8 Commits

Author SHA1 Message Date
269a9e9ccd add replace exception import 2023-05-28 22:31:46 +02:00
4c0ec09d91 move exception 2023-05-28 22:07:43 +02:00
42cfb30583 Merge pull request 'remove-thread' (#14) from remove-thread into master
Reviewed-on: #14
2023-05-26 22:18:19 +00:00
c76b20e64a add remove multithread 2023-05-27 00:16:41 +02:00
aff69bfcbc add multithread for remove 2023-05-27 00:06:11 +02:00
fd426f150d add variable 2023-05-26 17:50:57 +02:00
e21721cac1 move exception 2023-05-26 17:44:28 +02:00
69504687ef add count 2023-05-26 16:38:19 +02:00
3 changed files with 401 additions and 309 deletions

View File

@@ -10,8 +10,8 @@ from lib.WPImport import WPimport
from lib.WPExport import WPExport from lib.WPExport import WPExport
from lib.WPRemove import WPRemove from lib.WPRemove import WPRemove
def remove(args, basic, logger, ssl_wordpress): def remove(index, number, args, basic, logger, ssl_wordpress):
removeWp = WPRemove(basic=basic, wordpress="", logger=logger, ssl_wordpress=ssl_wordpress) removeWp = WPRemove(basic=basic, wordpress="", logger=logger, ssl_wordpress=ssl_wordpress, index_name=index, number_thread=number)
if args.remove == True: if args.remove == True:
for i in args.wordpress.split(","): for i in args.wordpress.split(","):
removeWp.setUrl(i) removeWp.setUrl(i)
@@ -280,5 +280,9 @@ if __name__ == '__main__':
if args.command == "remove": if args.command == "remove":
remove(args, basic, logger, ssl_wordpress) try:
with futures.ThreadPoolExecutor(max_workers=int(args.parallel)) as ex:
wait_for = [ ex.submit(remove, i, args.parallel, args, basic, logger, ssl_wordpress) for i in range(0, int(args.parallel)) ]
except Exception as err:
logger.error("Thread error for remove : {0}".format(err))
exit(0) exit(0)

View File

@@ -39,20 +39,22 @@ class WPimport:
for i in range(0, len(webpage)): for i in range(0, len(webpage)):
try: try:
r = self._request.get(webpage[i]) r = self._request.get(webpage[i])
except Exception as err: if r.status_code == 200:
self._logger.info("{0} : ({1}/{2}) : Page is importing : {3}".format(self._name, i+1, len(webpage), webpage[i]))
soup = BeautifulSoup(r.content, self._parser)
articlebody = soup.find_all("div", class_="articlebody")
if len(articlebody) > 0:
self._addOrUpdatePost(soup)
else:
self._addOrUpdateFeaturedMedia(soup)
else:
self._logger.error("{0} : Connection error for get url {1} with status code : {2}".format(self._name, webpage[i], r.status_code))
self._logger.debug("{0} : {1}".format(self._name, r.content))
except ConnectionError as err:
self._logger.error("{0} : Connection error for get url {1} : {2}".format(self._name, webpage[i], err)) self._logger.error("{0} : Connection error for get url {1} : {2}".format(self._name, webpage[i], err))
exit(1) exit(1)
if r.status_code == 200: except Exception as err:
self._logger.info("{0} : ({1}/{2}) : Page is importing : {3}".format(self._name, i+1, len(webpage), webpage[i])) self._logger.error("{0} : Exception error for get url {1} : {2}".format(self._name, webpage[i], err))
soup = BeautifulSoup(r.content, self._parser)
articlebody = soup.find_all("div", class_="articlebody")
if len(articlebody) > 0:
self._addOrUpdatePost(soup)
else:
self._addOrUpdateFeaturedMedia(soup)
else:
self._logger.error("{0} : Connection error for get url {1} with status code : {2}".format(self._name, webpage[i], r.status_code))
self._logger.debug("{0} : {1}".format(self._name, r.content))
def fromDirectory(self, directory="", number_thread=1, max_thread=1): def fromDirectory(self, directory="", number_thread=1, max_thread=1):
@@ -137,60 +139,65 @@ class WPimport:
params = {"search":h2, "type":"post"} params = {"search":h2, "type":"post"}
try: try:
page = self._request.get("{1}://{0}/wp-json/wp/v2/search".format(self._wordpress, self._protocol), auth=self._basic, params=params) page = self._request.get("{1}://{0}/wp-json/wp/v2/search".format(self._wordpress, self._protocol), auth=self._basic, params=params)
except Exception as err: if page.status_code == 200:
result = page.json()
if len(result) > 0:
if h2 == result[0]["title"]:
img = i.find_all("img")
if len(img) > 0:
img_src = img[0].get("src")
try:
page = self._request.get(img_src)
if page.status_code == 200:
name_img = img_src.replace("_q", "")
name_img = name_img.split("/")[len(name_img.split("/"))-1]
params = {"search": name_img}
try:
page = self._request.get("{1}://{0}/wp-json/wp/v2/media".format(self._wordpress, self._protocol), auth=self._basic, params=params)
if page.status_code == 200:
res = page.json()
if len(res) > 0:
id_media = res[0]["id"]
data = {"featured_media": id_media}
try:
r = self._request.post("{2}://{0}/wp-json/wp/v2/posts/{1}".format(self._wordpress, result[0]["id"], self._protocol), auth=self._basic, headers=self._headers_json, data=json.dumps(data))
if r.status_code == 200:
self._logger.info("{0} : Add media featured : {1}".format(self._name, r.json()["title"]["raw"]))
else:
self._logger.error("{0} : Connection error with status code for featured media : {1}".format(self._name, r.status_code))
self._logger.debug("{0} : {1}".format(self._name, r.content))
except ConnectionError as err:
self._logger.error("{0} : Connection error for post media featured : {1}".format(self._name, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for post media featured : {1}".format(self._name, err))
else:
self._logger.info("{0} : No media found for {1}".format(self._name, h2))
else:
self._logger.error("{0} : Connection error with status code for search featured media: {1}".format(self._name, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
except ConnectionError as err:
self._logger.error("{0} : Connection error search featured media : {1}".format(self._name, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error search featured media : {1}".format(self._name, err))
else:
self._logger.error("{0} : Connection error for get featured media with status code : {1}".format(self._name, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
except ConnectionError as err:
self._logger.error("{0} : Connection error for get featured media : {1}".format(self._name, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for get featured media : {1}".format(self._name, err))
else:
self._logger.error("{0} : Connection error with status code for featured media : {1}".format(self._name, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
except ConnectionError as err:
self._logger.error("{0} : Connection error : {1}".format(self._name, err)) self._logger.error("{0} : Connection error : {1}".format(self._name, err))
exit(1) exit(1)
if page.status_code == 200: except Exception as err:
result = page.json() self._logger.error("{0} : Connection error : {1}".format(self._name, err))
if len(result) > 0:
if h2 == result[0]["title"]:
img = i.find_all("img")
if len(img) > 0:
img_src = img[0].get("src")
try:
page = self._request.get(img_src)
except Exception as err:
self._logger.error("{0} : Connection error for get featured media : {1}".format(self._name, err))
exit(1)
if page.status_code == 200:
name_img = img_src.replace("_q", "")
name_img = name_img.split("/")[len(name_img.split("/"))-1]
params = {"search": name_img}
try:
page = self._request.get("{1}://{0}/wp-json/wp/v2/media".format(self._wordpress, self._protocol), auth=self._basic, params=params)
except Exception as err:
self._logger.error("{0} : Connection error search featured media : {1}".format(self._name, err))
exit(1)
if page.status_code == 200:
res = page.json()
if len(res) > 0:
id_media = res[0]["id"]
data = {"featured_media": id_media}
try:
r = self._request.post("{2}://{0}/wp-json/wp/v2/posts/{1}".format(self._wordpress, result[0]["id"], self._protocol), auth=self._basic, headers=self._headers_json, data=json.dumps(data))
except Exception as err:
self._logger.error("{0} : Connection error for post media featured : {1}".format(self._name, err))
exit(1)
if r.status_code == 200:
self._logger.info("{0} : Add media featured : {1}".format(self._name, r.json()["title"]["raw"]))
else:
self._logger.error("{0} : Connection error with status code for featured media : {1}".format(self._name, r.status_code))
self._logger.debug("{0} : {1}".format(self._name, r.content))
else:
self._logger.info("{0} : No media found for {1}".format(self._name, h2))
else:
self._logger.error("{0} : Connection error with status code for search featured media: {1}".format(self._name, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
else:
self._logger.error("{0} : Connection error for get featured media with status code : {1}".format(self._name, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
else:
self._logger.error("{0} : Connection error with status code for featured media : {1}".format(self._name, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
## Association image to post ## Association image to post
@@ -199,14 +206,16 @@ class WPimport:
data = {"post": post_id} data = {"post": post_id}
try: try:
r = self._request.post("{2}://{0}/wp-json/wp/v2/media/{1}".format(self._wordpress, i["id"], self._protocol), auth=self._basic, data=data) r = self._request.post("{2}://{0}/wp-json/wp/v2/media/{1}".format(self._wordpress, i["id"], self._protocol), auth=self._basic, data=data)
except Exception as err: if r.status_code == 200:
self._logger.info("{0} : Link image to post {1}".format(self._name, title))
else:
self._logger.error("{0} Connection error with status code for link image to post : {1}".format(self._name, r.status_code))
self._logger.debug("{0} : {1}".format(self._name, r.content))
except ConnectionError as err:
self._logger.error("{0} : Connection error for link image to post : {1}".format(self._name, err)) self._logger.error("{0} : Connection error for link image to post : {1}".format(self._name, err))
exit(1) exit(1)
if r.status_code == 200: except Exception as err:
self._logger.info("{0} : Link image to post {1}".format(self._name, title)) self._logger.error("{0} : Exception error for link image to post : {1}".format(self._name, err))
else:
self._logger.error("{0} Connection error with status code for link image to post : {1}".format(self._name, r.status_code))
self._logger.debug("{0} : {1}".format(self._name, r.content))
## Add or update img ## Add or update img
@@ -226,49 +235,57 @@ class WPimport:
params = { "search": img_name} params = { "search": img_name}
try: try:
r = self._request.get("{1}://{0}/wp-json/wp/v2/media".format(self._wordpress, self._protocol), auth=self._basic, params=params) r = self._request.get("{1}://{0}/wp-json/wp/v2/media".format(self._wordpress, self._protocol), auth=self._basic, params=params)
except Exception as err:
self._logger.debug("{0} : Search for image {1} and his status code {2}".format(self._name, img_name, r.status_code))
if r.status_code == 200:
res = r.json()
self._logger.debug("{0} : Number of image in search : {1}".format(self._name, len(res)))
if len(res) > 0:
params = {"force":1}
try:
r = self._request.delete("{2}://{0}/wp-json/wp/v2/media/{1}".format(self._wordpress, res[0]["id"], self._protocol), auth=self._basic, params=params)
if r.status_code == 200:
self._logger.info("{0} : Image removed {1}".format(self._name, img_name))
else:
self._logger.error("{0} : Image {1} not removed due status code : {2}".format(self._name, img_name, r.status_code))
