Mercurial > code > home > repos > homeauto
annotate service/mqtt_to_rdf/inference.py @ 1633:6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
author | drewp@bigasterisk.com |
---|---|
date | Sun, 12 Sep 2021 21:48:36 -0700 |
parents | bd79a2941cab |
children | ba59cfc3c747 |
rev | line source |
---|---|
1587 | 1 """ |
2 copied from reasoning 2021-08-29. probably same api. should | |
3 be able to lib/ this out | |
4 """ | |
1588
0757fafbfdab
WIP inferencer - partial var and function support
drewp@bigasterisk.com
parents:
1587
diff
changeset
|
5 import itertools |
1587 | 6 import logging |
1601 | 7 import time |
1594 | 8 from collections import defaultdict |
1626 | 9 from dataclasses import dataclass |
1633
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
10 from typing import Dict, Iterator, List, Optional, Sequence, Set, Tuple, Union, cast |
1588
0757fafbfdab
WIP inferencer - partial var and function support
drewp@bigasterisk.com
parents:
1587
diff
changeset
|
11 |
1631
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
12 from prometheus_client import Histogram, Summary |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
13 from rdflib import BNode, Graph, Namespace |
1589
5c1055be3c36
WIP more debugging, working towards bnode-matching support
drewp@bigasterisk.com
parents:
1588
diff
changeset
|
14 from rdflib.graph import ConjunctiveGraph, ReadOnlyGraphAggregate |
1631
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
15 from rdflib.term import Literal, Node, Variable |
1587 | 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) |
1631
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
19 from lhs_evaluation import Decimal, Evaluation, numericNode |
1605 | 20 |
1587 | 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 | 23 |
1626 | 24 INFER_CALLS = Summary('inference_infer_calls', 'calls') |
25 INFER_GRAPH_SIZE = Histogram('inference_graph_size', 'statements', buckets=[2**x for x in range(2, 20, 2)]) | |
1587 | 26 |
27 ROOM = Namespace("http://projects.bigasterisk.com/room/") | |
28 LOG = Namespace('http://www.w3.org/2000/10/swap/log#') | |
29 MATH = Namespace('http://www.w3.org/2000/10/swap/math#') | |
30 | |
31 | |
1631
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
32 def stmtTemplate(stmt: Triple) -> Tuple[Optional[Node], Optional[Node], Optional[Node]]: |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
33 return ( |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
34 None if isinstance(stmt[0], (Variable, BNode)) else stmt[0], |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
35 None if isinstance(stmt[1], (Variable, BNode)) else stmt[1], |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
36 None if isinstance(stmt[2], (Variable, BNode)) else stmt[2], |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
37 ) |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
38 |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
39 |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
40 class NoOptions(ValueError): |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
41 """stmtlooper has no possibilites to add to the binding; the whole rule must therefore not apply""" |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
42 |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
43 |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
44 class Inconsistent(ValueError): |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
45 """adding this stmt would be inconsistent with an existing binding""" |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
46 |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
47 |
1632 | 48 _stmtLooperShortId = itertools.count() |
49 | |
50 | |
1631
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
51 @dataclass |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
52 class StmtLooper: |
1632 | 53 """given one LHS stmt, iterate through the possible matches for it, |
54 returning what bindings they would imply. Only distinct bindings are | |
55 returned. The bindings build on any `prev` StmtLooper's results. | |
56 | |
57 This iterator is restartable.""" | |
1631
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
58 lhsStmt: Triple |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
59 prev: Optional['StmtLooper'] |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
60 workingSet: ReadOnlyWorkingSet |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
61 |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
62 def __repr__(self): |
1632 | 63 return f'StmtLooper{self._shortId}({graphDump([self.lhsStmt])} {"<pastEnd>" if self.pastEnd() else ""})' |
1631
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
64 |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
65 def __post_init__(self): |
1632 | 66 self._shortId = next(_stmtLooperShortId) |
1631
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
67 self._myWorkingSetMatches = self._myMatches(self.workingSet) |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
68 |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
69 self._current = CandidateBinding({}) |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
70 self._pastEnd = False |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
71 self._seenBindings: List[Dict[BindableTerm, Node]] = [] |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
72 self.restart() |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
73 |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
74 def _myMatches(self, g: Graph) -> List[Triple]: |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
75 template = stmtTemplate(self.lhsStmt) |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
76 |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
77 stmts = sorted(cast(Iterator[Triple], list(g.triples(template)))) |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
78 # plus new lhs possibilties... |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
79 # log.debug(f'{INDENT*6} {self} find {len(stmts)=} in {len(self.workingSet)=}') |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
80 |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
81 return stmts |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
82 |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
83 def _prevBindings(self) -> Dict[BindableTerm, Node]: |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
84 if not self.prev or self.prev.pastEnd(): |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
85 return {} |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
86 |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
87 return self.prev.currentBinding().binding |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
