Merge pull request 'webpage-file' (#20) from webpage-file into master

Reviewed-on: #20

.gitignore

@@ -1,4 +1,4 @@
-backup/
+backup*/
-backup1/
+wp-navigation
-web_scrap.log
+*.log
 __pycache__/

backup_canalblog.sh

@@ -3,8 +3,8 @@
 TAR=/usr/bin/tar
 PYTHON=/usr/bin/python3
 GZIP=/usr/bin/gzip
-SCRIPTDIR=/home/valentin/script
+SCRIPTDIR=/home/valentin/script/webscrap
-WEBSCRAP=${SCRIPTDIR}/web_scrap.py
+WEBSCRAP=${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}

import_export_canalblog.py

@@ -2,9 +2,130 @@
 from requests.auth import HTTPBasicAuth
 from getpass import getpass
 from urllib.parse import urlparse
-import argparse, logging
+from concurrent import futures
+from concurrent.futures import as_completed, wait, ALL_COMPLETED
+
+import argparse, logging, threading, os, glob
 from lib.WPImport import WPimport
 from lib.WPExport import WPExport
+from lib.WPRemove import WPRemove
+from lib.WPChange import WPChange
+
+def change(index, number, args, logger):
+    changeWp = WPChange(logger=logger, index_name=index, number_thread=number)
+    changeWp.fromDirectory(args.directory)
+
+    del changeWp
+
+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)
+            removeWp.cleanPosts()
+            removeWp.cleanTags()
+            removeWp.cleanCategories()
+            removeWp.cleanMedia()
+    else:
+        for i in args.wordpress.split(","):
+            removeWp.setUrl(i)
+            if args.posts == True:
+                removeWp.cleanPosts()
+            if args.categories == True:
+                removeWp.cleanCategories()
+            if args.tags == True:
+                removeWp.cleanTags()
+            if args.media == True:
+                removeWp.cleanMedia()
+    del removeWp
+
+
+def download(name_thread, max_thread, url, logger, parser, directory, html, img, ssl_canalblog, revert, tmp):
+    exportWp = WPExport(name="Thread-{0}".format(int(name_thread) + 1), url=url, logger=logger, parser=parser, directory=directory, ssl_canalblog=ssl_canalblog)    
+    if not revert:
+        exportWp.getUrlPage(name_thread, max_thread)
+    for i in ["article", "page"]:
+        for j in ["publications", "principal"]:
+            if html is False:
+                exportWp.downloadHTML(j, i)
+
+            if img is False:
+                exportWp.downloadImg(j, i)
+    del exportWp
+
+
+def importUrl(name_thread, max_thread, canalblog, logger, parser, wordpress, basic, serial, ssl_wordpress, ssl_canalblog, create, update, image, revert, tmp):
+    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=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, ssl_canalblog=ssl_canalblog, tmp=tmp)
+            if not revert:
+                exportWp.getUrlPage(name_thread, max_thread)
+            del exportWp
+            for j in wordpress:
+                importWp = WPimport(name=name, basic=basic, wordpress=j, logger=logger, parser=parser, ssl_wordpress=ssl_wordpress, no_create=create, no_update=update, no_image=image, tmp=tmp)
+                for k in ["article", "page"]:
+                    for l in ["publications", "principal"]:
+                        importWp.fromUrl(l, k)
+
+                del importWp
+    else:
+        if len(canalblog) != len(wordpress):
+            logger.error("{0} : ERREUR : Le nombre de dossier n'est pas equivalent au nombre d'URL wordpress".format(name))
+            exit(1)
+        for i in range(0, len(canalblog)-1):
+            try:
+                o = urlparse(canalblog[i])
+                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, ssl_canalblog=ssl_canalblog)
+            if not revert:
+                exportWp.getUrlPage(name_thread, max_thread)
+            del exportWp
+            importWp = WPimport(name=name, basic=basic, wordpress=wordpress[i], logger=logger, parser=parser, ssl_wordpress=ssl_wordpress, no_create=create, no_update=update, no_image=image, tmp=tmp)
+
+            for k in ["article", "page"]:
+                for l in ["publications", "principal"]:
+                    importWp.fromUrl(webpage[l][k])
+
+            del importWp
+    
+
+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)
+    directory = directory.split(",")
+    wordpress = wordpress.split(",")
+    if serial is False:
+        for i in wordpress:
+            importWp = WPimport(name=name, basic=basic, wordpress=i, logger=logger, parser=parser, ssl_wordpress=ssl_wordpress, no_create=create, no_update=update, no_image=image)
+            for j in directory:
+                importWp.fromDirectory(j, name_thread, max_thread)
+                del importWp
+
+    else:
+        if len(directory) != len(wordpress):
+            logger.error("{0} : Error : Number directory is different than wordpress".format(name))
+            exit(1)
+        for i in range(0, len(wordpress)-1):
+            importWp = WPimport(name=name, basic=basic, wordpress=wordpress[i], logger=logger, parser=parser, ssl_wordpress=ssl_wordpress, no_create=create, no_update=update, no_image=image)
+            importWp.fromDirectory(directory[i])
+            del importWp
+
 
 
 if __name__ == '__main__':
@@ -13,16 +134,42 @@ if __name__ == '__main__':
     parser.add_argument("--logfile", help="Log file",  default="")
     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="")
+    parser.add_argument("--revert", help="Restart a work from stopping work", action="store_true")
+    parser.add_argument("--tmp", help="directory tmp", default="/tmp/import_export_canablog")
 
     subparsers = parser.add_subparsers(dest="command")
 
     import_parser = subparsers.add_parser("import")
     import_parser.add_argument("--user", help="wordpress user", required=True)
+    import_parser.add_argument("--password", help="password wordpress's user", default="")
     import_parser.add_argument("--file", help="HTML file", default="")
     import_parser.add_argument("--directory", help="HTML directory", default="")
     import_parser.add_argument("--canalblog", help="URL Canalblog", default="")
     import_parser.add_argument("--wordpress", help="URL Wordpress", required=True)
     import_parser.add_argument("--serial", help="Serial execution", action="store_true")
+    import_parser.add_argument("--remove-all", dest="remove", help="Remove all", action="store_true")
+    import_parser.add_argument("--remove-posts", help="Remove all posts", dest="posts", 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-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.add_argument("--user", help="wordpress user", required=True)
+    remove_parser.add_argument("--password", help="password wordpress's user", default="")
+    remove_parser.add_argument("--wordpress", help="URL Wordpress", required=True)
+    remove_parser.add_argument("--all", dest="remove", help="Remove all (posts, media, tags, categories)", action="store_true")
+    remove_parser.add_argument("--posts", help="Remove all posts", action="store_true")
+    remove_parser.add_argument("--categories", help="Remove all categories", action="store_true")
+    remove_parser.add_argument("--tags", help="Remove all tags", action="store_true")
+    remove_parser.add_argument("--media", help="Remove all media", action="store_true")
+
 
 
     export_parser = subparsers.add_parser("export")
@@ -36,12 +183,27 @@ if __name__ == '__main__':
     export_parser.add_argument("--no-img", help="No img", dest="img",  action="store_true")
     export_parser.add_argument("--no-html", help="No HTML", dest="html", action="store_true")
 
