def insert(mddid,name,box_dict):
dbconn.insert("mfwmdd",
mddid = mddid,
name = name,
post_main = box_dict["post_main"],
box_box_photo = box_dict["box_box_photo"],
box_master = box_dict["box_master"],
box_box_map = box_dict["box_box_map"],
box_baike = box_dict["box_baike"],
box_box_discuss = box_dict["box_box_discuss"],
box_box_plan = box_dict["box_box_plan"],
box_other_city = box_dict["box_other_city"],
box_box_book = box_dict["box_box_book"],
box_box_tips = box_dict["box_box_tips"],
poi_nav = box_dict["poi_nav"],
box_postBox = box_dict["box_postBox"],
city_hd = box_dict["city_hd"],
box_mb10 = box_dict["box_mb10"],
box_mod_rec = box_dict["box_mod_rec"],
box_hotBox2 = box_dict["box_hotBox2"],
box_pathBox2 = box_dict["box_pathBox2"],
box_pathBox1 = box_dict["box_pathBox1"],
box_hotBox1 = box_dict["box_hotBox1"],
info_cate = box_dict["info_cate"],
box_other_mdd = box_dict["box_other_mdd"]
)
def fileToNickname():
comp = re.compile(u"(\d+)_(\d+).html")
for feedDir in os.listdir(tempDir):
if not os.path.isdir(tempDir +"/"+ feedDir):
continue
for filename in os.listdir(tempDir +"/"+ feedDir):
m = comp.search(filename)
if not m:
continue
if int(m.group(2)) == 1:
userid = int(m.group(1))
if len(dbconn.query("select * from mfwuser where userid = $userid",vars=dict(userid=userid))) == 0:
nickname = getNickName(userid)
dbconn.insert("mfwuser",userid=userid,nickname=nickname)
def getPagesAndCal(userid,lastpage):
actDictList = []
for pagenumber in range(1,lastpage+1):
html = open(tempDir + "/" + str(userid)[0:2] +"/" +str(userid) + "_" + str(pagenumber) + ".html").read()
actDictList.extend( getFeed(html) )
if len(actDictList) == 0:
dbconn.insert("mfwuserfeed",
userid = userid,
pageCount=lastpage,
sumCount=0
)
return
if len(actDictList) == 1:
dbconn.insert("mfwuserfeed",
userid = userid,
pageCount=lastpage,
sumCount=1,
firstAct = actDictList[-1][0],
firstActTime = actDictList[-1][1],
mostAct = actDictList[-1][0],
actSummaryString = "$" + str(actDictList[-1][0]) + "|1",
actDense = 1,
dateDense = 1
)
return
##registryTime = getRegistryTime()
sumCount = len(actDictList)
firstAct = actDictList[-1][0]
firstActTime = actDictList[-1][1]
actSummary,dateSummary = getActDateSummary(actDictList)
mostAct = actSummary[0][0]
actSummaryString = summaryToString(actSummary)
actDense = calDense(actSummary)
dateDense = calDense(dateSummary)
longestPeriod, mostPeriod, deviation, avgPerd, middlePerd = calRate(dateSummary)
print "sumCount:" + str(sumCount)
print "firstAct:" + str(firstAct)
print "firstActTime:" + str(firstActTime)
print "mostAct:" + str(mostAct)
print "actSummaryString:" + str(actSummaryString)
print "actDense:" + str(actDense)
print "dateDense:" + str(dateDense)
print "longestPeriod:" + str(longestPeriod)
print "mostPeriod:" + str(mostPeriod)
print "deviation:" + str(deviation)
print "avgPerd:" + str(avgPerd)
print "middlePerd:" + str(middlePerd)
dbconn.insert("mfwuserfeed",
userid = userid,
pageCount=lastpage,
sumCount=sumCount,
firstAct=firstAct,
firstActTime = firstActTime,
mostAct = mostAct,
actSummaryString = actSummaryString,
actDense = actDense,
dateDense = dateDense,
longestPeriod = longestPeriod,
mostPeriod = mostPeriod,
deviation = deviation,
avgPerd = avgPerd,
middlePerd = middlePerd
)
def cal(className):
if isinstance(className,list):
tempName = ""
for c in className:
tempName = tempName + " " +c
className = tempName
if class_dict.has_key(className):
class_dict[className] = class_dict[className] + 1
else:
class_dict.update({className:1})
##for mddDir in os.listdir(tempMddDir):
## if not os.path.isdir(tempMddDir +"/"+ mddDir):
## continue
for filename in os.listdir(tempMddDir +"/1"):
ana(open(tempMddDir+"/1/"+filename,"r").read())
f = open("d:/log/mmd.log","wb")
for d in dictToOrderList(class_dict):
f.write(d[0] + "," + str(d[1])+ "\r\n")
"""
res = dbconn.query("select distinct pid from mfwmdd where pid is not null")
comp = re.compile(u"<title>(.+)地区旅游地图")
for r in res:
pid = r.pid
html = open(tempMddDir+"/"+str(pid)[0]+"/"+str(pid)+".html","r").read()
soup = BeautifulSoup(html,from_encoding="utf8")
name = comp.search(unicode(soup.title)).group(1)
dbconn.insert("mfwpid",pid=pid,name=name)
# coding:utf-8
from bs4 import BeautifulSoup
import bs4, os
from publicsettings import tempDir, dbconn
moveDict = {}
htmlfiles = os.listdir(tempDir)
for htmlfile in htmlfiles:
html = open(tempDir + "/" + htmlfile)
soup = BeautifulSoup(html, from_encoding="utf8")
news_lists = soup.find_all("div", "news_list")
for news_list in news_lists:
con = news_list.find("div", "con")
##print con.string
text = ""
for i in con.children:
if isinstance(i, bs4.element.NavigableString):
i = i.strip().encode("utf8")
if i <> "":
text = text + "$" + i
if text <> "":
if moveDict.has_key(text):
moveDict[text] = moveDict[text] + 1
else:
moveDict.update({text: 1})
for m in moveDict.keys():
dbconn.insert("mfwaction", name=m, count=moveDict[m])