88 |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
89 def advance(self): |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
90 """update to a new set of bindings we haven't seen (since last restart), or go into pastEnd mode""" |
1633
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
91 if self._pastEnd: |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
92 raise NotImplementedError('need restart') |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
93 log.debug('') |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
94 augmentedWorkingSet: Sequence[Triple] = [] |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
95 if self.prev is None: |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
96 augmentedWorkingSet = self._myWorkingSetMatches |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
97 else: |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
98 augmentedWorkingSet = list(self.prev.currentBinding().apply(self._myWorkingSetMatches, |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
99 returnBoundStatementsOnly=False)) |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
100 |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
101 log.debug(f'{INDENT*6} {self} has {self._myWorkingSetMatches=}') |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
102 |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
103 log.debug(f'{INDENT*6} {self} mines {len(augmentedWorkingSet)} matching augmented statements') |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
104 for s in augmentedWorkingSet: |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
105 log.debug(f'{INDENT*7} {s}') |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
106 |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
107 for i, stmt in enumerate(augmentedWorkingSet): |
1631
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
108 try: |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
109 outBinding = self._totalBindingIfThisStmtWereTrue(stmt) |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
110 except Inconsistent: |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
111 log.debug(f'{INDENT*7} {self} - {stmt} would be inconsistent with prev bindings') |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
112 continue |
1632 | 113 |
114 log.debug(f'{INDENT*6} {outBinding=} {self._seenBindings=}') | |
1631
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
115 if outBinding.binding not in self._seenBindings: |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
116 self._seenBindings.append(outBinding.binding.copy()) |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
117 self._current = outBinding |
1632 | 118 log.debug(f'{INDENT*7} new binding from {self} -> {outBinding}') |
1631
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
119 return |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
120 log.debug(f'yes we saw') |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
121 |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
122 log.debug(f'{INDENT*6} {self} mines rules') |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
123 |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
124 if self.lhsStmt[1] == ROOM['asFarenheit']: |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
125 pb: Dict[BindableTerm, Node] = self._prevBindings() |
1632 | 126 log.debug(f'{INDENT*6} {self} consider ?x faren ?y where ?x={self.lhsStmt[0]} and {pb=}') |
127 | |
1631
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
128 if self.lhsStmt[0] in pb: |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
129 operands = [pb[cast(BindableTerm, self.lhsStmt[0])]] |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
130 f = cast(Literal, Literal(Decimal(numericNode(operands[0])) * 9 / 5 + 32)) |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
131 objVar = self.lhsStmt[2] |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
132 if not isinstance(objVar, Variable): |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
133 raise TypeError(f'expected Variable, got {objVar!r}') |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
134 newBindings = {cast(BindableTerm, objVar): cast(Node, f)} |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
135 self._current.addNewBindings(CandidateBinding(newBindings)) |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
136 if newBindings not in self._seenBindings: |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
137 self._seenBindings.append(newBindings) |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
138 self._current = CandidateBinding(newBindings) |
1633
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
139 return |
1631
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
140 |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
141 log.debug(f'{INDENT*6} {self} is past end') |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
142 self._pastEnd = True |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
143 |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
144 def _totalBindingIfThisStmtWereTrue(self, newStmt: Triple) -> CandidateBinding: |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
145 outBinding = self._prevBindings().copy() |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
146 for rt, ct in zip(self.lhsStmt, newStmt): |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
147 if isinstance(rt, (Variable, BNode)): |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
148 if rt in outBinding and outBinding[rt] != ct: |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
149 raise Inconsistent() |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
150 outBinding[rt] = ct |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
151 return CandidateBinding(outBinding) |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
152 |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
153 def currentBinding(self) -> CandidateBinding: |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
154 if self.pastEnd(): |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
155 raise NotImplementedError() |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
156 return self._current |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
157 |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
158 def newLhsStmts(self) -> List[Triple]: |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
159 """under the curent bindings, what new stmts beyond workingSet are also true? includes all `prev`""" |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
160 return [] |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