+    change_parser = subparsers.add_parser("change")
+    change_parser.add_argument("--directory",
+                      default="",
+                      help="Directory")
+    change_parser.add_argument("--file",
+                      default="",
+                      help="File")
 
     
     args = parser.parse_args()
 
     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()
@@ -64,80 +226,70 @@ if __name__ == '__main__':
         fileHandler.setFormatter(formatter)
         logger.addHandler(fileHandler)
 
-    if args.command == "import":
+    os.makedirs(args.tmp, exist_ok=True)
-        password = getpass()
+
-        if len(password) == 0:
+    if args.command == "import" or args.command == "remove":
-            logger.error("No password error !!! ")
+        password = args.password
-            exit(1)
+        if len(args.password) == 0:
+            password = getpass()
+            if len(password) == 0:
+                logger.error("No password error !!! ")
+                exit(1)
 
         basic = HTTPBasicAuth(args.user, password)
+    if args.command == "import":
         wordpress = args.wordpress.split(",")
-        importWp = WPimport(basic, "", logger, 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(args.file.split(","))
+                importWp.fromFile(files=args.file.split(","))
-            exit(0)
         if len(args.directory) > 0:
-            directory = args.directory.split(",")
+            try:
-            if args.serial is False:
+                with futures.ThreadPoolExecutor(max_workers=int(args.parallel)) as ex:
-                for i in wordpress:
+                    wait_for = [ ex.submit(remove, i, args.parallel, args, basic, logger, ssl_wordpress) for i in range(0, int(args.parallel)) ]
-                    importWp.setUrl(i)
+                    wait(wait_for, return_when=ALL_COMPLETED)
-                    for j in directory:
+                    wait_for = [
-                        importWp.fromDirectory(j)
+                    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)
-            else:
+                    for i in range(0, int(args.parallel))
-                if len(directory) != len(wordpress):
+                    ]
-                    logger.error("ERREUR : Le nombre de dossier n'est pas equivalent au nombre d'URL wordpress")
+            except Exception as err:
-                    exit(1)
+                logger.error("Threading error : {0}".format(err))
-                for i in range(0, len(wordpress)-1):
-                    importWp.setUrl(wordpress[i])
-                    importWp.fromDirectory(directory[i])
-            exit(0)
         if len(args.canalblog) > 0:
-            exportWp = WPExport("", logger, args.parser, args.directory)
+            try:
-            canalblog = args.canalblog.split(",")
+                with futures.ThreadPoolExecutor(max_workers=int(args.parallel)) as ex:
-            wordpress = args.wordpress.split(",")
+                    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)
+                    if args.revert is True:
+                        files_tmp = glob.glob("{0}/*.json".format(args.tmp))
+                        if len(files_tmp) == 0:
+                            logger.error("Error revert, because files not found")
+                            exit(1)
+                        if len(files_tmp) != int(args.parallel):
+                            for file_r in files_tmp:
+                                os.remove(file_r)
                         
-            if args.serial is False:
+                    wait_for = [
-                for canal in 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, args.revert, args.tmp)
-                    try:
+                    for i in range(0, int(args.parallel))
-                        o = urlparse(canal)
+                    ]
-                        o = o._replace(scheme="https")
-                        url = o.geturl().replace(":///", "://")
-                    except Exception as err:
-                        logger.error("parsing error : {0}".format(err))
-                        exit(1)
-                    exportWp.setUrl(url)
-                    webpage = exportWp.getUrlPage()
-                    for j in wordpress:
-                        importWp.setUrl(j)
-                        importWp.fromUrl(webpage)
-            else:
-                if len(canalblog) != len(wordpress):
-                    logger.error("ERREUR : Le nombre de dossier n'est pas equivalent au nombre d'URL wordpress")
-                    exit(1)
-                for i in range(0, len(canalblog)-1):
-                    try:
-                        o = urlparse(canalblog[i])
-                        o = o._replace(scheme="https")
-                        url = o.geturl().replace(":///", "://")
-                    except Exception as err:
-                        logger.error("parsing error : {0}".format(err))
-                        exit(1)
-                    exportWp.setUrl(url)
-                    webpage = exportWp.getUrlPage()
-                    importWp.setUrl(wordpress[i])
-                    importWp.fromUrl(webpage)
 
 
+            except Exception as err:
+                logger.error("Threading error : {0}".format(err))
+        exit(0)
+
 
     if args.command == "export":
         canalblog = args.url.split(",")
-        exportWp = WPExport("", logger, args.parser, 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))
@@ -148,12 +300,39 @@ if __name__ == '__main__':
     
             if args.css is False:
                 exportWp.downloadCss()
+            del exportWp
+
+
     
             if args.html is False or args.img is False:
-                webpage = exportWp.getUrlPage()
+                try:
-            if args.html is False:
+                    with futures.ThreadPoolExecutor(max_workers=int(args.parallel)) as ex:
-                exportWp.downloadHTML(webpage)
+                        wait_for = [
-    
+                        ex.submit(download, i, int(args.parallel), url, logger, args.parser, args.directory, args.html, args.img, ssl_canalblog, args.revert, args.tmp)
-            if args.img is False:
+                        for i in range(0, int(args.parallel))
-                exportWp.downloadImg(webpage)
+                        ]
+                except Exception as err:
+                    logger.error("Threading error : {0}".format(err))
+        exit(0)
+
+
+    if args.command == "remove":
+        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)
+
+    if args.command == "change":
+        if len(args.directory) > 0:
+            try:
+                with futures.ThreadPoolExecutor(max_workers=int(args.parallel)) as ex:
+                    wait_for = [ ex.submit(change, i, args.parallel, args, logger) for i in range(0, int(args.parallel)) ]
+            except Exception as err:
+                logger.error("Thread error for remove : {0}".format(err))
+        if len(args.file) > 0:
+            changeWp = WPChange(logger=logger)
+            for filei in args.file.split(","):
+                changeWp.fromFile(filei)
         exit(0)

