def main():
IDList = parserEdmund.readLoadedIDs()
threads = parserEdmund.PopulateFromIDList(IDList)
x = [(tr.pagesLoaded, tr.title, tr.ID, tr) for tr in threads]
biggest = sorted(x)[-50:]
tr = next(tu for tu in x if tu[2]=='ef0a892')[3]
# [(a[1], a[0], a[2]) for a in biggest if a[1].find('edan')!=-1]
#x ('GM News, New Models and Market Share - READ ONLY', 3200, 'f16697b')
# ('Midsize Sedans 2.0', 1805, 'f12d514'),
# ('Midsize Sedans Comparison Thread - READ ONLY', 1230, 'ef733f6'),
#x ('Entry Level Luxury Performance Sedans', 1603, 'ef0a892'),
# tr = biggest[1][3]
# ('High End Luxury Cars', 2473, 'ee9e5eb')]
print tr.title
comment = tr.title +" "+ tr.ID
parserEdmund.export2MizikInterview(tr.getPosts(),comment)
runMizikLDA(100)
cleanUp()
def main():
IDList = parserEdmund.readLoadedIDs()
threads = parserEdmund.PopulateFromIDList(IDList)
x = [(tr.pagesLoaded, tr.title, tr.ID, tr) for tr in threads]
biggest = sorted(x)[-50:]
tr = next(tu for tu in x if tu[2] == 'ef0a892')[3]
# [(a[1], a[0], a[2]) for a in biggest if a[1].find('edan')!=-1]
#x ('GM News, New Models and Market Share - READ ONLY', 3200, 'f16697b')
# ('Midsize Sedans 2.0', 1805, 'f12d514'),
# ('Midsize Sedans Comparison Thread - READ ONLY', 1230, 'ef733f6'),
#x ('Entry Level Luxury Performance Sedans', 1603, 'ef0a892'),
# tr = biggest[1][3]
# ('High End Luxury Cars', 2473, 'ee9e5eb')]
print tr.title
comment = tr.title + " " + tr.ID
parserEdmund.export2MizikInterview(tr.getPosts(), comment)
runMizikLDA(100)
cleanUp()
import parserEdmund
from datetime import datetime
t= datetime.now()
IDList = parserEdmund.readLoadedIDs()
stuff = parserEdmund.PopulateFromIDList(IDList)
print datetime.now()-t
t= datetime.now()
yearCount = dict()
for Ethread in stuff:
for post in Ethread.getPosts():
try:
yearCount[post.msgTime.tm_year] +=1
except KeyError:
yearCount[post.msgTime.tm_year] =1
print yearCount
print datetime.now()-t
def init():
IDList = parserEdmund.readLoadedIDs()
threads = parserEdmund.PopulateFromIDList(IDList)
threads = sorted(threads,key=lambda x: x.title)
return threads
def init():
IDList = parserEdmund.readLoadedIDs()
threads = parserEdmund.PopulateFromIDList(IDList)
threads = sorted(threads, key=lambda x: x.title)
return threads
import parserEdmund
from datetime import datetime
IDs = parserEdmund.readLoadedIDs()
items = parserEdmund.PopulateFromIDList(IDs)
PagesDownloaded = sum([I.pagesLoaded for I in items])
print '\nPagesDownloaded:'+str(PagesDownloaded)
PostsDownloaded1 = sum([I.pagesLoaded*10 for I in items if I.pages > I.pagesLoaded])
PostsDownloaded2 = sum([I.Nposts for I in items if I.pages<=I.pagesLoaded])
print 'PostsDownloaded:'+str(PostsDownloaded1)+"(active threads)+"+str(PostsDownloaded2)+"(done threads)"
active = sorted([(I.pagesLoaded,I.pages,I.title,I.ID ) for I in items if I.pages-1>I.pagesLoaded])
est = sum([(-x+y)*2.0/60/60 for (x, y, z,t) in active])
print '\n(nearly)Active Threads(+-1 page):'+str(len(active))
print str(int(est))+"h estimated to finish active"
for t in active: print t
done = sorted([(I.pagesLoaded,I.pages,I.title,I.ID) for I in items if I.pages<=I.pagesLoaded])
