49 Commits
3.1.0 ... 4.0.0

Author SHA1 Message Date
b4d0fe8aa0 Merge pull request 'album' (#23) from album into master
Reviewed-on: #23
2023-08-25 21:47:47 +00:00
6401692d0d finish menu with album import 2023-08-25 23:46:43 +02:00
1fc9c48d2c fix add menu with album 2023-08-24 22:43:15 +02:00
d9c20cedcb add menu 2023-08-24 00:01:46 +02:00
9d41e57379 add method getAuthor 2023-08-23 23:35:35 +02:00
d88ae7ed44 add author 2023-08-23 22:48:38 +02:00
50bf31d334 HTTP Post for album 2023-08-22 00:21:14 +02:00
9b58b45ae8 replace string 2023-08-20 21:41:16 +02:00
418bea3778 add album test wip 2023-08-17 00:18:45 +02:00
5959ab5b2e add galery wip 2023-08-12 00:06:09 +02:00
e4eb1b6b68 add image to media wordpress 2023-08-05 23:22:44 +02:00
9ed5ffe399 add variable 2023-08-05 12:28:33 +02:00
28b513e1b2 add private method album 2023-08-05 12:03:17 +02:00
02f0c20bd0 Merge pull request 'menu-api' (#22) from menu-api into master
Reviewed-on: #22
2023-08-04 21:22:38 +00:00
1655217050 add album article 2023-08-04 23:21:45 +02:00
49c1552062 add wpmenu 2023-07-31 23:57:35 +02:00
7f800c8f7b add menu 100% work 2023-07-31 23:39:40 +02:00
5399b12133 menu wordpress 75% 2023-07-27 00:39:15 +02:00
b7493206a2 add private method addItemMenu 2023-07-26 19:22:43 +02:00
8de7485775 return value menu 2023-07-26 19:18:18 +02:00
eee14e08f1 add update menu item 2023-07-24 23:51:41 +02:00
cf2c1aa617 add item menu 2023-07-24 23:32:20 +02:00
e17cace820 remove navigation 2023-07-24 22:51:40 +02:00
941776a7c1 Create menu 2023-07-24 22:43:52 +02:00
72fbe0a364 add menu WIP debug 2023-07-22 20:23:26 +02:00
1f8ea70b40 define content for navigation 2023-07-20 00:04:11 +02:00
c6894648d1 add link 2023-07-18 22:09:20 +02:00
d5ddf1601b add post fine 2023-07-17 23:20:47 +02:00
aa8ac9cfcb first test id from post 2023-07-15 14:00:23 +02:00
82f9acd658 get title from post 2023-07-15 13:55:01 +02:00
7593b68b6c getting post title 2023-07-15 11:51:15 +02:00
5fe4b1f786 add href 2023-07-13 23:53:54 +02:00
0445054dc8 fix get id menu 2023-07-13 22:35:38 +02:00
be9ad9b934 swap loop 2023-07-13 00:25:14 +02:00
e6328135da get id 2023-07-13 00:09:18 +02:00
dded126926 change return getId 2023-07-11 21:42:50 +02:00
bca529f3c3 parent and children array 2023-07-11 21:39:41 +02:00
5059a15826 menu and submenu 2023-07-11 00:17:24 +02:00
20c4adb3cf menu and sub-menu wip 2023-07-09 21:42:09 +02:00
74fa87ea73 list menu wip 2023-07-09 19:11:05 +02:00
03f833a2c3 Menu and sub-menu 2023-07-09 15:27:20 +02:00
9acb620f93 add condition if 2023-07-09 12:29:13 +02:00
c6ccf98b1b working menu 2023-07-06 00:42:47 +02:00
95f5203727 list menu with id 2023-07-04 00:26:27 +02:00
9bbf769b40 test WIP 2023-07-01 22:16:35 +02:00
f0b4ba5e27 add url and file 2023-07-01 21:56:10 +02:00
47f504beb5 list tag 2023-07-01 00:11:43 +02:00
4b6b06aade add public method from URL 2023-06-30 23:52:56 +02:00
d10867a983 add class WPMenu 2023-06-30 23:28:54 +02:00
4 changed files with 620 additions and 33 deletions

1
.gitignore vendored
View File

@@ -2,3 +2,4 @@ backup*/
wp-navigation
*.log
__pycache__/
wp-gallery

View File

@@ -10,6 +10,7 @@ from lib.WPImport import WPimport
from lib.WPExport import WPExport
from lib.WPRemove import WPRemove
from lib.WPChange import WPChange
from lib.WPMenu import WPMenu
def errorRevert(logger, revert, tmp):
if revert is True:
@@ -66,7 +67,7 @@ def download(name_thread, max_thread, url, logger, parser, directory, html, img,
del exportWp
def importUrl(name_thread, max_thread, canalblog, logger, parser, wordpress, basic, serial, ssl_wordpress, ssl_canalblog, create, update, image, revert, tmp):
def importUrl(name_thread, max_thread, canalblog, logger, parser, wordpress, basic, serial, ssl_wordpress, ssl_canalblog, create, update, image, revert, tmp, author):
canalblog = canalblog.split(",")
wordpress = wordpress.split(",")
name = "Thread-{0}".format(int(name_thread) + 1)
@@ -87,7 +88,7 @@ def importUrl(name_thread, max_thread, canalblog, logger, parser, wordpress, bas
exportWp.getUrlPage(name_thread, max_thread)
del exportWp
for j in wordpress:
importWp = WPimport(name=name, basic=basic, wordpress=j, logger=logger, parser=parser, ssl_wordpress=ssl_wordpress, no_create=create, no_update=update, no_image=image, tmp=tmp)
importWp = WPimport(name=name, basic=basic, wordpress=j, logger=logger, parser=parser, ssl_wordpress=ssl_wordpress, no_create=create, no_update=update, no_image=image, tmp=tmp, author=author)
for k in ["article", "page"]:
for l in ["publications", "principal"]:
importWp.fromUrl(l, k)
@@ -109,7 +110,7 @@ def importUrl(name_thread, max_thread, canalblog, logger, parser, wordpress, bas
if not revert:
exportWp.getUrlPage(name_thread, max_thread)
del exportWp
importWp = WPimport(name=name, basic=basic, wordpress=wordpress[i], logger=logger, parser=parser, ssl_wordpress=ssl_wordpress, no_create=create, no_update=update, no_image=image, tmp=tmp)
importWp = WPimport(name=name, basic=basic, wordpress=wordpress[i], logger=logger, parser=parser, ssl_wordpress=ssl_wordpress, no_create=create, no_update=update, no_image=image, tmp=tmp, author=author)
for k in ["article", "page"]:
for l in ["publications", "principal"]:
@@ -118,13 +119,13 @@ def importUrl(name_thread, max_thread, canalblog, logger, parser, wordpress, bas
del importWp
def importDirectory(name_thread, max_thread, directory, logger, parser, wordpress, basic, serial, ssl_wordpress, create, update, image, revert):
def importDirectory(name_thread, max_thread, directory, logger, parser, wordpress, basic, serial, ssl_wordpress, create, update, image, revert, author):
name = "Thread-{0}".format(int(name_thread) + 1)
directory = directory.split(",")
wordpress = wordpress.split(",")
if serial is False:
for i in wordpress:
importWp = WPimport(name=name, basic=basic, wordpress=i, logger=logger, parser=parser, ssl_wordpress=ssl_wordpress, no_create=create, no_update=update, no_image=image)
importWp = WPimport(name=name, basic=basic, wordpress=i, logger=logger, parser=parser, ssl_wordpress=ssl_wordpress, no_create=create, no_update=update, no_image=image, author=author)
for j in directory:
importWp.fromDirectory(j, name_thread, max_thread, revert)
del importWp
@@ -134,7 +135,7 @@ def importDirectory(name_thread, max_thread, directory, logger, parser, wordpres
logger.error("{0} : Error : Number directory is different than wordpress".format(name))
exit(1)
for i in range(0, len(wordpress)-1):
importWp = WPimport(name=name, basic=basic, wordpress=wordpress[i], logger=logger, parser=parser, ssl_wordpress=ssl_wordpress, no_create=create, no_update=update, no_image=image)
importWp = WPimport(name=name, basic=basic, wordpress=wordpress[i], logger=logger, parser=parser, ssl_wordpress=ssl_wordpress, no_create=create, no_update=update, no_image=image, author=author)
importWp.fromDirectory(directory[i], name_thread, max_thread, revert)
del importWp
@@ -169,6 +170,7 @@ if __name__ == '__main__':
import_parser.add_argument("--no-create", help="No create post", dest="create", default="store_false", action="store_true")
import_parser.add_argument("--no-update", help="No update post", dest="update", default="store_false", action="store_true")
import_parser.add_argument("--no-image", help="No image add or update", dest="image", default="store_false", action="store_true")
import_parser.add_argument("--author-album", dest=author, help="Define author for page album", default="")
@@ -202,8 +204,14 @@ if __name__ == '__main__':
change_parser.add_argument("--file",
default="",
help="File")
menu_parser = subparsers.add_parser("menu")
menu_parser.add_argument("--user", help="wordpress user", required=True)
menu_parser.add_argument("--password", help="password wordpress's user", default="")
menu_parser.add_argument("--file", help="HTML file", default="")
menu_parser.add_argument("--canalblog", help="URL Canalblog", default="")
menu_parser.add_argument("--wordpress", help="URL Wordpress", required=True)
args = parser.parse_args()
logger = logging.getLogger('import export canalblog')
@@ -240,7 +248,7 @@ if __name__ == '__main__':
os.makedirs(args.tmp, exist_ok=True)
if args.command == "import" or args.command == "remove":
if args.command == "import" or args.command == "remove" or args.command == "menu":
password = args.password
if len(args.password) == 0:
password = getpass()
@@ -251,11 +259,13 @@ if __name__ == '__main__':
basic = HTTPBasicAuth(args.user, password)
if args.command == "import":
wordpress = args.wordpress.split(",")
importWp = WPimport(basic=basic, wordpress="", logger=logger, parser=args.parser, ssl_wordpress=ssl_wordpress)
importWp = WPimport(basic=basic, wordpress="", logger=logger, parser=args.parser, ssl_wordpress=ssl_wordpress, author=args.author)
if len(args.file) > 0:
for i in wordpress:
importWp.setUrl(i)
importWp.fromFile(files=args.file.split(","))
menuWp = WPMenu(name="Thread-1", basic=basic, wordpress=args.wordpress, logger=logger, parser=args.parser, ssl_canalblog=ssl_canalblog, ssl_wordpress=ssl_wordpress)
menuWp.fromFile("{0}".format(args.file.split(",")[0]))
if len(args.directory) > 0:
try:
with futures.ThreadPoolExecutor(max_workers=int(args.parallel)) as ex:
@@ -263,9 +273,12 @@ if __name__ == '__main__':
wait(wait_for, return_when=ALL_COMPLETED)
errorRevert(logger, args.revert, args.tmp)
wait_for = [
ex.submit(importDirectory, i, int(args.parallel), args.directory, logger, args.parser, args.wordpress, basic, args.serial, ssl_wordpress, args.create, args.update, args.image, args.revert)
ex.submit(importDirectory, i, int(args.parallel), args.directory, logger, args.parser, args.wordpress, basic, args.serial, ssl_wordpress, args.create, args.update, args.image, args.revert, args.author)
for i in range(0, int(args.parallel))
]
wait(wait_for, return_when=ALL_COMPLETED)
menuWp = WPMenu(name="Thread-1", basic=basic, wordpress=args.wordpress, logger=logger, parser=args.parser, ssl_canalblog=ssl_canalblog, ssl_wordpress=ssl_wordpress)
menuWp.fromFile("{0}/index.html".format(args.directory))
except Exception as err:
logger.error("Threading error : {0}".format(err))
if len(args.canalblog) > 0:
@@ -275,9 +288,13 @@ if __name__ == '__main__':
wait(wait_for, return_when=ALL_COMPLETED)
errorRevert(logger, args.revert, args.tmp)
wait_for = [
ex.submit(importUrl, i, int(args.parallel), args.canalblog, logger, args.parser, args.wordpress, basic, args.serial, ssl_wordpress, ssl_canalblog, args.create, args.update, args.image, args.revert, args.tmp)
ex.submit(importUrl, i, int(args.parallel), args.canalblog, logger, args.parser, args.wordpress, basic, args.serial, ssl_wordpress, ssl_canalblog, args.create, args.update, args.image, args.revert, args.tmp, args.author)
for i in range(0, int(args.parallel))
]
wait(wait_for, return_when=ALL_COMPLETED)
menuWp = WPMenu(name="Thread-1", basic=basic, wordpress=args.wordpress, logger=logger, parser=args.parser, ssl_canalblog=ssl_canalblog, ssl_wordpress=ssl_wordpress)
menuWp.fromUrl(args.canalblog)
except Exception as err:
@@ -341,4 +358,12 @@ if __name__ == '__main__':
changeWp = WPChange(logger=logger)
for filei in args.file.split(","):
changeWp.fromFile(filei)
exit(0)
if args.command == "menu":
menuWp = WPMenu(name="Thread-1", basic=basic, wordpress=args.wordpress, logger=logger, parser=args.parser, ssl_canalblog=ssl_canalblog, ssl_wordpress=ssl_wordpress)
if len(args.file) > 0:
menuWp.fromFile(args.file)
if len(args.canalblog) > 0:
menuWp.fromUrl(args.canalblog)
exit(0)

View File

@@ -8,7 +8,7 @@ from requests.packages.urllib3.util.retry import Retry
class WPimport:
# Constructor
def __init__(self, name="Thread-0", basic=None, wordpress="", logger=None, parser="html.parser", ssl_wordpress=True, no_create=False, no_update=False, no_image=False, tmp="/tmp/import_export_canablog"):
def __init__(self, name="Thread-0", basic=None, wordpress="", logger=None, parser="html.parser", ssl_wordpress=True, no_create=False, no_update=False, no_image=False, tmp="/tmp/import_export_canablog", author=""):
self._name = name
self._basic = basic
self._wordpress = wordpress
@@ -29,6 +29,7 @@ class WPimport:
self._no_update = no_update
self._no_image = no_image
self._tmp = tmp
self._author = author
# Destructor
def __del__(self):
@@ -55,7 +56,11 @@ class WPimport:
if len(articlebody) > 0:
self._addOrUpdatePost(soup)
else:
self._addOrUpdateFeaturedMedia(soup)
albumbody = soup.find_all("div", class_="albumbody")
if len(albumbody) > 0:
self._addOrUpdateAlbum(soup)
else:
self._addOrUpdateFeaturedMedia(soup)
del webpage_content[first][second][i]
webpage_content = json.dumps(webpage_content)
open("{0}/{1}.json".format(self._tmp, self._name), "wt").write(webpage_content)
@@ -100,12 +105,192 @@ class WPimport:
if len(articlebody) > 0:
self._addOrUpdatePost(soup)
else:
self._addOrUpdateFeaturedMedia(soup)
albumbody = soup.find_all("div", class_="albumbody")
if len(albumbody) > 0:
self._addOrUpdateAlbum(soup)
else:
self._addOrUpdateFeaturedMedia(soup)
# Private method
def _getAuthor(self, author):
params = {"search":author, "per_page":100}
author = 0
try:
self._logger.info("{0} : Search author : {1}".format(self._name, author))