lib/WPChange.py

@@ -0,0 +1,128 @@
+from bs4 import BeautifulSoup
+from urllib.parse import urlparse
+import requests, os, logging, re, json
+
+class WPChange:
+    # 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))
+        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):
+        ext_img = ["png", "svg", "gif", "jpg", "jpeg"]
+        try:
+            with open(file, 'r') as f:
+                content = f.read()
+                soup = BeautifulSoup(content, self._parser)
+                img = soup.find_all("img")
+                for i in img:
+                    src = i.get("src")
+                    o = urlparse(src)
+                    if len(o.netloc) > 0:
+                        self._logger.info("{0} : Change source image {1} /img/{2}/{3}".format(self._name, src, o.netloc, o.path))
+                        content = content.replace(src, "/img/{0}/{1}".format(o.netloc, o.path))
+                script = soup.find_all("script", {"type": "text/javascript"})
+                for i in script:
+                    src = i.get("src")
+                    if src is not None:
+                        o = urlparse(src)
+                        if len(o.netloc) > 0:
+                            self._logger.info("{0} : Change source js {1} /dists/js/{2}/{3}".format(self._name, src, o.netloc, o.path))
+                            content = content.replace(src, "/dists/js/{0}/{1}".format(o.netloc, o.path))
+                link = soup.find_all("link", {"rel": "stylesheet"})
+                for i in link:
+                    href = i.get("href")
+                    if href is not None:
+                        o = urlparse(href)
+                        if len(o.netloc) > 0:
+                            self._logger.info("{0} : Change source css {1} /dists/css/{2}/{3}".format(self._name, href, o.netloc, o.path))
+                            content = content.replace(href, "/dists/css/{0}/{1}".format(o.netloc, o.path))
+                
+                a = soup.find_all("a", {"target": "_blank"})
+                for i in a:
+                    href = i.get("href")
+                    if href is not None:
+                        o = urlparse(href)
+                        if len(o.netloc) > 0:
+                            ext = o.path.split(".")[len(o.path.split("."))-1]
+                            if ext in ext_img:
+                                self._logger.info("{0} : Change a img {1} /img/{2}/{3}".format(self._name, href, o.netloc, o.path))
+                                content = content.replace(href, "/img/{0}/{1}".format(o.netloc, o.path))
+                try:
+                    with open(file, "w") as f:
+                        self._logger.info("{0} : File write : {1}".format(self._name, file))
+                        f.write(content)
+                except Exception as ex:
+                    self._logger.error("{0} : Error for write file {1} : {2}".format(self._name, file, ex))
+
+        except Exception as ex:
+            self._logger.error("{0} : Error for read file {1} : {2}".format(self._name, file, ex))
+
+
+
+

lib/WPExport.py

@@ -1,27 +1,41 @@
 #!/usr/bin/python3
 from bs4 import BeautifulSoup
 from urllib.parse import urlparse
-import requests, os, argparse, logging
+import requests, os, argparse, logging, json
 from requests.adapters import HTTPAdapter
 from requests.packages.urllib3.util.retry import Retry
 
 class WPExport:
-    def __init__(self, url, logger, parser, directory):
+    def __init__(self, name = "Thread-0", url = "", logger = None, parser = "html.parser", directory = "backup", ssl_canalblog=True, tmp="/tmp/import_export_canablog"):
         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))
+        self._tmp = tmp
 
+    # Destructor 
+    def __del__(self):
+        self._logger.info("{0} : Export finished for {1}".format(self._name, self._url))
 
     # Public method
 
+
+    # Set name
+
+    def setName(self, name):
+        self._name = "Thread-{0}".format(int(name) + 1)
+
     # Set URL
 
     def setUrl(self, url):
@@ -43,85 +57,119 @@ class WPExport:
 
     # Download HTML
 
-    def downloadHTML(self, webpage):
+    def downloadHTML(self, first, second):
-        self._downloadPage(webpage, self._dir)
+        self._downloadPage(webpage[first][second], self._dir)
 
     # Download Image
     
-    def downloadImg(self, webpage):
+    def downloadImg(self, first, second):
-        page_src = self._getImg(webpage)
+        page_src = self._getImg(webpage[first][second])
         o = urlparse(self._url)
         self._downloadPage(page_src, "{0}/{1}/{2}".format(self._dir, o.path, "img"))
 
 
      # Get URL
-    def getUrlPage(self):
+    def getUrlPage(self, index_thread, max_thread):
         try:
             page = self._request.get(self._url)
-        except Exception as err:
+        
-            self._logger.error("Connection error : {0}".format(err))
+            page_url = []
+            if page.status_code == 200:
+                soup = BeautifulSoup(page.text, self._parser)
+                ul = soup.find_all("ul", id="listsmooth")
+                for anchor in ul[0].find_all("a"):
+                    href = anchor.get('href', '/')
+                    if href != "#":
+                        page_url.append(href)
+            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)
-        page_url = []
+        except Exception as err:
-        if page.status_code == 200:
+            self._logger.error("{0} : Exception error : {1}".format(self._name, err))
-            soup = BeautifulSoup(page.text, self._parser)
-            ul = soup.find_all("ul", id="listsmooth")
-            for anchor in ul[0].find_all("a"):
-                href = anchor.get('href', '/')
-                if href != "#":
-                    page_url.append(href)
-        else:
-            self._logger.error("Url did not get due status code : {0}".format(page.status_code))
-            self._logger.debug(page.content)
 
-
+        webpage = {"principal": {"page":[], "article":[]}, "publications": {"page":[], "article":[]}}
-        webpage = []
         for i in page_url:
+            section = "publications"
+            o = urlparse(i)
+            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("Connection error : {0}".format(err))
-                exit(1)
-            if page.status_code == 200:
-                self._logger.info("page : {0}".format(i))
-                if i not in webpage:
-                    webpage.append(i)
-                soup = BeautifulSoup(page.text, self._parser)
-                class_div = pagingfirstline = soup.find_all("div", class_="pagingfirstline")
-                if len(class_div) > 0:
-                    pagingfirstline = class_div[0].find_all("a")
-                    if len(pagingfirstline) > 1:
-                        lastpage = pagingfirstline[len(pagingfirstline)-1].get("href", "/")
-                        element_lastpage = lastpage.split("/")[len(lastpage.split("/"))-1]
-                        number_page = element_lastpage.split("-")[0].split("p")[1]
-                        number_lastpage = int(number_page) / 10
-                        for j in range(1,int(number_lastpage)):
-                            paging = j * 10
-                            categorie = urlparse(i).path.split("/")
-                            url_paging = "{0}/archives/p{1}-10.html".format(self._url, paging)
-                            if len(categorie) > 2:
-                                url_paging = "{0}/archives/{1}/p{2}-10.html".format(self._url, categorie[2], paging)
-                            self._logger.info(url_paging)
-                            if url_paging not in webpage:
-                                webpage.append(url_paging)
-                            page = self._request.get(url_paging)
-                            if page.status_code == 200:
-                                soup = BeautifulSoup(page.text, self._parser)
-                                h2 = soup.find_all("h2")
-                                for title in h2:
-                                    href = title.find_all("a")[0].get("href", "/")
-                                    if href not in webpage:
-                                        try:
-                                            o = urlparse(href)
-                                            o = o._replace(scheme="https").geturl()
-                                        except Exception as err:
-                                            self._logger.error("parsing error : {0}".format(err))
-                                            exit(1)
-                                        webpage.append(o)
-            else:
-                self._logger.error("web didn't get due status code : {0}".format(page.status_code))
-                self._logger.debug(page.content)
             
