16 Commits

5 changed files with 626 additions and 495 deletions

View File

@@ -3,8 +3,8 @@
TAR=/usr/bin/tar
PYTHON=/usr/bin/python3
GZIP=/usr/bin/gzip
SCRIPTDIR=/home/valentin/script
WEBSCRAP=${SCRIPTDIR}/web_scrap.py
SCRIPTDIR=/home/valentin/script/webscrap
WEBSCRAP=${SCRIPTDIR}/import_export_canalblog.py
URL=www.clarissariviere.com
DATE=$(date +%Y%m%d)
DIRECTORY=/home/valentin/backup
@@ -24,7 +24,7 @@ else
fi
subject="${subject} ${URL} ${DATE}"
echo > ${BACKUPDIR}/${LOGFILE}
${PYTHON} ${WEBSCRAP} --url ${URL} --dir ${DIRECTORY} --quiet --logfile ${BACKUPDIR}/${LOGFILE}
${PYTHON} ${WEBSCRAP} --quiet --logfile ${BACKUPDIR}/${LOGFILE} --parallel 20 export --url ${URL} --directory ${DIRECTORY}
if [ ${?} -ne 0 ]; then
subject="${subject} echoue : recuperation page"
echo ${subject} | mail -s "${subject}" -A ${BACKUPDIR}/${LOGFILE} ${SENDER}

View File

@@ -10,8 +10,8 @@ from lib.WPImport import WPimport
from lib.WPExport import WPExport
from lib.WPRemove import WPRemove
def remove(args, basic, logger):
removeWp = WPRemove(basic=basic, wordpress="", logger=logger)
def remove(index, number, args, basic, logger, ssl_wordpress):
removeWp = WPRemove(basic=basic, wordpress="", logger=logger, ssl_wordpress=ssl_wordpress, index_name=index, number_thread=number)
if args.remove == True:
for i in args.wordpress.split(","):
removeWp.setUrl(i)
@@ -33,8 +33,8 @@ def remove(args, basic, logger):
del removeWp
def download(name_thread, max_thread, url, logger, parser, directory, html, img):
exportWp = WPExport(name="Thread-{0}".format(int(name_thread) + 1), url=url, logger=logger, parser=parser, directory=directory)
def download(name_thread, max_thread, url, logger, parser, directory, html, img, ssl_canalblog):
exportWp = WPExport(name="Thread-{0}".format(int(name_thread) + 1), url=url, logger=logger, parser=parser, directory=directory, ssl_canalblog=ssl_canalblog)
webpage = exportWp.getUrlPage(name_thread, max_thread)
for i in ["article", "page"]:
for j in ["publications", "principal"]:
@@ -47,25 +47,27 @@ def download(name_thread, max_thread, url, logger, parser, directory, html, img)
def importUrl(name_thread, max_thread, canalblog, logger, parser, wordpress, basic, serial):
def importUrl(name_thread, max_thread, canalblog, logger, parser, wordpress, basic, serial, ssl_wordpress, ssl_canalblog):
canalblog = canalblog.split(",")
wordpress = wordpress.split(",")
name = "Thread-{0}".format(int(name_thread) + 1)
protocol = "https"
if ssl_canalblog is False:
protocol = "http"
if serial is False:
for canal in canalblog:
try:
o = urlparse(canal)
o = o._replace(scheme="https")
o = o._replace(scheme=protocol)
url = o.geturl().replace(":///", "://")
except Exception as err:
logger.error("{0} : parsing error : {1}".format(name, err))
exit(1)
exportWp = WPExport(name="Thread-{0}".format(int(name_thread) + 1), url=url, logger=logger, parser=parser)
exportWp = WPExport(name="Thread-{0}".format(int(name_thread) + 1), url=url, logger=logger, parser=parser, ssl_canalblog=ssl_canalblog)
webpage = exportWp.getUrlPage(name_thread, max_thread)
del exportWp
for j in wordpress:
importWp = WPimport(name=name, basic=basic, wordpress=j, logger=logger, parser=parser)
importWp = WPimport(name=name, basic=basic, wordpress=j, logger=logger, parser=parser, ssl_wordpress=ssl_wordpress)
for k in ["article", "page"]:
for l in ["publications", "principal"]:
importWp.fromUrl(webpage[l][k])
@@ -78,15 +80,15 @@ def importUrl(name_thread, max_thread, canalblog, logger, parser, wordpress, bas
for i in range(0, len(canalblog)-1):
try:
o = urlparse(canalblog[i])
o = o._replace(scheme="https")
o = o._replace(scheme=protocol)
url = o.geturl().replace(":///", "://")
except Exception as err:
logger.error("parsing error : {0}".format(err))
exit(1)
exportWp = WPExport(name=name, url=url, logger=logger, parser=parser)
exportWp = WPExport(name=name, url=url, logger=logger, parser=parser, ssl_canalblog=ssl_canalblog)
webpage = exportWp.getUrlPage(name_thread, max_thread)
del exportWp
importWp = WPimport(name=name, basic=basic, wordpress=wordpress[i], logger=logger, parser=parser)
importWp = WPimport(name=name, basic=basic, wordpress=wordpress[i], logger=logger, parser=parser, ssl_wordpress=ssl_wordpress)
for k in ["article", "page"]:
for l in ["publications", "principal"]:
@@ -95,13 +97,13 @@ 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):
def importDirectory(name_thread, max_thread, directory, logger, parser, wordpress, basic, serial, ssl_wordpress):
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)
