Mercurial > code > home > repos > light9
annotate bin/rdfdb @ 811:b19cd005a491
just comments
Ignore-this: 850b0da90240ab6f764430505809fd6d
| author | drewp@bigasterisk.com |
|---|---|
| date | Sat, 29 Sep 2012 20:59:13 +0000 |
| parents | a631e075a5bf |
| children | 1ae8e6b287e3 |
| rev | line source |
|---|---|
| 796 | 1 #!bin/python |
| 2 """ | |
| 3 other tools POST themselves to here as subscribers to the graph. They | |
| 811 | 4 are providing a URL we can PUT to with graph updates. |
| 796 | 5 |
| 6 we immediately PUT them back all the contents of the graph as a bunch | |
| 7 of adds. | |
| 8 | |
| 811 | 9 later we PUT them back with patches (del/add lists) when there are |
| 796 | 10 changes. |
| 11 | |
| 12 If we fail to reach a registered caller, we forget about it for future | |
| 811 | 13 calls. We could PUT empty diffs as a heartbeat to notice disappearing |
| 796 | 14 callers faster. |
| 15 | |
| 811 | 16 A caller can submit a patch which we'll persist and broadcast to every |
| 17 other client. | |
| 796 | 18 |
| 19 Global data undo should probably happen within this service. | |
| 20 | |
| 21 Maybe some subgraphs are for transient data (e.g. current timecode, | |
| 22 mouse position in curvecalc) that only some listeners want to hear about. | |
| 23 | |
|
797
904913de4599
deletes are now quads. refactor files. named clients. auto client port
drewp@bigasterisk.com
parents:
796
diff
changeset
|
24 Deletes are graph-specific, so callers may be surprised to delete a |
|
904913de4599
deletes are now quads. refactor files. named clients. auto client port
drewp@bigasterisk.com
parents:
796
diff
changeset
|
25 stmt from one graph but then find that statement is still true. |
| 796 | 26 |
|
797
904913de4599
deletes are now quads. refactor files. named clients. auto client port
drewp@bigasterisk.com
parents:
796
diff
changeset
|
27 Alternate plan: would it help to insist that every patch is within |
|
904913de4599
deletes are now quads. refactor files. named clients. auto client port
drewp@bigasterisk.com
parents:
796
diff
changeset
|
28 only one subgraph? I think it's ok for them to span multiple ones. |
| 796 | 29 |
| 30 Inserts can be made on any subgraphs, and each subgraph is saved in | |
| 31 its own file. The file might not be in a format that can express | |
| 32 graphs, so I'm just going to not store the subgraph URI in any file. | |
| 33 | |
| 34 I don't support wildcard deletes, and there are race conditions where a | |
| 35 s-p could end up with unexpected multiple objects. Every client needs | |
| 36 to be ready for this. | |
| 37 | |
| 38 We watch the files and push their own changes back to the clients. | |
| 39 | |
| 40 Persist our client list, to survive restarts. In another rdf file? A | |
| 41 random json one? memcache? Also hold the recent changes. We're not | |
| 42 logging everything forever, though, since the output files and a VCS | |
| 43 shall be used for that | |
| 44 | |
| 45 Bnodes: this rdfdb graph might be able to track bnodes correctly, and | |
| 46 they make for more compact n3 files. I'm not sure if it's going to be | |
| 47 hard to keep the client bnodes in sync though. File rereads would be | |
| 811 | 48 hard, if ever a bnode was used across graphs, so that probably should |
| 796 | 49 not be allowed. |
| 50 | |
| 51 Our API: | |
| 52 | |
| 53 GET / ui | |