-        return webpage
+                if page.status_code == 200:
+                    self._logger.info("{0} : page : {1}".format(self._name, i))
+                    if i not in webpage[section]["page"]:
+                        webpage[section]["page"].append(i)
+                    soup = BeautifulSoup(page.text, self._parser)
+                    class_div = soup.find_all("div", class_="pagingfirstline")
+                    if len(class_div) > 0:
+                        pagingfirstline = class_div[0].find_all("a")
+                        if len(pagingfirstline) > 1:
+                            lastpage = pagingfirstline[len(pagingfirstline)-1].get("href", "/")
+                            self._logger.debug("{0} : Last page {1}".format(self._name, lastpage))
+
+                            element_lastpage = lastpage.split("/")[len(lastpage.split("/"))-1]
+                            number_page = element_lastpage.split("-")[0].split("p")[1]
+                            number_lastpage = int(number_page) / 10
+
+                            setPageDivided = int(number_lastpage) / max_thread
+                            if setPageDivided > int(setPageDivided):
+                                setPageDivided = setPageDivided + 1
+                            setPagePart = setPageDivided * (index_thread + 1)
+                            firstPagePart = (setPagePart - setPageDivided)
+
+                            self._logger.debug("{0} : Total page : {1}".format(self._name,int(number_lastpage)))
+                            self._logger.debug("{0} : First range : {1}".format(self._name, int(firstPagePart)))
+                            self._logger.debug("{0} : Last range : {1}".format(self._name, int(setPagePart)))
+
+                            for j in range(int(firstPagePart),int(setPagePart)+1):
+                                paging = j * 10
+                                categorie = urlparse(i).path.split("/")
+                                url_paging = "{0}/archives/p{1}-10.html".format(self._url, paging)
+                                if len(categorie) > 2:
+                                    url_paging = "{0}/archives/{1}/p{2}-10.html".format(self._url, categorie[2], paging)
+                                self._logger.info("{0} : {1}".format(self._name, url_paging))
+                                if url_paging not in webpage[section]["page"]:
+                                    webpage[section]["page"].append(url_paging)
+                                page = self._request.get(url_paging)
+                                if page.status_code == 200:
+                                    soup = BeautifulSoup(page.text, self._parser)
+                                    h2 = soup.find_all("h2")
+                                    self._logger.debug("{0} : {1} H2 : {2}".format(self._name, url_paging, h2))
+                                    for title in h2:
+                                        self._logger.debug("{0} : {1} a : {2}".format(self._name, url_paging, title.find_all("a")))
+                                        href = title.find_all("a")[0].get("href", "/")
+                                        if href not in webpage[section]["article"]:
+                                            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)
+                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)
+        try:
+            string_webpage = json.dumps(webpage)
+            open("{0}/{1}.json".format(self._tmp, self._name), "wt").write(string_webpage)
+        except Exception as ex:
+            self._logger.error("{0} : Error for writing webpage : {1}".format(self._name, ex))
+
+
+
+        
 
 
     # Private method
@@ -135,7 +183,7 @@ class WPExport:
                 makedir.append(i)
                 repath = "/".join(makedir)
                 if not os.path.exists(repath):
-                    self._logger.debug("Dossier crée : {0}".format(repath))
+                    self._logger.debug("{0} : Dossier crée : {1}".format(self._name, repath))
                     try:
                         if len(repath) > 0:
                             os.mkdir(repath)
@@ -149,47 +197,52 @@ 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))
+            page_url = []
-            exit(1)
+            if page.status_code == 200:
-        page_url = []
+                soup = BeautifulSoup(page.text, self._parser)
-        if page.status_code == 200:
+                if js is True:
-            soup = BeautifulSoup(page.text, self._parser)
+                    script = soup.find_all("script")
-            if js is True:
+                    for anchor in script:
-                script = soup.find_all("script")
+                        src = anchor.get("src", "/")
-                for anchor in script:
+                        if src != "/":
-                    src = anchor.get("src", "/")
-                    if src != "/":
-                        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())
-            if css is True:
-                link = soup.find_all("link")
-                for anchor in link:
-                    rel = anchor.get("rel")
-                    if rel[0] == "stylesheet":
-                        href = anchor.get("href", "/")
-                        if href != "/":
                             try:
                                 u = urlparse(self._url)
-                                o = urlparse(href)
+                                o = urlparse(src)
+                                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 o.netloc == "":
+                            
-                                o = o._replace(netloc=u.netloc)
+                if css is True:
-                                o = o._replace(scheme=u.scheme)
+                    link = soup.find_all("link")
-                            page_url.append(o.geturl())        
+                    for anchor in link:
-        else:
+                        rel = anchor.get("rel")
-            self._logger.error("JS or CSS did not get due status code : {0}".format(page.status_code))
+                        if rel[0] == "stylesheet":
-            self._logger.debug(page.content)
+                            href = anchor.get("href", "/")
+                            if href != "/":
+                                try:
+                                    u = urlparse(self._url)
+                                    o = urlparse(href)
+                                    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
 