importWp = WPimport(name=name, basic=basic, wordpress=i, logger=logger, parser=parser, ssl_wordpress=ssl_wordpress)
for j in directory:
importWp.fromDirectory(j, name_thread, max_thread)
del importWp
@@ -111,7 +113,7 @@ def importDirectory(name_thread, max_thread, directory, logger, parser, wordpres
logger.error("{0} : Error : Number directory is differant than wordpress".format(name))
exit(1)
for i in range(0, len(wordpress)-1):
importWp = WPimport(name=name, basic=basic, wordpress=wordpress[i], logger=logger, parser=parser)
importWp = WPimport(name=name, basic=basic, wordpress=wordpress[i], logger=logger, parser=parser, ssl_wordpress=ssl_wordpress)
importWp.fromDirectory(directory[i])
del importWp
@@ -124,6 +126,7 @@ if __name__ == '__main__':
parser.add_argument("--quiet", help="No console output", action="store_true")
parser.add_argument("--parser", help="Parser content", default="html.parser")
parser.add_argument("--parallel", help="Define number thread (default : 1)", default=1)
parser.add_argument("--no-ssl", help="No ssl for canalblog and/or wordpress (example wordpress,canalblog)", dest="ssl", default="")
subparsers = parser.add_subparsers(dest="command")
@@ -171,6 +174,14 @@ if __name__ == '__main__':
logger = logging.getLogger('import export canalblog')
formatter = logging.Formatter('%(asctime)s - %(name)s - %(levelname)s - %(message)s')
ssl_canalblog = True
ssl_wordpress = True
for i in args.ssl.split(","):
if i == "canalblog":
ssl_canalblog = False
if i == "wordpress":
ssl_wordpress = False
if args.quiet is False:
ch = logging.StreamHandler()
@@ -204,27 +215,27 @@ 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)
importWp = WPimport(basic=basic, wordpress="", logger=logger, parser=args.parser, ssl_wordpress=ssl_wordpress)
if len(args.file) > 0:
for i in wordpress:
importWp.setUrl(i)
importWp.fromFile(files=args.file.split(","))
if len(args.directory) > 0:
remove(args, basic, logger)
remove(args, basic, logger, ssl_wordpress)
try:
with futures.ThreadPoolExecutor(max_workers=int(args.parallel)) as ex:
wait_for = [
ex.submit(importDirectory, i, int(args.parallel), args.directory, logger, args.parser, args.wordpress, basic, args.serial)
ex.submit(importDirectory, i, int(args.parallel), args.directory, logger, args.parser, args.wordpress, basic, args.serial, ssl_wordpress)
for i in range(0, int(args.parallel))
]
except Exception as err:
logger.error("Threading error : {0}".format(err))
if len(args.canalblog) > 0:
remove(args, basic, logger)
remove(args, basic, logger, ssl_wordpress)
try:
with futures.ThreadPoolExecutor(max_workers=int(args.parallel)) as ex:
wait_for = [
ex.submit(importUrl, i, int(args.parallel), args.canalblog, logger, args.parser, args.wordpress, basic, args.serial)
ex.submit(importUrl, i, int(args.parallel), args.canalblog, logger, args.parser, args.wordpress, basic, args.serial, ssl_wordpress, ssl_canalblog)
for i in range(0, int(args.parallel))
]
except Exception as err:
@@ -234,11 +245,14 @@ if __name__ == '__main__':
if args.command == "export":
canalblog = args.url.split(",")
exportWp = WPExport(logger=logger, parser=args.parser, directory=args.directory)
protocol = "https"
if ssl_canalblog is False:
protocol = "http"
exportWp = WPExport(logger=logger, parser=args.parser, directory=args.directory, ssl_canalblog=ssl_canalblog)
for canal in canalblog:
try:
o = urlparse(canal)
o = o._replace(scheme="https")
o = o._replace(scheme=protocol)
url = o.geturl().replace(":///", "://")
except Exception as err:
logger.error("parsing error : {0}".format(err))
@@ -257,7 +271,7 @@ if __name__ == '__main__':
try:
with futures.ThreadPoolExecutor(max_workers=int(args.parallel)) as ex:
wait_for = [
ex.submit(download, i, int(args.parallel), url, logger, args.parser, args.directory, args.html, args.img)
ex.submit(download, i, int(args.parallel), url, logger, args.parser, args.directory, args.html, args.img, ssl_canalblog)
for i in range(0, int(args.parallel))
]
except Exception as err:
@@ -266,5 +280,9 @@ if __name__ == '__main__':
if args.command == "remove":
remove(args, basic, logger)
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)

View File

@@ -6,20 +6,22 @@ from requests.adapters import HTTPAdapter
from requests.packages.urllib3.util.retry import Retry
class WPExport:
def __init__(self, name = "Thread-0", url = "", logger = None, parser = "html.parser", directory = "backup"):
def __init__(self, name = "Thread-0", url = "", logger = None, parser = "html.parser", directory = "backup", ssl_canalblog=True):
self._url = url
self._logger = logger
self._parser = parser
self._dir = directory
self._name = name
self._protocol = "https"
if ssl_canalblog is False:
self._protocol = "http"
self._request = requests.Session()
retries = Retry(total=10,
status_forcelist=[429, 500, 502, 503, 504], backoff_factor=2)
self._request.mount('http://', HTTPAdapter(max_retries=retries))
self._request.mount('{0}://'.format(self._protocol), HTTPAdapter(max_retries=retries))
# Destructor
@@ -70,9 +72,7 @@ class WPExport:
def getUrlPage(self, index_thread, max_thread):
try:
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 = []
if page.status_code == 200:
soup = BeautifulSoup(page.text, self._parser)
@@ -84,20 +84,23 @@ class WPExport:
else:
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))
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":[]}}
for i in page_url:
section = "publications"
o = urlparse(i)
o = o._replace(scheme="https")
o = o._replace(scheme=self._protocol)
i = o.geturl().replace(":///", "://")
if i == "{0}/".format(self._url):
section = "principal"
try:
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:
self._logger.info("{0} : page : {1}".format(self._name, i))
if i not in webpage[section]["page"]:
@@ -145,13 +148,19 @@ class WPExport:
try:
o = urlparse(href)
o = o._replace(scheme="https").geturl()
webpage[section]["article"].append(o)
except Exception as err:
self._logger.error("parsing error : {0}".format(err))
exit(1)
webpage[section]["article"].append(o)
else:
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))
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
@@ -181,9 +190,7 @@ class WPExport:
def _getScriptCss(self, js, css):
try:
page = self._request.get(self._url)
except Exception as err:
self._logger.error("Connection error : {0}".format(err))
exit(1)
page_url = []
if page.status_code == 200:
soup = BeautifulSoup(page.text, self._parser)
@@ -195,13 +202,14 @@ class WPExport:
try:
u = urlparse(self._url)
o = urlparse(src)
except Exception as err:
self._logger.error("parsing error : {0}".format(err))
exit(1)
if o.netloc == "":
o = o._replace(netloc=u.netloc)
o = o._replace(scheme=u.scheme)
page_url.append(o.geturl())
except Exception as err:
self._logger.error("parsing error : {0}".format(err))
exit(1)
if css is True:
link = soup.find_all("link")
for anchor in link:
@@ -212,16 +220,22 @@ class WPExport:
try:
u = urlparse(self._url)
o = urlparse(href)
except Exception as err:
self._logger.error("parsing error : {0}".format(err))
exit(1)
if o.netloc == "":
o = o._replace(netloc=u.netloc)
o = o._replace(scheme=u.scheme)
page_url.append(o.geturl())
except Exception as err:
self._logger.error("parsing error : {0}".format(err))
exit(1)
else:
self._logger.error("JS or CSS did not get due status code : {0}".format(page.status_code))
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
@@ -232,9 +246,7 @@ class WPExport:
for i in webpage:
try:
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:
soup = BeautifulSoup(page.text, self._parser)
img = soup.find_all("img")
@@ -248,6 +260,11 @@ class WPExport:
else:
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))
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
@@ -258,9 +275,7 @@ class WPExport:
for i in range(0, len(webpage)):
try:
o = urlparse(webpage[i])
except Exception as err:
self._logger.error("parsing error : {0}".format(err))
exit(1)
path_web = o.path.split("/")
filePageWeb = path_web[len(path_web)-1]
path_web.pop(len(path_web)-1)
@@ -268,9 +283,7 @@ class WPExport:
self._mkdirPath("{0}/{1}/{2}".format(backup_dir, o.netloc, dir_page_web))
try:
r = self._request.get(webpage[i])
except Exception as err:
self._logger.error("Connection error : {0}".format(err))
exit(1)
if r.status_code == 200:
fileDownload = "{0}/{1}/index.html".format(backup_dir, o.netloc)
if len(dir_page_web) > 0 and len(filePageWeb) > 0:
@@ -284,3 +297,11 @@ class WPExport:
else:
self._logger.error("Not download due status code : {0}".format(r.status_code))
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

@@ -5,24 +5,25 @@ from urllib.parse import urlparse
import requests, os, logging, re, json
from requests.adapters import HTTPAdapter
from requests.packages.urllib3.util.retry import Retry
from slugify import slugify
class WPimport:
# Constructor
def __init__(self, name="Thread-0", basic=None, wordpress="", logger=None, parser="html.parser"):
def __init__(self, name="Thread-0", basic=None, wordpress="", logger=None, parser="html.parser", ssl_wordpress=True):
