18 Commits

Author SHA1 Message Date
c631909cb6 WPchange wip 2023-06-06 00:22:16 +02:00
3e76892676 add wpchange 2023-06-05 23:46:57 +02:00
3e75f05340 Merge pull request 'add-parameter' (#17) from add-parameter into master
Reviewed-on: #17
2023-06-05 20:58:51 +00:00
e48b262d7e add parameter no-image 2023-06-03 09:07:33 +02:00
2f1c081823 add parameter 2023-06-01 15:28:48 +02:00
4bd6f5c038 Merge pull request 'add wait' (#16) from wait_remove into master
Reviewed-on: #16
2023-05-29 21:36:38 +00:00
d3a03e1cb3 add wait 2023-05-29 23:36:11 +02:00
f507efce60 Merge pull request 'replace-exception' (#15) from replace-exception into master
Reviewed-on: #15
2023-05-29 21:29:18 +00:00
75c9fa0ad3 fix if 2023-05-28 22:42:38 +02:00
110ccc4bb1 replace exception for wpexport 2023-05-28 22:42:04 +02:00
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
5 changed files with 677 additions and 476 deletions

View File

@@ -3,15 +3,15 @@ from requests.auth import HTTPBasicAuth
from getpass import getpass from getpass import getpass
from urllib.parse import urlparse from urllib.parse import urlparse
from concurrent import futures from concurrent import futures
from concurrent.futures import as_completed, wait from concurrent.futures import as_completed, wait, ALL_COMPLETED
import argparse, logging, threading import argparse, logging, threading
from lib.WPImport import WPimport 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)
@@ -47,7 +47,7 @@ def download(name_thread, max_thread, url, logger, parser, directory, html, img,
def importUrl(name_thread, max_thread, canalblog, logger, parser, wordpress, basic, serial, ssl_wordpress, ssl_canalblog): def importUrl(name_thread, max_thread, canalblog, logger, parser, wordpress, basic, serial, ssl_wordpress, ssl_canalblog, create, update, image):
canalblog = canalblog.split(",") canalblog = canalblog.split(",")
wordpress = wordpress.split(",") wordpress = wordpress.split(",")
name = "Thread-{0}".format(int(name_thread) + 1) name = "Thread-{0}".format(int(name_thread) + 1)
@@ -67,7 +67,7 @@ def importUrl(name_thread, max_thread, canalblog, logger, parser, wordpress, bas
webpage = exportWp.getUrlPage(name_thread, max_thread) webpage = exportWp.getUrlPage(name_thread, max_thread)
del exportWp del exportWp
for j in wordpress: for j in wordpress:
importWp = WPimport(name=name, basic=basic, wordpress=j, logger=logger, parser=parser, ssl_wordpress=ssl_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)
for k in ["article", "page"]: for k in ["article", "page"]:
for l in ["publications", "principal"]: for l in ["publications", "principal"]:
importWp.fromUrl(webpage[l][k]) importWp.fromUrl(webpage[l][k])
@@ -88,7 +88,7 @@ def importUrl(name_thread, max_thread, canalblog, logger, parser, wordpress, bas
exportWp = WPExport(name=name, url=url, logger=logger, parser=parser, ssl_canalblog=ssl_canalblog) exportWp = WPExport(name=name, url=url, logger=logger, parser=parser, ssl_canalblog=ssl_canalblog)
webpage = exportWp.getUrlPage(name_thread, max_thread) webpage = exportWp.getUrlPage(name_thread, max_thread)
del exportWp del exportWp
importWp = WPimport(name=name, basic=basic, wordpress=wordpress[i], logger=logger, parser=parser, ssl_wordpress=ssl_wordpress) 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)
for k in ["article", "page"]: for k in ["article", "page"]:
for l in ["publications", "principal"]: for l in ["publications", "principal"]:
@@ -97,13 +97,13 @@ def importUrl(name_thread, max_thread, canalblog, logger, parser, wordpress, bas
del importWp del importWp
def importDirectory(name_thread, max_thread, directory, logger, parser, wordpress, basic, serial, ssl_wordpress): def importDirectory(name_thread, max_thread, directory, logger, parser, wordpress, basic, serial, ssl_wordpress, create, update, image):
name = "Thread-{0}".format(int(name_thread) + 1) name = "Thread-{0}".format(int(name_thread) + 1)
directory = directory.split(",") directory = directory.split(",")
wordpress = wordpress.split(",") wordpress = wordpress.split(",")
if serial is False: if serial is False:
for i in wordpress: for i in wordpress:
importWp = WPimport(name=name, basic=basic, wordpress=i, logger=logger, parser=parser, ssl_wordpress=ssl_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)
for j in directory: for j in directory:
importWp.fromDirectory(j, name_thread, max_thread) importWp.fromDirectory(j, name_thread, max_thread)
del importWp del importWp
@@ -113,7 +113,7 @@ def importDirectory(name_thread, max_thread, directory, logger, parser, wordpres
logger.error("{0} : Error : Number directory is differant than wordpress".format(name)) logger.error("{0} : Error : Number directory is differant than wordpress".format(name))
exit(1) exit(1)