@@ -200,22 +253,25 @@ class WPExport:
         for i in webpage:
             try:
                 page = self._request.get(i)
-            except Exception as err:
+            
-                self._logger.error("Connection error : {0}".format(err))
+                if page.status_code == 200:
+                    soup = BeautifulSoup(page.text, self._parser)
+                    img = soup.find_all("img")
+                    self._logger.info("{0} : image from page: {1} : ".format(self._name,i))
+                    for anchor in img:
+                        src = anchor.get("src", "/")
+                        if src != "/":
+                            if src not in page_img:
+                                self._logger.info("{0} : image: {1} : ".format(self._name, src))
+                                page_img.append(src)
+                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)
-            if page.status_code == 200:
+            except Exception as err:
-                soup = BeautifulSoup(page.text, self._parser)
+                self._logger.error("{0} : Exception error : {1}".format(self._name, err))
-                img = soup.find_all("img")
-                self._logger.info("image from page: {0} : ".format(i))
-                for anchor in img:
-                    src = anchor.get("src", "/")
-                    if src != "/":
-                        if src not in page_img:
-                            self._logger.info("image: {0} : ".format(src))
-                            page_img.append(src)
-            else:
-                self._logger.error("Image did not get due status code : {0}".format(page.status_code))
-                self._logger.debug(page.content)
 
         return page_img
 
@@ -226,29 +282,33 @@ class WPExport:
         for i in range(0, len(webpage)):
             try:
                 o = urlparse(webpage[i])
+            
+                path_web = o.path.split("/")
+                filePageWeb = path_web[len(path_web)-1]
+                path_web.pop(len(path_web)-1)
+                dir_page_web = "/".join(path_web)
+                self._mkdirPath("{0}/{1}/{2}".format(backup_dir, o.netloc, dir_page_web))
+                try:
+                    r = self._request.get(webpage[i])
+                
+                    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:
+                            fileDownload = "{0}/{1}{2}/{3}".format(backup_dir, o.netloc, dir_page_web, filePageWeb)
+                        self._logger.info("{0} : {1}/{2} : {3}".format(self._name, i+1, len(webpage), fileDownload))
+                        try:
+                            open(fileDownload, "wb").write(r.content)
+                        except Exception as err:
+                            self._logger.error("file error : {0}".format(err))
+                            exit(1)
+                    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)
-            path_web = o.path.split("/")
-            filePageWeb = path_web[len(path_web)-1]
-            path_web.pop(len(path_web)-1)
-            dir_page_web = "/".join(path_web)
-            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:
-                    fileDownload = "{0}/{1}{2}/{3}".format(backup_dir, o.netloc, dir_page_web, filePageWeb)
-                self._logger.info("{0}/{1} : {2}".format(i+1, len(webpage), fileDownload))
-                try:
-                    open(fileDownload, "wb").write(r.content)
-                except Exception as err:
-                    self._logger.error("file error : {0}".format(err))
-                    exit(1)
-            else:
-                self._logger.error("Not download due status code : {0}".format(r.status_code))
-                self._logger.debug(r.content)

lib/WPImport.py

@@ -8,67 +8,130 @@ from requests.packages.urllib3.util.retry import Retry
 
 class WPimport:
     # Constructor
-    def __init__(self, basic, wordpress, logger, parser):
+    def __init__(self, name="Thread-0", basic=None, wordpress="", logger=None, parser="html.parser", ssl_wordpress=True, no_create=False, no_update=False, no_image=False, tmp="/tmp/import_export_canablog"):
+        self._name = name
         self._basic = basic
         self._wordpress = wordpress
         self._logger = logger
         self._parser = parser
-        self._headers_json = {'Content-Type': 'application/json', 'Accept':'application/json'}
+        self._headers_json = {'Content-Type': 'application/json; charset=utf-8', 'Accept':'application/json'}
-
+        self._protocol = "https"
+        self._directory = "backup"
+        if ssl_wordpress is False:
+            self._protocol = "http"
         self._request = requests.Session()
 
-        retries = Retry(total=10,
+        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))
+        self._no_create = no_create
+        self._no_update = no_update
+        self._no_image = no_image
+        self._tmp = tmp
+    
+    # Destructor 
+    def __del__(self):
+        print("{0} : Import finished for {1}".format(self._name, self._wordpress))
     
     # Public method
 
     def setUrl(self, wordpress):
         self._wordpress = wordpress
 
-    def fromUrl(self, webpage):
+    def fromUrl(self, first, second):
-        for i in range(0, len(webpage)):
+        try:
-            try:
+            with open("{0}/{1}.json".format(self._tmp, self._name)) as file:
-                r = self._request.get(webpage[i])
+                webpage_content = json.loads(file.read())
-            except Exception as err:
+                self._logger.debug("{0} : size of webpage : {1}".format(self._name, len(webpage_content)))
-                self._logger.error("Connection error : {0}".format(err))
+                webpage = webpage_content[first][second]
-                exit(1)
+                for i in range(0, len(webpage)):
-            if r.status_code == 200:
+                    try:
-                self._logger.info("({0}/{1} : Page en cours d'import : {2}".format(i+1, len(webpage), webpage[i]))
+                        r = self._request.get(webpage[i])
-                soup = BeautifulSoup(r.content, self._parser)
+                        if r.status_code == 200:
-                articlebody = soup.find_all("div", class_="articlebody")
+                            self._logger.info("{0} : ({1}/{2}) : Page is importing : {3}".format(self._name, i+1, len(webpage), webpage[i]))
-                if len(articlebody) > 0:
+                            soup = BeautifulSoup(r.content, self._parser)
-                    self._addOrUpdatePost(soup)
+                            articlebody = soup.find_all("div", class_="articlebody")
-                else:
+                            if len(articlebody) > 0:
-                    self._addOrUpdateFeaturedMedia(soup)
+                                self._addOrUpdatePost(soup)
-            else:
+                            else:
-                self._logger.error("Connection error with status code : {0}".format(r.status_code))
+                                self._addOrUpdateFeaturedMedia(soup)
-                self._logger.debug(r.content)
+                            del webpage_content[first][second][i]
+                            webpage_content = json.dumps(webpage_content)
+                            open("{0}/{1}.json".format(self._tmp, self._name), "wt").write(webpage_content)
+                        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 IOError as err:
+                        self._logger.error("{0} : Connection error for IO 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))
+        except Exception as ex:
+            self._logger.error("{0} : Read file json from tmp : {1}".format(self._name, ex))
 
 
-    def fromDirectory(self, directory):
+    def fromDirectory(self, directory="", number_thread=1, max_thread=1):
+        self._directory = directory
         directory = "{0}/archives".format(directory)
         directories = self._getDirectories([], "{0}".format(directory))
-        files = self._getFiles(directories)
+        if len(directories) > 0:
-        self.fromFile(files)
+            files = self._getFiles(directories)
+            self.fromFile(files=files, number_thread=number_thread, max_thread=max_thread)
+        else:
+            self._logger.error("{0} : No files for {1}".format(self._name, directory))
     
