annotate service/reasoning/inputgraph.py @ 720:e157afd642b5

rewrite reasoning PutOutputs Ignore-this: 9c7c4b67b1f42992920572d147544a4f
author drewp@bigasterisk.com
date Wed, 05 Feb 2020 00:19:43 -0800
parents 53a2664f450a
children b87b6e9cedb2
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
303
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
1 import logging, time, sys
275
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
2
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
3 from rdflib import Graph, ConjunctiveGraph
303
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
4 from rdflib import Namespace, URIRef, Literal, RDF, RDFS
281
9728288c7f2f refactor
drewp@bigasterisk.com
parents: 275
diff changeset
5 from rdflib.parser import StringInputSource
275
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
6
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
7 from twisted.python.filepath import FilePath
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
8 from twisted.internet.defer import inlineCallbacks, gatherResults
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
9
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
10 from rdflibtrig import addTrig
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
11 from graphop import graphEqual
392
79d041273e26 mqtt has two devices now. various older cleanups.
drewp@bigasterisk.com
parents: 351
diff changeset
12 from greplin import scales
275
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
13
571
53a2664f450a build/import update for reasoning service
drewp@bigasterisk.com
parents: 463
diff changeset
14 from patchablegraph.patchsource import ReconnectingPatchSource
303
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
15
351
7716b1810d6c reasoning & collector move into docker images
drewp@bigasterisk.com
parents: 314
diff changeset
16 from rdfdb.rdflibpatch import patchQuads
303
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
17
275
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
18 log = logging.getLogger('fetch')
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
19
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
20 ROOM = Namespace("http://projects.bigasterisk.com/room/")
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
21 DEV = Namespace("http://projects.bigasterisk.com/device/")
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
22
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
23
392
79d041273e26 mqtt has two devices now. various older cleanups.
drewp@bigasterisk.com
parents: 351
diff changeset
24 STATS = scales.collection('/web',
79d041273e26 mqtt has two devices now. various older cleanups.
drewp@bigasterisk.com
parents: 351
diff changeset
25 scales.PmfStat('combineGraph'),
79d041273e26 mqtt has two devices now. various older cleanups.
drewp@bigasterisk.com
parents: 351
diff changeset
26 )
281
9728288c7f2f refactor
drewp@bigasterisk.com
parents: 275
diff changeset
27 def parseRdf(text, contentType):
9728288c7f2f refactor
drewp@bigasterisk.com
parents: 275
diff changeset
28 g = Graph()
9728288c7f2f refactor
drewp@bigasterisk.com
parents: 275
diff changeset
29 g.parse(StringInputSource(text), format={
9728288c7f2f refactor
drewp@bigasterisk.com
parents: 275
diff changeset
30 'text/n3': 'n3',
9728288c7f2f refactor
drewp@bigasterisk.com
parents: 275
diff changeset
31 }[contentType])
9728288c7f2f refactor
drewp@bigasterisk.com
parents: 275
diff changeset
32 return g
9728288c7f2f refactor
drewp@bigasterisk.com
parents: 275
diff changeset
33
9728288c7f2f refactor
drewp@bigasterisk.com
parents: 275
diff changeset
34
303
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
35 class RemoteData(object):
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
36 def __init__(self, onChange):
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
37 self.onChange = onChange
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
38 self.graph = ConjunctiveGraph()
392
79d041273e26 mqtt has two devices now. various older cleanups.
drewp@bigasterisk.com
parents: 351
diff changeset
39 self.patchSource = ReconnectingPatchSource(
79d041273e26 mqtt has two devices now. various older cleanups.
drewp@bigasterisk.com
parents: 351
diff changeset
40 URIRef('http://bang:9072/graph/home'),
79d041273e26 mqtt has two devices now. various older cleanups.
drewp@bigasterisk.com
parents: 351
diff changeset
41 #URIRef('http://frontdoor:10012/graph/events'),
463
1ceb26846eca add separate :matchPredicate support. some build and log edits.
drewp@bigasterisk.com
parents: 392
diff changeset
42 self.onPatch, reconnectSecs=10, agent='reasoning')
303
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
43
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
44 def onPatch(self, p, fullGraph):
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
45 if fullGraph:
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
46 self.graph = ConjunctiveGraph()
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
47 patchQuads(self.graph,
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
48 deleteQuads=p.delQuads,
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
49 addQuads=p.addQuads,
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
50 perfect=True)
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
51
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
52 ignorePredicates = [
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
53 ROOM['signalStrength'],
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
54 # perhaps anything with a number-datatype for its
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
55 # object should be filtered out, and you have to make
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
56 # an upstream quantization (e.g. 'temp high'/'temp
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
57 # low') if you want to do reasoning on the difference
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
58 URIRef("http://bigasterisk.com/map#lastSeenAgoSec"),
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
59 URIRef("http://bigasterisk.com/map#lastSeenAgo"),
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
60 ROOM['usingPower'],
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
61 ROOM['idleTimeMinutes'],
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
62 ROOM['idleTimeMs'],
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
63 ROOM['graphLoadMs'],
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
64 ROOM['localTimeToSecond'],
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
65 ROOM['history'],
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
66 ROOM['connectedAgo'],
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
67 RDFS['comment'],
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
68 ]
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
69 ignoreContexts = [
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
70 URIRef('http://bigasterisk.com/sse_collector/'),
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
71 ]
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
72 for affected in p.addQuads + p.delQuads:
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
73 if (affected[1] not in ignorePredicates and
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
74 affected[3] not in ignoreContexts):
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
75 log.debug(" remote graph changed")
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
76 self.onChange()
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
77 break
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
78 else:
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
79 log.debug(" remote graph has no changes to trigger rules")
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
80
275
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
81 class InputGraph(object):
303
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
82 def __init__(self, inputDirs, onChange):
275
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
83 """
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
84 this has one Graph that's made of:
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
85 - all .n3 files from inputDirs (read at startup)
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
86 - all the remote graphs, specified in the file graphs
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
87
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
88 call updateFileData or updateRemoteData to reread those
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
89 graphs. getGraph to access the combined graph.