| 811 | 54 GET /graph the whole graph, or a query from it (needed? just for ui browsing?) |
| 796 | 55 PUT /patches clients submit changes |
| 56 GET /patches (recent) patches from clients | |
| 57 POST /graphClients clientUpdate={uri} to subscribe | |
| 58 GET /graphClients current clients | |
| 59 | |
| 60 format: | |
| 61 json {"adds" : [[quads]...], | |
| 62 "deletes": [[quads]], | |
| 811 | 63 "senderUpdateUri" : tooluri, |
| 64 "created":tttt // maybe to help resolve some conflicts | |
| 796 | 65 } |
| 66 maybe use some http://json-ld.org/ in there. | |
| 67 | |
| 806 | 68 proposed rule feature: |
| 69 rdfdb should be able to watch a pair of (sourceFile, rulesFile) and | |
| 70 rerun the rules when either one changes. Should the sourceFile be able | |
| 71 to specify its own rules file? That would be easier | |
| 72 configuration. How do edits work? Not allowed? Patch the source only? | |
| 73 Also see the source graph loaded into a different ctx, and you can | |
| 74 edit that one and see the results in the output context? | |
| 75 | |
| 796 | 76 Our web ui: |
| 77 | |
| 811 | 78 sections |
| 79 | |
| 80 registered clients | |
| 796 | 81 |
| 811 | 82 recent patches, each one says what client it came from. You can reverse |
| 83 them here. We should be able to take patches that are close in time | |
| 84 and keep updating the same data (e.g. a stream of changes as the user | |
| 85 drags a slider) and collapse them into a single edit for clarity. | |
| 86 | |
| 87 Ways to display patches, using labels and creator/subj icons | |
| 88 where possible: | |
| 796 | 89 |
| 811 | 90 <creator> set <subj>'s <p> to <o> |
| 91 <creator> changed <subj>'s <pred> from <o1> to <o2> | |
| 92 <creator> added <o> to <s> <p> | |
| 93 | |
| 94 raw messages for debugging this client | |
| 806 | 95 |
| 811 | 96 ctx urls take you to-> |
| 97 files, who's dirty, have we seen external changes, notice big | |
| 98 files that are taking a long time to save | |
| 806 | 99 |
| 811 | 100 graph contents. plain rdf browser like an outliner or |
| 101 something. clicking any resource from the other displays takes you | |
| 102 to this, focused on that resource | |
| 803 | 103 |
| 796 | 104 """ |
| 105 from twisted.internet import reactor | |
|
797
904913de4599
deletes are now quads. refactor files. named clients. auto client port
drewp@bigasterisk.com
parents:
796
diff
changeset
|
106 import twisted.internet.error |
| 796 | 107 import sys, optparse, logging, json, os |
| 108 import cyclone.web, cyclone.httpclient, cyclone.websocket | |
| 109 sys.path.append(".") | |
|
808
a631e075a5bf
KC big rewrites, now multiple KC instances can sync with rdfdb
drewp@bigasterisk.com
parents:
806
diff
changeset
|
110 from light9 import networking, showconfig, prof |
| 796 | 111 from rdflib import ConjunctiveGraph, URIRef, Graph |
|
797
904913de4599
deletes are now quads. refactor files. named clients. auto client port
drewp@bigasterisk.com
parents:
796
diff
changeset
|
112 from light9.rdfdb.graphfile import GraphFile |
|
904913de4599
deletes are now quads. refactor files. named clients. auto client port
drewp@bigasterisk.com
parents:
796
diff
changeset
|
113 from light9.rdfdb.patch import Patch, ALLSTMTS |
|
798
5c158d37f1ce
autoretry websocket. fix rdflib quad patching. only rerun handlers that asked for the affected subj-preds.
drewp@bigasterisk.com
parents:
797
diff
changeset
|
114 from light9.rdfdb.rdflibpatch import patchQuads |
|
797
904913de4599
deletes are now quads. refactor files. named clients. auto client port
drewp@bigasterisk.com
parents:
796
diff
changeset
|
115 from light9.rdfdb import syncedgraph |
| 796 | 116 |
| 117 from twisted.internet.inotify import INotify | |
| 118 logging.basicConfig(level=logging.DEBUG) | |
| 119 log = logging.getLogger() | |
| 120 | |
| 121 try: | |
| 122 import sys | |
| 123 sys.path.append("../homeauto/lib") | |
| 124 from cycloneerr import PrettyErrorHandler | |
| 125 except ImportError: | |
| 126 class PrettyErrorHandler(object): | |
| 127 pass | |
| 128 | |
| 129 class Client(object): | |
|
797
904913de4599
deletes are now quads. refactor files. named clients. auto client port
drewp@bigasterisk.com
parents:
796
diff
changeset
|
130 """ |
|
904913de4599
deletes are now quads. refactor files. named clients. auto client port
drewp@bigasterisk.com
parents:
796
diff
changeset
|
131 one of our syncedgraph clients |
|
904913de4599
deletes are now quads. refactor files. named clients. auto client port
drewp@bigasterisk.com
parents:
796
diff
changeset
|
132 """ |
|
904913de4599
deletes are now quads. refactor files. named clients. auto client port
drewp@bigasterisk.com
parents:
796
diff
changeset
|
133 def __init__(self, updateUri, label, db): |
| 796 | 134 self.db = db |
|
797
904913de4599
deletes are now quads. refactor files. named clients. auto client port
drewp@bigasterisk.com
parents:
796
diff
changeset
|
135 self.label = label |
| 796 | 136 self.updateUri = updateUri |
| 137 self.sendAll() | |
| 138 | |
|
797
904913de4599
deletes are now quads. refactor files. named clients. auto client port
drewp@bigasterisk.com
parents:
796
diff
changeset
|
139 def __repr__(self): |
|
904913de4599
deletes are now quads. refactor files. named clients. auto client port
drewp@bigasterisk.com
parents:
796
diff
changeset
|
140 return "<%s client at %s>" % (self.label, self.updateUri) |
|
904913de4599
deletes are now quads. refactor files. named clients. auto client port
drewp@bigasterisk.com
parents:
796
diff
changeset
|
141 |
| 796 | 142 def sendAll(self): |
| 143 """send the client the whole graph contents""" | |
|
797
904913de4599
deletes are now quads. refactor files. named clients. auto client port
drewp@bigasterisk.com
parents:
796
diff
changeset
|
144 log.info("sending all graphs to %s at %s" % |
|
904913de4599
deletes are now quads. refactor files. named clients. auto client port
drewp@bigasterisk.com
parents:
796
diff
changeset
|
145 (self.label, self.updateUri)) |
|
904913de4599
deletes are now quads. refactor files. named clients. auto client port
drewp@bigasterisk.com
parents:
796
diff
changeset
|
146 self.sendPatch(Patch( |
|
904913de4599
deletes are now quads. refactor files. named clients. auto client port
drewp@bigasterisk.com
parents:
796
diff
changeset
|
147 addQuads=self.db.graph.quads(ALLSTMTS), |
|
904913de4599
deletes are now quads. refactor files. named clients. auto client port
drewp@bigasterisk.com
parents:
796
diff
changeset
|
148 delQuads=[])) |
| 796 | 149 |
| 150 def sendPatch(self, p): | |
|
797
904913de4599
deletes are now quads. refactor files. named clients. auto client port
drewp@bigasterisk.com
parents:
796
diff
changeset
|
151 return syncedgraph.sendPatch(self.updateUri, p) |
| 796 | 152 |
| 153 class Db(object): | |
| 811 | 154 """ |
| 155 the master graph, all the connected clients, all the files we're watching | |
| 156 """ | |
| 796 | 157 def __init__(self): |
| 158 self.clients = [] | |
| 159 self.graph = ConjunctiveGraph() | |
| 160 | |
| 161 notifier = INotify() | |
| 162 notifier.startReading() | |
| 163 | |
|
798
5c158d37f1ce
autoretry websocket. fix rdflib quad patching. only rerun handlers that asked for the affected subj-preds.
drewp@bigasterisk.com
parents:
797
diff
changeset
|
164 for inFile in [#"show/dance2012/config.n3", |
| 801 | 165 "show/dance2012/patch.n3", |
|
798
5c158d37f1ce
autoretry websocket. fix rdflib quad patching. only rerun handlers that asked for the affected subj-preds.