             
-    def fromFile(self, files):
+
-        for file in files:
+
-            if os.path.exists(file):
+        
-                self._logger.info("Fichier en cours de traitement : {0}".format(file))
+
-                with open(file, 'r') as f:
+    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]))
+                with open(files[i], 'r') as f:
                     content = f.read()
+                    self._logger.debug("{0} : Size of article : {1}".format(self._name, len(content)))
                     soup = BeautifulSoup(content, self._parser)
                     articlebody = soup.find_all("div", class_="articlebody")
+                    self._logger.debug("{0} : Number of article : {1}".format(self._name, len(articlebody)))
                     if len(articlebody) > 0:
                         self._addOrUpdatePost(soup)
                     else:
                         self._addOrUpdateFeaturedMedia(soup)
     
+    
     # Private method
 
+    ## replace caracter 
+
+    def _replaceCaracter(self, title_rendered):
+        list_replace = {'’': "'", '–': '-', '…': '...', '« ': '"', ' »': '"', '« ': '"', ' »': '"', '’': "'", '"‘': "'"}
+        for old, new in list_replace.items():
+            title_rendered = title_rendered.replace(old, new)
+        return title_rendered
+
+    ## remove space 
+
+    def _removeSpace(self, title):
+        if title[len(title)-1] == " ":
+            title = title[:-1]
+        if title[0] == " ":
+            title = title[1:]
+        return title
+
     ## Get all files
 
     def _getFiles(self, item):
@@ -97,61 +160,66 @@ 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)
+                page = self._request.get("{1}://{0}/wp-json/wp/v2/search".format(self._wordpress, self._protocol), auth=self._basic, params=params)
-            except Exception as err:
+                if page.status_code == 200:
-                self._logger.error("Connection error : {0}".format(err))
+                    result = page.json()
-                exit(1)
+                    if len(result) > 0:
-            if page.status_code == 200:
+                        if h2 == result[0]["title"]:
-                result = page.json()
+                            img = i.find_all("img")
-                if len(result) > 0:
+                            if len(img) > 0:
-                    if h2 == result[0]["title"]:
+                                img_src = img[0].get("src")
-                        img = i.find_all("img")
-                        if len(img) > 0:
-                            img_src = img[0].get("src")
-                            try:
-                                page = self._request.get(img_src)
-                            except Exception as err:
-                                self._logger.error("Connection error : {0}".format(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)
+                                    page = self._request.get(img_src)
-                                except Exception as err:
+                                    if page.status_code == 200:
-                                    self._logger.error("Connection error : {0}".format(err))
+                                        name_img = img_src.replace("_q", "")
-                                    exit(1)
+                                        name_img = name_img.split("/")[len(name_img.split("/"))-1]
-                                if page.status_code == 200:
+                                        params = {"search": name_img}
-                                    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))
+                                            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("Connection error : {0}".format(err))
+                                            if page.status_code == 200:
+                                                res = page.json()
+                                                if len(res) > 0:
+                                                    id_media = res[0]["id"]
+                                                    data = {"featured_media": id_media}
+                                                    try:
+                                                        r = self._request.post("{2}://{0}/wp-json/wp/v2/posts/{1}".format(self._wordpress, result[0]["id"], self._protocol), auth=self._basic, headers=self._headers_json, data=json.dumps(data))
+                                                        if r.status_code == 200:
+                                                            self._logger.info("{0} : Add media featured : {1}".format(self._name, r.json()["title"]["raw"]))
+                                                        else:
+                                                            self._logger.error("{0} : Connection error with status code for featured media : {1}".format(self._name, r.status_code))
+                                                            self._logger.debug("{0} : {1}".format(self._name, r.content))
+                                                    except ConnectionError as err:
+                                                        self._logger.error("{0} : Connection error for post media featured : {1}".format(self._name, err))
+                                                        exit(1)
+                                                    except Exception as err:
+                                                        self._logger.error("{0} : Exception error for post media featured : {1}".format(self._name, err))
+                                                else:
+                                                    self._logger.info("{0} : No media found for {1}".format(self._name, h2))
+                                            else:
+                                                self._logger.error("{0} : Connection error with status code for search featured media: {1}".format(self._name, page.status_code))
+                                                self._logger.debug("{0} : {1}".format(self._name, page.content))
+                                        except ConnectionError as err:
+                                            self._logger.error("{0} : Connection error search featured media : {1}".format(self._name, err))
                                             exit(1)
-                                        if r.status_code == 200:
+                                        except Exception as err:
-                                            self._logger.info("Ajout media featured : {0}".format(r.json()["title"]["raw"]))
+                                            self._logger.error("{0} : Exception error search featured media : {1}".format(self._name, err))
-                                        else:
-                                            self._logger.error("Connection error with status code : {0}".format(r.status_code))
-                                            self._logger.debug(r.content)
-
                                     else:
-                                        self._logger.info("Aucun media trouvé pour {0}".format(h2))
+                                        self._logger.error("{0} : Connection error for get featured media with status code : {1}".format(self._name, page.status_code))
-                                else:
+                                        self._logger.debug("{0} : {1}".format(self._name, page.content))
-                                    self._logger.error("Connection error with status code : {0}".format(page.status_code))
+                                except ConnectionError as err:
-                                    self._logger.debug(page.content)
+                                    self._logger.error("{0} : Connection error for get featured media : {1}".format(self._name, err))
-
+                                    exit(1)
-                            else:
+                                except Exception as err:
-                                self._logger.error("Connection error with status code : {0}".format(page.status_code))
+                                    self._logger.error("{0} : Exception error for get featured media : {1}".format(self._name, err))
-                                self._logger.debug(page.content)
+                else:
-
+                    self._logger.error("{0} : Connection error with status code for featured media : {1}".format(self._name, page.status_code))
-            else:
+                    self._logger.debug("{0} : {1}".format(self._name, page.content))
-                self._logger.error("Connection error with status code : {0}".format(page.status_code))
+            except ConnectionError as err:
-                self._logger.debug(page.content)
+                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
 
@@ -159,67 +227,87 @@ 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)
+                r = self._request.post("{2}://{0}/wp-json/wp/v2/media/{1}".format(self._wordpress, i["id"], self._protocol), auth=self._basic, data=data)
-            except Exception as err:
+                if r.status_code == 200:
-                self._logger.error("Connection error : {0}".format(err))
+                    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)
-            if r.status_code == 200:
+            except Exception as err:
-                self._logger.info("Association d'une image à l'article {0}".format(title))
+                self._logger.error("{0} : Exception error for link image to post : {1}".format(self._name, err))
-            else:
-                self._logger.error("Connection error with status code : {0}".format(r.status_code))
-                self._logger.debug(r.content)
 