self._name = name
self._basic = basic
self._wordpress = wordpress
self._logger = logger
self._parser = parser
self._headers_json = {'Content-Type': 'application/json; charset=utf-8', 'Accept':'application/json'}
self._protocol = "https"
if ssl_wordpress is False:
self._protocol = "http"
self._request = requests.Session()
retries = Retry(connect=10, read=10, redirect=5,
status_forcelist=[429, 500, 502, 503, 504], backoff_factor=2)
self._request.mount('http://', HTTPAdapter(max_retries=retries))
self._request.mount('{0}://'.format(self._protocol), HTTPAdapter(max_retries=retries))
# Destructor
@@ -38,9 +39,6 @@ class WPimport:
for i in range(0, len(webpage)):
try:
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:
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)
@@ -52,6 +50,11 @@ class WPimport:
else:
self._logger.error("{0} : Connection error for get url {1} with status code : {2}".format(self._name, webpage[i], r.status_code))
self._logger.debug("{0} : {1}".format(self._name, r.content))
except ConnectionError as err:
self._logger.error("{0} : Connection error for get url {1} : {2}".format(self._name, webpage[i], err))
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):
@@ -135,10 +138,7 @@ class WPimport:
h2 = i.find_all("h2")[0].text
params = {"search":h2, "type":"post"}
try:
page = self._request.get("http://{0}/wp-json/wp/v2/search".format(self._wordpress), auth=self._basic, params=params)
except Exception as err:
self._logger.error("{0} : Connection error : {1}".format(self._name, err))
exit(1)
page = self._request.get("{1}://{0}/wp-json/wp/v2/search".format(self._wordpress, self._protocol), auth=self._basic, params=params)
if page.status_code == 200:
result = page.json()
if len(result) > 0:
@@ -148,48 +148,56 @@ class WPimport:
img_src = img[0].get("src")
try:
page = self._request.get(img_src)
except Exception as err:
self._logger.error("{0} : Connection error for get featured media : {1}".format(self._name, err))
exit(1)
if page.status_code == 200:
name_img = img_src.replace("_q", "")
name_img = name_img.split("/")[len(name_img.split("/"))-1]
params = {"search": name_img}
try:
page = self._request.get("http://{0}/wp-json/wp/v2/media".format(self._wordpress), 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)
page = self._request.get("{1}://{0}/wp-json/wp/v2/media".format(self._wordpress, self._protocol), auth=self._basic, params=params)
if page.status_code == 200:
res = page.json()
if len(res) > 0:
id_media = res[0]["id"]
data = {"featured_media": id_media}
try:
r = self._request.post("http://{0}/wp-json/wp/v2/posts/{1}".format(self._wordpress, result[0]["id"]), 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)
r = self._request.post("{2}://{0}/wp-json/wp/v2/posts/{1}".format(self._wordpress, result[0]["id"], self._protocol), auth=self._basic, headers=self._headers_json, data=json.dumps(data))
if r.status_code == 200:
self._logger.info("{0} : Add media featured : {1}".format(self._name, r.json()["title"]["raw"]))
else:
self._logger.error("{0} : Connection error with status code for featured media : {1}".format(self._name, r.status_code))
self._logger.debug("{0} : {1}".format(self._name, r.content))
except ConnectionError as err:
self._logger.error("{0} : Connection error for post media featured : {1}".format(self._name, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for post media featured : {1}".format(self._name, err))
else:
self._logger.info("{0} : No media found for {1}".format(self._name, h2))
else:
self._logger.error("{0} : Connection error with status code for search featured media: {1}".format(self._name, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
except ConnectionError as err:
self._logger.error("{0} : Connection error search featured media : {1}".format(self._name, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error search featured media : {1}".format(self._name, err))
else:
self._logger.error("{0} : Connection error for get featured media with status code : {1}".format(self._name, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
except ConnectionError as err:
self._logger.error("{0} : Connection error for get featured media : {1}".format(self._name, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for get featured media : {1}".format(self._name, err))
else:
self._logger.error("{0} : Connection error with status code for featured media : {1}".format(self._name, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
except ConnectionError as err:
self._logger.error("{0} : Connection error : {1}".format(self._name, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Connection error : {1}".format(self._name, err))
## Association image to post
@@ -197,15 +205,17 @@ class WPimport:
for i in list_img:
data = {"post": post_id}
try:
r = self._request.post("http://{0}/wp-json/wp/v2/media/{1}".format(self._wordpress, i["id"]), 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)
r = self._request.post("{2}://{0}/wp-json/wp/v2/media/{1}".format(self._wordpress, i["id"], self._protocol), auth=self._basic, data=data)
if r.status_code == 200:
self._logger.info("{0} : Link image to post {1}".format(self._name, title))
else:
self._logger.error("{0} Connection error with status code for link image to post : {1}".format(self._name, r.status_code))
self._logger.debug("{0} : {1}".format(self._name, r.content))
except ConnectionError as err:
self._logger.error("{0} : Connection error for link image to post : {1}".format(self._name, err))
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
@@ -221,13 +231,11 @@ class WPimport:
self._logger.error("{0} : Element {1} is not image".format(self._name,img_name))
is_img = False
if is_img is True:
self._logger.debug("{0} : Search for image {1} with URL {2}".format(self._name, img_name, "http://{0}/wp-json/wp/v2/media".format(self._wordpress)))
self._logger.debug("{0} : Search for image {1} with URL {2}".format(self._name, img_name, "{1}://{0}/wp-json/wp/v2/media".format(self._wordpress, self._protocol)))
params = { "search": img_name}
try:
r = self._request.get("http://{0}/wp-json/wp/v2/media".format(self._wordpress), 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)
r = self._request.get("{1}://{0}/wp-json/wp/v2/media".format(self._wordpress, self._protocol), auth=self._basic, params=params)
self._logger.debug("{0} : Search for image {1} and his status code {2}".format(self._name, img_name, r.status_code))
if r.status_code == 200:
res = r.json()
@@ -235,15 +243,17 @@ class WPimport:
if len(res) > 0:
params = {"force":1}
try:
r = self._request.delete("http://{0}/wp-json/wp/v2/media/{1}".format(self._wordpress, res[0]["id"]), 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)
r = self._request.delete("{2}://{0}/wp-json/wp/v2/media/{1}".format(self._wordpress, res[0]["id"], self._protocol), auth=self._basic, params=params)
if r.status_code == 200:
self._logger.info("{0} : Image removed {1}".format(self._name, img_name))
else:
self._logger.error("{0} : Image {1} not removed due status code : {2}".format(self._name, img_name, r.status_code))
self._logger.debug("{0} : {1}".format(self._name, r.content))
except ConnectionError as err:
self._logger.error("{0} Connection error for delete image : {1}".format(self._name, err))
exit(1)
except Exception as err:
self._logger.error("{0} Exception error for delete image : {1}".format(self._name, err))
data = page.content
img_type = "image/{0}".format(img_type_file)
@@ -251,10 +261,8 @@ class WPimport:
img_type = "image/jpeg"
headers={ 'Content-Type': img_type,'Content-Disposition' : 'attachment; filename={0}'.format(img_name)}
try:
r = self._request.post("http://{0}/wp-json/wp/v2/media".format(self._wordpress), 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)
r = self._request.post("{1}://{0}/wp-json/wp/v2/media".format(self._wordpress, self._protocol), auth=self._basic, headers=headers, data=data)
if r.status_code == 201:
self._logger.info("{0} : Image added {1}".format(self._name, img_name))
res = r.json()
@@ -263,11 +271,21 @@ class WPimport:
else:
self._logger.error("{0} : Image {1}.{2} not added due status code : {3}".format(self._name, img_name, img_type, r.status_code))
self._logger.debug("{0} : {1}".format(self._name, r.content))
except ConnectionError as err:
self._logger.error("{0} : Connection error for add image : {1}".format(self._name, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for add image : {1}".format(self._name, err))
exit(1)
else:
self._logger.error("{0} : Connection error for search image with status code : {1}".format(self._name, r.status_code))
self._logger.debug("{0} : {1}".format(self._name, r.content))
except ConnectionError as err:
self._logger.error("{0} : Connection error for search media : {1}".format(self._name, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for search media : {1}".format(self._name, err))
return media
## Add or update comment
@@ -277,29 +295,35 @@ class WPimport:
try:
params = {"post": post, "author_name":i["author"], "date":i["date"]}
page = self._request.get("http://{0}/wp-json/wp/v2/comments".format(self._wordpress), 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)
page = self._request.get("{1}://{0}/wp-json/wp/v2/comments".format(self._wordpress, self._protocol), auth=self._basic, params=params)
if page.status_code == 200:
result = page.json()
for j in result:
try:
params = {"force":1}
page = self._request.delete("http://{0}/wp-json/wp/v2/comments/{1}".format(self._wordpress, j["id"]), 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)
page = self._request.delete("{2}://{0}/wp-json/wp/v2/comments/{1}".format(self._wordpress, j["id"], self._protocol), params=params, auth=self._basic)
if page.status_code == 200:
self._logger.info("{0} : Comment deleted for {1}".format(self._name, title))
self._logger.debug("{0} : Comment deleted : {1}".format(self._name, j))
else:
self._logger.error("{0} : Comment not deleted for {1} due status code : {2}".format(self._name, title, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
except ConnectionError as err:
self._logger.error("{0} : Connection error for delete comment : {1}".format(self._name, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for delete comment : {1}".format(self._name, err))
else:
self._logger.error("{0} : Comment not listed for {1} due status code : {2}".format(self._name, title, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
except ConnectionError as err:
self._logger.error("{0} : Connection error for search comment : {1}".format(self._name, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for search comment : {1}".format(self._name, err))
for i in comment:
data = {"post": post, "content": i["content"], "date": i["date"], "author_name": i["author"], "status": "approved"}
@@ -308,10 +332,8 @@ class WPimport:
parent_id = int(i["parent_id"])
params = {"post": post, "author_name":comment[parent_id]["author"], "date":comment[parent_id]["date"]}
try:
page = self._request.get("http://{0}/wp-json/wp/v2/comments".format(self._wordpress), 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)
page = self._request.get("{1}://{0}/wp-json/wp/v2/comments".format(self._wordpress, self._protocol), auth=self._basic, params=params)
if page.status_code == 200:
result = page.json()
if len(result) > 0:
@@ -319,18 +341,27 @@ class WPimport:
else:
self._logger.error("{0} : Connection error for parent comment with status code : {1}".format(self._name, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
except ConnectionError as err:
self._logger.error("{0} : Connection error for parent comment : {1}".format(self._name, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for parent comment : {1}".format(self._name, err))
try:
page = self._request.post("http://{0}/wp-json/wp/v2/comments".format(self._wordpress), 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)
page = self._request.post("{1}://{0}/wp-json/wp/v2/comments".format(self._wordpress, self._protocol), auth=self._basic, data=data)
if page.status_code == 201:
self._logger.info("{0} : Comment added for {1}".format(self._name, title))
self._logger.debug("{0} : Data : {1}".format(self._name, data))
else:
self._logger.error("{0} : Comment not added for {1} due status code : {2}".format(self._name, title, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
except ConnectionError as err:
self._logger.error("{0} : Connection error for add comment : {1}".format(self._name, err))
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
@@ -415,16 +446,17 @@ class WPimport:
new_img["old_href"]=href_a
try:
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:
href_img = href_a
try:
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))
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))
if page_img.status_code == 200:
media=self._addOrUpdateMedia(href_img, page_img)
@@ -439,6 +471,12 @@ class WPimport:
if page_img.status_code not in [200, 404]:
self._logger.error("{0} : Connection error with status code for get image : {1}".format(self._name, page_img.status_code))
self._logger.debug("{0} : {1}".format(self._name, page_img.content))
except ConnectionError as err:
self._logger.error("{0} : Connection error for get image : {1}".format(self._name, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for get image : {1}".format(self._name, err))
exit(1)
self._logger.debug("{0} : Number of image : {1}".format(self._name, len(list_img)))
comment_post = self._getComment(comment)
@@ -460,12 +498,8 @@ class WPimport:
self._logger.info("{0} : search {1} with index {2} : {3}".format(self._name, i, index, title_element))
try:
params = {"search":title_element, "per_page":"100", "page":index}
page = self._request.get("http://{0}/wp-json/wp/v2/{1}".format(self._wordpress, i), 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))
page = self._request.get("{2}://{0}/wp-json/wp/v2/{1}".format(self._wordpress, i, self._protocol), auth=self._basic, params=params)
if page.status_code == 200:
result = page.json()
self._logger.debug("{0} : content {3} {2} : {1}".format(self._name, result, title_element, i))
@@ -490,19 +524,19 @@ class WPimport:
else:
self._logger.error("{0} : {1} not found due status code : {2}".format(self._name, i, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
except ConnectionError as err:
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))
if element_exist == False:
data = {"name": title_element}
self._logger.info("{0} : Create {1} : {2}".format(self._name, i, title_element))
self._logger.debug("{0} : Data : {1}".format(self._name, data))
try:
page = self._request.post("http://{0}/wp-json/wp/v2/{1}".format(self._wordpress, i), 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))
page = self._request.post("{2}://{0}/wp-json/wp/v2/{1}".format(self._wordpress, i, self._protocol), auth=self._basic, headers=self._headers_json, data=json.dumps(data))
if page.status_code == 201:
self._logger.info("{0} : {1} created : {2}".format(self._name, i, j))
result = page.json()
@@ -510,6 +544,11 @@ class WPimport:
else:
self._logger.error("{0} : {1} not added due status code : {2}".format(self._name, i, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
except ConnectionError as err:
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
author = articleacreator[0].text.lower()
@@ -537,16 +576,9 @@ class WPimport:
params = {"search":author, "per_page":100}
try:
self._logger.info("{0} : Search author : {1}".format(self._name, author))
page = self._request.get("http://{0}/wp-json/wp/v2/users".format(self._wordpress), 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} : 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:
self._logger.info("{0} : Get author id : {1}".format(self._name, result))
result = page.json()
@@ -555,18 +587,18 @@ class WPimport:
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))
page_is_exist = False
for index in range(1,10):
params = {"search": title, "per_page":100, "page": index}
try:
self._logger.info("{0} : Search post with index {2} : {1}".format(self._name, title, index))
page = self._request.get("http://{0}/wp-json/wp/v2/posts".format(self._wordpress), 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))
page = self._request.get("{1}://{0}/wp-json/wp/v2/posts".format(self._wordpress, self._protocol), auth=self._basic, params=params, headers=self._headers_json)
if page.status_code == 200:
self._logger.debug("{0} : Encoding : {1}".format(self._name, page.encoding))
page.encoding = "utf-8"
@@ -590,29 +622,25 @@ class WPimport:
self._logger.info("{0} : Page {1} is double and going to delete".format(self._name, title))
try:
params = {"force":1}
page = self._request.delete("http://{0}/wp-json/wp/v2/posts/{1}".format(self._wordpress, post_id), 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))
page = self._request.delete("{2}://{0}/wp-json/wp/v2/posts/{1}".format(self._wordpress, post_id, self._protocol), auth=self._basic, headers=self._headers_json, params=params)
if page.status_code == 200:
self._logger.info("{0} : Post deleted : {1}".format(self._name, title))
else:
self._logger.error("{0} : Post not updated due status code : {1}".format(self._name, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
except ConnectionError as err:
self._logger.error("{0} : Connection error for deleted post : {1}".format(self._name, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for deleted post : {1}".format(self._name, err))
else:
self._logger.debug("{0} : Data for post to update : {1}".format(self._name, i))
self._logger.info("{0} : Page {1} already exist and going to update".format(self._name, title))
try:
page = self._request.post("http://{0}/wp-json/wp/v2/posts/{1}".format(self._wordpress, post_id), 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))
page = self._request.post("{2}://{0}/wp-json/wp/v2/posts/{1}".format(self._wordpress, post_id, self._protocol), auth=self._basic, headers=self._headers_json, data=json.dumps(data))
if page.status_code == 200:
result = page.json()
self._logger.info("{0} : Post updated : {1}".format(self._name, title))
@@ -621,6 +649,11 @@ class WPimport:
else:
self._logger.error("{0} : Post not updated due status code : {1}".format(self._name, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
except ConnectionError as err:
self._logger.error("{0} : Connection error for update post : {1}".format(self._name, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for update post : {1}".format(self._name, err))
if page.status_code == 400:
self._logger.error("{0} : Connection for update post unauthorized : {1}".format(self._name, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
@@ -628,16 +661,17 @@ class WPimport:
else:
self._logger.error("{0} : Connection for update post error with status code : {1}".format(self._name, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
except ConnectionError as err:
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:
try:
self._logger.info("{0} : Creating posts : {1}".format(self._name, data["title"]))
page = self._request.post("http://{0}/wp-json/wp/v2/posts".format(self._wordpress), 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))
page = self._request.post("{1}://{0}/wp-json/wp/v2/posts".format(self._wordpress, self._protocol), auth=self._basic, headers=self._headers_json, data=json.dumps(data))
if page.status_code == 201:
result = page.json()
self._logger.info("{0} : Post added : {1}".format(self._name, result["title"]["raw"]))
@@ -646,3 +680,8 @@ class WPimport:
else:
self._logger.error("{0} : Post not added due status code : {1}".format(self._name, r.status_code))
self._logger.debug("{0} : {1}".format(self._name, r.content))
except ConnectionError as err:
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

@@ -8,19 +8,23 @@ from requests.packages.urllib3.util.retry import Retry
class WPRemove:
# Constructor
def __init__(self, name="Thread-0", basic=None, wordpress="", logger=None):
self._name = name
def __init__(self, index_name=1, number_thread=1, basic=None, wordpress="", logger=None, ssl_wordpress=True):
self._basic = basic
self._wordpress = wordpress
self._logger = logger
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._number_thread = number_thread
if ssl_wordpress is False:
self._protocol = "http"
self._request = requests.Session()
retries = Retry(connect=10, read=10, redirect=5,
status_forcelist=[429, 500, 502, 503, 504], backoff_factor=2)
self._request.mount('http://', HTTPAdapter(max_retries=retries))
self._request.mount('{0}://'.format(self._protocol), HTTPAdapter(max_retries=retries))
# Destructor
@@ -29,6 +33,24 @@ class WPRemove:
# 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):
self._wordpress = wordpress
@@ -47,29 +69,60 @@ class WPRemove:
# Private method
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:
self._logger.info("{0} : List {2} to remove for url : {1}".format(self._name, self._wordpress, composant))
r = self._request.get("http://{0}/wp-json/wp/v2/{1}".format(self._wordpress, composant), 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))
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)
if r.status_code == 200:
result = r.json()
if len(result) > 0:
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}
try:
r = self._request.delete("http://{0}/wp-json/wp/v2/{1}/{2}".format(self._wordpress, composant, i["id"]), 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:
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:
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))
except Exception as err:
self._logger.error("{0} : Connection error for post {1} {2} {3} with status code {4}".format(self._name, self._wordpress, composant, title, r.status_code))
except ConnectionError as err:
self._logger.error("{0} : Connection error for {1} remove : {2}".format(self._name, composant, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for {1} remove : {2}".format(self._name, composant, err))
self._removeAll(composant)
if r.status_code == 400:
self._logger.error("{0} : No content for {1} to remove : {2}".format(self._name, composant, r.status_code))
else:
self._logger.error("{0} : Error for list to remove {1} due status code {2}".format(self._name, composant, r.status_code))
self._logger.debug("{0} : Content error for {1} : {2}".format(self._name, composant, r.content))
except ConnectionError as err:
self._logger.error("{0} : Connection error for list {1} to remove : {2}".format(self._name, composant, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for list {1} to remove : {2}".format(self._name, composant, err))