drewp@bigasterisk.com
parents:
797
diff
changeset
|
166 "show/dance2012/subs/bcools", |
|
799
fcf95ff23cc5
PersistentSubmaster split. keyboardcomposer now notices submaster changes
drewp@bigasterisk.com
parents:
798
diff
changeset
|
167 "show/dance2012/subs/bwarm", |
|
fcf95ff23cc5
PersistentSubmaster split. keyboardcomposer now notices submaster changes
drewp@bigasterisk.com
parents:
798
diff
changeset
|
168 "show/dance2012/subs/house", |
|
fcf95ff23cc5
PersistentSubmaster split. keyboardcomposer now notices submaster changes
drewp@bigasterisk.com
parents:
798
diff
changeset
|
169 "demo.n3", |
|
798
5c158d37f1ce
autoretry websocket. fix rdflib quad patching. only rerun handlers that asked for the affected subj-preds.
drewp@bigasterisk.com
parents:
797
diff
changeset
|
170 ]: |
| 796 | 171 self.g = GraphFile(notifier, |
| 172 inFile, | |
|
798
5c158d37f1ce
autoretry websocket. fix rdflib quad patching. only rerun handlers that asked for the affected subj-preds.
drewp@bigasterisk.com
parents:
797
diff
changeset
|
173 URIRef("http://example.com/file/%s" % |
| 796 | 174 os.path.basename(inFile)), |
| 175 self.patch, | |
| 176 self.getSubgraph) | |
| 177 | |
| 178 def patch(self, p): | |
| 179 """ | |
| 180 apply this patch to the master graph then notify everyone about it | |
| 811 | 181 |
| 182 dueToFileChange if this is a patch describing an edit we read | |
| 183 *from* the file (such that we shouldn't write it back to the file) | |
| 184 | |
| 185 if p has a senderUpdateUri attribute, we won't send this patch | |
| 186 back to the sender with that updateUri | |
| 796 | 187 """ |
|
798
5c158d37f1ce
autoretry websocket. fix rdflib quad patching. only rerun handlers that asked for the affected subj-preds.
drewp@bigasterisk.com
parents:
797
diff
changeset
|
188 log.info("patching graph -%d +%d" % (len(p.delQuads), len(p.addQuads))) |
|
797
904913de4599
deletes are now quads. refactor files. named clients. auto client port
drewp@bigasterisk.com
parents:
796
diff
changeset
|
189 |
|
798
5c158d37f1ce
autoretry websocket. fix rdflib quad patching. only rerun handlers that asked for the affected subj-preds.
drewp@bigasterisk.com
parents:
797
diff
changeset
|
190 patchQuads(self.graph, p.delQuads, p.addQuads, perfect=True) |
|
808
a631e075a5bf
KC big rewrites, now multiple KC instances can sync with rdfdb
drewp@bigasterisk.com
parents:
806
diff
changeset
|
191 senderUpdateUri = getattr(p, 'senderUpdateUri', None) |
| 796 | 192 self.summarizeToLog() |
| 193 for c in self.clients: | |
|
808
a631e075a5bf
KC big rewrites, now multiple KC instances can sync with rdfdb
drewp@bigasterisk.com
parents:
806
diff
changeset
|
194 print "send to %s? %s %s" % (c, c.updateUri, senderUpdateUri) |
|
a631e075a5bf
KC big rewrites, now multiple KC instances can sync with rdfdb
drewp@bigasterisk.com
parents:
806
diff
changeset
|
195 if c.updateUri == senderUpdateUri: |
|
a631e075a5bf
KC big rewrites, now multiple KC instances can sync with rdfdb
drewp@bigasterisk.com
parents:
806
diff
changeset
|
196 # this client has self-applied the patch already |
|
a631e075a5bf
KC big rewrites, now multiple KC instances can sync with rdfdb
drewp@bigasterisk.com
parents:
806
diff
changeset
|
197 continue |
|
798
5c158d37f1ce
autoretry websocket. fix rdflib quad patching. only rerun handlers that asked for the affected subj-preds.