     
     ## Add or update img 
 
     def _addOrUpdateMedia(self, href_img, page):
+        media_authorized = ["png", "jpg", "jpeg", "svg", "gif"]
         media = {"id":"", "rendered":""}
         split_fileimg = href_img.split("/")
         img_name = split_fileimg[len(split_fileimg)-1]
-        params = { "search": img_name}
+        img_type_file = img_name.split(".")[len(img_name.split("."))-1]
-        try:
+        is_img = True
-            r = self._request.get("http://{0}/wp-json/wp/v2/media".format(self._wordpress), auth=self._basic, params=params)
+        if img_type_file not in media_authorized:
-        except Exception as err:
+            self._logger.error("{0} : Element {1} is not image".format(self._name,img_name))
-            self._logger.error("Connection error : {0}".format(err))
+            is_img = False
-            exit(1)
+        if is_img is True:
-        if r.status_code == 200:
+            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)))
-            res = r.json()
+            params = { "search": img_name}
-            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("Connection error : {0}".format(err))
-                    exit(1)
-                if r.status_code == 200:
-                    self._logger.info("Image removed {0}".format(img_name))
-                else:
-                    self._logger.error("Image not removed due status code : {0}".format(r.status_code))
-                    self._logger.debug(r.content)
-
-            data = page.content
-            img_type = "image/png"
-            if img_name.split(".")[1] == "jpg" or img_name.split(".")[1] == "jpeg":
-                img_type = "image/jpg"
-            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)
+                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("Connection error : {0}".format(err))
+                self._logger.debug("{0} : Search for image {1} and his status code {2}".format(self._name, img_name, r.status_code))
+                if r.status_code == 200:
+                    res = r.json()
+                    self._logger.debug("{0} : Number of image in search : {1}".format(self._name, len(res)))
+                    if len(res) > 0:
+                        params = {"force":1}
+                        try:
+                            r = self._request.delete("{2}://{0}/wp-json/wp/v2/media/{1}".format(self._wordpress, res[0]["id"], self._protocol), auth=self._basic, params=params)
+                            if r.status_code == 200:
+                                self._logger.info("{0} : Image removed {1}".format(self._name, img_name))
+                            else:
+                                self._logger.error("{0} : Image {1} not removed due status code : {2}".format(self._name, img_name, r.status_code))
+                                self._logger.debug("{0} : {1}".format(self._name, r.content))
+                        except ConnectionError as err:
+                            self._logger.error("{0} Connection error for delete image : {1}".format(self._name, err))
+                            exit(1)
+                        except Exception as err:
+                            self._logger.error("{0} Exception error for delete image : {1}".format(self._name, err))
+
+                    data = page.content
+                    img_type = "image/{0}".format(img_type_file)
+                    if img_type_file == "jpg":
+                        img_type = "image/jpeg"
+                    headers={ 'Content-Type': img_type,'Content-Disposition' : 'attachment; filename={0}'.format(img_name)}
+                    try:
+                        r = self._request.post("{1}://{0}/wp-json/wp/v2/media".format(self._wordpress, self._protocol), auth=self._basic, headers=headers, data=data)
+                    
+                        if r.status_code == 201:
+                            self._logger.info("{0} : Image added {1}".format(self._name, img_name))
+                            res = r.json()
+                            media["id"] = res["id"]
+                            media["rendered"] = res["guid"]["rendered"]
+                        else:
+                            self._logger.error("{0} : Image {1}.{2} not added due status code : {3}".format(self._name, img_name, img_type, r.status_code))
+                            self._logger.debug("{0} : {1}".format(self._name, r.content))
+                    except ConnectionError as err:
+                        self._logger.error("{0} : Connection error for add image : {1}".format(self._name, err))
+                        exit(1)
+                    except Exception as err:
+                        self._logger.error("{0} : Exception error for add image : {1}".format(self._name, err))
+                        exit(1)
+
+                else:
+                    self._logger.error("{0} : Connection error for search image with status code : {1}".format(self._name, r.status_code))
+                    self._logger.debug("{0} : {1}".format(self._name, r.content))
+            except ConnectionError as err:
+                self._logger.error("{0} : Connection error for search media : {1}".format(self._name, err))
                 exit(1)
-            if r.status_code == 201:
+            except Exception as err:
-                self._logger.info("Image added {0}".format(img_name))
+                self._logger.error("{0} : Exception error for search media : {1}".format(self._name, err))
-                res = r.json()
-                media["id"] = res["id"]
-                media["rendered"] = res["guid"]["rendered"]
-            else:
-                self._logger.error("Image not added due status code : {0}".format(r.status_code))
-                self._logger.debug(r.content)
-
-        else:
-            self._logger.error("Connection error with status code : {0}".format(r.status_code))
-            self._logger.debug(r.content)
-
         return media
 
     ## Add or update comment
@@ -229,29 +317,35 @@ 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)
+                    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("Connection error : {0}".format(err))
-                    exit(1)
-                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("Connection error : {0}".format(err))
-                            exit(1)
-                        if page.status_code == 200:
-                            self._logger.info("Comment deleted for {0}".format(title))
-                            self._logger.debug("Comment deleted : {0}".format(j))
-                        else:
-                            self._logger.error("Comment not deleted for {0} due status code : {1}".format(title, page.status_code))
-                            self._logger.debug(page.content)
                 
-                else:
+                    if page.status_code == 200:
-                    self._logger.error("Comment not listed for {0} due status code : {1}".format(title, page.status_code))
+                        result = page.json()
-                    self._logger.debug(page.content)
+                        for j in result:
+                            try:
+                                params = {"force":1}
+                                page = self._request.delete("{2}://{0}/wp-json/wp/v2/comments/{1}".format(self._wordpress, j["id"], self._protocol), params=params, auth=self._basic)
+                            
+                                if page.status_code == 200:
+                                    self._logger.info("{0} : Comment deleted for {1}".format(self._name, title))
+                                    self._logger.debug("{0} : Comment deleted : {1}".format(self._name, j))
+                                else:
+                                    self._logger.error("{0} : Comment not deleted for {1} due status code : {2}".format(self._name, title, page.status_code))
+                                    self._logger.debug("{0} : {1}".format(self._name, page.content))
+                            except ConnectionError as err:
+                                self._logger.error("{0} : Connection error for delete comment : {1}".format(self._name, err))
+                                exit(1)
+                            except Exception as err:
+                                self._logger.error("{0} : Exception error for delete comment : {1}".format(self._name, err))
+
+                    else:
+                        self._logger.error("{0} : Comment not listed for {1} due status code : {2}".format(self._name, title, page.status_code))
+                        self._logger.debug("{0} : {1}".format(self._name, page.content))
+                except ConnectionError as err:
+                    self._logger.error("{0} : Connection error for search comment : {1}".format(self._name, err))
+                    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"}
@@ -260,29 +354,36 @@ 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)
+                        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("Connection error : {0}".format(err))
+                        if page.status_code == 200:
+                            result = page.json()
+                            if len(result) > 0:
+                                data["parent"]=result[0]["id"]
+                        else:
+                            self._logger.error("{0} : Connection error for parent comment with status code : {1}".format(self._name, page.status_code))
+                            self._logger.debug("{0} : {1}".format(self._name, page.content))
+                    except ConnectionError as err:
+                        self._logger.error("{0} : Connection error for parent comment : {1}".format(self._name, err))
                         exit(1)
-                    if page.status_code == 200:
+                    except Exception as err:
-                        result = page.json()
+                        self._logger.error("{0} : Exception error for parent comment : {1}".format(self._name, err))
-                        if len(result) > 0:
-                            data["parent"]=result[0]["id"]
-                    else:
-                        self._logger.error("Connection error for parent comment with status code : {0}".format(page.status_code))
-                        self._logger.debug(page.content)
 
