def crawl(self):
_pageSource = self.navigation()
# read desired values
self.Volume = self.extractUlLi(_pageSource, "Volume")
yearRange = self.extractUlLi(_pageSource,
"52Wk") # Expected format Low-high
if '-' in yearRange:
self.YearHigh = tools.cast(yearRange.split('-')[1].strip(), float)
self.YearLow = tools.cast(yearRange.split('-')[0].strip(), float)
PE_EPS = self.extractUlLi(_pageSource,
"P/E") # expected format P/E (EPS)
if '(' in PE_EPS:
self.PE = tools.cast(PE_EPS.split('(')[0].strip(), float)
self.EPS = tools.cast(
PE_EPS.split('(')[1].replace(')', '').strip(), float)
self.Price = self.readPrice(_pageSource)
return
def readPagingLength(self, pageSource):
# read the pagination section which is represented as a ul HTML element
# with nav-page class name
_soup = soup.Helper()
_block = _soup.elemSelector("ul", {"class": "nav-page"}, pageSource)
# expected format
# <ul>
# <li> <button value="page number">
# ...
# </ul>
_max = 0
_pageNum = 0
# find the biggest paging number
for _li in _block.find_all("li"):
_btn = _li.find("button")
if _btn is not None:
_pageNum = tools.cast(_btn["value"], int, 0)
if _pageNum > _max:
_max = _pageNum
self.PAGINGLENGTH = _max
return
def isMyFinancialReport(self, financialObj):
# reported quarterly or yearly data belong to the last quarter or year
_myDate = tools.cast(self.Date, datetime)
if type(_myDate) != datetime:
print("Date format is incorrect")
return False
# define paterns
regex_quarter = r"(^\d{4}) ([Q][1-4]$)" # Year Q[1-4]
if re.search(regex_quarter, financialObj.Period):
# check if the financial report is maximum 3 month older than the price date
_year = financialObj.Period[:4]
_quarter = financialObj.Period[4:].strip()
if _quarter.lower() == "q1":
return str(_myDate.year) == _year and _myDate.month - 3 <= 6
elif _quarter.lower() == "q2":
return str(_myDate.year) == _year and _myDate.month - 6 <= 6
elif _quarter.lower() == "q3":
return str(_myDate.year -
1) == _year and _myDate.month - 9 <= 6
elif _quarter.lower() == "q4":
return str(_myDate.year -
1) == _year and 12 - _myDate.month >= 6
return False
def readPrice(self, pageSource):
_soup = soup.Helper()
_block = _soup.convertToSoup(pageSource)
_priceElem = _block.find("span", {"class": "currentval"})
return tools.cast(_priceElem.text, float)
def setPPEBV(self):
self.PPEBV = round(
tools.cast(self.PBV, float, 0) *
tools.cast(self.PriceInfo.PE, float, 0))
def setPBV(self):
if self.BookValue == None:
return
self.PBV = round((tools.cast(self.PriceInfo.Price, float, 0) /
tools.cast(self.BookValue, float, 1)), 2)
return