homeauto: service/mqtt_to_rdf/inference/stmt

annotate service/mqtt_to_rdf/inference/stmt_chunk.py @ 1727:23e6154e6c11

file moves

author	drewp@bigasterisk.com
date	Tue, 20 Jun 2023 23:26:24 -0700
parents	service/mqtt_to_rdf/stmt_chunk.py@88f6e9bf69d1
children

rev	line source
1651 20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	1 import itertools
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	2 import logging
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	3 from dataclasses import dataclass
1673 80f4e741ca4f redo RHS bnode processing drewp@bigasterisk.com parents: 1669 diff changeset	4 from typing import Iterable, Iterator, List, Optional, Set, Tuple, Type, Union, cast
1651 20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	5
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	6 from rdflib.graph import Graph
1727 23e6154e6c11 file moves drewp@bigasterisk.com parents: 1697 diff changeset	7 from rdflib import RDF
1673 80f4e741ca4f redo RHS bnode processing drewp@bigasterisk.com parents: 1669 diff changeset	8 from rdflib.term import Literal, Node, URIRef, Variable
1651 20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	9
1727 23e6154e6c11 file moves drewp@bigasterisk.com parents: 1697 diff changeset	10 from inference.candidate_binding import CandidateBinding
23e6154e6c11 file moves drewp@bigasterisk.com parents: 1697 diff changeset	11 from inference.inference_types import Inconsistent, RuleUnboundBnode, WorkingSetBnode
1651 20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	12
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	13 log = logging.getLogger('infer')
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	14
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	15 INDENT = ' '
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	16
1660 31f7dab6a60b function evaluation uses Chunk lists now and runs fast. Only a few edge cases still broken drewp@bigasterisk.com parents: 1659 diff changeset	17 ChunkPrimaryTriple = Tuple[Optional[Node], Node, Optional[Node]]
31f7dab6a60b function evaluation uses Chunk lists now and runs fast. Only a few edge cases still broken drewp@bigasterisk.com parents: 1659 diff changeset	18
1651 20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	19
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	20 @dataclass
1664 1a7c1261302c logic fix- some bindings were being returned 2+; some 0 times drewp@bigasterisk.com parents: 1661 diff changeset	21 class AlignedRuleChunk:
1668 89e53cb8a01c fix some harder tests. Mostly, _advanceTheStack needed to spin the odometer rings starting from the other side, to get all the right combos drewp@bigasterisk.com parents: 1664 diff changeset	22 """a possible association between a rule chunk and a workingSet chunk. You can test
89e53cb8a01c fix some harder tests. Mostly, _advanceTheStack needed to spin the odometer rings starting from the other side, to get all the right combos drewp@bigasterisk.com parents: 1664 diff changeset	23 whether the association would still be possible under various additional bindings."""
1664 1a7c1261302c logic fix- some bindings were being returned 2+; some 0 times drewp@bigasterisk.com parents: 1661 diff changeset	24 ruleChunk: 'Chunk'
1a7c1261302c logic fix- some bindings were being returned 2+; some 0 times drewp@bigasterisk.com parents: 1661 diff changeset	25 workingSetChunk: 'Chunk'
1a7c1261302c logic fix- some bindings were being returned 2+; some 0 times drewp@bigasterisk.com parents: 1661 diff changeset	26
1668 89e53cb8a01c fix some harder tests. Mostly, _advanceTheStack needed to spin the odometer rings starting from the other side, to get all the right combos drewp@bigasterisk.com parents: 1664 diff changeset	27 def __post_init__(self):
89e53cb8a01c fix some harder tests. Mostly, _advanceTheStack needed to spin the odometer rings starting from the other side, to get all the right combos drewp@bigasterisk.com parents: 1664 diff changeset	28 if not self.matches():
89e53cb8a01c fix some harder tests. Mostly, _advanceTheStack needed to spin the odometer rings starting from the other side, to get all the right combos drewp@bigasterisk.com parents: 1664 diff changeset	29 raise Inconsistent()
89e53cb8a01c fix some harder tests. Mostly, _advanceTheStack needed to spin the odometer rings starting from the other side, to get all the right combos drewp@bigasterisk.com parents: 1664 diff changeset	30
89e53cb8a01c fix some harder tests. Mostly, _advanceTheStack needed to spin the odometer rings starting from the other side, to get all the right combos drewp@bigasterisk.com parents: 1664 diff changeset	31 def newBindingIfMatched(self, prevBindings: CandidateBinding) -> CandidateBinding:
89e53cb8a01c fix some harder tests. Mostly, _advanceTheStack needed to spin the odometer rings starting from the other side, to get all the right combos drewp@bigasterisk.com parents: 1664 diff changeset	32 """supposing this rule did match the statement, what new bindings would
89e53cb8a01c fix some harder tests. Mostly, _advanceTheStack needed to spin the odometer rings starting from the other side, to get all the right combos drewp@bigasterisk.com parents: 1664 diff changeset	33 that produce?
89e53cb8a01c fix some harder tests. Mostly, _advanceTheStack needed to spin the odometer rings starting from the other side, to get all the right combos drewp@bigasterisk.com parents: 1664 diff changeset	34
89e53cb8a01c fix some harder tests. Mostly, _advanceTheStack needed to spin the odometer rings starting from the other side, to get all the right combos drewp@bigasterisk.com parents: 1664 diff changeset	35 raises Inconsistent if the existing bindings mean that our aligned
89e53cb8a01c fix some harder tests. Mostly, _advanceTheStack needed to spin the odometer rings starting from the other side, to get all the right combos drewp@bigasterisk.com parents: 1664 diff changeset	36 chunks can no longer match.
89e53cb8a01c fix some harder tests. Mostly, _advanceTheStack needed to spin the odometer rings starting from the other side, to get all the right combos drewp@bigasterisk.com parents: 1664 diff changeset	37 """
89e53cb8a01c fix some harder tests. Mostly, _advanceTheStack needed to spin the odometer rings starting from the other side, to get all the right combos drewp@bigasterisk.com parents: 1664 diff changeset	38 outBinding = CandidateBinding({})
1664 1a7c1261302c logic fix- some bindings were being returned 2+; some 0 times drewp@bigasterisk.com parents: 1661 diff changeset	39 for rt, ct in zip(self.ruleChunk._allTerms(), self.workingSetChunk._allTerms()):
1673 80f4e741ca4f redo RHS bnode processing drewp@bigasterisk.com parents: 1669 diff changeset	40 if isinstance(rt, (Variable, RuleUnboundBnode)):
1668 89e53cb8a01c fix some harder tests. Mostly, _advanceTheStack needed to spin the odometer rings starting from the other side, to get all the right combos drewp@bigasterisk.com parents: 1664 diff changeset	41 if prevBindings.contains(rt) and prevBindings.applyTerm(rt) != ct:
89e53cb8a01c fix some harder tests. Mostly, _advanceTheStack needed to spin the odometer rings starting from the other side, to get all the right combos drewp@bigasterisk.com parents: 1664 diff changeset	42 msg = f'{rt=} {ct=} {prevBindings=}' if log.isEnabledFor(logging.DEBUG) else ''
89e53cb8a01c fix some harder tests. Mostly, _advanceTheStack needed to spin the odometer rings starting from the other side, to get all the right combos drewp@bigasterisk.com parents: 1664 diff changeset	43 raise Inconsistent(msg)
1664 1a7c1261302c logic fix- some bindings were being returned 2+; some 0 times drewp@bigasterisk.com parents: 1661 diff changeset	44 if outBinding.contains(rt) and outBinding.applyTerm(rt) != ct:
1668 89e53cb8a01c fix some harder tests. Mostly, _advanceTheStack needed to spin the odometer rings starting from the other side, to get all the right combos drewp@bigasterisk.com parents: 1664 diff changeset	45 # maybe this can happen, for stmts like ?x :a ?x .
89e53cb8a01c fix some harder tests. Mostly, _advanceTheStack needed to spin the odometer rings starting from the other side, to get all the right combos drewp@bigasterisk.com parents: 1664 diff changeset	46 raise Inconsistent("outBinding inconsistent with itself")
1664 1a7c1261302c logic fix- some bindings were being returned 2+; some 0 times drewp@bigasterisk.com parents: 1661 diff changeset	47 outBinding.addNewBindings(CandidateBinding({rt: ct}))
1668 89e53cb8a01c fix some harder tests. Mostly, _advanceTheStack needed to spin the odometer rings starting from the other side, to get all the right combos drewp@bigasterisk.com parents: 1664 diff changeset	48 else:
89e53cb8a01c fix some harder tests. Mostly, _advanceTheStack needed to spin the odometer rings starting from the other side, to get all the right combos drewp@bigasterisk.com parents: 1664 diff changeset	49 if rt != ct:
89e53cb8a01c fix some harder tests. Mostly, _advanceTheStack needed to spin the odometer rings starting from the other side, to get all the right combos drewp@bigasterisk.com parents: 1664 diff changeset	50 # getting here means prevBindings was set to something our
89e53cb8a01c fix some harder tests. Mostly, _advanceTheStack needed to spin the odometer rings starting from the other side, to get all the right combos drewp@bigasterisk.com parents: 1664 diff changeset	51 # rule statement disagrees with.
89e53cb8a01c fix some harder tests. Mostly, _advanceTheStack needed to spin the odometer rings starting from the other side, to get all the right combos drewp@bigasterisk.com parents: 1664 diff changeset	52 raise Inconsistent(f'{rt=} != {ct=}')
1664 1a7c1261302c logic fix- some bindings were being returned 2+; some 0 times drewp@bigasterisk.com parents: 1661 diff changeset	53 return outBinding
1a7c1261302c logic fix- some bindings were being returned 2+; some 0 times drewp@bigasterisk.com parents: 1661 diff changeset	54
1a7c1261302c logic fix- some bindings were being returned 2+; some 0 times drewp@bigasterisk.com parents: 1661 diff changeset	55 def matches(self) -> bool:
1a7c1261302c logic fix- some bindings were being returned 2+; some 0 times drewp@bigasterisk.com parents: 1661 diff changeset	56 """could this rule, with its BindableTerm wildcards, match workingSetChunk?"""
1a7c1261302c logic fix- some bindings were being returned 2+; some 0 times drewp@bigasterisk.com parents: 1661 diff changeset	57 for selfTerm, otherTerm in zip(self.ruleChunk._allTerms(), self.workingSetChunk._allTerms()):
1673 80f4e741ca4f redo RHS bnode processing drewp@bigasterisk.com parents: 1669 diff changeset	58 if not isinstance(selfTerm, (Variable, RuleUnboundBnode)) and selfTerm != otherTerm:
1664 1a7c1261302c logic fix- some bindings were being returned 2+; some 0 times drewp@bigasterisk.com parents: 1661 diff changeset	59 return False
1694 73abfd4cf5d0 new html log and other refactoring as i work on the advanceTheStack problems drewp@bigasterisk.com parents: 1677 diff changeset	60
1664 1a7c1261302c logic fix- some bindings were being returned 2+; some 0 times drewp@bigasterisk.com parents: 1661 diff changeset	61 return True
1a7c1261302c logic fix- some bindings were being returned 2+; some 0 times drewp@bigasterisk.com parents: 1661 diff changeset	62
1a7c1261302c logic fix- some bindings were being returned 2+; some 0 times drewp@bigasterisk.com parents: 1661 diff changeset	63
1a7c1261302c logic fix- some bindings were being returned 2+; some 0 times drewp@bigasterisk.com parents: 1661 diff changeset	64 @dataclass
1651 20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	65 class Chunk: # rename this
1661 00a5624d1d14 cleanups and optimizations drewp@bigasterisk.com parents: 1660 diff changeset	66 """A statement, maybe with variables in it, except *the subject or object
00a5624d1d14 cleanups and optimizations drewp@bigasterisk.com parents: 1660 diff changeset	67 can be rdf lists*. This is done to optimize list comparisons (a lot) at the
00a5624d1d14 cleanups and optimizations drewp@bigasterisk.com parents: 1660 diff changeset	68 very minor expense of not handling certain exotic cases, such as a branching
00a5624d1d14 cleanups and optimizations drewp@bigasterisk.com parents: 1660 diff changeset	69 list.
1651 20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	70
1661 00a5624d1d14 cleanups and optimizations drewp@bigasterisk.com parents: 1660 diff changeset	71 Example: (?x ?y) math:sum ?z . <-- this becomes one Chunk.
1651 20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	72
1661 00a5624d1d14 cleanups and optimizations drewp@bigasterisk.com parents: 1660 diff changeset	73 A function call in a rule is always contained in exactly one chunk.