self._logger.debug("{0} : {1}".format(self._name, r.content))
except ConnectionError as err:
self._logger.error("{0} Connection error for delete image : {1}".format(self._name, err))
exit(1)
except Exception as err:
self._logger.error("{0} Exception error for delete image : {1}".format(self._name, err))
data = page.content
img_type = "image/{0}".format(img_type_file)
if img_type_file == "jpg":
img_type = "image/jpeg"
headers={ 'Content-Type': img_type,'Content-Disposition' : 'attachment; filename={0}'.format(img_name)}
try:
r = self._request.post("{1}://{0}/wp-json/wp/v2/media".format(self._wordpress, self._protocol), auth=self._basic, headers=headers, data=data)
if r.status_code == 201:
self._logger.info("{0} : Image added {1}".format(self._name, img_name))
res = r.json()
media["id"] = res["id"]
media["rendered"] = res["guid"]["rendered"]
else:
self._logger.error("{0} : Image {1}.{2} not added due status code : {3}".format(self._name, img_name, img_type, r.status_code))
self._logger.debug("{0} : {1}".format(self._name, r.content))
except ConnectionError as err:
self._logger.error("{0} : Connection error for add image : {1}".format(self._name, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for add image : {1}".format(self._name, err))
exit(1)
else:
self._logger.error("{0} : Connection error for search image with status code : {1}".format(self._name, r.status_code))
self._logger.debug("{0} : {1}".format(self._name, r.content))
except ConnectionError as err:
self._logger.error("{0} : Connection error for search media : {1}".format(self._name, err)) self._logger.error("{0} : Connection error for search media : {1}".format(self._name, err))
exit(1) exit(1)
self._logger.debug("{0} : Search for image {1} and his status code {2}".format(self._name, img_name, r.status_code)) except Exception as err:
if r.status_code == 200: self._logger.error("{0} : Exception error for search media : {1}".format(self._name, err))
res = r.json()
self._logger.debug("{0} : Number of image in search : {1}".format(self._name, len(res)))
if len(res) > 0:
params = {"force":1}
try:
r = self._request.delete("{2}://{0}/wp-json/wp/v2/media/{1}".format(self._wordpress, res[0]["id"], self._protocol), auth=self._basic, params=params)
except Exception as err:
self._logger.error("{0} Connection error for delete image : {1}".format(self._name, err))
exit(1)
if r.status_code == 200:
self._logger.info("{0} : Image removed {1}".format(self._name, img_name))
else:
self._logger.error("{0} : Image {1} not removed due status code : {2}".format(self._name, img_name, r.status_code))
self._logger.debug("{0} : {1}".format(self._name, r.content))
data = page.content
img_type = "image/{0}".format(img_type_file)
if img_type_file == "jpg":
img_type = "image/jpeg"
headers={ 'Content-Type': img_type,'Content-Disposition' : 'attachment; filename={0}'.format(img_name)}
try:
r = self._request.post("{1}://{0}/wp-json/wp/v2/media".format(self._wordpress, self._protocol), auth=self._basic, headers=headers, data=data)
except Exception as err:
self._logger.error("{0} : Connection error for add image : {1}".format(self._name, err))
exit(1)
if r.status_code == 201:
self._logger.info("{0} : Image added {1}".format(self._name, img_name))
res = r.json()
media["id"] = res["id"]
media["rendered"] = res["guid"]["rendered"]
else:
self._logger.error("{0} : Image {1}.{2} not added due status code : {3}".format(self._name, img_name, img_type, r.status_code))
self._logger.debug("{0} : {1}".format(self._name, r.content))
else:
self._logger.error("{0} : Connection error for search image with status code : {1}".format(self._name, r.status_code))
self._logger.debug("{0} : {1}".format(self._name, r.content))
return media return media
## Add or update comment ## Add or update comment
@@ -279,28 +296,34 @@ class WPimport:
try: try:
params = {"post": post, "author_name":i["author"], "date":i["date"]} params = {"post": post, "author_name":i["author"], "date":i["date"]}
page = self._request.get("{1}://{0}/wp-json/wp/v2/comments".format(self._wordpress, self._protocol), auth=self._basic, params=params) page = self._request.get("{1}://{0}/wp-json/wp/v2/comments".format(self._wordpress, self._protocol), auth=self._basic, params=params)
except Exception as err:
if page.status_code == 200:
result = page.json()
for j in result:
try:
params = {"force":1}
page = self._request.delete("{2}://{0}/wp-json/wp/v2/comments/{1}".format(self._wordpress, j["id"], self._protocol), params=params, auth=self._basic)
if page.status_code == 200:
self._logger.info("{0} : Comment deleted for {1}".format(self._name, title))
self._logger.debug("{0} : Comment deleted : {1}".format(self._name, j))
else:
self._logger.error("{0} : Comment not deleted for {1} due status code : {2}".format(self._name, title, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
except ConnectionError as err:
self._logger.error("{0} : Connection error for delete comment : {1}".format(self._name, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for delete comment : {1}".format(self._name, err))
else:
self._logger.error("{0} : Comment not listed for {1} due status code : {2}".format(self._name, title, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
except ConnectionError as err:
self._logger.error("{0} : Connection error for search comment : {1}".format(self._name, err)) self._logger.error("{0} : Connection error for search comment : {1}".format(self._name, err))
exit(1) exit(1)
if page.status_code == 200: except Exception as err:
result = page.json() self._logger.error("{0} : Exception error for search comment : {1}".format(self._name, err))
