def run(self):
ie=PAMIE()
urlre=re.compile('docdetail_.*html')
while True:
try:
host=hosts.get()
try:
#url=urllib2.urlopen(host)
ie.navigate(host)
chunk=ie.getPageText()
except:
try:
ie.navigate(host)
chunk=ie.getPageText()
except:
ie.navigate(host)
chunk=ie.getPageText()
print 'souping+'+host+'\n'
links=urlre.findall(chunk)
for link in links:
totallink='http://www.microbell.com/'+link
if not self.isindexed(totallink):
pages.put(totallink)
#soup=BeautifulSoup(chunk,fromEncoding='GBK')
#areas=soup.findAll('div',{"class":"classbaogao_sousuo_new"})
#print '##########################'
#for area in areas:
# link=area.find('a',href=True)['href']
# totallink='http://www.microbell.com/'+link
# if not self.isindexed(totallink):
# pages.put(totallink)
# print 'END***************'
except:
writeindex=file('hostsErro.txt','a+')
writeindex.write(host+'\n')
writeindex.close()
continue
ie.quit()
def run(self):
ie1=PAMIE()
while True:
try:
TotalAbstract=''
page=pages.get()
try:
ie1.navigate(page)
chunk=ie1.getPageText()
except:
try:
ie1.navigate(page)
chunk=ie1.getPageText()
except:
ie1.navigate(page)
chunk=ie1.getPageText()
print 'souping+'+page+'\n'
soup=BeautifulSoup(chunk,fromEncoding='GBK')
soup1=soup.find('div',{"class":"baogaonews"})
chunks=soup1.findAll('td')
stockName=chunks[1].find('b').string #股票名称
if stockName==None:
stockName=str(stockName)
print stockName+'\n'
reportAuthor=chunks[3].find('b').string #研究报告作者
if reportAuthor==None:
reportAuthor=str(reportAuthor)
print reportAuthor+'\n'
stockCode=chunks[6].find('b').string #股票代码
if stockCode==None:
stockCode=str(stockCode)
print stockCode+'\n'
reportBelongCompany=chunks[8].find('b').string #研究报告出处
if reportBelongCompany==None:
reportBelongCompany=str(reportBelongCompany)
print reportBelongCompany+'\n'
rateLevel=chunks[11].find('b').string #推荐评级
if rateLevel==None:
rateLevel=str(rateLevel)
print rateLevel+'\n'
reportType=chunks[13].find('b').string #研究报告栏目
if reportType==None:
reportType=str(reportType)
print reportType+'\n'
Title=soup.find('strong').string
print Title+'\n'
time=chunks[18].find('b').string
tmptimes=time.split(' ')
time=tmptimes[0]
timechunks=time.split('-')
if len(timechunks)==3:
if len(timechunks[1])==1:
timechunks[1]='0'+timechunks[1]
if len(timechunks[2])==1:
timechunks[2]='0'+timechunks[2]
time=timechunks[0]+timechunks[1]+timechunks[2]
Title=Title
ContentAbstractArea=soup.find('div',{"class":"new_content"})
if ContentAbstractArea !=None:
tmp=ContentAbstractArea.find('td')
tmpStr=str(tmp)
abstract=tmpStr[34:-12].replace('<font style="DISPLAY:none;"><a href="http://www.microbell.com/">www.microbell.com</a>(\xe8\xbf\x88\xe5\x8d\x9a\xe6\xb1\x87\xe9\x87\x91)\xe3\x80\x82</font>','\n')
abstract=abstract.replace(' ','')
abstract=abstract.replace('<br />','')
abstract=abstract.replace('\n','')
TotalAbstract=TotalAbstract+abstract+'\n'
##INSERT THE INFO TYPE:GBK
print '#################################'
try:
self.cursor.execute("insert into MBStrategy values('%s','%s','%s','%s','%s','%s','%s','%s','%s')"%(page,Title,TotalAbstract.decode('utf-8'),time,reportAuthor,stockCode,reportBelongCompany,rateLevel,reportType))
self.conn.commit()
except Exception,e:
print e
print '************************************'
except:
def OperateACWebPage(self, bCA, lst):
#global Obj_WaitDialog_Cancel
#g_Group_Idx
#define 'ie' object to control web page
ie = PAMIE()
#check version of IE. different version have different operate.
nVer = self.getIEVersion(ie)
if nVer == -1:
log_public(WEB_ERR_NO_0002 + dstURL[KEY_URL])
self.m_ERROR_MSG = WEB_ERR_NO_0002 + dstURL[KEY_URL]
return False
#KillCA = True
KillCA = bCA
#loop to control every web page,all the control information storage at lst.
#we loop to get page information at 'lst' to control web pages.
#We call it one page_operate.
lstLen = len(lst)
for i in range(lstLen):
#define 'lstSub',this list go to storage web page information.
lstSub = []
lstSub.extend(lst[i])
#loop to control every control_components of one web page.
#we loop to get control_components at 'lstSub' to control control_components of one web pages.
#We call it line_operation.
#Obj_WaitDialog_Cancel = 0 #if dialog popup after you click a button.you need to use WAIT DIALOG EVENT. This variable will starage the object.
lstSubLen = len(lstSub)
for j in range(0, lstSubLen):
#get URL path
#URL path is storage at '1' index of 'lstSubLen',so judge 'j' wether equal 1.
if j == 0:
#define 'dstURL' as dictionary to storage URL and its value.
dstURL = {}
dstURL.update(lstSub[j])
#URL Value can be None sometime,we will not go to the page when the value is 'NONE'.