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} : Debug requests : {1}".format(self._name, page.content))
if page.status_code == 200:
self._logger.info("{0} : Get author : {1}".format(self._name, author))
result = page.json()
for a in result:
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:
self._logger.error("{0} : Connection error for get author : {1}".format(self._name, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for get author : {1}".format(self._name, err))
return author
def _addOrUpdateAlbum(self, soup):
self._logger.info("{0} : Add/Update Album".format(self._name))
albumbody = soup.find("div", class_="albumbody")
albumtitle = albumbody.find("h2").get_text()
self._logger.debug("{0} : Title of the album : {1}".format(self._name, albumtitle))
albumdesc = albumbody.find("div", class_="albumdesc").find("p")
img_a = albumbody.find_all("img")
list_img = []
page_is_exist = False
if self._no_image is False:
self._logger.debug("{0} : Number of image's tag : {1}".format(self._name, len(img_a)))
for i in img_a:
new_img = {}
href_img = i.get("src")
href_img_o = urlparse(href_img)
try:
if len(href_img_o.netloc) > 0:
img_ok = False
href_img = href_img.replace("_q", "_o")
page_img = self._request.get(href_img)
if page_img.status_code == 200:
img_ok = True
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
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)
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} content img : {1}".format(self._name, list_img))
content_html = ""
if len(list_img) > 0:
content_html = "<!-- wp:paragraph -->\n{0}\n<!-- /wp:paragraph -->\n\n".format(albumdesc)
for i in range(0, len(list_img)):
content_html = content_html + "<!-- wp:image {\"id\":id-image,\"sizeSlug\":\"large\",\"linkDestination\":\"none\"} --><figure class=\"wp-block-image size-large\"><img src=\"src-image\" alt=\"\" class=\"wp-image-id-image\"/></figure><!-- /wp:image -->\n\n".replace("id-image", str(list_img[i]["id"])).replace("src-image", list_img[i]["new_src"])
self._logger.info("{0} : content html : {1}".format(self._name, content_html))
if len(content_html) > 0:
data = {"title":albumtitle, "content":content_html, "status":"publish"}
if len(self._author) > 0:
author = self._getAuthor(self._author)
if author != 0:
data = {"title":albumtitle, "content":content_html, "status":"publish", "author":author}
self._logger.debug("{0} : data for album page : {1}".format(self._name, data))
for index in range(1,10):
params = {"search": albumtitle, "per_page":100, "page": index}
try:
self._logger.info("{0} : Search post with index {2} : {1}".format(self._name, albumtitle, index))
page = self._request.get("{1}://{0}/wp-json/wp/v2/pages".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 pages for |{2}| : |{1}|".format(self._name, title_rendered, albumtitle))
if len(title_rendered) != len(albumtitle):
title_rendered = self._replaceCaracter(title_rendered)
self._logger.debug("{0} : Search title pages for |{2}| : |{1}|".format(self._name, title_rendered, albumtitle))
self._logger.debug("{0} : SIze of title : {1} - {2}".format(self._name, len(albumtitle), len(title_rendered)))
if title_rendered == albumtitle:
if self._no_update is False:
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, albumtitle))
try:
params = {"force":1}
page = self._request.delete("{2}://{0}/wp-json/wp/v2/pages/{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} : Page deleted : {1}".format(self._name, albumtitle))
else:
self._logger.error("{0} : Page 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 page : {1}".format(self._name, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for deleted page : {1}".format(self._name, err))
else:
self._logger.debug("{0} : Data for page to update : {1}".format(self._name, i))
self._logger.info("{0} : Page {1} already exist and going to update".format(self._name, albumtitle))
try:
page = self._request.post("{2}://{0}/wp-json/wp/v2/pages/{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} : page updated : {1}".format(self._name, albumtitle))
else:
self._logger.error("{0} : page 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 page : {1}".format(self._name, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for update page : {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 page 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:
self._logger.error("{0} : Connection error for search page : {1}".format(self._name, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for search page : {1}".format(self._name, err))
if page_is_exist is False and self._no_create is False:
try:
self._logger.info("{0} : Creating page : {1}".format(self._name, data["title"]))
page = self._request.post("{1}://{0}/wp-json/wp/v2/pages".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} : page added : {1}".format(self._name, result["title"]["raw"]))
else:
self._logger.error("{0} : page 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:
self._logger.error("{0} : Connection error for create page : {1}".format(self._name, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for create page : {1}".format(self._name, err))
def _fromFileTmp(self):
try:
with open("{0}/{1}.json".format(self._tmp, self._name)) as file:
@@ -643,25 +828,7 @@ class WPimport:
data = {"title":title, "content":bodyhtml, "status":"publish", "date": "{0}-{1}-{2}T{3}:00".format(time[2],month[time[1]],time[0], hour), "tags": listelement["tags"], "categories": listelement["categories"]}
self._logger.debug("{0} : Data for post : |{1}| : {2}" .format(self._name, title, data))
params = {"search":author, "per_page":100}
try:
self._logger.info("{0} : Search author : {1}".format(self._name, author))
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} : 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:
self._logger.error("{0} : Connection error for get author : {1}".format(self._name, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for get author : {1}".format(self._name, err))
data["author"] = self._getAuthor(author)
page_is_exist = False
for index in range(1,10):

394
lib/WPMenu.py Normal file
View File

@@ -0,0 +1,394 @@
#!/usr/bin/python3
from bs4 import BeautifulSoup
from urllib.parse import urlparse
import requests, os, logging, re, json
from requests.adapters import HTTPAdapter
from requests.packages.urllib3.util.retry import Retry
class WPMenu:
# Constructor
def __init__(self, name="Thread-0", basic=None, canalblog="", wordpress="", logger=None, parser="html.parser", ssl_canalblog=True, ssl_wordpress=True):
self._name = name
self._basic = basic
self._canalblog = canalblog
self._wordpress = wordpress
self._logger = logger
self._parser = parser
self._headers_json = {'Content-Type': 'application/json; charset=utf-8', 'Accept':'application/json'}
self._protocol_wordpress = "https"
self._protocol_canalblog = "https"
self._directory = "backup"
if ssl_wordpress is False:
self._protocol_wordpress = "http"
if ssl_canalblog is False:
self._protocol_canalblog = "http"
self._request_canalblog = requests.Session()
self._request_wordpress = requests.Session()
retries = Retry(connect=10, read=10, redirect=5,
status_forcelist=[429, 500, 502, 503, 504], backoff_factor=2)
self._request_canalblog.mount('{0}://'.format(self._protocol_canalblog), HTTPAdapter(max_retries=retries))
self._request_wordpress.mount('{0}://'.format(self._protocol_wordpress), HTTPAdapter(max_retries=retries))
# Destructor
def __del__(self):
print("{0} : Import finished for {1}".format(self._name, self._wordpress))
# 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, canalblog):
self._canalblog = canalblog
try:
o = urlparse(canalblog)
o = o._replace(scheme=self._protocol_canalblog)
i = o.geturl().replace(":///", "://")
page = self._request_canalblog.get(i)
if page.status_code == 200:
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))
except ConnectionError as err:
self._logger.error("{0} : Connection error for get url {1} : {2}".format(self._name, canalblog, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for get url {1} : {2}".format(self._name, canalblog, err))
## replace caracter
def _replaceCaracter(self, title_rendered):
list_replace = {'&rsquo;': "'", '&#8211;': '-', '&#8230;': '...', '« ': '"', ' »': '"', '« ': '"', ' »': '"', '&#8217;': "'", '"&lsquo;': "'"}
for old, new in list_replace.items():
title_rendered = title_rendered.replace(old, new)
return title_rendered
def _getIdfromTitlePost(self, content):
idMenu = {"id":0, "type":"", "link":""}
soup = BeautifulSoup(content, self._parser)
articletitle = soup.find_all("h2", class_="articletitle")
if len(articletitle) > 0:
articletitle = articletitle[0].get_text()
search = "posts"
post_type = "post"
