annotate service/reasoning/inputgraph.py @ 1080:4d16fa39d54a

refactor inputgraph Ignore-this: 9931e669c180d8141a51fb6f7927db0a darcs-hash:3b3358d6aec2e6242eba7e73c9d753b267ac0b85
author drewp <drewp@bigasterisk.com>
date Fri, 06 May 2016 15:42:04 -0700
parents
children 9728288c7f2f
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
1080
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
1 import logging, time
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
2
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
3 from rdflib import Graph, ConjunctiveGraph
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
4 from rdflib import Namespace, URIRef, Literal, RDF
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
5
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
6 from twisted.python.filepath import FilePath
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
7 from twisted.internet.defer import inlineCallbacks, gatherResults
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
8
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
9 from rdflibtrig import addTrig
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
10 from graphop import graphEqual
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
11
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
12 log = logging.getLogger('fetch')
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
13
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
14 ROOM = Namespace("http://projects.bigasterisk.com/room/")
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
15 DEV = Namespace("http://projects.bigasterisk.com/device/")
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
16
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
17
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
18 class InputGraph(object):
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
19 def __init__(self, inputDirs, onChange, sourceSubstr=None):
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
20 """
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
21 this has one Graph that's made of:
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
22 - all .n3 files from inputDirs (read at startup)
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
23 - all the remote graphs, specified in the file graphs
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
24
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
25 call updateFileData or updateRemoteData to reread those
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
26 graphs. getGraph to access the combined graph.
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
27
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
28 onChange(self) is called if the contents of the full graph
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
29 change (in an interesting way) during updateFileData or
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
30 updateRemoteData. Interesting means statements other than the
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
31 ones with the predicates on the boring list. onChange(self,
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
32 oneShot=True) means: don't store the result of this change
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
33 anywhere; it needs to be processed only once
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
34
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
35 sourceSubstr filters to only pull from sources containing the
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
36 string (for debugging).
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
37 """
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
38 self.inputDirs = inputDirs
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
39 self.onChange = onChange
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
40 self.sourceSubstr = sourceSubstr
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
41 self._fileGraph = Graph()
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
42 self._remoteGraph = None
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
43 self._combinedGraph = None
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
44 self._oneShotAdditionGraph = None
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
45 self._lastErrLog = {} # source: error
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
46
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
47 def updateFileData(self):
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
48 """
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
49 make sure we contain the correct data from the files in inputDirs
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
50 """
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
51 # this sample one is actually only needed for the output, but I don't
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
52 # think I want to have a separate graph for the output
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
53 # handling
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
54 log.debug("read file graphs")
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
55 for fp in FilePath("input").walk():
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
56 if fp.isdir():
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
57 continue
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
58 if fp.splitext()[1] != '.n3':
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
59 continue
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
60 log.debug("read %s", fp)
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
61 # todo: if this fails, leave the report in the graph
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
62 self._fileGraph.parse(fp.open(), format="n3")
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
63 self._combinedGraph = None
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
64
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
65 self.onChange(self)
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
66
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
67 @inlineCallbacks
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
68 def updateRemoteData(self):
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
69 """
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
70 read all remote graphs (which are themselves enumerated within
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
71 the file data)
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
72 """
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
73 t1 = time.time()
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
74 log.debug("read remote graphs")
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
75 g = ConjunctiveGraph()
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
76
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
77 @inlineCallbacks
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
78 def fetchOne(source):
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
79 try:
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
80 fetchTime = yield addTrig(g, source, timeout=5)
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
81 except Exception, e:
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
82 e = str(e)
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
83 if self._lastErrLog.get(source) != e:
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
84 log.error(" can't add source %s: %s", source, e)
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
85 self._lastErrLog[source] = e
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
86 g.add((URIRef(source), ROOM['graphLoadError'], Literal(e)))
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
87 g.add((URIRef(source), RDF.type, ROOM['FailedGraphLoad']))
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
88 else:
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
89 if self._lastErrLog.get(source):
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
90 log.warning(" source %s is back", source)
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
91 self._lastErrLog[source] = None
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
92 g.add((URIRef(source), ROOM['graphLoadMs'],
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
93 Literal(round(fetchTime * 1000, 1))))
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
94
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
95 fetchDone = []
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
96 filtered = 0
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
97 for source in self._fileGraph.objects(ROOM['reasoning'],
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
98 ROOM['source']):
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
99 if self.sourceSubstr and self.sourceSubstr not in source:
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
100 filtered += 1
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
101 continue
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
102 fetchDone.append(fetchOne(source))
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
103 yield gatherResults(fetchDone, consumeErrors=True)
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
104 log.debug("loaded %s (skipping %s) in %.1f ms", len(fetchDone),
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
105 filtered, 1000 * (time.time() - t1))
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
106
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
107 prevGraph = self._remoteGraph
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
108 self._remoteGraph = g
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
109 self._combinedGraph = None
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
110 if (prevGraph is None or
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
111 not graphEqual(g, prevGraph, ignorePredicates=[
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
112 ROOM['signalStrength'],
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
113 # perhaps anything with a number-datatype for its
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
114 # object should be filtered out, and you have to make
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
115 # an upstream quantization (e.g. 'temp high'/'temp
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
116 # low') if you want to do reasoning on the difference
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
117 URIRef("http://bigasterisk.com/map#lastSeenAgoSec"),
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
118 URIRef("http://bigasterisk.com/map#lastSeenAgo"),
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
119 ROOM['usingPower'],
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
120 ROOM['idleTimeMinutes'],
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
121 ROOM['idleTimeMs'],
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
122 ROOM['graphLoadMs'],
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
123 ROOM['localTimeToSecond'],
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
124 ROOM['history'],
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
125 ROOM['temperatureF'],
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
126 ROOM['connectedAgo'],
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
127 ])):
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
128 log.debug(" remote graph changed")
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
129 self.onChange(self)
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
130 else:
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
131 log.debug(" remote graph has no changes to trigger rules")
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
132
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
133 def addOneShot(self, g):
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
134 """
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
135 add this graph to the total, call onChange, and then revert
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
136 the addition of this graph
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
137 """
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
138 self._oneShotAdditionGraph = g
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
139 self._combinedGraph = None
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
140 try:
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
141 self.onChange(self, oneShot=True, oneShotGraph=g)
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
142 finally:
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
143 self._oneShotAdditionGraph = None
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
144 self._combinedGraph = None
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
145
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
146 def getGraph(self):
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
147 """rdflib Graph with the file+remote contents of the input graph"""
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
148 # this could be much faster with the combined readonly graph
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
149 # view from rdflib
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
150 if self._combinedGraph is None:
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
151 self._combinedGraph = Graph()
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
152 if self._fileGraph:
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
153 for s in self._fileGraph:
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
154 self._combinedGraph.add(s)
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
155 if self._remoteGraph:
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
156 for s in self._remoteGraph:
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
157 self._combinedGraph.add(s)
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
158 if self._oneShotAdditionGraph:
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
159 for s in self._oneShotAdditionGraph:
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
160 self._combinedGraph.add(s)
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
161
4d16fa39d54a refactor inputgraph
drewp <drewp@bigasterisk.com>
parents:
diff changeset
162 return self._combinedGraph