161 |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
162 def pastEnd(self) -> bool: |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
163 return self._pastEnd |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
164 |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
165 def restart(self): |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
166 self._pastEnd = False |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
167 self._seenBindings = [] |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
168 self.advance() |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
169 if self.pastEnd(): |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
170 raise NoOptions() |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
171 |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
172 |
1594 | 173 @dataclass |
1593
b0df43d5494c
big rewrite- more classes, smaller methods, more typesafe, all current tests passing
drewp@bigasterisk.com
parents:
1592
diff
changeset
|
174 class Lhs: |
1594 | 175 graph: Graph |
176 | |
177 def __post_init__(self): | |
1608 | 178 # do precomputation in here that's not specific to the workingSet |
1631
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
179 # self.staticRuleStmts = Graph() |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
180 # self.nonStaticRuleStmts = 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
|
181 |
1631
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
182 # self.lhsBindables: Set[BindableTerm] = set() |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
183 # self.lhsBnodes: Set[BNode] = set() |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
184 # for ruleStmt in self.graph: |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
185 # varsAndBnodesInStmt = [term for term in ruleStmt if isinstance(term, (Variable, BNode))] |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
186 # self.lhsBindables.update(varsAndBnodesInStmt) |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
187 # self.lhsBnodes.update(x for x in varsAndBnodesInStmt if isinstance(x, BNode)) |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
188 # if not varsAndBnodesInStmt: |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
189 # self.staticRuleStmts.add(ruleStmt) |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
190 # else: |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
191 # 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
|
192 |
1631
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
193 # self.nonStaticRuleStmtsSet = set(self.nonStaticRuleStmts) |
1615
bcfa368e5498
change a Graph.__sub__ to Set.difference in verify() for a big speedup
drewp@bigasterisk.com
parents:
1614
diff
changeset
|
194 |
1602 | 195 self.evaluations = list(Evaluation.findEvals(self.graph)) |
196 | |
1609
34f2817320cc
new tests for a smaller part of the inner loop
drewp@bigasterisk.com
parents:
1608
diff
changeset
|
197 def __repr__(self): |
34f2817320cc
new tests for a smaller part of the inner loop
drewp@bigasterisk.com
parents:
1608
diff
changeset
|
198 return f"Lhs({graphDump(self.graph)})" |
34f2817320cc
new tests for a smaller part of the inner loop
drewp@bigasterisk.com
parents:
1608
diff
changeset
|
199 |
1621 | 200 def findCandidateBindings(self, knownTrue: 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
|
201 """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
|
202 from LHS""" |
1633
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
203 if self.graph.__len__() == 0: |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
204 # special case- no LHS! |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
205 yield BoundLhs(self, CandidateBinding({})) |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
206 return |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
207 |
1631
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
208 log.debug(f'{INDENT*4} build new StmtLooper stack') |
1593
b0df43d5494c
big rewrite- more classes, smaller methods, more typesafe, all current tests passing
drewp@bigasterisk.com
parents:
1592
diff
changeset
|
209 |
1631
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
210 try: |
1633
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
211 stmtStack = self._assembleRings(knownTrue) |
1631
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
212 except NoOptions: |
1632 | 213 log.debug(f'{INDENT*5} start up with no options; 0 bindings') |
1593
b0df43d5494c
big rewrite- more classes, smaller methods, more typesafe, all current tests passing
drewp@bigasterisk.com
parents:
1592
diff
changeset
|
214 return |
1632 | 215 self._debugStmtStack('initial odometer', stmtStack) |
1633
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
216 self._assertAllRingsAreValid(stmtStack) |
1631
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
217 |
1633
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
218 lastRing = stmtStack[-1] |
1631
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
219 iterCount = 0 |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
220 while True: |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
221 iterCount += 1 |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
222 if iterCount > 10: |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
223 raise ValueError('stuck') |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
224 |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
225 log.debug(f'{INDENT*4} vv findCandBindings iteration {iterCount}') |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
226 |
1633
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
227 yield BoundLhs(self, lastRing.currentBinding()) |
1631
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
228 |
1632 | 229 self._debugStmtStack('odometer', stmtStack) |
1607
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
230 |
1631
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
231 done = self._advanceAll(stmtStack) |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
232 |
1632 | 233 self._debugStmtStack('odometer after ({done=})', stmtStack) |
1631
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
234 |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
235 log.debug(f'{INDENT*4} ^^ findCandBindings iteration done') |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
236 if done: |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
237 break |
1607
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
238 |
1632 | 239 def _debugStmtStack(self, label, stmtStack): |
240 log.debug(f'{INDENT*5} {label}:') | |
241 for l in stmtStack: | |
242 log.debug(f'{INDENT*6} {l} curbind={l.currentBinding() if not l.pastEnd() else "<end>"}') | |