00a5624d1d14 cleanups and optimizations drewp@bigasterisk.com parents: 1660 diff changeset	74
00a5624d1d14 cleanups and optimizations drewp@bigasterisk.com parents: 1660 diff changeset	75 https://www.w3.org/TeamSubmission/n3/#:~:text=Implementations%20may%20treat%20list%20as%20a%20data%20type
1651 20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	76 """
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	77 # all immutable
1660 31f7dab6a60b function evaluation uses Chunk lists now and runs fast. Only a few edge cases still broken drewp@bigasterisk.com parents: 1659 diff changeset	78 primary: ChunkPrimaryTriple
1653 e7d594c065d4 minor refactoring drewp@bigasterisk.com parents: 1652 diff changeset	79 subjList: Optional[List[Node]] = None
e7d594c065d4 minor refactoring drewp@bigasterisk.com parents: 1652 diff changeset	80 objList: Optional[List[Node]] = None
1651 20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	81
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	82 def __post_init__(self):
1659 15e84c71beee parse lists from graph into the Chunks drewp@bigasterisk.com parents: 1654 diff changeset	83 if not (((self.primary[0] is not None) ^ (self.subjList is not None)) and
15e84c71beee parse lists from graph into the Chunks drewp@bigasterisk.com parents: 1654 diff changeset	84 ((self.primary[2] is not None) ^ (self.objList is not None))):
15e84c71beee parse lists from graph into the Chunks drewp@bigasterisk.com parents: 1654 diff changeset	85 raise TypeError("invalid chunk init")
1651 20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	86 self.predicate = self.primary[1]
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	87 self.sortKey = (self.primary, tuple(self.subjList or []), tuple(self.objList or []))
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	88
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	89 def __hash__(self):
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	90 return hash(self.sortKey)
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	91
1664 1a7c1261302c logic fix- some bindings were being returned 2+; some 0 times drewp@bigasterisk.com parents: 1661 diff changeset	92 def __lt__(self, other):
1a7c1261302c logic fix- some bindings were being returned 2+; some 0 times drewp@bigasterisk.com parents: 1661 diff changeset	93 return self.sortKey < other.sortKey
1651 20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	94
1660 31f7dab6a60b function evaluation uses Chunk lists now and runs fast. Only a few edge cases still broken drewp@bigasterisk.com parents: 1659 diff changeset	95 def _allTerms(self) -> Iterator[Node]:
31f7dab6a60b function evaluation uses Chunk lists now and runs fast. Only a few edge cases still broken drewp@bigasterisk.com parents: 1659 diff changeset	96 """the terms in `primary` plus the lists. Output order is undefined but stable between same-sized Chunks"""
31f7dab6a60b function evaluation uses Chunk lists now and runs fast. Only a few edge cases still broken drewp@bigasterisk.com parents: 1659 diff changeset	97 yield self.primary[1]
31f7dab6a60b function evaluation uses Chunk lists now and runs fast. Only a few edge cases still broken drewp@bigasterisk.com parents: 1659 diff changeset	98 if self.primary[0] is not None:
31f7dab6a60b function evaluation uses Chunk lists now and runs fast. Only a few edge cases still broken drewp@bigasterisk.com parents: 1659 diff changeset	99 yield self.primary[0]
31f7dab6a60b function evaluation uses Chunk lists now and runs fast. Only a few edge cases still broken drewp@bigasterisk.com parents: 1659 diff changeset	100 else:
31f7dab6a60b function evaluation uses Chunk lists now and runs fast. Only a few edge cases still broken drewp@bigasterisk.com parents: 1659 diff changeset	101 yield from cast(List[Node], self.subjList)
31f7dab6a60b function evaluation uses Chunk lists now and runs fast. Only a few edge cases still broken drewp@bigasterisk.com parents: 1659 diff changeset	102 if self.primary[2] is not None:
31f7dab6a60b function evaluation uses Chunk lists now and runs fast. Only a few edge cases still broken drewp@bigasterisk.com parents: 1659 diff changeset	103 yield self.primary[2]
31f7dab6a60b function evaluation uses Chunk lists now and runs fast. Only a few edge cases still broken drewp@bigasterisk.com parents: 1659 diff changeset	104 else:
31f7dab6a60b function evaluation uses Chunk lists now and runs fast. Only a few edge cases still broken drewp@bigasterisk.com parents: 1659 diff changeset	105 yield from cast(List[Node], self.objList)
1651 20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	106
1664 1a7c1261302c logic fix- some bindings were being returned 2+; some 0 times drewp@bigasterisk.com parents: 1661 diff changeset	107 def ruleMatchesFrom(self, workingSet: 'ChunkedGraph') -> Iterator[AlignedRuleChunk]:
1a7c1261302c logic fix- some bindings were being returned 2+; some 0 times drewp@bigasterisk.com parents: 1661 diff changeset	108 """Chunks from workingSet where self, which may have BindableTerm wildcards, could match that workingSet Chunk."""
