from retrieval import article
import sys
__author__="Dan"
__date__ ="$16-Mar-2010 21:40:24$"
if __name__ == "__main__":
todays_date = datetime.datetime.today()
two_week_difference = datetime.timedelta(days=-14)
week_difference = datetime.timedelta(days=-7)
two_day_difference = datetime.timedelta(days=-3)
week_ago_date_mysql = (todays_date + week_difference).strftime("%Y-%m-%d")
days_ago_date_mysql = (todays_date + two_day_difference).strftime("%Y-%m-%d")
article_list = article.db_get_all_articles("NOT isnull(clusterid) AND EXISTS (SELECT articleid FROM articleswithterms WHERE articleid = articles.id) AND `datepublished` >= '%(weekago)s'" % {"weekago" : week_ago_date_mysql})
models,inv_index = articleunigram.db_load_models(article_list)
mergers_query = "SELECT * FROM proposedmergers WHERE `approved` = 1"
db = database.connect_to_database()
cur = db.cursor(cursorclass = MySQLdb.cursors.DictCursor)
clusterer = ClusterMaker(inv_index = inv_index)
cur.execute(mergers_query)
result = cur.fetchall()
proposed_mergers = dict()
for row in result:
clusters_query = "SELECT `clusterid` FROM proposedmergers_clusters WHERE `mergerid` = %d" % (row['id'])
cur.execute(clusters_query)
clusters = cur.fetchall()
proposed_mergers[row['id']] = [clustermodel.ClusterModel(inv_index = inv_index,id = cl['clusterid'],load_full_article = True) for cl in clusters]
for p in proposed_mergers.items():
print "Setting cluster type to " + str(self.cluster_type)
def get_cluster_info(self,e = None):
if e:
self.selected_cluster = int(self.lb_clusters.GetSelection(int(e.GetSelection())))
print str(self.selected_cluster)
def start_clustering(self,e = None):
if e:
clusterer = clustering.ClusterMaker(cluster_type = self.cluster_type)
cluster_thread = clustering.ClustererThread(clusterer,self.models,self.inv_index,float(self.tc_threshold.GetValue()),self.cluster_method,on_change=self.update)
cluster_thread.start()
# clusterer.cluster_articles(self.models,self.inv_index,float(self.tc_threshold.GetValue()),self.cluster_method,on_change=self.update)
# clusterer.run(self.models,self.inv_index,float(self.tc_threshold.GetValue()),self.cluster_method,on_change=self.update,name = "whatever")
def update(self,clusters=None,new_cluster=None,remove_model=None):
if clusters:
self.tc_num_clusters.SetValue(str(len(clusters)))
self.tc_num_clusters.Update()
self.clusters = clusters
if new_cluster:
self.lb_clusters.AppendAndEnsureVisible(str(new_cluster))
self.lb_clusters.Update()
theVar = 1
app = wx.App(False)
man_articles = article.db_get_all_articles("NOT trainingcluster = 0")
models,inv_index = articleunigram.db_load_models(man_articles)
frame = ClusterMonitor(None,'Article Browser',models,inv_index)
app.MainLoop()
