annotate lib/patchsource.py @ 427:db031d9ec28e

don't use 'connected' for time and for network. add rdf:type. Ignore-this: 59899796d4785300f40137de2c2c44d5
author drewp@bigasterisk.com
date Thu, 04 Apr 2019 02:19:10 -0700
parents 7716b1810d6c
children fcd2c026f51e
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
312
170dc9b1e789 fix input graph web display by dirtying combinedGraph better.
drewp@bigasterisk.com
parents: 306
diff changeset
1 import sys, logging
302
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
2 import traceback
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
3 from twisted.internet import reactor, defer
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
4 from twisted_sse_demo.eventsource import EventSource
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
5 from rdflib import ConjunctiveGraph
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
6 from rdflib.parser import StringInputSource
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
7
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
8 sys.path.append("../../lib")
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
9 from patchablegraph import patchFromJson
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
10
351
7716b1810d6c reasoning & collector move into docker images
drewp@bigasterisk.com
parents: 312
diff changeset
11 sys.path.append("/my/proj/rdfdb")
7716b1810d6c reasoning & collector move into docker images
drewp@bigasterisk.com
parents: 312
diff changeset
12 from rdfdb.patch import Patch
302
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
13
312
170dc9b1e789 fix input graph web display by dirtying combinedGraph better.
drewp@bigasterisk.com
parents: 306
diff changeset
14 log = logging.getLogger('fetch')
302
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
15
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
16 class PatchSource(object):
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
17 """wrap EventSource so it emits Patch objects and has an explicit stop method."""
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
18 def __init__(self, url):
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
19 self.url = url
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
20
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
21 # add callbacks to these to learn if we failed to connect
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
22 # (approximately) or if the ccnnection was unexpectedly lost
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
23 self.connectionFailed = defer.Deferred()
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
24 self.connectionLost = defer.Deferred()
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
25
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
26 self._listeners = set()
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
27 log.info('start read from %s', url)
312
170dc9b1e789 fix input graph web display by dirtying combinedGraph better.
drewp@bigasterisk.com
parents: 306
diff changeset
28 # note: fullGraphReceived isn't guaranteed- the stream could
170dc9b1e789 fix input graph web display by dirtying combinedGraph better.
drewp@bigasterisk.com
parents: 306
diff changeset
29 # start with patches
302
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
30 self._fullGraphReceived = False
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
31 self._eventSource = EventSource(url.toPython().encode('utf8'))
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
32 self._eventSource.protocol.delimiter = '\n'
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
33
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
34 self._eventSource.addEventListener('fullGraph', self._onFullGraph)
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
35 self._eventSource.addEventListener('patch', self._onPatch)
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
36 self._eventSource.onerror(self._onError)
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
37
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
38 origSet = self._eventSource.protocol.setFinishedDeferred
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
39 def sfd(d):
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
40 origSet(d)
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
41 d.addCallback(self._onDisconnect)
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
42 self._eventSource.protocol.setFinishedDeferred = sfd
306
6aad04b34231 sse_collector stats page
drewp@bigasterisk.com
parents: 302
diff changeset
43
6aad04b34231 sse_collector stats page
drewp@bigasterisk.com
parents: 302
diff changeset
44 def stats(self):
6aad04b34231 sse_collector stats page
drewp@bigasterisk.com
parents: 302
diff changeset
45 return {
6aad04b34231 sse_collector stats page
drewp@bigasterisk.com
parents: 302
diff changeset
46 'url': self.url,
6aad04b34231 sse_collector stats page
drewp@bigasterisk.com
parents: 302
diff changeset
47 'fullGraphReceived': self._fullGraphReceived,
6aad04b34231 sse_collector stats page
drewp@bigasterisk.com
parents: 302
diff changeset
48 }
302
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
49
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
50 def addPatchListener(self, func):
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
51 """
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
52 func(patch, fullGraph=[true if the patch is the initial fullgraph])
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
53 """
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
54 self._listeners.add(func)
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
55
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
56 def stop(self):
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
57 log.info('stop read from %s', self.url)
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
58 try:
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
59 self._eventSource.protocol.stopProducing() # needed?
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
60 except AttributeError:
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
61 pass
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
62 self._eventSource = None
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
63
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
64 def _onDisconnect(self, a):
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
65 log.debug('PatchSource._onDisconnect from %s', self.url)
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
66 # skip this if we're doing a stop?
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
67 self.connectionLost.callback(None)
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
68
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
69 def _onError(self, msg):
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
70 log.debug('PatchSource._onError from %s %r', self.url, msg)
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
71 if not self._fullGraphReceived:
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
72 self.connectionFailed.callback(msg)
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
73 else:
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
74 self.connectionLost.callback(msg)
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
75
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
76 def _onFullGraph(self, message):
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
77 try:
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
78 g = ConjunctiveGraph()
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
79 g.parse(StringInputSource(message), format='json-ld')
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
80 p = Patch(addGraph=g)
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
81 self._sendPatch(p, fullGraph=True)
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
82 except:
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
83 log.error(traceback.format_exc())
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
84 raise
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
85 self._fullGraphReceived = True
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
86
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
87 def _onPatch(self, message):
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
88 try:
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
89 p = patchFromJson(message)
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
90 self._sendPatch(p, fullGraph=False)
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
91 except:
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
92 log.error(traceback.format_exc())
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
93 raise
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
94
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
95 def _sendPatch(self, p, fullGraph):
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
96 log.debug('PatchSource %s received patch %s (fullGraph=%s)', self.url, p.shortSummary(), fullGraph)
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
97 for lis in self._listeners:
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
98 lis(p, fullGraph=fullGraph)
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
99
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
100 def __del__(self):
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
101 if self._eventSource:
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
102 raise ValueError
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
103
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
104 class ReconnectingPatchSource(object):
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
105 """
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
106 PatchSource api, but auto-reconnects internally and takes listener
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
107 at init time to not miss any patches. You'll get another
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
108 fullGraph=True patch if we have to reconnect.
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
109
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
110 todo: generate connection stmts in here
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
111 """
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
112 def __init__(self, url, listener):
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
113 self.url = url
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
114 self._stopped = False
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
115 self._listener = listener
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
116 self._reconnect()
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
117
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
118 def _reconnect(self):
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
119 if self._stopped:
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
120 return
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
121 self._ps = PatchSource(self.url)
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
122 self._ps.addPatchListener(self._onPatch)
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
123 self._ps.connectionFailed.addCallback(self._onConnectionFailed)
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
124 self._ps.connectionLost.addCallback(self._onConnectionLost)
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
125
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
126 def _onPatch(self, p, fullGraph):
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
127 self._listener(p, fullGraph=fullGraph)
306
6aad04b34231 sse_collector stats page
drewp@bigasterisk.com
parents: 302
diff changeset
128
6aad04b34231 sse_collector stats page
drewp@bigasterisk.com
parents: 302
diff changeset
129 def stats(self):
6aad04b34231 sse_collector stats page
drewp@bigasterisk.com
parents: 302
diff changeset
130 return {
6aad04b34231 sse_collector stats page
drewp@bigasterisk.com
parents: 302
diff changeset
131 'reconnectedPatchSource': self._ps.stats(),
6aad04b34231 sse_collector stats page
drewp@bigasterisk.com
parents: 302
diff changeset
132 }
302
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
133
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
134 def stop(self):
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
135 self._stopped = True
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
136 self._ps.stop()
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
137
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
138 def _onConnectionFailed(self, arg):
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
139 reactor.callLater(60, self._reconnect)
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
140
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
141 def _onConnectionLost(self, arg):
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
142 reactor.callLater(60, self._reconnect)
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
143