1a7c1261302c logic fix- some bindings were being returned 2+; some 0 times drewp@bigasterisk.com parents: 1661 diff changeset	109 # if log.isEnabledFor(logging.DEBUG):
1a7c1261302c logic fix- some bindings were being returned 2+; some 0 times drewp@bigasterisk.com parents: 1661 diff changeset	110 # log.debug(f'{INDENT*6} computing {self}.ruleMatchesFrom({workingSet}')
1a7c1261302c logic fix- some bindings were being returned 2+; some 0 times drewp@bigasterisk.com parents: 1661 diff changeset	111 allChunksIter = workingSet.allChunks()
1697 88f6e9bf69d1 stats and non-debug mode speedups drewp@bigasterisk.com parents: 1694 diff changeset	112 if log.isEnabledFor(logging.DEBUG): # makes failures a bit more stable, but shows up in profiling
1664 1a7c1261302c logic fix- some bindings were being returned 2+; some 0 times drewp@bigasterisk.com parents: 1661 diff changeset	113 allChunksIter = sorted(allChunksIter)
1a7c1261302c logic fix- some bindings were being returned 2+; some 0 times drewp@bigasterisk.com parents: 1661 diff changeset	114 for chunk in allChunksIter:
1668 89e53cb8a01c fix some harder tests. Mostly, _advanceTheStack needed to spin the odometer rings starting from the other side, to get all the right combos drewp@bigasterisk.com parents: 1664 diff changeset	115 try:
89e53cb8a01c fix some harder tests. Mostly, _advanceTheStack needed to spin the odometer rings starting from the other side, to get all the right combos drewp@bigasterisk.com parents: 1664 diff changeset	116 aligned = AlignedRuleChunk(self, chunk)
89e53cb8a01c fix some harder tests. Mostly, _advanceTheStack needed to spin the odometer rings starting from the other side, to get all the right combos drewp@bigasterisk.com parents: 1664 diff changeset	117 except Inconsistent:
89e53cb8a01c fix some harder tests. Mostly, _advanceTheStack needed to spin the odometer rings starting from the other side, to get all the right combos drewp@bigasterisk.com parents: 1664 diff changeset	118 continue
89e53cb8a01c fix some harder tests. Mostly, _advanceTheStack needed to spin the odometer rings starting from the other side, to get all the right combos drewp@bigasterisk.com parents: 1664 diff changeset	119 yield aligned
1651 20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	120
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	121 def __repr__(self):
1677 aa35ae7a1acc add new bug test (no fix yet) drewp@bigasterisk.com parents: 1673 diff changeset	122 pre = ('+'.join(repr(elem) for elem in self.subjList) + '+' if self.subjList else '')
aa35ae7a1acc add new bug test (no fix yet) drewp@bigasterisk.com parents: 1673 diff changeset	123 post = ('+' + '+'.join(repr(elem) for elem in self.objList) if self.objList else '')
1659 15e84c71beee parse lists from graph into the Chunks drewp@bigasterisk.com parents: 1654 diff changeset	124 return pre + repr(self.primary) + post
1651 20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	125
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	126 def isFunctionCall(self, functionsFor) -> bool:
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	127 return bool(list(functionsFor(cast(URIRef, self.predicate))))
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	128
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	129 def isStatic(self) -> bool:
1660 31f7dab6a60b function evaluation uses Chunk lists now and runs fast. Only a few edge cases still broken drewp@bigasterisk.com parents: 1659 diff changeset	130 return all(_termIsStatic(s) for s in self._allTerms())
31f7dab6a60b function evaluation uses Chunk lists now and runs fast. Only a few edge cases still broken drewp@bigasterisk.com parents: 1659 diff changeset	131
1669 9d00adef0b22 rm used parameter drewp@bigasterisk.com parents: 1668 diff changeset	132 def apply(self, cb: CandidateBinding) -> 'Chunk':
1660 31f7dab6a60b function evaluation uses Chunk lists now and runs fast. Only a few edge cases still broken drewp@bigasterisk.com parents: 1659 diff changeset	133 """Chunk like this one but with cb substitutions applied. If the flag is
31f7dab6a60b function evaluation uses Chunk lists now and runs fast. Only a few edge cases still broken drewp@bigasterisk.com parents: 1659 diff changeset	134 True, we raise BindingUnknown instead of leaving a term unbound"""