243 | |
1633
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
244 def _assembleRings(self, knownTrue: ReadOnlyWorkingSet) -> List[StmtLooper]: |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
245 """make StmtLooper for each stmt in our LHS graph, but do it in a way that they all |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
246 start out valid (or else raise NoOptions)""" |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
247 |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
248 stmtsToAdd = list(self.graph) |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
249 |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
250 for perm in itertools.permutations(stmtsToAdd): |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
251 stmtStack: List[StmtLooper] = [] |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
252 prev: Optional[StmtLooper] = None |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
253 log.debug(f'{INDENT*5} try stmts in this order: {" -> ".join(graphDump([p]) for p in perm)}') |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
254 |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
255 for s in perm: |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
256 try: |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
257 elem = StmtLooper(s, prev, knownTrue) |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
258 except NoOptions: |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
259 log.debug(f'{INDENT*6} permutation didnt work, try another') |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
260 break |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
261 stmtStack.append(elem) |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
262 prev = stmtStack[-1] |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
263 else: |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
264 return stmtStack |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
265 log.debug(f'{INDENT*6} no perms worked- rule cannot match anything') |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
266 |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
267 raise NoOptions() |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
268 |
1631
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
269 def _advanceAll(self, stmtStack: List[StmtLooper]) -> bool: |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
270 carry = True # 1st elem always must advance |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
271 for i, ring in enumerate(stmtStack): |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
272 # unlike normal odometer, advancing any earlier ring could invalidate later ones |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
273 if carry: |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
274 log.debug(f'{INDENT*5} advanceAll [{i}] {ring} carry/advance') |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
275 ring.advance() |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
276 carry = False |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
277 if ring.pastEnd(): |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
278 if ring is stmtStack[-1]: |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
279 log.debug(f'{INDENT*5} advanceAll [{i}] {ring} says we done') |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
280 return True |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
281 log.debug(f'{INDENT*5} advanceAll [{i}] {ring} restart') |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
282 ring.restart() |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
283 carry = True |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
284 return False |
1607
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
285 |
1633
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
286 def _assertAllRingsAreValid(self, stmtStack): |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
287 if any(ring.pastEnd() for ring in stmtStack): # this is an unexpected debug assertion |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
288 log.debug(f'{INDENT*5} some rings started at pastEnd {stmtStack}') |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
289 raise NoOptions() |
6107603ed455
fix farenheit rule case, fix some others that depend on rings order, but this breaks some performance because of itertools.perm
drewp@bigasterisk.com
parents:
1632
diff
changeset
|
290 |
1621 | 291 def _allStaticStatementsMatch(self, knownTrue: ReadOnlyWorkingSet) -> bool: |
1613
03ed8c9abd5b
forget GRAPH_ID optimization in this case
drewp@bigasterisk.com
parents:
1612
diff
changeset
|
292 # 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
|
293 # static stmt is matched by a non-static stmt in the rule itself |
1608 | 294 for ruleStmt in self.staticRuleStmts: |
1621 | 295 if ruleStmt not in knownTrue: |
1608 | 296 log.debug(f'{INDENT*3} {ruleStmt} not in working set- skip rule') |
297 return False | |
298 return True | |
299 | |
1607
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
300 def _possibleBindings(self, workingSet, stats) -> Iterator['BoundLhs']: |
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
301 """this yields at least the working bindings, and possibly others""" |
1600 | 302 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
|
303 for bindRow in self._product(candidateTermMatches): |
97b2c3cfdb83
refactor: inline an odometer algorithm in place of itertools.product
drewp@bigasterisk.com
parents:
1613
diff
changeset
|
304 try: |
97b2c3cfdb83
refactor: inline an odometer algorithm in place of itertools.product
drewp@bigasterisk.com
parents:
1613
diff
changeset
|
305 yield BoundLhs(self, bindRow) |
97b2c3cfdb83
refactor: inline an odometer algorithm in place of itertools.product
drewp@bigasterisk.com
parents:
1613
diff
changeset
|
306 except EvaluationFailed: |
97b2c3cfdb83
refactor: inline an odometer algorithm in place of itertools.product
drewp@bigasterisk.com
parents:
1613
diff
changeset
|
307 stats['permCountFailingEval'] += 1 |
1593
b0df43d5494c
big rewrite- more classes, smaller methods, more typesafe, all current tests passing
drewp@bigasterisk.com
parents:
1592
diff
changeset
|
308 |
1600 | 309 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
|
310 """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
|
311 |
1593
b0df43d5494c
big rewrite- more classes, smaller methods, more typesafe, all current tests passing
drewp@bigasterisk.com
parents:
1592
diff
changeset
|
312 candidateTermMatches: Dict[BindableTerm, Set[Node]] = defaultdict(set) |
1594 | 313 for lhsStmt in self.graph: |
