def do_many_days_s3_APPS(argv):
if len(argv) < 3:
print "usage: start_day end_day"
print "usage: 2011-10-7 2011-10-13"
sys.exit()
QUERY_FILES = [
'euospcomp07.osp_query.log',
'euospcomp08.osp_query.log',
'euospsch01.osp_query.log',
'euospsch03.osp_query.log',
'euospsch01.2.osp_query.log',
'euospsch03.2.osp_query.log',
]
CLICK_FILES = [
'euospsch03.osp_click.log',
'euospcomp08.osp_click.log'
]
#QUERY_FILES = QUERY_FILES[:2]
daylist = handolUtil.get_day_list(argv[1],argv[2])
analyzer = LogAnalyzer("APPS", "%s-%s" %(daylist[0], daylist[-1]))
stopwatch = handolUtil.StopWatch()
for daystr in daylist:
for q in QUERY_FILES:
s3_file = "DEVELOPING/app/7nmc1m75ij/apps-log/query_log/%s/%s.%s" % (daystr, q, daystr)
analyzer.querylog_from_s3("sch-emr", s3_file)
print "Loading & ETL: %f sec" % (stopwatch.laptime())
analyzer.get_stats()
print "Calc Stats: %f sec" % (stopwatch.laptime())
analyzer.write_info()
def get_stats_from_qcmatched(self, fname):
""" calculate statistics from Query/Click matched log
"""
stopwatch = handolUtil.StopWatch()
stopwatch.start()
qcmatched_list = QcLogAnalyzer.loadfile_qc_matched(fname)
print "Loading : %f sec" % (stopwatch.laptime())
stopwatch.start()
for v in qcmatched_list:
im = v[0]
num_clicks = int(v[1])
click_pos = int(v[2])
tot = int(v[3])
devtype = v[4]
dos = v[5]
dm = v[6]
shop = v[7]
keyword = v[8]
if tot == 0: nores = 1
else: nores = 0
if num_clicks > 0:
anyclick = 1
else:
anyclick = 0
if is_non_ascii(keyword):
english = 0
else:
english = 1
os_shop = "%s_%s" % (dos, shop)
shop_keyw = (shop, keyword)
self.langD.add(english,
[1, anyclick, nores, click_pos, num_clicks])
self.imD.add(im,
[1, anyclick, nores, click_pos, num_clicks, english
]) # query count, click
self.keywordD.add(
keyword, [1, anyclick, nores, click_pos, num_clicks, english])
self.shopkeywordD.add(shop_keyw,
[1, anyclick, nores, click_pos, num_clicks])
self.shopD.add(
shop, [1, anyclick, nores, click_pos, num_clicks, english])
self.osshopD.add(
os_shop, [1, anyclick, nores, click_pos, num_clicks, english])
self.devtypeD.add(
devtype, [1, anyclick, nores, click_pos, num_clicks, english])
self.dosD.add(dos,
[1, anyclick, nores, click_pos, num_clicks, english])
self.dmD.add(dm,
[1, anyclick, nores, click_pos, num_clicks, english])
#self.diffsecD.add(diffsec, [1, anyclick, nores, click_pos, len(clicklist)])
if tot == 0:
self.zeroD.add(keyword, [1, english])
#self.zeroshopD.add(shop, [1])
#self.zeroshopkeywordD.add(shop_keyw, [1])
print "Calculating : %f sec" % (stopwatch.laptime())
def test_w_local_LHRHVH():
analyzer = LogAnalyzer("LHRHVH", "any")
stopwatch = handolUtil.StopWatch()
analyzer.querylog_from_file("osp_query.log.20130306")
print "Loading & ETL: %f sec" % (stopwatch.laptime())
analyzer.get_stats()
print "Calc Stats: %f sec" % (stopwatch.laptime())
analyzer.write_info()
def test_w_local_APPS():
analyzer = LogAnalyzer("APPS", "any")
stopwatch = handolUtil.StopWatch()
analyzer.querylog_from_file("euospsch03.2.osp_query.log.20130215")
#analyzer.querylog_from_file("a.query.log")
print "Loading & ETL: %f sec" % (stopwatch.laptime())
analyzer.get_stats()
print "Calc Stats: %f sec" % (stopwatch.laptime())
analyzer.write_info()
def main(self, bucket, cfiles, qfiles, outfile):
stopwatch = handolUtil.StopWatch()
for cfile in cfiles:
self.add_click_file(cfile, bucket)
print "Loading Click - %s: %f sec" % (cfile, stopwatch.laptime())
qclistAll = []
for qfile in qfiles:
qclist = self.add_query_file(qfile, bucket)
qclistAll += qclist
print "Load Query & Match - %s: %f sec" % (qfile,
stopwatch.laptime())
QueryClickMatcher.save_qc_match(qclistAll, outfile)
def load_qcmatched(org_log_dir, daystr):