for j in result:
try:
params = {"force":1}
page = self._request.delete("{2}://{0}/wp-json/wp/v2/comments/{1}".format(self._wordpress, j["id"], self._protocol), params=params, auth=self._basic)
except Exception as err:
self._logger.error("{0} : Connection error for delete comment : {1}".format(self._name, err))
exit(1)
if page.status_code == 200:
self._logger.info("{0} : Comment deleted for {1}".format(self._name, title))
self._logger.debug("{0} : Comment deleted : {1}".format(self._name, j))
else:
self._logger.error("{0} : Comment not deleted for {1} due status code : {2}".format(self._name, title, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
else:
self._logger.error("{0} : Comment not listed for {1} due status code : {2}".format(self._name, title, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
for i in comment: for i in comment:
data = {"post": post, "content": i["content"], "date": i["date"], "author_name": i["author"], "status": "approved"} data = {"post": post, "content": i["content"], "date": i["date"], "author_name": i["author"], "status": "approved"}
@@ -310,28 +333,35 @@ class WPimport:
params = {"post": post, "author_name":comment[parent_id]["author"], "date":comment[parent_id]["date"]} params = {"post": post, "author_name":comment[parent_id]["author"], "date":comment[parent_id]["date"]}
try: try:
page = self._request.get("{1}://{0}/wp-json/wp/v2/comments".format(self._wordpress, self._protocol), auth=self._basic, params=params) page = self._request.get("{1}://{0}/wp-json/wp/v2/comments".format(self._wordpress, self._protocol), auth=self._basic, params=params)
except Exception as err:
if page.status_code == 200:
result = page.json()
if len(result) > 0:
data["parent"]=result[0]["id"]
else:
self._logger.error("{0} : Connection error for parent comment with status code : {1}".format(self._name, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
except ConnectionError as err:
self._logger.error("{0} : Connection error for parent comment : {1}".format(self._name, err)) self._logger.error("{0} : Connection error for parent comment : {1}".format(self._name, err))
exit(1) exit(1)
if page.status_code == 200: except Exception as err:
result = page.json() self._logger.error("{0} : Exception error for parent comment : {1}".format(self._name, err))
if len(result) > 0:
data["parent"]=result[0]["id"]
else:
self._logger.error("{0} : Connection error for parent comment with status code : {1}".format(self._name, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
try: try:
page = self._request.post("{1}://{0}/wp-json/wp/v2/comments".format(self._wordpress, self._protocol), auth=self._basic, data=data) page = self._request.post("{1}://{0}/wp-json/wp/v2/comments".format(self._wordpress, self._protocol), auth=self._basic, data=data)
except Exception as err:
if page.status_code == 201:
self._logger.info("{0} : Comment added for {1}".format(self._name, title))
self._logger.debug("{0} : Data : {1}".format(self._name, data))
else:
self._logger.error("{0} : Comment not added for {1} due status code : {2}".format(self._name, title, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
except ConnectionError as err:
self._logger.error("{0} : Connection error for add comment : {1}".format(self._name, err)) self._logger.error("{0} : Connection error for add comment : {1}".format(self._name, err))
exit(1) exit(1)
if page.status_code == 201: except Exception as err:
self._logger.info("{0} : Comment added for {1}".format(self._name, title)) self._logger.error("{0} : Exception error for add comment : {1}".format(self._name, err))
self._logger.debug("{0} : Data : {1}".format(self._name, data)) exit(1)
else:
self._logger.error("{0} : Comment not added for {1} due status code : {2}".format(self._name, title, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
## Check class name ## Check class name
@@ -416,30 +446,37 @@ class WPimport:
new_img["old_href"]=href_a new_img["old_href"]=href_a
try: try:
page_img = self._request.get(href_img) page_img = self._request.get(href_img)
except Exception as err:
self._logger.error("{0} : Connection error for get image : {1}".format(self._name, err)) if page_img.status_code == 404:
exit(1) href_img = href_a
if page_img.status_code == 404: try:
href_img = href_a page_img = self._request.get(href_a)
try: except ConnectionError as err:
page_img = self._request.get(href_a) self._logger.error("{0} : Connection error for get image : {1}".format(self._name, err))
except Exception as err: exit(1)
self._logger.error("{0} : Connection error for get image : {1}".format(self._name, err)) except Exception as err:
exit(1) self._logger.error("{0} : Exception error for get image : {1}".format(self._name, err))
self._logger.debug("{0} : Status code for image {1} : {2}".format(self._name, href_img, page_img.status_code)) exit(1)
if page_img.status_code == 200: self._logger.debug("{0} : Status code for image {1} : {2}".format(self._name, href_img, page_img.status_code))
media=self._addOrUpdateMedia(href_img, page_img) if page_img.status_code == 200:
new_img["id"]=media["id"] media=self._addOrUpdateMedia(href_img, page_img)
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["id"]=media["id"]
new_img["new_src"]=media["rendered"] new_img["new_src"]=media["rendered"]
list_img.append(new_img) list_img.append(new_img)
if page_img.status_code not in [200, 404]: if href_img != href_a:
self._logger.error("{0} : Connection error with status code for get image : {1}".format(self._name, page_img.status_code)) media=self._addOrUpdateMedia(href_a, page_img)