1597 | 314 log.debug(f'{INDENT*4} possibles for this lhs stmt: {lhsStmt}') |
1608 | 315 for i, trueStmt in enumerate(workingSet): |
1597 | 316 # 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
|
317 |
1594 | 318 for v, vals in self._bindingsFromStatement(lhsStmt, trueStmt): |
319 candidateTermMatches[v].update(vals) | |
320 | |
1593
b0df43d5494c
big rewrite- more classes, smaller methods, more typesafe, all current tests passing
drewp@bigasterisk.com
parents:
1592
diff
changeset
|
321 return candidateTermMatches |
b0df43d5494c
big rewrite- more classes, smaller methods, more typesafe, all current tests passing
drewp@bigasterisk.com
parents:
1592
diff
changeset
|
322 |
1594 | 323 def _bindingsFromStatement(self, stmt1: Triple, stmt2: Triple) -> Iterator[Tuple[Variable, Set[Node]]]: |
324 """if these stmts match otherwise, what BNode or Variable mappings do we learn? | |
325 | |
326 e.g. stmt1=(?x B ?y) and stmt2=(A B C), then we yield (?x, {A}) and (?y, {C}) | |
327 or stmt1=(_:x B C) and stmt2=(A B C), then we yield (_:x, {A}) | |
328 or stmt1=(?x B C) and stmt2=(A B D), then we yield nothing | |
329 """ | |
330 bindingsFromStatement = {} | |
331 for term1, term2 in zip(stmt1, stmt2): | |
332 if isinstance(term1, (BNode, Variable)): | |
333 bindingsFromStatement.setdefault(term1, set()).add(term2) | |
334 elif term1 != term2: | |
335 break | |
336 else: | |
337 for v, vals in bindingsFromStatement.items(): | |
1597 | 338 log.debug(f'{INDENT*5} {v=} {vals=}') |
1594 | 339 yield v, vals |
340 | |
1627
ea559a846714
some shuffling, i don't know- i'm about to rewrite again
drewp@bigasterisk.com
parents:
1626
diff
changeset
|
341 def _product(self, candidateTermMatches: Dict[BindableTerm, Set[Node]]) -> Iterator[CandidateBinding]: |
ea559a846714
some shuffling, i don't know- i'm about to rewrite again
drewp@bigasterisk.com
parents:
1626
diff
changeset
|
342 orderedVars, orderedValueSets = _organize(candidateTermMatches) |
ea559a846714
some shuffling, i don't know- i'm about to rewrite again
drewp@bigasterisk.com
parents:
1626
diff
changeset
|
343 self._logCandidates(orderedVars, orderedValueSets) |
ea559a846714
some shuffling, i don't know- i'm about to rewrite again
drewp@bigasterisk.com
parents:
1626
diff
changeset
|
344 log.debug(f'{INDENT*3} trying all permutations:') |
ea559a846714
some shuffling, i don't know- i'm about to rewrite again
drewp@bigasterisk.com
parents:
1626
diff
changeset
|
345 if not orderedValueSets: |
ea559a846714
some shuffling, i don't know- i'm about to rewrite again
drewp@bigasterisk.com
parents:
1626
diff
changeset
|
346 yield CandidateBinding({}) |
ea559a846714
some shuffling, i don't know- i'm about to rewrite again
drewp@bigasterisk.com
parents:
1626
diff
changeset
|
347 return |
ea559a846714
some shuffling, i don't know- i'm about to rewrite again
drewp@bigasterisk.com
parents:
1626
diff
changeset
|
348 |
ea559a846714
some shuffling, i don't know- i'm about to rewrite again
drewp@bigasterisk.com
parents:
1626
diff
changeset
|
349 if not orderedValueSets or not all(orderedValueSets): |
ea559a846714
some shuffling, i don't know- i'm about to rewrite again
drewp@bigasterisk.com
parents:
1626
diff
changeset
|
350 # some var or bnode has no options at all |
ea559a846714
some shuffling, i don't know- i'm about to rewrite again
drewp@bigasterisk.com
parents:
1626
diff
changeset
|
351 return |
ea559a846714
some shuffling, i don't know- i'm about to rewrite again
drewp@bigasterisk.com
parents:
1626
diff
changeset
|
352 rings: List[Iterator[Node]] = [itertools.cycle(valSet) for valSet in orderedValueSets] |
ea559a846714
some shuffling, i don't know- i'm about to rewrite again
drewp@bigasterisk.com
parents:
1626
diff
changeset
|
353 currentSet: List[Node] = [next(ring) for ring in rings] |
ea559a846714
some shuffling, i don't know- i'm about to rewrite again
drewp@bigasterisk.com
parents:
1626
diff
changeset
|
354 starts = [valSet[-1] for valSet in orderedValueSets] |
ea559a846714
some shuffling, i don't know- i'm about to rewrite again
drewp@bigasterisk.com
parents:
1626
diff
changeset
|
355 while True: |
ea559a846714
some shuffling, i don't know- i'm about to rewrite again
drewp@bigasterisk.com
parents:
1626
diff
changeset
|
356 for col, curr in enumerate(currentSet): |
ea559a846714
some shuffling, i don't know- i'm about to rewrite again
drewp@bigasterisk.com
parents:
1626
diff
changeset
|
357 currentSet[col] = next(rings[col]) |
ea559a846714
some shuffling, i don't know- i'm about to rewrite again
drewp@bigasterisk.com
parents:
1626
diff
changeset
|
358 log.debug(f'{INDENT*4} currentSet: {repr(currentSet)}') |
ea559a846714
some shuffling, i don't know- i'm about to rewrite again
drewp@bigasterisk.com
parents:
1626
diff
changeset
|
359 yield CandidateBinding(dict(zip(orderedVars, currentSet))) |
ea559a846714
some shuffling, i don't know- i'm about to rewrite again
drewp@bigasterisk.com
parents:
1626
diff
changeset
|
360 if curr is not starts[col]: |
ea559a846714
some shuffling, i don't know- i'm about to rewrite again
drewp@bigasterisk.com
parents:
1626
diff
changeset
|
361 break |
ea559a846714
some shuffling, i don't know- i'm about to rewrite again
drewp@bigasterisk.com
parents:
1626
diff
changeset
|
362 if col == len(orderedValueSets) - 1: |
ea559a846714
some shuffling, i don't know- i'm about to rewrite again
drewp@bigasterisk.com
parents:
1626
diff
changeset
|
363 return |
ea559a846714
some shuffling, i don't know- i'm about to rewrite again
drewp@bigasterisk.com
parents:
1626
diff
changeset
|
364 |
1600 | 365 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
|
366 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
|
367 return |
1597 | 368 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
|
369 for v, vals in zip(orderedVars, orderedValueSets): |
1597 | 370 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
|
371 for val in vals: |
1597 | 372 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
|
373 |
d7b66234064b
pure reordering of funcs to make the next diffs smaller
drewp@bigasterisk.com
parents:
1591
diff
changeset
|
374 |
1622
38bd8ef9ef67
add CandidateTermMatches, unused so far
drewp@bigasterisk.com
parents:
1621
diff
changeset
|
375 @dataclass |
1607
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
376 class BoundLhs: |
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
377 lhs: Lhs |
1610 | 378 binding: CandidateBinding |
1607
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
379 |
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
380 def __post_init__(self): |
1613
03ed8c9abd5b
forget GRAPH_ID optimization in this case
drewp@bigasterisk.com
parents:
1612
diff
changeset
|
381 self.usedByFuncs = Graph() |
1631
