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

add image to media wordpress

Browse Source
This commit is contained in:
Valentin CZERYBA 2023-08-05 23:22:44 +02:00
parent 9ed5ffe399
commit e4eb1b6b68
1 changed files with 61 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

61
lib/WPImport.py
View File

@ -120,6 +120,67 @@ class WPimport:
albumtitle = albumbody.find("h2").get_text() albumtitle = albumbody.find("h2").get_text()
self._logger.debug("{0} : Title of the album : {1}".format(self._name, albumtitle)) self._logger.debug("{0} : Title of the album : {1}".format(self._name, albumtitle))
albumdesc = albumbody.find("div", class_="albumdesc").find("p") albumdesc = albumbody.find("div", class_="albumdesc").find("p")
img_a = albumbody.find_all("a", {"target": "_blank"})
if self._no_image is False:
for i in img_a:
new_img = {}
img = i.find_all("img")
self._logger.debug("{0} : Number of image's tag : {1}".format(self._name, len(img)))
if len(img) > 0:
href_a = i.get("href")
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_href"]=href_a
try:
if len(href_img_o.netloc) > 0:
img_ok = False
page_img = self._request.get(href_img)
if page_img.status_code == 404:
href_img = href_a
try:
page_img = self._request.get(href_a)
if page_img.status_code == 200:
img_ok = True
except ConnectionError as err:
self._logger.error("{0} : Connection error for get image : {1}".format(self._name, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for get image : {1}".format(self._name, err))
exit(1)
else:
if os.path.exists("{0}/..{1}".format(self._directory, href_img)):
page_img = open("{0}/..{1}".format(self._directory, href_img), "r")
img_ok = True
else:
if os.path.exists("{0}/..{1}".format(self._directory, href_a)):
page_img = open("{0}/..{1}".format(self._directory, href_a), "r")
img_ok = True
self._logger.debug("{0} : Status code for image {1} : {2}".format(self._name, href_img, page_img.status_code))
if img_ok is True:
media=self._addOrUpdateMedia(href_img, page_img)
new_img["id"]=media["id"]
new_img["new_src"]=media["rendered"]
list_img.append(new_img)
if href_img != href_a:
media=self._addOrUpdateMedia(href_a, page_img)
new_img["id"]=media["id"]
new_img["new_src"]=media["rendered"]
list_img.append(new_img)
if page_img.status_code not in [200, 404]:
self._logger.error("{0} : Connection error with status code for get image : {1}".format(self._name, page_img.status_code))
self._logger.debug("{0} : {1}".format(self._name, page_img.content))
except ConnectionError as err:
self._logger.error("{0} : Connection error for get image : {1}".format(self._name, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for get image : {1}".format(self._name, err))
exit(1)
def _fromFileTmp(self): def _fromFileTmp(self):
try: try: