def weightplot(self,clip=1):
dtb = db.WikiDatabase()
pages = dtb.getallfetched()
print "got", len(pages)
import itertools
possw = list(itertools.product([0,1], repeat=6))
shuffle(pages)
for r in pages[:clip]:
article = r[0].decode('utf-8')
pageid = r[1]
domain = r[2]
xaxisname = "Username"
yaxisname = "Edit share",
title = "Editors of " + safefilename(article)
title += ", " + domain + str(pageid) + ", by edit count"
filename = domain + str(pageid) + "editcW"
dbdata = dtb.getuserchange2(pageid,domain)
dat = WikiDataHandle()
xlz = []
ylz = [[],[]]
for w in possw:
for d in (0,1):
ww = list(w + (d,))
f = filename + str(ww)
xlabels, ypoints = dat.editsharedata(pageid,domain,weights=ww,
namesort=True,dbdata=dbdata)
xlz = xlabels
ylz[d].append(ypoints)
realy = [[],[]]
for i,y in enumerate(ylz):
yy = zip(*y)
yyy = [sum(e) for e in yy]
assert len(yyy) == len(xlz)
realy[i] = yyy
xf = []
yf = [[],[]]
for i,y in enumerate(realy):
f = filename + "average" + "G" + str(i)
comb = zip(xlz,y)
comb = sorted(comb, key = lambda x: x[1])
xf, yf[i] = zip(*comb)
xpoints = list(range(len(xf)))
print self.specialtrajectory(xpoints, yf[0], xpoints,
yf[1], None, None, None,
xaxisname, yaxisname, None,
title, filename, width=20, height=12)
def talkplots(self):
dat = WikiDataHandle()
titles, pids, tids, domains = dat.talkpages()
for i in range(len(titles)):
domain = domains[i]
pageid = pids[i]
talkid = tids[i]
title = titles[i]
print title, pageid
try:
x1, t1, _ = dat.trajectorydata(pageid, domain, normalise=False)
x2, t2, growth = dat.trajectorydata(talkid, domain, normalise=False)
except:
continue
t1sum = max(t1)
if t1sum:
for i in xrange(len(t1)):
t1[i] /= t1sum
t2sum = max(t2)
if t2sum:
for i in xrange(len(t2)):
t2[i] /= t2sum
gsum = max(growth)
if gsum:
for i in xrange(len(growth)):
growth[i] /= gsum
creation = min(x1[0],x2[0])
for x in (x1, x2):
for i in xrange(len(x)):
x[i] = (x[i]-creation).total_seconds()/3600
title = title.decode('utf-8').replace('/','') + " article vs talk page trajectories"
xaxisname = "Hours since creation"
taxisname = "Change"
gaxisname = "Article page size"
filename = domain + str(pageid) + " Special Combo"
print self.talktrajectory(x1, t1, x2, t2, growth,
xaxisname, taxisname, gaxisname,
title, filename, width=20, height=12)
def plot(self, title, pageid, domain, trajectory=True,
editcount=True, editshare=True, weights=None):
filenames = []
dat = WikiDataHandle()
try:
title = title.decode('utf-8')
except:
title = "id" + str(pageid)
domain = domain.decode('utf-8')
if trajectory:
xpoints, tpoints, gpoints = dat.trajectorydata(pageid, domain)
filenames.append(self.trajectory(xpoints, tpoints,
gpoints,
"Hours since creation",
"Edit distance from final",
"Article size",
"Trajectory of " + safefilename(title) + ", " +\
domain + str(pageid),
domain + str(pageid) + "traj",
width=13, height=8))
print "plotted", filenames[-1]
logDebug("plotted "+ filenames[-1])
if editcount:
xlabels, ypoints = dat.editcountdata(pageid, domain)
filenames.append(self.barchart(xlabels, ypoints,
"Username", "Edit count",
"Editors of " + safefilename(title) + ", " +
domain + str(pageid) + ", by edit count",
domain + str(pageid) + "editc",
width = 13, height = 8))
print "plotted", filenames[-1]
logDebug("plotted "+ filenames[-1])
if editshare:
xlabels, ypoints = dat.editsharedata(pageid, domain, self.weights)
filenames.append(self.barchart(xlabels, ypoints,
"Username", "Edit share",
"Editors of " + safefilename(title) + ", " + domain\
+ str(pageid) + ", by share",
domain + str(pageid) + "share",
width = 13, height = 8))
print "plotted", filenames[-1]
logDebug("plotted "+ filenames[-1])
return filenames
