v4l3n71n/web_scrap
1
0
Fork 0
Code Issues Pull Requests Projects Releases 10 Wiki Activity

fix webpage

Browse Source
This commit is contained in:
Valentin CZERYBA
2023-09-05 22:06:14 +02:00
parent 3059f785c2
commit eaec1ba9d4
2 changed files with 17 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

20
lib/WPExport.py
View File

@@ -58,14 +58,26 @@ class WPExport:
# Download HTML
def downloadHTML(self, first, second):
self._downloadPage(webpage[first][second], self._dir)
try:
with open("{0}/{1}.json".format(self._tmp, self._name)) as file:
webpage = json.loads(file.read())
self._downloadPage(webpage[first][second], self._dir)
except Exception as ex:
self._logger.error("{0} : Read file json from tmp : {1}".format(self._name, ex))
# Download Image
def downloadImg(self, first, second):
page_src = self._getImg(webpage[first][second])
o = urlparse(self._url)
self._downloadPage(page_src, "{0}/{1}/{2}".format(self._dir, o.path, "img"))
try:
with open("{0}/{1}.json".format(self._tmp, self._name)) as file:
webpage = json.loads(file.read())
page_src = self._getImg(webpage[first][second])
o = urlparse(self._url)
self._downloadPage(page_src, "{0}/{1}/{2}".format(self._dir, o.path, "img"))
except Exception as ex:
self._logger.error("{0} : Read file json from tmp : {1}".format(self._name, ex))
# Get URL