annotate lib/patchablegraph/patchsource.py @ 514:495f573af4f4

make patchablegraph release Ignore-this: f55c9a56b052797ff23a80630714b51a
author drewp@bigasterisk.com
date Mon, 22 Apr 2019 23:29:19 -0700
parents lib/patchsource.py@b5abd4fc65a4
children 2d13254fdd44
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
514
495f573af4f4 make patchablegraph release
drewp@bigasterisk.com
parents: 482
diff changeset
1 import 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 rdflib import ConjunctiveGraph
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
4 from rdflib.parser import StringInputSource
514
495f573af4f4 make patchablegraph release
drewp@bigasterisk.com
parents: 482
diff changeset
5 from twisted.internet import reactor, defer
302
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
6
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
7 from patchablegraph import patchFromJson
351
7716b1810d6c reasoning & collector move into docker images
drewp@bigasterisk.com
parents: 312
diff changeset
8 from rdfdb.patch import Patch
514
495f573af4f4 make patchablegraph release
drewp@bigasterisk.com
parents: 482
diff changeset
9 from twisted_sse_demo.eventsource import EventSource
302
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
10
312
170dc9b1e789 fix input graph web display by dirtying combinedGraph better.
drewp@bigasterisk.com
parents: 306
diff changeset
11 log = logging.getLogger('fetch')
302
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
12
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
13 class PatchSource(object):
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
14 """wrap EventSource so it emits Patch objects and has an explicit stop method."""
482
b5abd4fc65a4 UA support, some rewrites from twisted_sse_demo work
drewp@bigasterisk.com
parents: 447
diff changeset
15 def __init__(self, url, agent):
b5abd4fc65a4 UA support, some rewrites from twisted_sse_demo work
drewp@bigasterisk.com
parents: 447
diff changeset
16 self.url = str(url)
302
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
17
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
18 # add callbacks to these to learn if we failed to connect
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
19 # (approximately) or if the ccnnection was unexpectedly lost
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
20 self.connectionFailed = defer.Deferred()
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
21 self.connectionLost = defer.Deferred()
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
22
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
23 self._listeners = set()
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
24 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
25 # 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
26 # start with patches
302
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
27 self._fullGraphReceived = False
482
b5abd4fc65a4 UA support, some rewrites from twisted_sse_demo work
drewp@bigasterisk.com
parents: 447
diff changeset
28 self._eventSource = EventSource(url.toPython().encode('utf8'),
b5abd4fc65a4 UA support, some rewrites from twisted_sse_demo work
drewp@bigasterisk.com
parents: 447
diff changeset
29 userAgent=agent)
302
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
30
482
b5abd4fc65a4 UA support, some rewrites from twisted_sse_demo work
drewp@bigasterisk.com
parents: 447
diff changeset
31 self._eventSource.addEventListener(b'fullGraph', self._onFullGraph)
b5abd4fc65a4 UA support, some rewrites from twisted_sse_demo work
drewp@bigasterisk.com
parents: 447
diff changeset
32 self._eventSource.addEventListener(b'patch', self._onPatch)
302
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
33 self._eventSource.onerror(self._onError)
482
b5abd4fc65a4 UA support, some rewrites from twisted_sse_demo work
drewp@bigasterisk.com
parents: 447
diff changeset
34 self._eventSource.onConnectionLost = self._onDisconnect
306
6aad04b34231 sse_collector stats page
drewp@bigasterisk.com
parents: 302
diff changeset
35
439
124c921ad52d stats->state to make room for greplin stats
drewp@bigasterisk.com
parents: 429
diff changeset
36 def state(self):
306
6aad04b34231 sse_collector stats page
drewp@bigasterisk.com
parents: 302
diff changeset
37 return {
6aad04b34231 sse_collector stats page
drewp@bigasterisk.com
parents: 302
diff changeset
38 'url': self.url,
6aad04b34231 sse_collector stats page
drewp@bigasterisk.com
parents: 302
diff changeset
39 'fullGraphReceived': self._fullGraphReceived,
6aad04b34231 sse_collector stats page
drewp@bigasterisk.com
parents: 302
diff changeset
40 }
302
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
41
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
42 def addPatchListener(self, func):
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
43 """
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
44 func(patch, fullGraph=[true if the patch is the initial fullgraph])
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
45 """
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
46 self._listeners.add(func)
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
47
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
48 def stop(self):
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
49 log.info('stop read from %s', self.url)
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
50 try:
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
51 self._eventSource.protocol.stopProducing() # needed?
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
52 except AttributeError:
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
53 pass
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
54 self._eventSource = None
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
55
482
b5abd4fc65a4 UA support, some rewrites from twisted_sse_demo work
drewp@bigasterisk.com
parents: 447
diff changeset
56 def _onDisconnect(self, reason):
b5abd4fc65a4 UA support, some rewrites from twisted_sse_demo work
drewp@bigasterisk.com
parents: 447
diff changeset
57 log.debug('PatchSource._onDisconnect from %s (%s)', self.url, reason)
302
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
58 # skip this if we're doing a stop?