1669 9d00adef0b22 rm used parameter drewp@bigasterisk.com parents: 1668 diff changeset	135 fn = lambda t: cb.applyTerm(t, failUnbound=False)
1660 31f7dab6a60b function evaluation uses Chunk lists now and runs fast. Only a few edge cases still broken drewp@bigasterisk.com parents: 1659 diff changeset	136 return Chunk(
31f7dab6a60b function evaluation uses Chunk lists now and runs fast. Only a few edge cases still broken drewp@bigasterisk.com parents: 1659 diff changeset	137 (
31f7dab6a60b function evaluation uses Chunk lists now and runs fast. Only a few edge cases still broken drewp@bigasterisk.com parents: 1659 diff changeset	138 fn(self.primary[0]) if self.primary[0] is not None else None, #
31f7dab6a60b function evaluation uses Chunk lists now and runs fast. Only a few edge cases still broken drewp@bigasterisk.com parents: 1659 diff changeset	139 fn(self.primary[1]), #
31f7dab6a60b function evaluation uses Chunk lists now and runs fast. Only a few edge cases still broken drewp@bigasterisk.com parents: 1659 diff changeset	140 fn(self.primary[2]) if self.primary[2] is not None else None),
31f7dab6a60b function evaluation uses Chunk lists now and runs fast. Only a few edge cases still broken drewp@bigasterisk.com parents: 1659 diff changeset	141 subjList=[fn(t) for t in self.subjList] if self.subjList else None,
31f7dab6a60b function evaluation uses Chunk lists now and runs fast. Only a few edge cases still broken drewp@bigasterisk.com parents: 1659 diff changeset	142 objList=[fn(t) for t in self.objList] if self.objList else None,
31f7dab6a60b function evaluation uses Chunk lists now and runs fast. Only a few edge cases still broken drewp@bigasterisk.com parents: 1659 diff changeset	143 )
1651 20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	144
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	145
1660 31f7dab6a60b function evaluation uses Chunk lists now and runs fast. Only a few edge cases still broken drewp@bigasterisk.com parents: 1659 diff changeset	146 def _termIsStatic(term: Optional[Node]) -> bool:
1659 15e84c71beee parse lists from graph into the Chunks drewp@bigasterisk.com parents: 1654 diff changeset	147 return isinstance(term, (URIRef, Literal)) or term is None
1651 20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	148
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	149
1694 73abfd4cf5d0 new html log and other refactoring as i work on the advanceTheStack problems drewp@bigasterisk.com parents: 1677 diff changeset	150 def applyChunky(cb: CandidateBinding, g: Iterable[AlignedRuleChunk]) -> Iterator[AlignedRuleChunk]:
1664 1a7c1261302c logic fix- some bindings were being returned 2+; some 0 times drewp@bigasterisk.com parents: 1661 diff changeset	151 for aligned in g:
1669 9d00adef0b22 rm used parameter drewp@bigasterisk.com parents: 1668 diff changeset	152 bound = aligned.ruleChunk.apply(cb)
1651 20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	153 try:
1668 89e53cb8a01c fix some harder tests. Mostly, _advanceTheStack needed to spin the odometer rings starting from the other side, to get all the right combos drewp@bigasterisk.com parents: 1664 diff changeset	154 yield AlignedRuleChunk(bound, aligned.workingSetChunk)
89e53cb8a01c fix some harder tests. Mostly, _advanceTheStack needed to spin the odometer rings starting from the other side, to get all the right combos drewp@bigasterisk.com parents: 1664 diff changeset	155 except Inconsistent:
89e53cb8a01c fix some harder tests. Mostly, _advanceTheStack needed to spin the odometer rings starting from the other side, to get all the right combos drewp@bigasterisk.com parents: 1664 diff changeset	156 pass
1651 20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	157
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	158
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	159 class ChunkedGraph:
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	160 """a Graph converts 1-to-1 with a ChunkedGraph, where the Chunks have
1652 dddfa09ea0b9 debug logging and comments drewp@bigasterisk.com parents: 1651 diff changeset	161 combined some statements together. (The only exception is that bnodes for
1651 20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	162 rdf lists are lost)"""
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	163
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	164 def __init__(
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	165 self,
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	166 graph: Graph,
1673 80f4e741ca4f redo RHS bnode processing drewp@bigasterisk.com parents: 1669 diff changeset	167 bnodeType: Union[Type[RuleUnboundBnode], Type[WorkingSetBnode]],
1651 20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	168 functionsFor # get rid of this- i'm just working around a circular import
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	169 ):
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	170 self.chunksUsedByFuncs: Set[Chunk] = set()
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	171 self.staticChunks: Set[Chunk] = set()
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	172 self.patternChunks: Set[Chunk] = set()
1659 15e84c71beee parse lists from graph into the Chunks drewp@bigasterisk.com parents: 1654 diff changeset	173
15e84c71beee parse lists from graph into the Chunks drewp@bigasterisk.com parents: 1654 diff changeset	174 firstNodes = {}
15e84c71beee parse lists from graph into the Chunks drewp@bigasterisk.com parents: 1654 diff changeset	175 restNodes = {}
15e84c71beee parse lists from graph into the Chunks drewp@bigasterisk.com parents: 1654 diff changeset	176 graphStmts = set()
15e84c71beee parse lists from graph into the Chunks drewp@bigasterisk.com parents: 1654 diff changeset	177 for s, p, o in graph:
15e84c71beee parse lists from graph into the Chunks drewp@bigasterisk.com parents: 1654 diff changeset	178 if p == RDF['first']:
15e84c71beee parse lists from graph into the Chunks drewp@bigasterisk.com parents: 1654 diff changeset	179 firstNodes[s] = o
15e84c71beee parse lists from graph into the Chunks drewp@bigasterisk.com parents: 1654 diff changeset	180 elif p == RDF['rest']:
15e84c71beee parse lists from graph into the Chunks drewp@bigasterisk.com parents: 1654 diff changeset	181 restNodes[s] = o
15e84c71beee parse lists from graph into the Chunks drewp@bigasterisk.com parents: 1654 diff changeset	182 else:
15e84c71beee parse lists from graph into the Chunks drewp@bigasterisk.com parents: 1654 diff changeset	183 graphStmts.add((s, p, o))
15e84c71beee parse lists from graph into the Chunks drewp@bigasterisk.com parents: 1654 diff changeset	184
15e84c71beee parse lists from graph into the Chunks drewp@bigasterisk.com parents: 1654 diff changeset	185 def gatherList(start):
15e84c71beee parse lists from graph into the Chunks drewp@bigasterisk.com parents: 1654 diff changeset	186 lst = []
15e84c71beee parse lists from graph into the Chunks drewp@bigasterisk.com parents: 1654 diff changeset	187 cur = start
15e84c71beee parse lists from graph into the Chunks drewp@bigasterisk.com parents: 1654 diff changeset	188 while cur != RDF['nil']:
15e84c71beee parse lists from graph into the Chunks drewp@bigasterisk.com parents: 1654 diff changeset	189 lst.append(firstNodes[cur])
15e84c71beee parse lists from graph into the Chunks drewp@bigasterisk.com parents: 1654 diff changeset	190 cur = restNodes[cur]
15e84c71beee parse lists from graph into the Chunks drewp@bigasterisk.com parents: 1654 diff changeset	191 return lst
15e84c71beee parse lists from graph into the Chunks drewp@bigasterisk.com parents: 1654 diff changeset	192
15e84c71beee parse lists from graph into the Chunks drewp@bigasterisk.com parents: 1654 diff changeset	193 for s, p, o in graphStmts:
15e84c71beee parse lists from graph into the Chunks drewp@bigasterisk.com parents: 1654 diff changeset	194 subjList = objList = None
15e84c71beee parse lists from graph into the Chunks drewp@bigasterisk.com parents: 1654 diff changeset	195 if s in firstNodes:
15e84c71beee parse lists from graph into the Chunks drewp@bigasterisk.com parents: 1654 diff changeset	196 subjList = gatherList(s)
15e84c71beee parse lists from graph into the Chunks drewp@bigasterisk.com parents: 1654 diff changeset	197 s = None
15e84c71beee parse lists from graph into the Chunks drewp@bigasterisk.com parents: 1654 diff changeset	198 if o in firstNodes:
15e84c71beee parse lists from graph into the Chunks drewp@bigasterisk.com parents: 1654 diff changeset	199 objList = gatherList(o)
15e84c71beee parse lists from graph into the Chunks drewp@bigasterisk.com parents: 1654 diff changeset	200 o = None
1673 80f4e741ca4f redo RHS bnode processing drewp@bigasterisk.com parents: 1669 diff changeset	201 from rdflib import BNode
1694 73abfd4cf5d0 new html log and other refactoring as i work on the advanceTheStack problems drewp@bigasterisk.com parents: 1677 diff changeset	202 if isinstance(s, BNode):
73abfd4cf5d0 new html log and other refactoring as i work on the advanceTheStack problems drewp@bigasterisk.com parents: 1677 diff changeset	203 s = bnodeType(s)
73abfd4cf5d0 new html log and other refactoring as i work on the advanceTheStack problems drewp@bigasterisk.com parents: 1677 diff changeset	204 if isinstance(p, BNode):
73abfd4cf5d0 new html log and other refactoring as i work on the advanceTheStack problems drewp@bigasterisk.com parents: 1677 diff changeset	205 p = bnodeType(p)
73abfd4cf5d0 new html log and other refactoring as i work on the advanceTheStack problems drewp@bigasterisk.com parents: 1677 diff changeset	206 if isinstance(o, BNode):
73abfd4cf5d0 new html log and other refactoring as i work on the advanceTheStack problems drewp@bigasterisk.com parents: 1677 diff changeset	207 o = bnodeType(o)
1673 80f4e741ca4f redo RHS bnode processing drewp@bigasterisk.com parents: 1669 diff changeset	208
1659 15e84c71beee parse lists from graph into the Chunks drewp@bigasterisk.com parents: 1654 diff changeset	209 c = Chunk((s, p, o), subjList=subjList, objList=objList)
15e84c71beee parse lists from graph into the Chunks drewp@bigasterisk.com parents: 1654 diff changeset	210
1651 20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	211 if c.isFunctionCall(functionsFor):
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	212 self.chunksUsedByFuncs.add(c)
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	213 elif c.isStatic():
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	214 self.staticChunks.add(c)
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	215 else:
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	216 self.patternChunks.add(c)
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	217
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	218 def allPredicatesExceptFunctions(self) -> Set[Node]:
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	219 return set(ch.predicate for ch in itertools.chain(self.staticChunks, self.patternChunks))
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	220
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	221 def noPredicatesAppear(self, preds: Iterable[Node]) -> bool:
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	222 return self.allPredicatesExceptFunctions().isdisjoint(preds)
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	223
1654 d47832373b34 __nonzero__ is called __bool__ in py3! thanks for nothing, linters drewp@bigasterisk.com parents: 1653 diff changeset	224 def __bool__(self):
1651 20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	225 return bool(self.chunksUsedByFuncs) or bool(self.staticChunks) or bool(self.patternChunks)
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	226
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	227 def __repr__(self):
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	228 return f'ChunkedGraph({self.__dict__})'
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	229
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	230 def allChunks(self) -> Iterable[Chunk]:
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	231 yield from itertools.chain(self.staticChunks, self.patternChunks, self.chunksUsedByFuncs)
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	232
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	233 def __contains__(self, ch: Chunk) -> bool:
20474ad4968e WIP - functions are broken as i move most layers to work in Chunks not Triples drewp@bigasterisk.com parents: diff changeset	234 return ch in self.allChunks()

Mercurial > code > home > repos > homeauto

annotate service/mqtt_to_rdf/inference/stmt_chunk.py @ 1727:23e6154e6c11