self._logger.debug("{0} : {1}".format(self._name, page_img.content)) 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)
self._logger.debug("{0} : Number of image : {1}".format(self._name, len(list_img))) self._logger.debug("{0} : Number of image : {1}".format(self._name, len(list_img)))
comment_post = self._getComment(comment) comment_post = self._getComment(comment)
@@ -462,36 +499,36 @@ class WPimport:
try: try:
params = {"search":title_element, "per_page":"100", "page":index} params = {"search":title_element, "per_page":"100", "page":index}
page = self._request.get("{2}://{0}/wp-json/wp/v2/{1}".format(self._wordpress, i, self._protocol), auth=self._basic, params=params) page = self._request.get("{2}://{0}/wp-json/wp/v2/{1}".format(self._wordpress, i, self._protocol), auth=self._basic, params=params)
if page.status_code == 200:
result = page.json()
self._logger.debug("{0} : content {3} {2} : {1}".format(self._name, result, title_element, i))
if len(result) > 0:
for k in result:
title_rendered = k["name"]
self._logger.debug("{0} : content {2} : {1}".format(self._name, title_rendered, i))
self._logger.debug("{0} : size of content {3} : {2} - {1}".format(self._name, len(title_rendered), len(title_element), i))
if len(title_element) != len(title_rendered):
title_rendered = self._replaceCaracter(title_rendered)
if title_element == title_rendered:
self._logger.info("{0} : {1} found : {2}".format(self._name, i, title_rendered))
element_exist = True
listelement[i].append(k["id"])
else:
break
if page.status_code == 400:
self._logger.error("{0} : {1} not found due status code : {2}".format(self._name, i, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
break
else:
self._logger.error("{0} : {1} not found due status code : {2}".format(self._name, i, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
except ConnectionError as err: except ConnectionError as err:
self._logger.error("{0} : Connection error for {1} : {2}".format(self._name, i, err)) self._logger.error("{0} : Connection error for {1} : {2}".format(self._name, i, err))
exit(1) exit(1)
except Exception as err: except Exception as err:
self._logger.error("{0} : Exception error for {1} : {2}".format(self._name, i, err)) self._logger.error("{0} : Exception error for {1} : {2}".format(self._name, i, err))
if page.status_code == 200:
result = page.json()
self._logger.debug("{0} : content {3} {2} : {1}".format(self._name, result, title_element, i))
if len(result) > 0:
for k in result:
title_rendered = k["name"]
self._logger.debug("{0} : content {2} : {1}".format(self._name, title_rendered, i))
self._logger.debug("{0} : size of content {3} : {2} - {1}".format(self._name, len(title_rendered), len(title_element), i))
if len(title_element) != len(title_rendered):
title_rendered = self._replaceCaracter(title_rendered)
if title_element == title_rendered:
self._logger.info("{0} : {1} found : {2}".format(self._name, i, title_rendered))
element_exist = True
listelement[i].append(k["id"])
else:
break
if page.status_code == 400:
self._logger.error("{0} : {1} not found due status code : {2}".format(self._name, i, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
break
else:
self._logger.error("{0} : {1} not found due status code : {2}".format(self._name, i, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
self._logger.debug("{0} : Element {3} {2} is {1}".format(self._name, element_exist, title_element, i)) self._logger.debug("{0} : Element {3} {2} is {1}".format(self._name, element_exist, title_element, i))
if element_exist == False: if element_exist == False:
data = {"name": title_element} data = {"name": title_element}
@@ -499,18 +536,19 @@ class WPimport:
self._logger.debug("{0} : Data : {1}".format(self._name, data)) self._logger.debug("{0} : Data : {1}".format(self._name, data))
try: try:
page = self._request.post("{2}://{0}/wp-json/wp/v2/{1}".format(self._wordpress, i, self._protocol), auth=self._basic, headers=self._headers_json, data=json.dumps(data)) page = self._request.post("{2}://{0}/wp-json/wp/v2/{1}".format(self._wordpress, i, self._protocol), auth=self._basic, headers=self._headers_json, data=json.dumps(data))
if page.status_code == 201:
self._logger.info("{0} : {1} created : {2}".format(self._name, i, j))
result = page.json()
listelement[i].append(result["id"])
else:
self._logger.error("{0} : {1} not added due status code : {2}".format(self._name, i, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
except ConnectionError as err: except ConnectionError as err:
self._logger.error("{0} : Connection error for post {1} : {2}".format(self._name, i, err)) self._logger.error("{0} : Connection error for post {1} : {2}".format(self._name, i, err))
exit(1) exit(1)
except Exception as err: except Exception as err:
self._logger.error("{0} : Exception error for post {1} : {2}".format(self._name, i, err)) self._logger.error("{0} : Exception error for post {1} : {2}".format(self._name, i, err))
if page.status_code == 201:
self._logger.info("{0} : {1} created : {2}".format(self._name, i, j))
result = page.json()
listelement[i].append(result["id"])
else:
self._logger.error("{0} : {1} not added due status code : {2}".format(self._name, i, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
title = articletitle[0].text title = articletitle[0].text
author = articleacreator[0].text.lower() author = articleacreator[0].text.lower()
@@ -541,21 +579,19 @@ class WPimport:
page = self._request.get("{1}://{0}/wp-json/wp/v2/users".format(self._wordpress, self._protocol), auth=self._basic, headers=self._headers_json, params=params) page = self._request.get("{1}://{0}/wp-json/wp/v2/users".format(self._wordpress, self._protocol), auth=self._basic, headers=self._headers_json, params=params)