for i in range(0, len(wordpress)-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) 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.fromDirectory(directory[i]) importWp.fromDirectory(directory[i])
del importWp del importWp
@@ -143,6 +143,10 @@ if __name__ == '__main__':
import_parser.add_argument("--remove-categories", help="Remove all categories", dest="categories", action="store_true") import_parser.add_argument("--remove-categories", help="Remove all categories", dest="categories", action="store_true")
import_parser.add_argument("--remove-tags", help="Remove all tags", dest="tags", action="store_true") import_parser.add_argument("--remove-tags", help="Remove all tags", dest="tags", action="store_true")
import_parser.add_argument("--remove-media", help="Remove all media", dest="media", action="store_true") import_parser.add_argument("--remove-media", help="Remove all media", dest="media", action="store_true")
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")
remove_parser = subparsers.add_parser("remove") remove_parser = subparsers.add_parser("remove")
@@ -221,23 +225,27 @@ if __name__ == '__main__':
importWp.setUrl(i) importWp.setUrl(i)
importWp.fromFile(files=args.file.split(",")) importWp.fromFile(files=args.file.split(","))
if len(args.directory) > 0: if len(args.directory) > 0:
remove(args, basic, logger, ssl_wordpress)
try: try:
with futures.ThreadPoolExecutor(max_workers=int(args.parallel)) as ex: 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)) ]
wait(wait_for, return_when=ALL_COMPLETED)
wait_for = [ wait_for = [
ex.submit(importDirectory, i, int(args.parallel), args.directory, logger, args.parser, args.wordpress, basic, args.serial, ssl_wordpress) 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)
for i in range(0, int(args.parallel)) for i in range(0, int(args.parallel))
] ]
except Exception as err: except Exception as err:
logger.error("Threading error : {0}".format(err)) logger.error("Threading error : {0}".format(err))
if len(args.canalblog) > 0: if len(args.canalblog) > 0:
remove(args, basic, logger, ssl_wordpress)
try: try:
with futures.ThreadPoolExecutor(max_workers=int(args.parallel)) as ex: 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)) ]
wait(wait_for, return_when=ALL_COMPLETED)
wait_for = [ wait_for = [
ex.submit(importUrl, i, int(args.parallel), args.canalblog, logger, args.parser, args.wordpress, basic, args.serial, ssl_wordpress, ssl_canalblog) 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)
for i in range(0, int(args.parallel)) for i in range(0, int(args.parallel))
] ]
except Exception as err: except Exception as err:
logger.error("Threading error : {0}".format(err)) logger.error("Threading error : {0}".format(err))
exit(0) exit(0)
@@ -280,5 +288,9 @@ 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)

80
lib/WPChange.py Normal file
View File

@@ -0,0 +1,80 @@
from bs4 import BeautifulSoup
from urllib.parse import urlparse
import requests, os, logging, re, json
class WPRemove:
# Constructor
def __init__(self, index_name=1, number_thread=1, logger=None, parser="html.parser"):
self._name = "Thread-{0}".format(index_name)
self._logger = logger
self._number_thread = number_thread
self._parser = parser
# Destructor
def __del__(self):
print("{0} : Import finished".format(self._name))
# Public method
## from file
def fromFile(self, files=[], number_thread=1, max_thread=1):
divFiles = int(len(files) / max_thread)
currentRangeFiles = int(divFiles * (number_thread+1))
firstRange = int(currentRangeFiles - divFiles)
self._logger.debug("{0} : index : {1}".format(self._name,number_thread))
self._logger.debug("{0} : first range : {1}".format(self._name,firstRange))
self._logger.debug("{0} : last range : {1}".format(self._name,currentRangeFiles))
for i in range(firstRange, currentRangeFiles):
if os.path.exists(files[i]):
self._logger.info("{0} : ({1}/{2}) File is being processed : {3}".format(self._name, i+1, currentRangeFiles + 1, files[i]))
self._change(files[i])
## From directory
def fromDirectory(self, directory="", number_thread=1, max_thread=1):
directory = "{0}/archives".format(directory)
directories = self._getDirectories([], "{0}".format(directory))
if len(directories) > 0:
files = self._getFiles(directories)
self.fromFile(files, number_thread, max_thread)
else:
self._logger.error("{0} : No files for {1}".format(self._name, directory))
# Private method
## Get all files
def _getFiles(self, item):
files = []
for i in item:
for j in os.listdir(i):
if os.path.isfile("{0}/{1}".format(i, j)):
files.append("{0}/{1}".format(i, j))
return files
## Get directories
def _getDirectories(self, subdirectory, item):
sub = subdirectory
for i in os.listdir(item):
if os.path.isdir("{0}/{1}".format(item, i)):
sub.append("{0}/{1}".format(item, i))
subdirectory = self._getDirectories(sub, "{0}/{1}".format(item, i))
return subdirectory
## Change path img file
def _change(self, file):
with open(file, 'r') as f:
content = f.read()
soup = BeautifulSoup(content, self._parser)
img = soup.find_all("img")

View File

@@ -72,9 +72,7 @@ class WPExport:
def getUrlPage(self, index_thread, max_thread): def getUrlPage(self, index_thread, max_thread):
try: try:
page = self._request.get(self._url) page = self._request.get(self._url)
except Exception as err:
self._logger.error("{0} : Connection error : {1}".format(self._name, err))
exit(1)
page_url = [] page_url = []
if page.status_code == 200: if page.status_code == 200:
soup = BeautifulSoup(page.text, self._parser) soup = BeautifulSoup(page.text, self._parser)
@@ -86,6 +84,11 @@ class WPExport:
else: else:
self._logger.error("{0} : URL did not get due status code : {1}".format(self._name, page.status_code)) self._logger.error("{0} : URL did not get due status code : {1}".format(self._name, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content)) 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))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error : {1}".format(self._name, err))
webpage = {"principal": {"page":[], "article":[]}, "publications": {"page":[], "article":[]}} webpage = {"principal": {"page":[], "article":[]}, "publications": {"page":[], "article":[]}}
for i in page_url: for i in page_url:
@@ -97,9 +100,7 @@ class WPExport:
section = "principal" section = "principal"
try: try:
page = self._request.get(i) page = self._request.get(i)
except Exception as err:
self._logger.error("{0} : Connection error : {1}".format(self._name, err))
exit(1)
if page.status_code == 200: if page.status_code == 200:
self._logger.info("{0} : page : {1}".format(self._name, i)) self._logger.info("{0} : page : {1}".format(self._name, i))
if i not in webpage[section]["page"]: if i not in webpage[section]["page"]:
@@ -147,13 +148,19 @@ class WPExport:
try: try:
o = urlparse(href) o = urlparse(href)
o = o._replace(scheme="https").geturl() o = o._replace(scheme="https").geturl()
webpage[section]["article"].append(o)
except Exception as err: except Exception as err:
self._logger.error("parsing error : {0}".format(err)) self._logger.error("parsing error : {0}".format(err))
exit(1) exit(1)
webpage[section]["article"].append(o)
else: else:
self._logger.error("{0} : web didn't get due status code : {1}".format(self._name, page.status_code)) self._logger.error("{0} : web didn't get due status code : {1}".format(self._name, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content)) 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))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error : {1}".format(self._name, err))
exit(1)
return webpage return webpage
@@ -183,9 +190,7 @@ class WPExport:
def _getScriptCss(self, js, css): def _getScriptCss(self, js, css):
try: try:
page = self._request.get(self._url) page = self._request.get(self._url)
except Exception as err:
self._logger.error("Connection error : {0}".format(err))
exit(1)
page_url = [] page_url = []
if page.status_code == 200: if page.status_code == 200:
soup = BeautifulSoup(page.text, self._parser) soup = BeautifulSoup(page.text, self._parser)
@@ -197,13 +202,14 @@ class WPExport:
try: try:
u = urlparse(self._url) u = urlparse(self._url)
o = urlparse(src) o = urlparse(src)
except Exception as err:
self._logger.error("parsing error : {0}".format(err))
exit(1)
if o.netloc == "": if o.netloc == "":
o = o._replace(netloc=u.netloc) o = o._replace(netloc=u.netloc)
o = o._replace(scheme=u.scheme) o = o._replace(scheme=u.scheme)
page_url.append(o.geturl()) page_url.append(o.geturl())
except Exception as err:
self._logger.error("parsing error : {0}".format(err))
exit(1)
if css is True: if css is True:
link = soup.find_all("link") link = soup.find_all("link")
for anchor in link: for anchor in link:
@@ -214,16 +220,22 @@ class WPExport:
try: try:
u = urlparse(self._url) u = urlparse(self._url)
o = urlparse(href) o = urlparse(href)
except Exception as err:
self._logger.error("parsing error : {0}".format(err))
exit(1)
if o.netloc == "": if o.netloc == "":
o = o._replace(netloc=u.netloc) o = o._replace(netloc=u.netloc)
o = o._replace(scheme=u.scheme) o = o._replace(scheme=u.scheme)
page_url.append(o.geturl()) page_url.append(o.geturl())
except Exception as err:
self._logger.error("parsing error : {0}".format(err))
exit(1)
else: else:
self._logger.error("JS or CSS did not get due status code : {0}".format(page.status_code)) self._logger.error("JS or CSS did not get due status code : {0}".format(page.status_code))
self._logger.debug(page.content) self._logger.debug(page.content)
except ConnectionError as err:
self._logger.error("Connection error : {0}".format(err))
exit(1)
except Exception as err:
self._logger.error("Exception error : {0}".format(err))