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
90
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
91 onChange(self) is called if the contents of the full graph
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
92 change (in an interesting way) during updateFileData or
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
93 updateRemoteData. Interesting means statements other than the
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
94 ones with the predicates on the boring list. onChange(self,
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
95 oneShot=True) means: don't store the result of this change
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
96 anywhere; it needs to be processed only once
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
97 """
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
98 self.inputDirs = inputDirs
312
170dc9b1e789 fix input graph web display by dirtying combinedGraph better.
drewp@bigasterisk.com
parents: 303
diff changeset
99 self._onChange = onChange
275
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
100 self._fileGraph = Graph()
312
170dc9b1e789 fix input graph web display by dirtying combinedGraph better.
drewp@bigasterisk.com
parents: 303
diff changeset
101 self._remoteData = RemoteData(lambda: self.onChangeLocal())
275
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
102 self._combinedGraph = None
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
103 self._oneShotAdditionGraph = None
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
104
312
170dc9b1e789 fix input graph web display by dirtying combinedGraph better.
drewp@bigasterisk.com
parents: 303
diff changeset
105 def onChangeLocal(self, oneShot=False, oneShotGraph=None):
170dc9b1e789 fix input graph web display by dirtying combinedGraph better.
drewp@bigasterisk.com
parents: 303
diff changeset
106 self._combinedGraph = None
170dc9b1e789 fix input graph web display by dirtying combinedGraph better.
drewp@bigasterisk.com
parents: 303
diff changeset
107 self._onChange(self, oneShot=oneShot, oneShotGraph=oneShotGraph)
170dc9b1e789 fix input graph web display by dirtying combinedGraph better.
drewp@bigasterisk.com
parents: 303
diff changeset
108
275
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
109 def updateFileData(self):
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
110 """
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
111 make sure we contain the correct data from the files in inputDirs
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
112 """
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
113 # this sample one is actually only needed for the output, but I don't
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
114 # think I want to have a separate graph for the output
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
115 # handling
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
116 log.debug("read file graphs")
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
117 for fp in FilePath("input").walk():
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
118 if fp.isdir():
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
119 continue
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
120 if fp.splitext()[1] != '.n3':
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
121 continue
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
122 log.debug("read %s", fp)
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
123 # todo: if this fails, leave the report in the graph
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
124 self._fileGraph.parse(fp.open(), format="n3")
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
125 self._combinedGraph = None
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
126
312
170dc9b1e789 fix input graph web display by dirtying combinedGraph better.
drewp@bigasterisk.com
parents: 303
diff changeset
127 self.onChangeLocal()
275
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
128
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
129 def addOneShot(self, g):
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
130 """
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
131 add this graph to the total, call onChange, and then revert
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
132 the addition of this graph
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
133 """
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
134 self._oneShotAdditionGraph = g
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
135 self._combinedGraph = None
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
136 try:
314
c51075f267bc fix oneshot
drewp@bigasterisk.com
parents: 312
diff changeset
137 self.onChangeLocal(oneShot=True, oneShotGraph=g)
275
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
138 finally:
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
139 self._oneShotAdditionGraph = None
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
140 self._combinedGraph = None
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
141
281
9728288c7f2f refactor
drewp@bigasterisk.com
parents: 275
diff changeset
142 def addOneShotFromString(self, body, contentType):
9728288c7f2f refactor
drewp@bigasterisk.com
parents: 275
diff changeset
143 g = parseRdf(body, contentType)
9728288c7f2f refactor
drewp@bigasterisk.com
parents: 275
diff changeset
144 if not len(g):
9728288c7f2f refactor
drewp@bigasterisk.com
parents: 275
diff changeset
145 log.warn("incoming oneshot graph had no statements: %r", body)
9728288c7f2f refactor
drewp@bigasterisk.com
parents: 275
diff changeset
146 return 0
9728288c7f2f refactor
drewp@bigasterisk.com
parents: 275
diff changeset
147 t1 = time.time()
9728288c7f2f refactor
drewp@bigasterisk.com
parents: 275
diff changeset
148 self.addOneShot(g)
9728288c7f2f refactor
drewp@bigasterisk.com
parents: 275
diff changeset
149 return time.time() - t1
392
79d041273e26 mqtt has two devices now. various older cleanups.
drewp@bigasterisk.com
parents: 351
diff changeset
150
79d041273e26 mqtt has two devices now. various older cleanups.
drewp@bigasterisk.com
parents: 351
diff changeset
151 @STATS.combineGraph.time()
275
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
152 def getGraph(self):
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
153 """rdflib Graph with the file+remote contents of the input graph"""
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
154 # this could be much faster with the combined readonly graph
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
155 # view from rdflib
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
156 if self._combinedGraph is None:
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
157 self._combinedGraph = Graph()
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
158 if self._fileGraph:
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
159 for s in self._fileGraph:
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
160 self._combinedGraph.add(s)
303
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
161 for s in self._remoteData.graph:
66fe7a93753d reasoning uses sse_collector
drewp@bigasterisk.com
parents: 281
diff changeset
162 self._combinedGraph.add(s)
275
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
163 if self._oneShotAdditionGraph:
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
164 for s in self._oneShotAdditionGraph:
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
165 self._combinedGraph.add(s)
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
166
d3733587e749 refactor inputgraph
drewp@bigasterisk.com
parents:
diff changeset
167 return self._combinedGraph