self._logger.debug("{0} : End Search author : {1}".format(self._name, author)) self._logger.debug("{0} : End Search author : {1}".format(self._name, author))
self._logger.debug("{0} : Debug requests : {1}".format(self._name, page.content)) self._logger.debug("{0} : Debug requests : {1}".format(self._name, page.content))
if page.status_code == 200:
self._logger.info("{0} : Get author id : {1}".format(self._name, result))
result = page.json()
for a in result:
data["author"] = a["id"]
else:
self._logger.error("{0} : Connection error with status code for get author : {1}".format(self._name, page.status_code))
self._logger.debug("{0} : {1}".format(page.content))
except ConnectionError as err: except ConnectionError as err:
self._logger.error("{0} : Connection error for get author : {1}".format(self._name, err)) self._logger.error("{0} : Connection error for get author : {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 author : {1}".format(self._name, err)) self._logger.error("{0} : Exception error for get author : {1}".format(self._name, err))
if page.status_code == 200:
self._logger.info("{0} : Get author id : {1}".format(self._name, result))
result = page.json()
for a in result:
data["author"] = a["id"]
else:
self._logger.error("{0} : Connection error with status code for get author : {1}".format(self._name, page.status_code))
self._logger.debug("{0} : {1}".format(page.content))
page_is_exist = False page_is_exist = False
for index in range(1,10): for index in range(1,10):
@@ -563,87 +599,89 @@ class WPimport:
try: try:
self._logger.info("{0} : Search post with index {2} : {1}".format(self._name, title, index)) self._logger.info("{0} : Search post with index {2} : {1}".format(self._name, title, index))
page = self._request.get("{1}://{0}/wp-json/wp/v2/posts".format(self._wordpress, self._protocol), auth=self._basic, params=params, headers=self._headers_json) page = self._request.get("{1}://{0}/wp-json/wp/v2/posts".format(self._wordpress, self._protocol), auth=self._basic, params=params, headers=self._headers_json)
if page.status_code == 200:
self._logger.debug("{0} : Encoding : {1}".format(self._name, page.encoding))
page.encoding = "utf-8"
result = page.json()
if len(result) == 0:
break
self._logger.info("{0} : Number result posts : {1}".format(self._name, len(result)))
count = 0
for i in result:
title_rendered = i["title"]["rendered"]
self._logger.info("{0} : Search title posts for |{2}| : |{1}|".format(self._name, title_rendered, title))
if len(title_rendered) != len(title):
title_rendered = self._replaceCaracter(title_rendered)
self._logger.debug("{0} : Search title posts for |{2}| : |{1}|".format(self._name, title_rendered, title))
self._logger.debug("{0} : SIze of title : {1} - {2}".format(self._name, len(title), len(title_rendered)))
if title_rendered == title:
page_is_exist = True
post_id = i["id"]
count = count + 1
if count > 1:
self._logger.info("{0} : Page {1} is double and going to delete".format(self._name, title))
try:
params = {"force":1}
page = self._request.delete("{2}://{0}/wp-json/wp/v2/posts/{1}".format(self._wordpress, post_id, self._protocol), auth=self._basic, headers=self._headers_json, params=params)
if page.status_code == 200:
self._logger.info("{0} : Post deleted : {1}".format(self._name, title))
else:
self._logger.error("{0} : Post not updated due status code : {1}".format(self._name, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
except ConnectionError as err:
self._logger.error("{0} : Connection error for deleted post : {1}".format(self._name, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for deleted post : {1}".format(self._name, err))
else:
self._logger.debug("{0} : Data for post to update : {1}".format(self._name, i))
self._logger.info("{0} : Page {1} already exist and going to update".format(self._name, title))
try:
page = self._request.post("{2}://{0}/wp-json/wp/v2/posts/{1}".format(self._wordpress, post_id, self._protocol), auth=self._basic, headers=self._headers_json, data=json.dumps(data))
if page.status_code == 200:
result = page.json()
self._logger.info("{0} : Post updated : {1}".format(self._name, title))
self._addOrUpdateComment(result["id"], comment_post, result["title"]["raw"])
self._linkImgPost(result["title"]["raw"], list_img, result["id"])
else:
self._logger.error("{0} : Post not updated due status code : {1}".format(self._name, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
except ConnectionError as err:
self._logger.error("{0} : Connection error for update post : {1}".format(self._name, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for update post : {1}".format(self._name, err))
if page.status_code == 400:
self._logger.error("{0} : Connection for update post unauthorized : {1}".format(self._name, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
break
else:
self._logger.error("{0} : Connection for update post error with status code : {1}".format(self._name, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
except ConnectionError as err: except ConnectionError as err:
self._logger.error("{0} : Connection error for search post : {1}".format(self._name, err)) self._logger.error("{0} : Connection error for search post : {1}".format(self._name, err))
exit(1) exit(1)
except Exception as err: except Exception as err:
self._logger.error("{0} : Exception error for search post : {1}".format(self._name, err)) self._logger.error("{0} : Exception error for search post : {1}".format(self._name, err))
if page.status_code == 200:
self._logger.debug("{0} : Encoding : {1}".format(self._name, page.encoding))
