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

diff path network

Browse Source
This commit is contained in:
Valentin CZERYBA 2023-06-13 22:00:51 +02:00
parent f01a69a1e7
commit 04da5bc5f6
1 changed files with 17 additions and 11 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

28
lib/WPImport.py
View File

@ -445,21 +445,27 @@ class WPimport:
if len(img) > 0: if len(img) > 0:
href_a = i.get("href") href_a = i.get("href")
href_img = img[0].get("src") href_img = img[0].get("src")
href_a_o = urlparse(href_a)
href_img_o = urlparse(href_img)
new_img["old_src"]=href_img new_img["old_src"]=href_img
new_img["old_href"]=href_a new_img["old_href"]=href_a
try: try:
page_img = self._request.get(href_img) if len(href_img_o.netloc) > 0:
page_img = self._request.get(href_img)
if page_img.status_code == 404: if page_img.status_code == 404:
href_img = href_a href_img = href_a
try: try:
page_img = self._request.get(href_a) page_img = self._request.get(href_a)
except ConnectionError as err: except ConnectionError as err:
self._logger.error("{0} : Connection error for get image : {1}".format(self._name, err)) self._logger.error("{0} : Connection error for get image : {1}".format(self._name, err))
exit(1) exit(1)
except Exception as err: except Exception as err:
self._logger.error("{0} : Exception error for get image : {1}".format(self._name, err)) self._logger.error("{0} : Exception error for get image : {1}".format(self._name, err))
exit(1) exit(1)
else:
if os.path.exists(href_img):
page_img = open(href_img, "r")
self._logger.debug("{0} : Status code for image {1} : {2}".format(self._name, href_img, page_img.status_code)) self._logger.debug("{0} : Status code for image {1} : {2}".format(self._name, href_img, page_img.status_code))
if page_img.status_code == 200: if page_img.status_code == 200:
media=self._addOrUpdateMedia(href_img, page_img) media=self._addOrUpdateMedia(href_img, page_img)