drewp@bigasterisk.com
parents:
797
diff
changeset
|
198 d = c.sendPatch(p) |
|
797
904913de4599
deletes are now quads. refactor files. named clients. auto client port
drewp@bigasterisk.com
parents:
796
diff
changeset
|
199 d.addErrback(self.clientErrored, c) |
| 796 | 200 sendToLiveClients(asJson=p.jsonRepr) |
| 201 | |
|
797
904913de4599
deletes are now quads. refactor files. named clients. auto client port
drewp@bigasterisk.com
parents:
796
diff
changeset
|
202 def clientErrored(self, err, c): |
|
904913de4599
deletes are now quads. refactor files. named clients. auto client port
drewp@bigasterisk.com
parents:
796
diff
changeset
|
203 err.trap(twisted.internet.error.ConnectError) |
|
904913de4599
deletes are now quads. refactor files. named clients. auto client port
drewp@bigasterisk.com
parents:
796
diff
changeset
|
204 log.info("connection error- dropping client %r" % c) |
|
904913de4599
deletes are now quads. refactor files. named clients. auto client port
drewp@bigasterisk.com
parents:
796
diff
changeset
|
205 self.clients.remove(c) |
|
798
5c158d37f1ce
autoretry websocket. fix rdflib quad patching. only rerun handlers that asked for the affected subj-preds.
drewp@bigasterisk.com
parents:
797
diff
changeset
|
206 self.sendClientsToAllLivePages() |
|
797
904913de4599
deletes are now quads. refactor files. named clients. auto client port
drewp@bigasterisk.com
parents:
796
diff
changeset
|
207 |
| 796 | 208 def summarizeToLog(self): |
|
798
5c158d37f1ce
autoretry websocket. fix rdflib quad patching. only rerun handlers that asked for the affected subj-preds.
drewp@bigasterisk.com
parents:
797
diff
changeset
|
209 log.info("contexts in graph (%s total stmts):" % len(self.graph)) |
| 796 | 210 for c in self.graph.contexts(): |
| 211 log.info(" %s: %s statements" % | |
| 212 (c.identifier, len(self.getSubgraph(c.identifier)))) | |
| 213 | |
| 214 def getSubgraph(self, uri): | |
| 811 | 215 """ |
| 216 this is meant to return a live view of the given subgraph, but | |
| 217 if i'm still working around an rdflib bug, it might return a | |
| 218 copy | |
| 219 | |
| 220 and it's returning triples, but I think quads would be better | |
| 221 """ | |
| 796 | 222 # this is returning an empty Graph :( |
| 223 #return self.graph.get_context(uri) | |
| 224 | |
| 225 g = Graph() | |
|
797
904913de4599
deletes are now quads. refactor files. named clients. auto client port
drewp@bigasterisk.com
parents:
796
diff
changeset
|
226 for s in self.graph.triples(ALLSTMTS, uri): |
| 796 | 227 g.add(s) |
| 228 return g | |
| 229 | |
|
797
904913de4599
deletes are now quads. refactor files. named clients. auto client port
drewp@bigasterisk.com
parents:
796
diff
changeset
|
230 def addClient(self, updateUri, label): |
| 796 | 231 [self.clients.remove(c) |
| 232 for c in self.clients if c.updateUri == updateUri] | |
| 233 | |
|
797
904913de4599
deletes are now quads. refactor files. named clients. auto client port
drewp@bigasterisk.com
parents:
796
diff
changeset
|
234 log.info("new client %s at %s" % (label, updateUri)) |
|
904913de4599
deletes are now quads. refactor files. named clients. auto client port
drewp@bigasterisk.com
parents:
796
diff
changeset
|
235 self.clients.append(Client(updateUri, label, self)) |
| 796 | 236 self.sendClientsToAllLivePages() |
| 237 | |
| 238 def sendClientsToAllLivePages(self): | |
|
797
904913de4599
deletes are now quads. refactor files. named clients. auto client port
drewp@bigasterisk.com
parents:
796
diff
changeset
|
239 sendToLiveClients({"clients":[ |
|
904913de4599
deletes are now quads. refactor files. named clients. auto client port
drewp@bigasterisk.com
parents:
796
diff
changeset
|
240 dict(updateUri=c.updateUri, label=c.label) |
|
904913de4599
deletes are now quads. refactor files. named clients. auto client port
drewp@bigasterisk.com
parents:
796
diff
changeset
|
241 for c in self.clients]}) |
| 796 | 242 |
| 243 class Index(PrettyErrorHandler, cyclone.web.RequestHandler): | |
| 244 def get(self): | |
| 245 self.set_header("Content-Type", "application/xhtml+xml") | |
| 246 self.write(open("light9/rdfdb.xhtml").read()) | |
| 247 | |
| 248 class GraphResource(PrettyErrorHandler, cyclone.web.RequestHandler): | |
| 249 def get(self): | |
| 250 pass | |
| 251 | |
| 252 class Patches(PrettyErrorHandler, cyclone.web.RequestHandler): | |
| 253 def __init__(self, *args, **kw): | |
| 254 cyclone.web.RequestHandler.__init__(self, *args, **kw) | |
|
797
904913de4599
deletes are now quads. refactor files. named clients. auto client port
drewp@bigasterisk.com
parents:
796
diff
changeset
|
255 p = syncedgraph.makePatchEndpointPutMethod(self.settings.db.patch) |
| 796 | 256 self.put = lambda: p(self) |
| 257 | |
| 258 def get(self): | |
| 259 pass | |
| 260 | |
| 261 | |
| 262 class GraphClients(PrettyErrorHandler, cyclone.web.RequestHandler): | |
| 263 def get(self): | |
| 264 pass | |
| 265 | |
| 266 def post(self): | |
| 267 upd = self.get_argument("clientUpdate") | |
| 268 try: | |
|
797
904913de4599
deletes are now quads. refactor files. named clients. auto client port
drewp@bigasterisk.com
parents:
796
diff
changeset
|
269 self.settings.db.addClient(upd, self.get_argument("label")) |
| 796 | 270 except: |
| 271 import traceback | |
| 272 traceback.print_exc() | |
| 273 raise | |
| 274 | |
| 275 liveClients = set() | |
| 276 def sendToLiveClients(d=None, asJson=None): | |
| 277 j = asJson or json.dumps(d) | |
| 278 for c in liveClients: | |
| 279 c.sendMessage(j) | |
| 280 | |
| 281 class Live(cyclone.websocket.WebSocketHandler): | |
| 282 | |
| 283 def connectionMade(self, *args, **kwargs): | |
| 811 | 284 log.info("websocket opened") |
| 796 | 285 liveClients.add(self) |
| 286 self.settings.db.sendClientsToAllLivePages() | |
| 287 | |
| 288 def connectionLost(self, reason): | |
| 811 | 289 log.info("websocket closed") |
| 796 | 290 liveClients.remove(self) |
| 291 | |
| 292 def messageReceived(self, message): | |
| 293 log.info("got message %s" % message) | |
| 294 self.sendMessage(message) | |
| 295 | |
| 296 if __name__ == "__main__": | |
| 297 logging.basicConfig() | |
| 298 log = logging.getLogger() | |
| 299 | |
| 300 parser = optparse.OptionParser() | |
| 301 parser.add_option('--show', | |
| 302 help='show URI, like http://light9.bigasterisk.com/show/dance2008', | |
| 303 default=showconfig.showUri()) | |
| 304 parser.add_option("-v", "--verbose", action="store_true", | |
| 305 help="logging.DEBUG") | |
| 306 (options, args) = parser.parse_args() | |
| 307 | |
| 308 log.setLevel(logging.DEBUG if options.verbose else logging.INFO) | |
| 309 | |
| 310 if not options.show: | |
| 311 raise ValueError("missing --show http://...") | |
| 312 | |
| 313 db = Db() | |
| 314 | |
| 315 port = 8051 | |
| 316 reactor.listenTCP(port, cyclone.web.Application(handlers=[ | |
| 317 (r'/', Index), | |
| 318 (r'/live', Live), | |
| 319 (r'/graph', GraphResource), | |
| 320 (r'/patches', Patches), | |
| 321 (r'/graphClients', GraphClients), | |
| 322 | |
| 323 (r"/(jquery-1\.7\.2\.min\.js)", cyclone.web.StaticFileHandler, | |
| 324 dict(path='lib')), | |
| 325 | |
| 326 ], db=db)) | |
| 327 log.info("serving on %s" % port) | |
|
808
a631e075a5bf
KC big rewrites, now multiple KC instances can sync with rdfdb
drewp@bigasterisk.com
parents:
806
diff
changeset
|
328 prof.run(reactor.run, profile=False) |