page.encoding = "utf-8"
result = page.json()
if len(result) == 0:
break
self._logger.info("{0} : Number result posts : {1}".format(self._name, len(result)))
count = 0
for i in result:
title_rendered = i["title"]["rendered"]
self._logger.info("{0} : Search title posts for |{2}| : |{1}|".format(self._name, title_rendered, title))
if len(title_rendered) != len(title):
title_rendered = self._replaceCaracter(title_rendered)
self._logger.debug("{0} : Search title posts for |{2}| : |{1}|".format(self._name, title_rendered, title))
self._logger.debug("{0} : SIze of title : {1} - {2}".format(self._name, len(title), len(title_rendered)))
if title_rendered == title:
page_is_exist = True
post_id = i["id"]
count = count + 1
if count > 1:
self._logger.info("{0} : Page {1} is double and going to delete".format(self._name, title))
try:
params = {"force":1}
page = self._request.delete("{2}://{0}/wp-json/wp/v2/posts/{1}".format(self._wordpress, post_id, self._protocol), auth=self._basic, headers=self._headers_json, params=params)
except ConnectionError as err:
self._logger.error("{0} : Connection error for deleted post : {1}".format(self._name, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for deleted post : {1}".format(self._name, err))
if page.status_code == 200:
self._logger.info("{0} : Post deleted : {1}".format(self._name, title))
else:
self._logger.error("{0} : Post not updated due status code : {1}".format(self._name, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
else:
self._logger.debug("{0} : Data for post to update : {1}".format(self._name, i))
self._logger.info("{0} : Page {1} already exist and going to update".format(self._name, title))
try:
page = self._request.post("{2}://{0}/wp-json/wp/v2/posts/{1}".format(self._wordpress, post_id, self._protocol), auth=self._basic, headers=self._headers_json, data=json.dumps(data))
except ConnectionError as err:
self._logger.error("{0} : Connection error for update post : {1}".format(self._name, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for update post : {1}".format(self._name, err))
if page.status_code == 200:
result = page.json()
self._logger.info("{0} : Post updated : {1}".format(self._name, title))
self._addOrUpdateComment(result["id"], comment_post, result["title"]["raw"])
self._linkImgPost(result["title"]["raw"], list_img, result["id"])
else:
self._logger.error("{0} : Post not updated due status code : {1}".format(self._name, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
if page.status_code == 400:
self._logger.error("{0} : Connection for update post unauthorized : {1}".format(self._name, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
break
else:
self._logger.error("{0} : Connection for update post error with status code : {1}".format(self._name, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
if page_is_exist is False: if page_is_exist is False:
try: try:
self._logger.info("{0} : Creating posts : {1}".format(self._name, data["title"])) self._logger.info("{0} : Creating posts : {1}".format(self._name, data["title"]))
page = self._request.post("{1}://{0}/wp-json/wp/v2/posts".format(self._wordpress, self._protocol), auth=self._basic, headers=self._headers_json, data=json.dumps(data)) page = self._request.post("{1}://{0}/wp-json/wp/v2/posts".format(self._wordpress, self._protocol), auth=self._basic, headers=self._headers_json, data=json.dumps(data))
if page.status_code == 201:
result = page.json()
self._logger.info("{0} : Post added : {1}".format(self._name, result["title"]["raw"]))
self._addOrUpdateComment(result["id"], comment_post, result["title"]["raw"])
self._linkImgPost(result["title"]["raw"], list_img, result["id"])
else:
self._logger.error("{0} : Post not added due status code : {1}".format(self._name, r.status_code))
self._logger.debug("{0} : {1}".format(self._name, r.content))
except ConnectionError as err: except ConnectionError as err:
self._logger.error("{0} : Connection error for create post : {1}".format(self._name, err)) self._logger.error("{0} : Connection error for create post : {1}".format(self._name, err))
exit(1) exit(1)
except Exception as err: except Exception as err:
self._logger.error("{0} : Exception error for create post : {1}".format(self._name, err)) self._logger.error("{0} : Exception error for create post : {1}".format(self._name, err))
if page.status_code == 201:
result = page.json()
self._logger.info("{0} : Post added : {1}".format(self._name, result["title"]["raw"]))
self._addOrUpdateComment(result["id"], comment_post, result["title"]["raw"])
self._linkImgPost(result["title"]["raw"], list_img, result["id"])
else:
self._logger.error("{0} : Post not added due status code : {1}".format(self._name, r.status_code))
self._logger.debug("{0} : {1}".format(self._name, r.content))

View File

@@ -8,14 +8,15 @@ from requests.packages.urllib3.util.retry import Retry
class WPRemove: class WPRemove:
# Constructor # Constructor
def __init__(self, name="Thread-0", basic=None, wordpress="", logger=None, ssl_wordpress=True): def __init__(self, index_name=1, number_thread=1, basic=None, wordpress="", logger=None, ssl_wordpress=True):
self._name = name
self._basic = basic self._basic = basic
self._wordpress = wordpress self._wordpress = wordpress
self._logger = logger self._logger = logger
self._headers_json = {'Content-Type': 'application/json', 'Accept':'application/json'} self._headers_json = {'Content-Type': 'application/json', 'Accept':'application/json'}
self._name = "Thread-{0}".format(index_name)
self._index_thread = index_name
self._protocol = "https" self._protocol = "https"
self._number_thread = number_thread
