add url and file
This commit is contained in:
@@ -34,23 +34,32 @@ class WPMenu:
|
||||
def __del__(self):
|
||||
print("{0} : Import finished for {1}".format(self._name, self._wordpress))
|
||||
|
||||
# Public method
|
||||
# Public method
|
||||
|
||||
## From file
|
||||
|
||||
def fromFile(self, files):
|
||||
if os.path.exists(files):
|
||||
with open(files, 'r') as f:
|
||||
self._logger.info("{0} : File is being processed : {1}".format(self._name, files))
|
||||
content = f.read()
|
||||
self._menu(content)
|
||||
else:
|
||||
self._logger.error("{0} : File isn't exist : {1}".format(self._name, files))
|
||||
|
||||
|
||||
|
||||
## Get from URL
|
||||
|
||||
def fromUrl(self):
|
||||
def fromUrl(self, canalblog):
|
||||
try:
|
||||
o = urlparse(self._canalblog)
|
||||
o = urlparse(canalblog)
|
||||
o = o._replace(scheme=self._protocol_canalblog)
|
||||
i = o.geturl().replace(":///", "://")
|
||||
page = self._request.get(i)
|
||||
if page.status_code == 200:
|
||||
soup = BeautifulSoup(page.text, self._parser)
|
||||
ul = soup.find_all("ul", id="listsmooth")
|
||||
for anchor in ul[0].find_all("li"):
|
||||
li = anchor.find_all("li")
|
||||
for content in li:
|
||||
a = content.find("a")
|
||||
self._logger.info("{0} : Page web is being processed : {1}".format(self._name, i))
|
||||
self._menu(page.content)
|
||||
else:
|
||||
self._logger.error("{0} : index didn't get due status code : {1}".format(self._name, page.status_code))
|
||||
self._logger.debug("{0} : {1}".format(self._name, page.content))
|
||||
@@ -58,4 +67,14 @@ class WPMenu:
|
||||
self._logger.error("{0} : Connection error for get url {1} : {2}".format(self._name, webpage[i], err))
|
||||
exit(1)
|
||||
except Exception as err:
|
||||
self._logger.error("{0} : Exception error for get url {1} : {2}".format(self._name, webpage[i], err))
|
||||
self._logger.error("{0} : Exception error for get url {1} : {2}".format(self._name, webpage[i], err))
|
||||
|
||||
|
||||
def _menu(self, content):
|
||||
|
||||
soup = BeautifulSoup(page.text, self._parser)
|
||||
ul = soup.find_all("ul", id="listsmooth")
|
||||
for anchor in ul[0].find_all("li"):
|
||||
li = anchor.find_all("li")
|
||||
for content in li:
|
||||
a = content.find("a")
|
||||
|
Reference in New Issue
Block a user