#if the value of 'KEY_URL' not NONE,we will got the page by 'navigate' function.
if dstURL[KEY_URL] != KEY_URL_NONE:
ret = ie.navigate(dstURL[KEY_URL])
if ret == False:
ie.quit()
#write logging
log_public(WEB_ERR_NO_0001 + dstURL[KEY_URL])
self.m_ERROR_MSG = WEB_ERR_NO_0001 + dstURL[KEY_URL]
return False
if KillCA == 'CAON':
#IE 8 need to click 'overridelink'
#IE 6 need to click 'CA authentication dialog'
if nVer == 8:
time.sleep(1)
ret = ie.clickLink('overridelink')
#encode
if ret == False:
ie.quit()
#write logging
log_public(WEB_ERR_NO_0001 +
dstURL[KEY_URL])
self.m_ERROR_MSG = WEB_ERR_NO_0001 + dstURL[
KEY_URL]
return False
KillCA = False #not need to kill next time open URL
elif nVer == 6:
time.sleep(1)
KillCA = False
else:
#get line_operation.
dstComb = {}
dstComb.update(lstSub[j])
#if 'ASSIST' have '//',This line_operation will be ignore.
if dstComb[KEY_ASSIST] == KEY_COMMENT:
continue
#dispatch line_operation.include kinds(set text box, set check box,select list box,...)
Value = dstComb[KEY_CONTROLTYPE]
if Value == KEY_WEB_TYPE_TEXTBOX:
if ie.setTextBox(dstComb[KEY_CONTROLNAME],
dstComb[KEY_CONTROLVALUE]) == False:
log_public(WEB_ERR_NO_0003 +
dstComb[KEY_CONTROLNAME])
self.m_ERROR_MSG = WEB_ERR_NO_0003 + dstComb[
KEY_CONTROLNAME]
return False
elif Value == KEY_WEB_TYPE_CHECKBOX:
if self.MainCheckBox(
ie, dstComb[KEY_ASSIST],
dstComb[KEY_CONTROLNAME],
dstComb[KEY_CONTROLVALUE]) == False:
log_public(WEB_ERR_NO_0004 +
dstComb[KEY_CONTROLNAME])
self.m_ERROR_MSG = WEB_ERR_NO_0004 + dstComb[
KEY_CONTROLNAME]
return False
elif Value == KEY_WEB_TYPE_LISTBOX:
if self.MainListBox(
ie, dstComb[KEY_ASSIST],
dstComb[KEY_CONTROLNAME],
dstComb[KEY_CONTROLVALUE]) == False:
log_public(WEB_ERR_NO_0005 +
dstComb[KEY_CONTROLNAME])
self.m_ERROR_MSG = WEB_ERR_NO_0005 + dstComb[
KEY_CONTROLNAME]
return False
elif Value == KEY_WEB_TYPE_BUTTON:
if self.MainButton(ie, dstComb[KEY_ASSIST],
dstComb[KEY_CONTROLNAME]) == False:
log_public(WEB_ERR_NO_0006 +
dstComb[KEY_CONTROLNAME])
self.m_ERROR_MSG = WEB_ERR_NO_0006 + dstComb[
KEY_CONTROLNAME]
False
elif Value == KEY_WEB_TYPE_WAIT:
if self.waitEvent(dstComb[KEY_CONTROLNAME],
dstComb[KEY_CONTROLVALUE]) == False:
log_public(WEB_ERR_NO_0007 +
dstComb[KEY_CONTROLNAME])
self.m_ERROR_MSG = WEB_ERR_NO_0007 + dstComb[
KEY_CONTROLNAME]
return False
elif Value == KEY_WEB_TYPE_JAVASCRIPT:
if self.MainJavaScript(
ie, dstComb[KEY_ASSIST],
dstComb[KEY_CONTROLNAME]) == False:
log_public(WEB_ERR_NO_0008 +
dstComb[KEY_CONTROLNAME])
self.m_ERROR_MSG = WEB_ERR_NO_0008 + dstComb[
KEY_CONTROLNAME]
return False
elif Value == KEY_WEB_TYPE_GROUP_CHECKBOX:
if self.GetGoupCheckBoxIndex(
ie, dstComb[KEY_CONTROLNAME]) == False:
log_public(WEB_ERR_NO_0009 +
dstComb[KEY_CONTROLNAME])
self.m_ERROR_MSG = WEB_ERR_NO_0009 + dstComb[
KEY_CONTROLNAME]
return False
#close ie after operation.
ie.quit()
from _winreg import *
def setProxy(proxy):
keyVal = 'Software\Microsoft\Windows\CurrentVersion\Internet Settings'
key = OpenKey(HKEY_CURRENT_USER, keyVal, 0, KEY_ALL_ACCESS)
SetValueEx(key, 'ProxyEnable', 0, REG_SZ, "1")
SetValueEx(key, 'ProxyServer', 0, REG_SZ, proxy)
CloseKey(key)
setProxy('64.212.73.53:8080')
from PAM30 import PAMIE
import cModalPopUp
import winGuiAuto
ie = PAMIE()
ie.navigate("http://gzjd.shenyangsheying.com/View.asp?id=20")
ie.clickImage("give_ticket.jpg")
ok = cModalPopUp.handlePopup("Confirm", "È·¶¨")
ok.start()
ie.quit()
'''
hwnd = winGuiAuto.findTopWindows("Windows Internet Explorer")
for it in hwnd :
control=winGuiAuto.dumpWindow(it)
for item in control:
if item[1]=='È·¶¨':
winGuiAuto.clickButton(item[0])
'''
def open(self, url, title):
PAMIE(url)
time.sleep(self.wait)
win32gui.EnumWindows(self.enumHandler, title)
self.isOpen = True
