. minor cleanups.
drewp@bigasterisk.com
parents: 1611
diff changeset
293 for i, rule in enumerate(self.rules):
272f78d4671a mark skipped tests. move applyRule into Rule. minor cleanups.
drewp@bigasterisk.com
parents: 1611
diff changeset
294 self._logRuleApplicationHeader(workingSet, i, rule)
272f78d4671a mark skipped tests. move applyRule into Rule. minor cleanups.
drewp@bigasterisk.com
parents: 1611
diff changeset
295 rule.applyRule(workingSet, implied, stats)
1587
9a3a18c494f9 WIP new inferencer. no vars yet.
drewp@bigasterisk.com
parents:
diff changeset
296
1607
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
297 def _logRuleApplicationHeader(self, workingSet, i, r: Rule):
1594
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
298 if not log.isEnabledFor(logging.DEBUG):
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
299 return
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
300
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
301 log.debug('')
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
302 log.debug(f'{INDENT*2} workingSet:')
1597
387a9cb66517 logging adjustments
drewp@bigasterisk.com
parents: 1596
diff changeset
303 for j, stmt in enumerate(sorted(workingSet)):
387a9cb66517 logging adjustments
drewp@bigasterisk.com
parents: 1596
diff changeset
304 log.debug(f'{INDENT*3} ({j}) {stmt}')
1594
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
305
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
306 log.debug('')
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
307 log.debug(f'{INDENT*2}-applying rule {i}')
1607
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
308 log.debug(f'{INDENT*3} rule def lhs: {graphDump(r.lhsGraph)}')
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
309 log.debug(f'{INDENT*3} rule def rhs: {graphDump(r.rhsGraph)}')
1594
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
310
1588
0757fafbfdab WIP inferencer - partial var and function support
drewp@bigasterisk.com
parents: 1587
diff changeset
311
1590
327202020892 WIP inference- getting into more degenerate test cases
drewp@bigasterisk.com
parents: 1589
diff changeset
312 def graphDump(g: Union[Graph, List[Triple]]):
327202020892 WIP inference- getting into more degenerate test cases
drewp@bigasterisk.com
parents: 1589
diff changeset
313 if not isinstance(g, Graph):
1607
b21885181e35 more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents: 1605
diff changeset
314 log.warning(f"it's a {type(g)}")
1590
327202020892 WIP inference- getting into more degenerate test cases
drewp@bigasterisk.com
parents: 1589
diff changeset
315 g2 = Graph()
1594
e58bcfa66093 cleanups and a few fixed cases
drewp@bigasterisk.com
parents: 1593
diff changeset
316 g2 += g
1590
327202020892 WIP inference- getting into more degenerate test cases
drewp@bigasterisk.com
parents: 1589
diff changeset
317 g = g2
1589
5c1055be3c36 WIP more debugging, working towards bnode-matching support
drewp@bigasterisk.com
parents: 1588
diff changeset
318 g.bind('', ROOM)
5c1055be3c36 WIP more debugging, working towards bnode-matching support
drewp@bigasterisk.com
parents: 1588
diff changeset
319 g.bind('ex', Namespace('http://example.com/'))
5c1055be3c36 WIP more debugging, working towards bnode-matching support
drewp@bigasterisk.com
parents: 1588
diff changeset
320 lines = cast(bytes, g.serialize(format='n3')).decode('utf8').splitlines()
1612
272f78d4671a mark skipped tests. move applyRule into Rule. minor cleanups.
drewp@bigasterisk.com
parents: 1611
diff changeset
321 lines = [line.strip() for line in lines if not line.startswith('@prefix')]
1589
5c1055be3c36 WIP more debugging, working towards bnode-matching support
drewp@bigasterisk.com
parents: 1588
diff changeset
322 return ' '.join(lines)
5c1055be3c36 WIP more debugging, working towards bnode-matching support
drewp@bigasterisk.com
parents: 1588
diff changeset
323
5c1055be3c36 WIP more debugging, working towards bnode-matching support
drewp@bigasterisk.com
parents: 1588
diff changeset
324
1600
89a50242cb5e cleanup internal names, imports
drewp@bigasterisk.com
parents: 1599
diff changeset
325 def _organize(candidateTermMatches: Dict[BindableTerm, Set[Node]]) -> Tuple[List[BindableTerm], List[List[Node]]]:
1592
d7b66234064b pure reordering of funcs to make the next diffs smaller
drewp@bigasterisk.com
parents: 1591
diff changeset
326 items = list(candidateTermMatches.items())
d7b66234064b pure reordering of funcs to make the next diffs smaller
drewp@bigasterisk.com
parents: 1591
diff changeset
327 items.sort()
d7b66234064b pure reordering of funcs to make the next diffs smaller
drewp@bigasterisk.com
parents: 1591
diff changeset
328 orderedVars: List[BindableTerm] = []
d7b66234064b pure reordering of funcs to make the next diffs smaller
drewp@bigasterisk.com
parents: 1591
diff changeset
329 orderedValueSets: List[List[Node]] = []
d7b66234064b pure reordering of funcs to make the next diffs smaller
drewp@bigasterisk.com
parents: 1591
diff changeset
330 for v, vals in items:
d7b66234064b pure reordering of funcs to make the next diffs smaller
drewp@bigasterisk.com
parents: 1591
diff changeset
331 orderedVars.append(v)
d7b66234064b pure reordering of funcs to make the next diffs smaller
drewp@bigasterisk.com
parents: 1591
diff changeset
332 orderedValues: List[Node] = list(vals)
d7b66234064b pure reordering of funcs to make the next diffs smaller
drewp@bigasterisk.com
parents: 1591
diff changeset
333 orderedValues.sort(key=str)
d7b66234064b pure reordering of funcs to make the next diffs smaller
drewp@bigasterisk.com
parents: 1591
diff changeset
334 orderedValueSets.append(orderedValues)
1588
0757fafbfdab WIP inferencer - partial var and function support
drewp@bigasterisk.com
parents: 1587
diff changeset
335
1592
d7b66234064b pure reordering of funcs to make the next diffs smaller
drewp@bigasterisk.com
parents: 1591
diff changeset
336 return orderedVars, orderedValueSets