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
382 # self._applyFunctions() |
1607
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
383 |
1616
3a6ed545357f
optimization: stream stmts instead of building a Graph
drewp@bigasterisk.com
parents:
1615
diff
changeset
|
384 def lhsStmtsWithoutEvals(self): |
3a6ed545357f
optimization: stream stmts instead of building a Graph
drewp@bigasterisk.com
parents:
1615
diff
changeset
|
385 for stmt in self.lhs.graph: |
3a6ed545357f
optimization: stream stmts instead of building a Graph
drewp@bigasterisk.com
parents:
1615
diff
changeset
|
386 if stmt in self.usedByFuncs: |
3a6ed545357f
optimization: stream stmts instead of building a Graph
drewp@bigasterisk.com
parents:
1615
diff
changeset
|
387 continue |
3a6ed545357f
optimization: stream stmts instead of building a Graph
drewp@bigasterisk.com
parents:
1615
diff
changeset
|
388 yield stmt |
1607
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
389 |
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
390 def _applyFunctions(self): |
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
391 """may grow the binding with some results""" |
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
392 while True: |
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
393 delta = self._applyFunctionsIteration() |
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
394 if delta == 0: |
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
395 break |
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
396 |
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
397 def _applyFunctionsIteration(self): |
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
398 before = len(self.binding.binding) |
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
399 delta = 0 |
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
400 for ev in self.lhs.evaluations: |
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
401 newBindings, usedGraph = ev.resultBindings(self.binding) |
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
402 self.usedByFuncs += usedGraph |
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
403 self.binding.addNewBindings(newBindings) |
1608 | 404 delta = len(self.binding.binding) - before |
405 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
|
406 return delta |
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
407 |
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
408 def verify(self, workingSet: ReadOnlyWorkingSet) -> bool: |
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
409 """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
|
410 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
|
411 boundLhs = self.binding.apply(rem) |
1607
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
412 |
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
|
413 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
|
414 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
|
415 self._logVerifyBanner(boundLhs, workingSet) |
1607
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
416 |
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
417 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
|
418 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
|
419 |
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
|
420 if stmt not in workingSet: |
1607
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
421 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
|
422 return False |
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
423 return True |
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
424 |
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
|
425 def _logVerifyBanner(self, boundLhs, workingSet: ReadOnlyWorkingSet): |
1607
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
426 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
|
427 for stmt in sorted(boundLhs): |
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
428 log.debug(f'{INDENT*4}|{INDENT} {stmt}') |
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
429 |
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
430 # 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
|
431 # for stmt in sorted(workingSet): |
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
432 # log.debug(f'{INDENT*4}|{INDENT} {stmt}') |
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
433 |
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
434 log.debug(f'{INDENT*4}\\') |
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
435 |
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
436 |
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
437 @dataclass |
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
438 class Rule: |
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
439 lhsGraph: Graph |
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
440 rhsGraph: Graph |
1632 | 441 |
1607
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
442 def __post_init__(self): |
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
443 self.lhs = Lhs(self.lhsGraph) |
1632 | 444 # |
1631
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
445 self.rhsBnodeMap = {} |
1607
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
446 |
1612
272f78d4671a
mark skipped tests. move applyRule into Rule. minor cleanups.
drewp@bigasterisk.com
parents:
1611
diff
changeset
|
447 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
|
448 for bound in self.lhs.findCandidateBindings(ReadOnlyGraphAggregate([workingSet]), stats): |
1631
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
449 log.debug(f'{INDENT*5} +rule has a working binding: {bound}') |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
450 |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
451 # rhs could have more bnodes, and they just need to be distinct per rule-firing that we do |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
452 existingRhsBnodes = set() |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
453 for stmt in self.rhsGraph: |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
454 for t in stmt: |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
455 if isinstance(t, BNode): |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
456 existingRhsBnodes.add(t) |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
457 # if existingRhsBnodes: |
1632 | 458 # log.debug(f'{INDENT*6} mapping rhs bnodes {existingRhsBnodes} to new ones') |
1631
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
459 |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
460 for b in existingRhsBnodes: |
1612
272f78d4671a
mark skipped tests. move applyRule into Rule. minor cleanups.