if len(articletitle) == 0:
articletitle = soup.find_all("div", class_="albumbody")
if len(articletitle) > 0:
articletitle = articletitle[0].find("h2").get_text()
search = "pages"
post_type = "page"
exist = False
for index in range(1,10):
if exist is False:
params = {"search":articletitle, "per_page":100, "page":index}
try:
self._logger.debug("{0} : Get Url for {3} : {1} {2}".format(self._name, "{1}://{0}/wp-json/wp/v2/{2}".format(self._wordpress, self._protocol_wordpress, search), params, search))
page = self._request_wordpress.get("{1}://{0}/wp-json/wp/v2/{2}".format(self._wordpress, self._protocol_wordpress, search), auth=self._basic, params=params)
if page.status_code == 200:
result = page.json()
self._logger.info("{0} : Get content {2} : {1}".format(self._name, len(result), search))
if len(result) > 0:
for i in result:
title_rendered = i["title"]["rendered"]
if len(articletitle) != len(title_rendered):
title_rendered = self._replaceCaracter(title_rendered)
self._logger.debug("{0} : comparaison debug {1} {2}".format(self._name, articletitle, title_rendered))
if articletitle == title_rendered:
self._logger.debug("{0} : get {2} id : {1}".format(self._name, i, search))
idMenu = {"id":i["id"], "type":post_type, "link": i["link"]}
exist = True
else:
self._logger.debug("{0} : {2} {1}".format(self._name, result, len(result)))
break
elif page.status_code == 400:
self._logger.debug("{0} : {2} {1}".format(self._name, page.content, page.status_code))
break
else:
self._logger.error("{0} : Post didn't get 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 get content : {1}".format(self._name, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for get content : {1} ".format(self._name, err))
return idMenu
def _getIdFromPost(self, href):
idMenu = {"id":0, "type":"", "link":""}
o = urlparse(href)
if len(o.netloc) > 0:
try:
page = self._request_canalblog.get(href)
if page.status_code == 200:
self._logger.info("{0} : Get content : {1}".format(self._name, href))
idMenu = self._getIdfromTitlePost(page.content)
else:
self._logger.error("{0} : {2} didn't get due status code : {1}".format(self._name, page.status_code, href))
self._logger.debug("{0} : {1}".format(self._name, page.content))
except ConnectionError as err:
self._logger.error("{0} : Connection error for get url {1} : {2}".format(self._name, href, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for get url {1} : {2}".format(self._name, href, err))
else:
if os.path.exists("{0}/..{1}".format(self._directory, o.path)):
try:
content = open("{0}/..{1}".format(self._directory, o.path), "r")
idMenu = self._getIdfromTitlePost(content)
except Exception as err:
self._logger.error("{0} : Exception error for get file content {1} : {2}".format(self._name, href, err))
return idMenu
def _getIdFromReverse(self, title, href):
self._logger.info("{0} : get title {1} from href {2}".format(self._name, title, href))
idMenu = {"id":0, "type":"", "link":""}
if href != "#":
title = href[::-1]
second_title = title.split("/")[2]
second_title = second_title[::-1]
link = title.split("/")[0]
link = link[::-1]
title = title.split("/")[1]
title = title[::-1]
self._logger.info("{0} link {1} title {2}".format(self._name, link, title))
if link == "index.html":
if second_title == "albums":
idMenu = self._getIdFromPost(href)
else:
idMenu = self._getId(title)
else:
idMenu = self._getIdFromPost(href)
return idMenu
def _getId(self, title):
idMenu = {"id": 0, "type":"", "link":""}
exist = False
if exist is False:
for i in ["categories", "tags"]:
typeId = "category"
if i == "tags":
typeId = "tag"
for index in range(1,10):
try:
params = {"search":title, "per_page":"100", "page":index}
self._logger.info("{0} Get menu {1} {2} {3}".format(self._name, "{2}://{0}/wp-json/wp/v2/{1}".format(self._wordpress, i, self._protocol_wordpress), index, title))
page = self._request_wordpress.get("{2}://{0}/wp-json/wp/v2/{1}".format(self._wordpress, i, self._protocol_wordpress), auth=self._basic, params=params)
if page.status_code == 200:
result = page.json()
if len(result) > 0:
for j in result:
self._logger.info("{0} info : {1} {2} {3}".format(self._name, j["name"], j["slug"], title))
if j["name"] == title or j["slug"] == title:
self._logger.info("{0} : comparaison ok : {1} {2}".format(self._name, j["id"], i))
idMenu = {"id": j["id"], "type": typeId, "link": j["link"]}
exist = True
