def __init__(self, gui, stream, mainDir):
StoppableThread.__init__(self, "Model")
self.__iter = stream.__iter__()
self.__elem = None
self.__gui = gui
self.__softPause = True
self.__urlBuilder = UrlBuilder()
self.__userBuilder = UserBuilder()
streamDir = os.path.join(mainDir, "stream")
userDir = os.path.join(mainDir, "user")
self.__tweetResolvedListener = ResolvedTweetQueue(
streamDir=streamDir, userDir=userDir, userBuilder=self.__userBuilder, urlBuilder=self.__urlBuilder
)
self.__urlResolver = UrlResolverManager(
os.path.join(streamDir, "urlResolverCache.db2"), self.__tweetResolvedListener
)
self.__refreshGui = Event()
self.__refreshStatusBar = Event()
self.__showProbDist = Event()
self.__probDistUrl = None
Publisher.subscribe(self.onPauseJob, "model.pause")
Publisher.subscribe(self.onResumeJob, "model.start")
Publisher.subscribe(self.onRefreshGui, "model.refreshGui")
Publisher.subscribe(self.onRefreshStatusBar, "model.refreshStatusBar")
Publisher.subscribe(self.onProbDist, "model.prob_dist")
Publisher.subscribe(self.onShowTreeMap, "model.showTreeMap")
self.doPauseJob()
self.start()
class Model(StoppableThread):
def __init__(self, gui, stream, mainDir):
StoppableThread.__init__(self, "Model")
self.__iter = stream.__iter__()
self.__elem = None
self.__gui = gui
self.__softPause = True
self.__urlBuilder = UrlBuilder()
self.__userBuilder = UserBuilder()
streamDir = os.path.join(mainDir, "stream")
userDir = os.path.join(mainDir, "user")
self.__tweetResolvedListener = ResolvedTweetQueue(
streamDir=streamDir, userDir=userDir, userBuilder=self.__userBuilder, urlBuilder=self.__urlBuilder
)
self.__urlResolver = UrlResolverManager(
os.path.join(streamDir, "urlResolverCache.db2"), self.__tweetResolvedListener
)
self.__refreshGui = Event()
self.__refreshStatusBar = Event()
self.__showProbDist = Event()
self.__probDistUrl = None
Publisher.subscribe(self.onPauseJob, "model.pause")
Publisher.subscribe(self.onResumeJob, "model.start")
Publisher.subscribe(self.onRefreshGui, "model.refreshGui")
Publisher.subscribe(self.onRefreshStatusBar, "model.refreshStatusBar")
Publisher.subscribe(self.onProbDist, "model.prob_dist")
Publisher.subscribe(self.onShowTreeMap, "model.showTreeMap")
self.doPauseJob()
self.start()
def onRefreshGui(self, msg):
self.__refreshGui.set()
def onRefreshStatusBar(self, msg):
self.__refreshStatusBar.set()
def onPauseJob(self, msg):
d = msg.data
if self.__softPause:
self.__softPause = d["soft"] if d and "soft" in d else False
self.doPauseJob()
def onProbDist(self, msg):
self.__showProbDist.set()
self.__probDistUrl = msg.data
def onShowTreeMap(self, msg):
import webbrowser
webbrowser.open(self.__tweetResolvedListener.getServerUrl())
def doPauseJob(self):
self.pauseJob()
self.__urlResolver.pauseWorkers()
Publisher.sendMessage("model.paused")
def onResumeJob(self, msg):
d = msg.data
softResume = d["soft"] if d and "soft" in d else False
userResume = not softResume
if userResume or self.__softPause:
logger.info("Continue job " + unicode(softResume) + unicode(self.__softPause))
self.doContinueJob()
else:
logger.info("Ignore resume request")
def doContinueJob(self):
self.continueJob()
self.__urlResolver.continueWorkers()
Publisher.sendMessage("model.started")
def atBegin(self):
logger.info("Preparing model...")
self.__urlResolver.start()
logger.info("Start analyzing tweets")
self.__tweetResolvedListener.start()
def runPart(self):
try:
s = self.__elem or self.__iter.next()
self.__elem = s
if u"text" in s:
try:
retweeted = (
TweetText(s[u"retweeted_status"], self.__urlBuilder, self.__userBuilder, None)
if s.has_key(u"retweeted_status")
else None
)
if retweeted:
for url in retweeted.urls():
self.__urlResolver.addUrlToQueue(url)
tweet = TweetText(s, self.__urlBuilder, self.__userBuilder, retweeted.id() if retweeted else None)
for url in tweet.urls():
self.__urlResolver.addUrlToQueue(url)
except UrlException as e:
logger.warn(u"Cannot build url: " + str(e))
self._doSmthElse()
self.__elem = None
except Full:
return
except StopIteration:
raise NothingToDo()
def onPause(self):
self._doSmthElse()
def _doSmthElse(self):
if self.__refreshGui.isSet():
logger.info("Send data to GUI")
self.__refreshGui.clear()
data = {}
data["urls"] = self.__tweetResolvedListener.finalUrls()
Publisher.sendMessage("update.urls", data=data)
if self.__showProbDist.isSet():
url = self.__probDistUrl
self.__showProbDist.clear()
self.__probDistUrl = None
probDistI = TxtClassificatorWrapper.instance().probDist(url.getText())
if self.__refreshStatusBar.isSet():
self.__refreshStatusBar.clear()
data = {}
data["cache"] = self.__urlResolver.cacheHitRate()
data["position"] = self.__iter.position()
data["position_end"] = self.__iter.count()
data["current_file_c"] = self.__iter.currentFile()
data["last_file_c"] = self.__iter.filesCount()
Publisher.sendMessage("update.statusBar", data=data)
def stop(self):
StoppableThread.stop(self)
self.__urlResolver.stop()
self.__tweetResolvedListener.stop()
logger.info("Wait for thread: " + ",".join([t.name for t in threading.enumerate()]))