return page_url return page_url
@@ -234,9 +246,7 @@ class WPExport:
for i in webpage: for i in webpage:
try: try:
page = self._request.get(i) page = self._request.get(i)
except Exception as err:
self._logger.error("{0} : Connection error : {1}".format(self._name, err))
exit(1)
if page.status_code == 200: if page.status_code == 200:
soup = BeautifulSoup(page.text, self._parser) soup = BeautifulSoup(page.text, self._parser)
img = soup.find_all("img") img = soup.find_all("img")
@@ -250,6 +260,11 @@ class WPExport:
else: else:
self._logger.error("{0} : Image did not get due status code : {1}".format(self._name, page.status_code)) self._logger.error("{0} : Image did not get due status code : {1}".format(self._name, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content)) 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))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error : {1}".format(self._name, err))
return page_img return page_img
@@ -260,9 +275,7 @@ class WPExport:
for i in range(0, len(webpage)): for i in range(0, len(webpage)):
try: try:
o = urlparse(webpage[i]) o = urlparse(webpage[i])
except Exception as err:
self._logger.error("parsing error : {0}".format(err))
exit(1)
path_web = o.path.split("/") path_web = o.path.split("/")
filePageWeb = path_web[len(path_web)-1] filePageWeb = path_web[len(path_web)-1]
path_web.pop(len(path_web)-1) path_web.pop(len(path_web)-1)
@@ -270,11 +283,7 @@ class WPExport:
self._mkdirPath("{0}/{1}/{2}".format(backup_dir, o.netloc, dir_page_web)) self._mkdirPath("{0}/{1}/{2}".format(backup_dir, o.netloc, dir_page_web))
try: try:
r = self._request.get(webpage[i]) r = self._request.get(webpage[i])
except ConnectionError as err:
self._logger.error("{0} : Connection error : {1}".format(self._name, err))
exit(1)
except Exception as err:
self._logger.error("{0} Exception error : {1}".format(self._name, err))
if r.status_code == 200: if r.status_code == 200:
fileDownload = "{0}/{1}/index.html".format(backup_dir, o.netloc) fileDownload = "{0}/{1}/index.html".format(backup_dir, o.netloc)
if len(dir_page_web) > 0 and len(filePageWeb) > 0: if len(dir_page_web) > 0 and len(filePageWeb) > 0:
@@ -288,3 +297,11 @@ class WPExport:
else: else:
self._logger.error("Not download due status code : {0}".format(r.status_code)) self._logger.error("Not download due status code : {0}".format(r.status_code))
self._logger.debug(r.content) self._logger.debug(r.content)
except ConnectionError as err:
self._logger.error("{0} : Connection error : {1}".format(self._name, err))
exit(1)
except Exception as err:
self._logger.error("{0} Exception error : {1}".format(self._name, err))
except Exception as err:
self._logger.error("parsing error : {0}".format(err))
exit(1)

View File

@@ -8,7 +8,7 @@ from requests.packages.urllib3.util.retry import Retry
class WPimport: class WPimport:
# Constructor # Constructor
def __init__(self, name="Thread-0", basic=None, wordpress="", logger=None, parser="html.parser", ssl_wordpress=True): 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):
self._name = name self._name = name
self._basic = basic self._basic = basic
self._wordpress = wordpress self._wordpress = wordpress
@@ -24,7 +24,9 @@ class WPimport:
status_forcelist=[429, 500, 502, 503, 504], backoff_factor=2) status_forcelist=[429, 500, 502, 503, 504], backoff_factor=2)
self._request.mount('{0}://'.format(self._protocol), HTTPAdapter(max_retries=retries)) self._request.mount('{0}://'.format(self._protocol), HTTPAdapter(max_retries=retries))
self._no_create = no_create
self._no_update = no_update
self._no_image = no_image
# Destructor # Destructor
def __del__(self): def __del__(self):
@@ -39,9 +41,6 @@ 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:
self._logger.error("{0} : Connection error for get url {1} : {2}".format(self._name, webpage[i], err))
exit(1)
if r.status_code == 200: if r.status_code == 200:
self._logger.info("{0} : ({1}/{2}) : Page is importing : {3}".format(self._name, i+1, len(webpage), webpage[i])) 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) soup = BeautifulSoup(r.content, self._parser)
@@ -53,6 +52,11 @@ class WPimport:
else: 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.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)) 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))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for get url {1} : {2}".format(self._name, webpage[i], err))
def fromDirectory(self, directory="", number_thread=1, max_thread=1): def fromDirectory(self, directory="", number_thread=1, max_thread=1):
@@ -137,9 +141,6 @@ 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:
self._logger.error("{0} : Connection error : {1}".format(self._name, err))
exit(1)
if page.status_code == 200: if page.status_code == 200:
result = page.json() result = page.json()