os.chdir(org_log_dir)
print "DIR:", os.getcwd()
print "DAY:", daystr
stopwatch = handolUtil.StopWatch()
stopwatch.start()
analyzer = QcLogAnalyzer(daystr)
QCLOG = "qc_matched.all.%s.log" % daystr
analyzer.get_stats_from_qcmatched(QCLOG)
print "Loading & Statistics: %f sec" % (stopwatch.laptime())
stopwatch.start()
analyzer.print_stats('%s.all.csv' % (daystr))
print "Saving output: %f sec" % (stopwatch.laptime())
def do_many_days_s3_LHRHVH(argv):
if len(argv) < 3:
print "usage: start_day end_day"
print "usage: 2011-10-7 2011-10-13"
sys.exit()
daylist = handolUtil.get_day_list(argv[1],argv[2])
analyzer = LogAnalyzer("LHRHVH", "%s-%s" %(daylist[0], daylist[-1]))
stopwatch = handolUtil.StopWatch()
for daystr in daylist:
s3_file = "DEVELOPING/app/7nmc1m75ij/hubs1-log/query_log/osp_query.log.%s" % (daystr)
analyzer.querylog_from_s3("sch-emr", s3_file)
print "Loading & ETL: %f sec" % (stopwatch.laptime())
s3_file = "DEVELOPING/app/7nmc1m75ij/hubs2-log/query_log/osp_query.log.%s" % (daystr)
analyzer.querylog_from_s3("sch-emr", s3_file)
print "Loading & ETL: %f sec" % (stopwatch.laptime())
analyzer.get_stats()
print "Calc Stats: %f sec" % (stopwatch.laptime())
analyzer.write_info()
def load_keywords_stat(fname, resfname, wikif):
"""
"""
watch = handolUtil.StopWatch()
print "Loading ...", wikif
wikiStat = {}
fp = codecs.open(wikif, 'rb', encoding='utf-8')
for line in fp:
flds = line.split()
w = flds[0].lower().replace('_', ' ')
n = int(flds[1])
if not wikiStat.has_key(w):
wikiStat[w] = n
fp.close()
print "Loaded: %f" % (watch.laptime())
"""
== keywords.stat
3997563 270333 6410 facebook
2399787 186241 45449 whatsapp
1433213 71764 1458 whats app
1315718 130860 2040 temple run
1004372 163723 272 games
947351 87185 19846 skype
829801 316731 834 angry birds
529552 64385 1416 fruit ninja
"""
print "Loading ...", fname
highKeywords = DictDict()
noresKeywords = []
fp = codecs.open(fname, 'rb', encoding='utf-8')
for line in fp:
flds = line.split('\t')
q = int(flds[0])
c = int(flds[1])
nores = int(flds[2])
keyword = flds[3].strip()
kflds = keyword.split()
if len(kflds) > 2: continue
noresratio = (nores * 100) / q
if q > 10 and noresratio > 50:
if not wikiStat.has_key(keyword):
noresKeywords.append([keyword, q, noresratio])
if q > 150 and noresratio < 4:
highKeywords.insert(keyword, [q, noresratio])
fp.close()
print "Loaded: %f" % (watch.laptime())
print "noresKeywords: %d" % (len(noresKeywords))
print "highKeywords: %d" % (len(highKeywords.D))
fp = codecs.open(resfname, 'wb', encoding='utf-8')
for noresK in noresKeywords:
nearest = highKeywords.searchNearest(noresK[0], noresK[1])
if nearest:
prnformat = u'\t'.join(
map(lambda (d, x, y, z): "%d\t%s\t%d\t%d" % (d, x, y, z),
nearest))
fp.write("%s\t%d\t%d\t%s\n" %
(noresK[0], noresK[1], noresK[2], prnformat))
fp.close()
print "Processed: %f" % (watch.laptime())
stopwatch = handolUtil.StopWatch()
stopwatch.start()
analyzer = QcLogAnalyzer(daystr)
QCLOG = "qc_matched.all.%s.log" % daystr
analyzer.get_stats_from_qcmatched(QCLOG)
print "Loading & Statistics: %f sec" % (stopwatch.laptime())
stopwatch.start()
analyzer.print_stats('%s.all.csv' % (daystr))
print "Saving output: %f sec" % (stopwatch.laptime())
#analyzer.print_info('%s.csv' % (daystr))
#analyzer.save_to_mongo()
#analyzer.save_keywords()
if __name__ == "__main__":
import os
import sys
#os.chdir(sys.argv[1])
#daystr = sys.argv[2]
org_log_dir = u"D:/0001/9 org logs/20130215"
daystr = '20130215'
stopwatch = handolUtil.StopWatch()
stopwatch.start()
#save_qcmatched(org_log_dir, daystr)
load_qcmatched(org_log_dir, daystr)
print is_non_ascii('avcd')
print is_non_ascii('微信')