else:
break
elif page.status_code == 400:
break
else:
self._logger.error("{0} : {2} didn't get due status code : {1}".format(self._name, page.status_code, i))
self._logger.debug("{0} : {1}".format(self._name, page.content))
except ConnectionError as err:
self._logger.error("{0} : Connection error for get url {1} : {2}".format(self._name, "{2}://{0}/wp-json/wp/v2/{1}".format(self._wordpress, i, self._protocol_wordpress), err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for get url {1} : {2}".format(self._name, "{2}://{0}/wp-json/wp/v2/{1}".format(self._wordpress, i, self._protocol_wordpress), err))
return idMenu
def _menu(self, content):
soup = BeautifulSoup(content, self._parser)
ul = soup.find("ul", id="listsmooth")
menu = list()
children = list()
for anchor in ul.find_all("li"):
parent = anchor.find("a").get_text().replace(" \xa0", "")
href = anchor.find("a").get("href")
if href == "{0}://{1}/".format(self._protocol_canalblog, self._canalblog):
parent = "home"
itemMenu = {"id":"", "type":"", "title": parent, "link":"", "href":href, "children":list()}
if href == "#":
li = anchor.find("ul").find_all("li")
for child in li:
a = child.find("a")
href = a.get("href")
self._logger.info("{0} Parent {1} : Child {2}".format(self._name, parent, a.get_text()))
children.append({"title": a.get_text(), "parent": parent, "href":href, "link":""})
menu.append(itemMenu)
for i in range(0, len(children)):
self._logger.info("{0} : Child {1} {2}".format(self._name, children[i], i))
for j in range(0, len(menu)):
if j < len(menu):
if menu[j]["title"] == children[i]["title"]:
self._logger.info("{0} : Parent {1} {2}".format(self._name, menu[j], j))
del menu[j]
for j in range(0, len(menu)):
self._logger.info("{0} : Children for : {1}".format(self._name, menu[j]["title"]))
if menu[j]["title"] == children[i]["parent"]:
menu[j]["children"].append({"id":"", "type":"", "title":children[i]["title"], "parent": children[i]["parent"], "link":"", "href":children[i]["href"]})
for i in range(0, len(menu)):
self._logger.info("{0} : Menu {1} {2}".format(self._name, menu[i]["title"], len(menu[i]["children"])))
if menu[i]["title"] != "home":
for j in range(0, len(menu[i]["children"])):
idMenu = self._getId(menu[i]["children"][j]["title"])
if idMenu["id"] == 0:
self._logger.debug("{0} : content children {1}".format(self._name, menu[i]["children"][j]))
idMenu = self._getIdFromReverse(menu[i]["children"][j]["title"], menu[i]["children"][j]["href"])
if idMenu["id"] != 0:
menu[i]["children"][j] = {"id":idMenu["id"], "type": idMenu["type"], "link": idMenu["link"], "title": menu[i]["children"][j]["title"], "parent": menu[i]["children"][j]["parent"]}
idMenu = self._getId(menu[i]["title"])
self._logger.debug("{0} : content parent {1}".format(self._name, menu[i]))
self._logger.debug("{0} : content idMenu {1}".format(self._name, idMenu))
if idMenu["id"] == 0:
idMenu = self._getIdFromReverse(menu[i]["title"], menu[i]["href"])
if idMenu["id"] != 0:
menu[i] = {"id":idMenu["id"], "type": idMenu["type"], "title":menu[i]["title"], "link":idMenu["link"], "children": menu[i]["children"]}
self._createMenu(menu)
def _createItemMenu(self, idMenu, itemMenu, parent):
idItemMenu = 0
self._logger.info("{0} : Create item menu from API Wordpress : {1}".format(self._name, self._wordpress))
try:
params = {"search": itemMenu["title"], "menus": idMenu}
page = self._request_wordpress.get("{1}://{0}/wp-json/wp/v2/menu-items".format(self._wordpress, self._protocol_wordpress), auth=self._basic, params=params)
if page.status_code == 200:
result = page.json()
for i in result:
if self._replaceCaracter(i["title"]["rendered"]) == itemMenu["title"]:
idItemMenu = int(i["id"])
self._logger.info("{0} : Length of result for item menus : {1}".format(self._name, len(result)))
url = "{1}://{0}/wp-json/wp/v2/menu-items".format(self._wordpress, self._protocol_wordpress)
if idItemMenu != 0:
url = "{1}://{0}/wp-json/wp/v2/menu-items/{2}".format(self._wordpress, self._protocol_wordpress, idItemMenu)
try:
objectt = itemMenu["type"]
if objectt == "tag":
objectt = "post_tag"
data = {"title": itemMenu["title"], "status": "publish", "parent":parent, "menus":idMenu, "url":"#"}
if itemMenu["title"] == "home":
data = {"title": itemMenu["title"], "status": "publish", "parent":parent, "menus":idMenu, "url":"{0}://{1}".format(self._protocol_wordpress, self._wordpress)}