if ssl_wordpress is False: if ssl_wordpress is False:
self._protocol = "http" self._protocol = "http"
self._request = requests.Session() self._request = requests.Session()
@@ -32,6 +33,24 @@ class WPRemove:
# Public method # Public method
def _getCount(self, composant):
count = 0
try:
params = {"per_page":1}
self._logger.info("{0} : Get count {2} to remove for url : {1}".format(self._name, self._wordpress, composant))
r = self._request.get("{2}://{0}/wp-json/wp/v2/{1}".format(self._wordpress, composant, self._protocol), params=params, auth=self._basic, headers=self._headers_json)
if r.status_code == 200:
count = int(r.headers["X-WP-Total"])
else:
self._logger.error("{0} : Error for list to remove {1} due status code {2}".format(self._name, composant, r.status_code))
self._logger.debug("{0} : Content error for {1} : {2}".format(self._name, composant, r.content))
except ConnectionError as err:
self._logger.error("{0} : Connection error for list {1} to remove : {2}".format(self._name, composant, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for list {1} to remove : {2}".format(self._name, composant, err))
return count
def setUrl(self, wordpress): def setUrl(self, wordpress):
self._wordpress = wordpress self._wordpress = wordpress
@@ -50,29 +69,60 @@ class WPRemove:
# Private method # Private method
def _removeAll(self, composant): def _removeAll(self, composant):
params = {"per_page":100} count = self._getCount(composant)
try: self._logger.debug("{0} : Count for {1} : {2}".format(self._name, composant, count))
self._logger.info("{0} : List {2} to remove for url : {1}".format(self._name, self._wordpress, composant)) if count > 0:
self._logger.debug("{0} : Number thread for {1} : {2}".format(self._name, composant, self._number_thread))
page = count / int(self._number_thread)
self._logger.debug("{0} : Page for {1} : {2}".format(self._name, composant, page))
if page > int(page):
page = int(page) + 1
if page > 100:
page = 100
params = {"per_page":page, "page":self._index_thread}
self._logger.info("{0} : Params for {1} : {2}".format(self._name, composant, params))
try:
self._logger.info("{0} : List {2} to remove for url : {1}".format(self._name, self._wordpress, composant))
r = self._request.get("{2}://{0}/wp-json/wp/v2/{1}".format(self._wordpress, composant, self._protocol), auth=self._basic, params=params, headers=self._headers_json) r = self._request.get("{2}://{0}/wp-json/wp/v2/{1}".format(self._wordpress, composant, self._protocol), auth=self._basic, params=params, headers=self._headers_json)
except Exception as err:
self._logger.error("{0} : Connection error for list {1} to remove : {2}".format(self._name, composant, err)) if r.status_code == 200:
if r.status_code == 200: result = r.json()
result = r.json() if len(result) > 0:
if len(result) > 0: for i in result:
for i in result: is_delete = True
self._logger.info("{0} : Remove {2} for url {1} : {3}".format(self._name, self._wordpress, composant, i["title"]["rendered"])) self._logger.info(i["slug"])
params = {"force":1} if i["slug"] == "non-classe":
try: is_delete = False
r = self._request.delete("{3}://{0}/wp-json/wp/v2/{1}/{2}".format(self._wordpress, composant, i["id"], self._protocol), auth=self._basic, headers=self._headers_json , params=params) if is_delete is True:
if r.status_code == 200: if composant == "tags" or composant == "categories":
self._logger.info("{0} : Post removed for URL {1} {2} : {3}".format(self._name, self._wordpress, composant, i["title"]["rendered"])) title = i["name"]
else: else:
self._logger.error("{0} : Connection error for post {1} {2} {3} with status code {4}".format(self._name, self._wordpress, composant, i["title"]["rendered"], r.status_code)) title = i["title"]["rendered"]
except Exception as err: self._logger.info("{0} : Remove {2} for url {1} : {3}".format(self._name, self._wordpress, composant, title))
self._logger.error("{0} : Connection error for {1} remove : {2}".format(self._name, composant, err)) params = {"force":1}
exit(1) try:
self._removeAll(composant) r = self._request.delete("{3}://{0}/wp-json/wp/v2/{1}/{2}".format(self._wordpress, composant, i["id"], self._protocol), auth=self._basic, headers=self._headers_json , params=params)
else: if r.status_code == 200:
self._logger.error("{0} : Error for list to remove {1} due status code {2}".format(self._name, composant, r.status_code)) self._logger.info("{0} : Post removed for URL {1} {2} : {3}".format(self._name, self._wordpress, composant, title))
self._logger.debug("{0} : Content error for {1} : {2}".format(self._name, composant, r.content)) else:
self._logger.error("{0} : Connection error for post {1} {2} {3} with status code {4}".format(self._name, self._wordpress, composant, title, r.status_code))
except ConnectionError as err:
self._logger.error("{0} : Connection error for {1} remove : {2}".format(self._name, composant, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for {1} remove : {2}".format(self._name, composant, err))
self._removeAll(composant)
if r.status_code == 400:
self._logger.error("{0} : No content for {1} to remove : {2}".format(self._name, composant, r.status_code))
else:
self._logger.error("{0} : Error for list to remove {1} due status code {2}".format(self._name, composant, r.status_code))
self._logger.debug("{0} : Content error for {1} : {2}".format(self._name, composant, r.content))
except ConnectionError as err:
self._logger.error("{0} : Connection error for list {1} to remove : {2}".format(self._name, composant, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for list {1} to remove : {2}".format(self._name, composant, err))