drewp@bigasterisk.com
parents:
1611
diff
changeset
|
461 |
1631
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
462 key = tuple(sorted(bound.binding.binding.items())), b |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
463 self.rhsBnodeMap.setdefault(key, BNode()) |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
464 |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
465 bound.binding.addNewBindings(CandidateBinding({b: self.rhsBnodeMap[key]})) |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
466 |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
467 # for lhsBoundStmt in bound.binding.apply(bound.lhsStmtsWithoutEvals()): |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
468 # log.debug(f'{INDENT*6} adding to workingSet {lhsBoundStmt=}') |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
469 # workingSet.add(lhsBoundStmt) |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
470 # log.debug(f'{INDENT*6} rhsGraph is good: {list(self.rhsGraph)}') |
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
471 |
1612
272f78d4671a
mark skipped tests. move applyRule into Rule. minor cleanups.
drewp@bigasterisk.com
parents:
1611
diff
changeset
|
472 for newStmt in bound.binding.apply(self.rhsGraph): |
1631
2c85a4f5dd9c
big rewrite of infer() using statements not variables as the things to iterate over
drewp@bigasterisk.com
parents:
1627
diff
changeset
|
473 # log.debug(f'{INDENT*6} adding {newStmt=}') |
1612
272f78d4671a
mark skipped tests. move applyRule into Rule. minor cleanups.
drewp@bigasterisk.com
parents:
1611
diff
changeset
|
474 workingSet.add(newStmt) |
272f78d4671a
mark skipped tests. move applyRule into Rule. minor cleanups.
drewp@bigasterisk.com
parents:
1611
diff
changeset
|
475 implied.add(newStmt) |
272f78d4671a
mark skipped tests. move applyRule into Rule. minor cleanups.
drewp@bigasterisk.com
parents:
1611
diff
changeset
|
476 |
1607
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
477 |
1587 | 478 class Inference: |
479 | |
480 def __init__(self) -> None: | |
1607
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
481 self.rules = [] |
1587 | 482 |
483 def setRules(self, g: ConjunctiveGraph): | |
1607
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
484 self.rules: List[Rule] = [] |
1599
abbf0eb0e640
fix a bug with a slightly moer complicated set of rules
drewp@bigasterisk.com
parents:
1598
diff
changeset
|
485 for stmt in g: |
abbf0eb0e640
fix a bug with a slightly moer complicated set of rules
drewp@bigasterisk.com
parents:
1598
diff
changeset
|
486 if stmt[1] == LOG['implies']: |
1607
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
487 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
|
488 # other stmts should go to a default working set? |
1587 | 489 |
1601 | 490 @INFER_CALLS.time() |
1587 | 491 def infer(self, graph: Graph): |
492 """ | |
493 returns new graph of inferred statements. | |
494 """ | |
1626 | 495 n = graph.__len__() |
496 INFER_GRAPH_SIZE.observe(n) | |
497 log.info(f'{INDENT*0} Begin inference of graph len={n} with rules len={len(self.rules)}:') | |
1601 | 498 startTime = time.time() |
1607
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
499 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
|
500 # 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
|
501 workingSet = Graph() |
b0df43d5494c
big rewrite- more classes, smaller methods, more typesafe, all current tests passing
drewp@bigasterisk.com
parents:
1592
diff
changeset
|
502 workingSet += graph |
1587 | 503 |
1594 | 504 # just the statements that came from RHS's of rules that fired. |
1587 | 505 implied = ConjunctiveGraph() |
506 | |
507 bailout_iterations = 100 | |
508 delta = 1 | |
1607
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
509 stats['initWorkingSet'] = cast(int, workingSet.__len__()) |
1587 | 510 while delta > 0 and bailout_iterations > 0: |
1620 | 511 log.debug('') |
1597 | 512 log.info(f'{INDENT*1}*iteration ({bailout_iterations} left)') |
1587 | 513 bailout_iterations -= 1 |
514 delta = -len(implied) | |
1607
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
515 self._iterateAllRules(workingSet, implied, stats) |
1587 | 516 delta += len(implied) |
1607
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
517 stats['iterations'] += 1 |
1597 | 518 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
|
519 stats['timeSpent'] = round(time.time() - startTime, 3) |
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
520 stats['impliedStmts'] = len(implied) |
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
521 log.info(f'{INDENT*0} Inference done {dict(stats)}. Implied:') |
1587 | 522 for st in implied: |
1597 | 523 log.info(f'{INDENT*1} {st}') |
1587 | 524 return implied |
525 | |
1607
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
526 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
|
527 for i, rule in enumerate(self.rules): |
272f78d4671a
mark skipped tests. move applyRule into Rule. minor cleanups.
drewp@bigasterisk.com
parents:
1611
diff
changeset
|
528 self._logRuleApplicationHeader(workingSet, i, rule) |
272f78d4671a
mark skipped tests. move applyRule into Rule. minor cleanups.