if type(itemMenu["id"]) is str:
if len(itemMenu["id"]) > 0:
data = {"title": itemMenu["title"], "status": "publish", "parent":parent, "url": itemMenu["link"], "menus":idMenu, "object":objectt, "object_id":int(itemMenu["id"])}
elif type(itemMenu["id"]) is int:
data = {"title": itemMenu["title"], "status": "publish", "parent":parent, "url": itemMenu["link"], "menus":idMenu, "object":objectt, "object_id":itemMenu["id"]}
self._logger.debug("{0} : data for create/update : {1}".format(self._name, data))
page = self._request_wordpress.post(url, auth=self._basic, headers=self._headers_json, data=json.dumps(data))
if page.status_code in [201, 200]:
result = page.json()
idItemMenu = int(result["id"])
self._logger.info("{0} : create/update item menu : {1}".format(self._name, itemMenu["title"]))
else:
self._logger.error("{0} : Create menu items for {2} didn't get due status code : {1}".format(self._name, page.status_code, itemMenu["title"]))
self._logger.debug("{0} : {1} {2}".format(self._name, page.content, itemMenu))
except ConnectionError as err:
self._logger.error("{0} : Connection error for create item menu {1} : {2}".format(self._name, "{1}://{0}/wp-json/wp/v2/menu-items".format(self._wordpress, self._protocol_wordpress), err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for create item menu {1} : {2}".format(self._name, "{1}://{0}/wp-json/wp/v2/menu-items".format(self._wordpress, self._protocol_wordpress), err))
else:
self._logger.error("{0} : Get menu items didn't get 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 get item menus {1} : {2}".format(self._name, "{1}://{0}/wp-json/wp/v2/menu-items".format(self._wordpress, self._protocol_wordpress), err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for get item menus {1} : {2}".format(self._name, "{1}://{0}/wp-json/wp/v2/menu-items".format(self._wordpress, self._protocol_wordpress), err))
return idItemMenu
def _createMenu(self, menu):
title = "Menu {0}".format(self._wordpress)
self._logger.info("{0} : Create menu from API Wordpress : {1}".format(self._name, title))
try:
params = {"search": title}
page = self._request_wordpress.get("{1}://{0}/wp-json/wp/v2/menus".format(self._wordpress, self._protocol_wordpress), auth=self._basic, params=params)
if page.status_code == 200:
result = page.json()
self._logger.info("{0} : Get content menus : {1}".format(self._name, len(result)))
idMenu = 0
if len(result) == 0:
self._logger.info("{0} : Create menu : {1}".format(self._name, title))
data = {"name": title}
try:
page = self._request_wordpress.post("{1}://{0}/wp-json/wp/v2/menus".format(self._wordpress, self._protocol_wordpress), auth=self._basic, headers=self._headers_json, data=json.dumps(data))
if page.status_code == 201:
result = page.json()
self._logger.debug("{0} : Get menus : {1}".format(self._name, result))
if len(result) > 0:
idMenu = result["id"]
else:
self._logger.error("{0} : Post menu didn't get 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 create menu {1} : {2}".format(self._name, "{1}://{0}/wp-json/wp/v2/menus".format(self._wordpress, self._protocol_wordpress), err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for get menu {1} : {2}".format(self._name, "{1}://{0}/wp-json/wp/v2/menus".format(self._wordpress, self._protocol_wordpress), err))
else:
self._logger.debug("{0} : Get menus : {1}".format(self._name, result))
for i in result:
self._logger.debug("{0} : List menus : {1}".format(self._name, i))
if i["name"] == title:
idMenu = i["id"]
self._logger.info("{0} : Get ID menus : {1}".format(self._name, idMenu))
self._addItemMenu(menu, idMenu)
else:
self._logger.error("{0} : Get menu didn't get 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 get menu {1} : {2}".format(self._name, "{1}://{0}/wp-json/wp/v2/menus".format(self._wordpress, self._protocol_wordpress), err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for get menu {1} : {2}".format(self._name, "{1}://{0}/wp-json/wp/v2/menus".format(self._wordpress, self._protocol_wordpress), err))
def _addItemMenu(self, menu, idMenu):
self._logger.info("{0} : add item to menu : {1}".format(self._name, idMenu))
parent = 0
for i in menu:
parent = 0
self._logger.debug("{0} : debug create item menu : {1}".format(self._name, i))
parent = self._createItemMenu(idMenu, i, parent)
for j in i["children"]:
self._createItemMenu(idMenu, j, parent)