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
59 self.connectionLost.callback(None)
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
60
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
61 def _onError(self, msg):
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
62 log.debug('PatchSource._onError from %s %r', self.url, msg)
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
63 if not self._fullGraphReceived:
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
64 self.connectionFailed.callback(msg)
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
65 else:
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
66 self.connectionLost.callback(msg)
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
67
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
68 def _onFullGraph(self, message):
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
69 try:
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
70 g = ConjunctiveGraph()
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
71 g.parse(StringInputSource(message), format='json-ld')
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
72 p = Patch(addGraph=g)
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
73 self._sendPatch(p, fullGraph=True)
482
b5abd4fc65a4 UA support, some rewrites from twisted_sse_demo work
drewp@bigasterisk.com
parents: 447
diff changeset
74 except Exception:
302
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
75 log.error(traceback.format_exc())
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
76 raise
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
77 self._fullGraphReceived = True
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
78
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
79 def _onPatch(self, message):
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
80 try:
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
81 p = patchFromJson(message)
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
82 self._sendPatch(p, fullGraph=False)
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
83 except:
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
84 log.error(traceback.format_exc())
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
85 raise
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
86
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
87 def _sendPatch(self, p, fullGraph):
482
b5abd4fc65a4 UA support, some rewrites from twisted_sse_demo work
drewp@bigasterisk.com
parents: 447
diff changeset
88 log.debug('PatchSource %s received patch %s (fullGraph=%s)',
b5abd4fc65a4 UA support, some rewrites from twisted_sse_demo work
drewp@bigasterisk.com
parents: 447
diff changeset
89 self.url, p.shortSummary(), fullGraph)
302
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
90 for lis in self._listeners:
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
91 lis(p, fullGraph=fullGraph)
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
92
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
93 def __del__(self):
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
94 if self._eventSource:
482
b5abd4fc65a4 UA support, some rewrites from twisted_sse_demo work
drewp@bigasterisk.com
parents: 447
diff changeset
95 raise ValueError("PatchSource wasn't stopped before del")
302
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
96
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
97 class ReconnectingPatchSource(object):
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
98 """
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
99 PatchSource api, but auto-reconnects internally and takes listener
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
100 at init time to not miss any patches. You'll get another
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
101 fullGraph=True patch if we have to reconnect.
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
102
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
103 todo: generate connection stmts in here
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
104 """
482
b5abd4fc65a4 UA support, some rewrites from twisted_sse_demo work
drewp@bigasterisk.com
parents: 447
diff changeset
105 def __init__(self, url, listener, reconnectSecs=60, agent='unset'):
b5abd4fc65a4 UA support, some rewrites from twisted_sse_demo work
drewp@bigasterisk.com
parents: 447
diff changeset
106 # type: (str, Any, Any, str)
302
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
107 self.url = url
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
108 self._stopped = False
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
109 self._listener = listener
429
fcd2c026f51e exportStats for sending scales data to influxdb
drewp@bigasterisk.com
parents: 351
diff changeset
110 self.reconnectSecs = reconnectSecs
482
b5abd4fc65a4 UA support, some rewrites from twisted_sse_demo work
drewp@bigasterisk.com
parents: 447
diff changeset
111 self.agent = agent
302
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
112 self._reconnect()
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
113
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
114 def _reconnect(self):
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
115 if self._stopped:
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
116 return
482
b5abd4fc65a4 UA support, some rewrites from twisted_sse_demo work
drewp@bigasterisk.com
parents: 447
diff changeset
117 self._ps = PatchSource(self.url, agent=self.agent)
302
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
118 self._ps.addPatchListener(self._onPatch)
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
119 self._ps.connectionFailed.addCallback(self._onConnectionFailed)
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
120 self._ps.connectionLost.addCallback(self._onConnectionLost)
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
121
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
122 def _onPatch(self, p, fullGraph):
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
123 self._listener(p, fullGraph=fullGraph)
306
6aad04b34231 sse_collector stats page
drewp@bigasterisk.com
parents: 302
diff changeset
124
439
124c921ad52d stats->state to make room for greplin stats
drewp@bigasterisk.com
parents: 429
diff changeset
125 def state(self):
306
6aad04b34231 sse_collector stats page
drewp@bigasterisk.com
parents: 302
diff changeset
126 return {
439
124c921ad52d stats->state to make room for greplin stats
drewp@bigasterisk.com
parents: 429
diff changeset
127 'reconnectedPatchSource': self._ps.state(),
306
6aad04b34231 sse_collector stats page
drewp@bigasterisk.com
parents: 302
diff changeset
128 }
302
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
129
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
130 def stop(self):
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
131 self._stopped = True
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
132 self._ps.stop()
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
133
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
134 def _onConnectionFailed(self, arg):
429
fcd2c026f51e exportStats for sending scales data to influxdb
drewp@bigasterisk.com
parents: 351
diff changeset
135 reactor.callLater(self.reconnectSecs, self._reconnect)
302
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
136
46c5fae89823 factor out patchsource
drewp@bigasterisk.com
parents:
diff changeset
137 def _onConnectionLost(self, arg):
429
fcd2c026f51e exportStats for sending scales data to influxdb
drewp@bigasterisk.com
parents: 351
diff changeset
138 reactor.callLater(self.reconnectSecs, self._reconnect)
482
b5abd4fc65a4 UA support, some rewrites from twisted_sse_demo work
drewp@bigasterisk.com
parents: 447
diff changeset
139