drewp@bigasterisk.com
parents:
1611
diff
changeset
|
529 rule.applyRule(workingSet, implied, stats) |
1587 | 530 |
1607
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
531 def _logRuleApplicationHeader(self, workingSet, i, r: Rule): |
1594 | 532 if not log.isEnabledFor(logging.DEBUG): |
533 return | |
534 | |
535 log.debug('') | |
536 log.debug(f'{INDENT*2} workingSet:') | |
1597 | 537 for j, stmt in enumerate(sorted(workingSet)): |
538 log.debug(f'{INDENT*3} ({j}) {stmt}') | |
1594 | 539 |
540 log.debug('') | |
541 log.debug(f'{INDENT*2}-applying rule {i}') | |
1632 | 542 log.debug(f'{INDENT*3} rule def lhs:') |
543 for stmt in r.lhsGraph: | |
544 log.debug(f'{INDENT*4} {stmt}') | |
1607
b21885181e35
more modules, types. Maybe less repeated computation on BoundLhs
drewp@bigasterisk.com
parents:
1605
diff
changeset
|
545 log.debug(f'{INDENT*3} rule def rhs: {graphDump(r.rhsGraph)}') |
1594 | 546 |
1588
0757fafbfdab
WIP inferencer - partial var and function support
drewp@bigasterisk.com
parents:
1587
diff
changeset
|
547 |
1590
327202020892
WIP inference- getting into more degenerate test cases
drewp@bigasterisk.com
parents:
1589
diff
changeset
|
548 def graphDump(g: Union[Graph, List[Triple]]): |
327202020892
WIP inference- getting into more degenerate test cases
drewp@bigasterisk.com
parents:
1589
diff
changeset
|
549 if not isinstance(g, Graph): |
327202020892
WIP inference- getting into more degenerate test cases
drewp@bigasterisk.com
parents:
1589
diff
changeset
|
550 g2 = Graph() |
1594 | 551 g2 += g |
1590
327202020892
WIP inference- getting into more degenerate test cases
drewp@bigasterisk.com
parents:
1589
diff
changeset
|
552 g = g2 |
1589
5c1055be3c36
WIP more debugging, working towards bnode-matching support
drewp@bigasterisk.com
parents:
1588
diff
changeset
|
553 g.bind('', ROOM) |
5c1055be3c36
WIP more debugging, working towards bnode-matching support
drewp@bigasterisk.com
parents:
1588
diff
changeset
|
554 g.bind('ex', Namespace('http://example.com/')) |
5c1055be3c36
WIP more debugging, working towards bnode-matching support
drewp@bigasterisk.com
parents:
1588
diff
changeset
|
555 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
|
556 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
|
557 return ' '.join(lines) |
5c1055be3c36
WIP more debugging, working towards bnode-matching support
drewp@bigasterisk.com
parents:
1588
diff
changeset
|
558 |
5c1055be3c36
WIP more debugging, working towards bnode-matching support
drewp@bigasterisk.com
parents:
1588
diff
changeset
|
559 |
1600 | 560 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
|
561 items = list(candidateTermMatches.items()) |
d7b66234064b
pure reordering of funcs to make the next diffs smaller
drewp@bigasterisk.com
parents:
1591
diff
changeset
|
562 items.sort() |
d7b66234064b
pure reordering of funcs to make the next diffs smaller
drewp@bigasterisk.com
parents:
1591
diff
changeset
|
563 orderedVars: List[BindableTerm] = [] |
d7b66234064b
pure reordering of funcs to make the next diffs smaller
drewp@bigasterisk.com
parents:
1591
diff
changeset
|
564 orderedValueSets: List[List[Node]] = [] |
d7b66234064b
pure reordering of funcs to make the next diffs smaller
drewp@bigasterisk.com
parents:
1591
diff
changeset
|
565 for v, vals in items: |
d7b66234064b
pure reordering of funcs to make the next diffs smaller
drewp@bigasterisk.com
parents:
1591
diff
changeset
|
566 orderedVars.append(v) |
d7b66234064b
pure reordering of funcs to make the next diffs smaller
drewp@bigasterisk.com
parents:
1591
diff
changeset
|
567 orderedValues: List[Node] = list(vals) |
d7b66234064b
pure reordering of funcs to make the next diffs smaller
drewp@bigasterisk.com
parents:
1591
diff
changeset
|
568 orderedValues.sort(key=str) |
d7b66234064b
pure reordering of funcs to make the next diffs smaller
drewp@bigasterisk.com
parents:
1591
diff
changeset
|
569 orderedValueSets.append(orderedValues) |
1588
0757fafbfdab
WIP inferencer - partial var and function support
drewp@bigasterisk.com
parents:
1587
diff
changeset
|
570 |
1592
d7b66234064b
pure reordering of funcs to make the next diffs smaller
drewp@bigasterisk.com
parents:
1591
diff
changeset
|
571 return orderedVars, orderedValueSets |