if len(result) > 0: if len(result) > 0:
@@ -149,18 +150,13 @@ class WPimport:
img_src = img[0].get("src") img_src = img[0].get("src")
try: try:
page = self._request.get(img_src) 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: if page.status_code == 200:
name_img = img_src.replace("_q", "") name_img = img_src.replace("_q", "")
name_img = name_img.split("/")[len(name_img.split("/"))-1] name_img = name_img.split("/")[len(name_img.split("/"))-1]
params = {"search": name_img} params = {"search": name_img}
try: try:
page = self._request.get("{1}://{0}/wp-json/wp/v2/media".format(self._wordpress, self._protocol), auth=self._basic, params=params) 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: if page.status_code == 200:
res = page.json() res = page.json()
if len(res) > 0: if len(res) > 0:
@@ -168,29 +164,42 @@ class WPimport:
data = {"featured_media": id_media} data = {"featured_media": id_media}
try: 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)) 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: if r.status_code == 200:
self._logger.info("{0} : Add media featured : {1}".format(self._name, r.json()["title"]["raw"])) self._logger.info("{0} : Add media featured : {1}".format(self._name, r.json()["title"]["raw"]))
else: else:
self._logger.error("{0} : Connection error with status code for featured media : {1}".format(self._name, r.status_code)) 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)) 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: else:
self._logger.info("{0} : No media found for {1}".format(self._name, h2)) self._logger.info("{0} : No media found for {1}".format(self._name, h2))
else: else:
self._logger.error("{0} : Connection error with status code for search featured media: {1}".format(self._name, page.status_code)) 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)) 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: else:
self._logger.error("{0} : Connection error for get featured media with status code : {1}".format(self._name, page.status_code)) 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)) 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: else:
self._logger.error("{0} : Connection error with status code for featured media : {1}".format(self._name, page.status_code)) 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)) 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))
exit(1)
except Exception as err:
self._logger.error("{0} : Connection error : {1}".format(self._name, err))
## Association image to post ## Association image to post
@@ -199,14 +208,16 @@ 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:
self._logger.error("{0} : Connection error for link image to post : {1}".format(self._name, err))
exit(1)
if r.status_code == 200: if r.status_code == 200:
self._logger.info("{0} : Link image to post {1}".format(self._name, title)) self._logger.info("{0} : Link image to post {1}".format(self._name, title))
else: else:
self._logger.error("{0} Connection error with status code for link image to post : {1}".format(self._name, r.status_code)) 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)) 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))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for link image to post : {1}".format(self._name, err))
## Add or update img ## Add or update img
@@ -226,9 +237,7 @@ 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.error("{0} : Connection error for search media : {1}".format(self._name, err))
exit(1)
self._logger.debug("{0} : Search for image {1} and his status code {2}".format(self._name, img_name, r.status_code)) 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: if r.status_code == 200:
res = r.json() res = r.json()
@@ -237,14 +246,16 @@ class WPimport:
params = {"force":1} params = {"force":1}
try: 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) 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: if r.status_code == 200:
self._logger.info("{0} : Image removed {1}".format(self._name, img_name)) self._logger.info("{0} : Image removed {1}".format(self._name, img_name))
else: else:
self._logger.error("{0} : Image {1} not removed due status code : {2}".format(self._name, img_name, r.status_code)) 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)) 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 data = page.content
img_type = "image/{0}".format(img_type_file) img_type = "image/{0}".format(img_type_file)
@@ -253,9 +264,7 @@ class WPimport:
headers={ 'Content-Type': img_type,'Content-Disposition' : 'attachment; filename={0}'.format(img_name)} headers={ 'Content-Type': img_type,'Content-Disposition' : 'attachment; filename={0}'.format(img_name)}
try: try:
r = self._request.post("{1}://{0}/wp-json/wp/v2/media".format(self._wordpress, self._protocol), auth=self._basic, headers=headers, data=data) 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: if r.status_code == 201:
self._logger.info("{0} : Image added {1}".format(self._name, img_name)) self._logger.info("{0} : Image added {1}".format(self._name, img_name))
res = r.json() res = r.json()
@@ -264,11 +273,21 @@ class WPimport:
else: 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.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)) 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: else:
self._logger.error("{0} : Connection error for search image with status code : {1}".format(self._name, r.status_code)) 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)) 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))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for search media : {1}".format(self._name, err))
return media return media
## Add or update comment ## Add or update comment
@@ -279,28 +298,34 @@ 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:
self._logger.error("{0} : Connection error for search comment : {1}".format(self._name, err))
exit(1)
if page.status_code == 200: if page.status_code == 200:
result = page.json() result = page.json()
for j in result: for j in result:
try: try:
params = {"force":1} 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) 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: if page.status_code == 200:
self._logger.info("{0} : Comment deleted for {1}".format(self._name, title)) self._logger.info("{0} : Comment deleted for {1}".format(self._name, title))
self._logger.debug("{0} : Comment deleted : {1}".format(self._name, j)) self._logger.debug("{0} : Comment deleted : {1}".format(self._name, j))
else: else:
self._logger.error("{0} : Comment not deleted for {1} due status code : {2}".format(self._name, title, page.status_code)) 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)) 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: else:
self._logger.error("{0} : Comment not listed for {1} due status code : {2}".format(self._name, title, page.status_code)) 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)) 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))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for search comment : {1}".format(self._name, err))
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"}
@@ -310,9 +335,7 @@ 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:
self._logger.error("{0} : Connection error for parent comment : {1}".format(self._name, err))
exit(1)
if page.status_code == 200: if page.status_code == 200:
result = page.json() result = page.json()
if len(result) > 0: if len(result) > 0:
@@ -320,18 +343,27 @@ class WPimport:
else: else:
self._logger.error("{0} : Connection error for parent comment with status code : {1}".format(self._name, page.status_code)) 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)) 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))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for parent comment : {1}".format(self._name, err))
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:
self._logger.error("{0} : Connection error for add comment : {1}".format(self._name, err))
exit(1)
if page.status_code == 201: if page.status_code == 201:
self._logger.info("{0} : Comment added for {1}".format(self._name, title)) self._logger.info("{0} : Comment added for {1}".format(self._name, title))
self._logger.debug("{0} : Data : {1}".format(self._name, data)) self._logger.debug("{0} : Data : {1}".format(self._name, data))
else: else:
self._logger.error("{0} : Comment not added for {1} due status code : {2}".format(self._name, title, page.status_code)) 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)) 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))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for add comment : {1}".format(self._name, err))
exit(1)
## Check class name ## Check class name
@@ -405,6 +437,7 @@ class WPimport:
img_a = articlebody[0].find_all("a", {"target": "_blank"}) img_a = articlebody[0].find_all("a", {"target": "_blank"})
self._logger.debug("{0} : Number of image's link : {1}".format(self._name, len(img_a))) self._logger.debug("{0} : Number of image's link : {1}".format(self._name, len(img_a)))
list_img = [] list_img = []
if self._no_image is False:
for i in img_a: for i in img_a:
new_img = {} new_img = {}
img = i.find_all("img") img = i.find_all("img")
@@ -416,16 +449,17 @@ 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))
exit(1)
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 Exception 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:
self._logger.error("{0} : Exception error for get image : {1}".format(self._name, err))
exit(1)
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)
@@ -440,6 +474,12 @@ class WPimport:
if page_img.status_code not in [200, 404]: 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.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)) 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)
@@ -462,11 +502,7 @@ 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)
except ConnectionError as err:
self._logger.error("{0} : Connection error for {1} : {2}".format(self._name, i, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for {1} : {2}".format(self._name, i, err))
if page.status_code == 200: if page.status_code == 200:
result = page.json() result = page.json()
self._logger.debug("{0} : content {3} {2} : {1}".format(self._name, result, title_element, i)) self._logger.debug("{0} : content {3} {2} : {1}".format(self._name, result, title_element, i))
@@ -491,7 +527,11 @@ class WPimport:
else: else:
self._logger.error("{0} : {1} not found due status code : {2}".format(self._name, i, page.status_code)) 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} : {1}".format(self._name, page.content))
except ConnectionError as err:
self._logger.error("{0} : Connection error for {1} : {2}".format(self._name, i, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for {1} : {2}".format(self._name, i, err))
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}
@@ -499,11 +539,7 @@ 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))
except ConnectionError as err:
self._logger.error("{0} : Connection error for post {1} : {2}".format(self._name, i, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for post {1} : {2}".format(self._name, i, err))
if page.status_code == 201: if page.status_code == 201:
self._logger.info("{0} : {1} created : {2}".format(self._name, i, j)) self._logger.info("{0} : {1} created : {2}".format(self._name, i, j))
result = page.json() result = page.json()
@@ -511,6 +547,11 @@ class WPimport:
else: else:
self._logger.error("{0} : {1} not added due status code : {2}".format(self._name, i, page.status_code)) 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)) self._logger.debug("{0} : {1}".format(self._name, page.content))
except ConnectionError as err:
self._logger.error("{0} : Connection error for post {1} : {2}".format(self._name, i, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for post {1} : {2}".format(self._name, i, err))
title = articletitle[0].text title = articletitle[0].text
author = articleacreator[0].text.lower() author = articleacreator[0].text.lower()
@@ -541,13 +582,6 @@ 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))
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))
if page.status_code == 200: if page.status_code == 200:
self._logger.info("{0} : Get author id : {1}".format(self._name, result)) self._logger.info("{0} : Get author id : {1}".format(self._name, result))
result = page.json() result = page.json()
@@ -556,6 +590,11 @@ class WPimport:
else: else:
self._logger.error("{0} : Connection error with status code for get author : {1}".format(self._name, page.status_code)) 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)) 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))
page_is_exist = False page_is_exist = False
for index in range(1,10): for index in range(1,10):
@@ -563,11 +602,6 @@ 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)
except ConnectionError as err:
self._logger.error("{0} : Connection error for search post : {1}".format(self._name, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for search post : {1}".format(self._name, err))
if page.status_code == 200: if page.status_code == 200:
self._logger.debug("{0} : Encoding : {1}".format(self._name, page.encoding)) self._logger.debug("{0} : Encoding : {1}".format(self._name, page.encoding))
page.encoding = "utf-8" page.encoding = "utf-8"
@@ -584,6 +618,7 @@ class WPimport:
self._logger.debug("{0} : Search title posts for |{2}| : |{1}|".format(self._name, title_rendered, title)) 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))) self._logger.debug("{0} : SIze of title : {1} - {2}".format(self._name, len(title), len(title_rendered)))
if title_rendered == title: if title_rendered == title:
if self._no_update is False:
page_is_exist = True page_is_exist = True
post_id = i["id"] post_id = i["id"]
count = count + 1 count = count + 1
@@ -592,16 +627,16 @@ class WPimport:
try: try:
params = {"force":1} 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) 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: if page.status_code == 200:
self._logger.info("{0} : Post deleted : {1}".format(self._name, title)) self._logger.info("{0} : Post deleted : {1}".format(self._name, title))
else: else:
self._logger.error("{0} : Post not updated due status code : {1}".format(self._name, page.status_code)) 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)) 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: else:
self._logger.debug("{0} : Data for post to update : {1}".format(self._name, i)) self._logger.debug("{0} : Data for post to update : {1}".format(self._name, i))
@@ -609,11 +644,7 @@ class WPimport:
try: 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)) 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: if page.status_code == 200:
result = page.json() result = page.json()
self._logger.info("{0} : Post updated : {1}".format(self._name, title)) self._logger.info("{0} : Post updated : {1}".format(self._name, title))
@@ -622,6 +653,11 @@ class WPimport:
else: else:
self._logger.error("{0} : Post not updated due status code : {1}".format(self._name, page.status_code)) 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)) 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: if page.status_code == 400:
self._logger.error("{0} : Connection for update post unauthorized : {1}".format(self._name, page.status_code)) 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)) self._logger.debug("{0} : {1}".format(self._name, page.content))
@@ -629,16 +665,17 @@ class WPimport:
else: else:
self._logger.error("{0} : Connection for update post error with status code : {1}".format(self._name, page.status_code)) 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)) self._logger.debug("{0} : {1}".format(self._name, page.content))
except ConnectionError as err:
self._logger.error("{0} : Connection error for search post : {1}".format(self._name, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for search post : {1}".format(self._name, err))
if page_is_exist is False: if page_is_exist is False and self._no_create 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))
except ConnectionError as err:
self._logger.error("{0} : Connection error for create post : {1}".format(self._name, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for create post : {1}".format(self._name, err))
if page.status_code == 201: if page.status_code == 201:
result = page.json() result = page.json()
self._logger.info("{0} : Post added : {1}".format(self._name, result["title"]["raw"])) self._logger.info("{0} : Post added : {1}".format(self._name, result["title"]["raw"]))
@@ -647,3 +684,8 @@ class WPimport:
else: else:
self._logger.error("{0} : Post not added due status code : {1}".format(self._name, r.status_code)) 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)) self._logger.debug("{0} : {1}".format(self._name, r.content))
except ConnectionError as err:
self._logger.error("{0} : Connection error for create post : {1}".format(self._name, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for create post : {1}".format(self._name, err))

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)
self._logger.debug("{0} : Count for {1} : {2}".format(self._name, composant, count))
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: try:
self._logger.info("{0} : List {2} to remove for url : {1}".format(self._name, self._wordpress, composant)) 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:
self._logger.info("{0} : Remove {2} for url {1} : {3}".format(self._name, self._wordpress, composant, i["title"]["rendered"])) is_delete = True
self._logger.info(i["slug"])
if i["slug"] == "non-classe":
is_delete = False
if is_delete is True:
if composant == "tags" or composant == "categories":
title = i["name"]
else:
title = i["title"]["rendered"]
self._logger.info("{0} : Remove {2} for url {1} : {3}".format(self._name, self._wordpress, composant, title))
params = {"force":1} params = {"force":1}
try: try:
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) 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 r.status_code == 200: if r.status_code == 200:
self._logger.info("{0} : Post removed for URL {1} {2} : {3}".format(self._name, self._wordpress, composant, i["title"]["rendered"])) self._logger.info("{0} : Post removed for URL {1} {2} : {3}".format(self._name, self._wordpress, composant, title))
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)) 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 Exception as err: except ConnectionError as err:
self._logger.error("{0} : Connection error for {1} remove : {2}".format(self._name, composant, err)) self._logger.error("{0} : Connection error for {1} remove : {2}".format(self._name, composant, err))
exit(1) exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for {1} remove : {2}".format(self._name, composant, err))
self._removeAll(composant) 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: else:
self._logger.error("{0} : Error for list to remove {1} due status code {2}".format(self._name, composant, r.status_code)) 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)) 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))