                 try:
-                    page = self._request.post("http://{0}/wp-json/wp/v2/comments".format(self._wordpress), 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("Connection error : {0}".format(err))
+                    if page.status_code == 201:
+                        self._logger.info("{0} : Comment added for {1}".format(self._name, title))
+                        self._logger.debug("{0} : Data : {1}".format(self._name, data))
+                    else:
+                        self._logger.error("{0} : Comment not added for {1} due status code : {2}".format(self._name, title, page.status_code))
+                        self._logger.debug("{0} : {1}".format(self._name, page.content))
+                except ConnectionError as err:
+                    self._logger.error("{0} : Connection error for add comment : {1}".format(self._name, err))
+                    exit(1)
+                except Exception as err:
+                    self._logger.error("{0} : Exception error for add comment : {1}".format(self._name, err))
                     exit(1)
-                if page.status_code == 201:
-                    self._logger.info("Comment added for {0}".format(title))
-                    self._logger.debug("Data : {0}".format(data))
-                else:
-                    self._logger.error("Comment not added for {0} due status code : {1}".format(title, page.status_code))
-                    self._logger.debug(page.content)
     
     ## Check class name 
 
@@ -346,6 +447,7 @@ class WPimport:
             listelement[i] = []
 
         articletitle = soup.find_all("h2", class_="articletitle")
+        self._logger.debug("{0} : Title of the article : {1}".format(self._name, articletitle))
         articlebody = soup.find_all("div", class_="articlebody")
         articledate = soup.find_all("span", class_="articledate")
         articleacreator = soup.find_all("span", class_="articlecreator")
@@ -353,43 +455,68 @@ class WPimport:
         itemfooter = soup.find_all("div", class_="itemfooter")
         comment = soup.find_all("li", class_="comment")
         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)))
         list_img = []
-        for i in img_a:
+        if self._no_image is False:
-            new_img = {}
+            for i in img_a:
-            img = i.find_all("img")
+                new_img = {}
-            if len(img) > 0:
+                img = i.find_all("img")
-                href_a = i.get("href")
+                self._logger.debug("{0} : Number of image's tag : {1}".format(self._name, len(img)))
-                href_img = img[0].get("src")
+                if len(img) > 0:
-                new_img["old_src"]=href_img
+                    href_a = i.get("href")
-                new_img["old_href"]=href_a
+                    href_img = img[0].get("src")
-                try:
+                    href_a_o = urlparse(href_a)
-                    page_img = self._request.get(href_img)
+                    href_img_o = urlparse(href_img)
-                except Exception as err:
+                    new_img["old_src"]=href_img
-                    self._logger.error("Connection error : {0}".format(err))
+                    new_img["old_href"]=href_a
-                    exit(1)
-                if page_img.status_code == 404:
-                    href_img = href_a
                     try:
-                        page_img = self._request.get(href_a)
+                        if len(href_img_o.netloc) > 0:
+                            img_ok = False
+                            page_img = self._request.get(href_img)
+                
+                            if page_img.status_code == 404:
+                                href_img = href_a
+                                try:
+                                    page_img = self._request.get(href_a)
+                                    if page_img.status_code == 200:
+                                        img_ok = True
+                                except ConnectionError as err:
+                                    self._logger.error("{0} : Connection error for get image : {1}".format(self._name, err))
+                                    exit(1)
+                                except Exception as err:
+                                    self._logger.error("{0} : Exception error for get image : {1}".format(self._name, err))
+                                    exit(1)
+                            
+                        else:
+                            if os.path.exists("{0}/..{1}".format(self._directory, href_img)):
+                                page_img = open("{0}/..{1}".format(self._directory, href_img), "r")
+                                img_ok = True
+                            else:
+                                if os.path.exists("{0}/..{1}".format(self._directory, href_a)):
+                                    page_img = open("{0}/..{1}".format(self._directory, href_a), "r")
+                                    img_ok = True
+                        self._logger.debug("{0} : Status code for image {1} : {2}".format(self._name, href_img, page_img.status_code))
+                        if img_ok is True:
+                            media=self._addOrUpdateMedia(href_img, page_img)
+                            new_img["id"]=media["id"]
+                            new_img["new_src"]=media["rendered"]
+                            list_img.append(new_img)
+                            if href_img != href_a:
+                                media=self._addOrUpdateMedia(href_a, page_img)
+                                new_img["id"]=media["id"]
+                                new_img["new_src"]=media["rendered"]
+                                list_img.append(new_img)
+                        if page_img.status_code not in [200, 404]:
+                            self._logger.error("{0} : Connection error with status code for get image : {1}".format(self._name, page_img.status_code))
+                            self._logger.debug("{0} : {1}".format(self._name, page_img.content))
+                    except ConnectionError as err:
+                        self._logger.error("{0} : Connection error for get image : {1}".format(self._name, err))
+                        exit(1)
                     except Exception as err:
-                        self._logger.error("Connection error : {0}".format(err))
+                        self._logger.error("{0} : Exception error for get image : {1}".format(self._name, err))
                         exit(1)
 
-                if page_img.status_code == 200:
+        self._logger.debug("{0} : Number of image : {1}".format(self._name, len(list_img)))
-                    media=self._addOrUpdateMedia(href_img, page_img)
-                    new_img["id"]=media["id"]
-                    new_img["new_src"]=media["rendered"]
-                    list_img.append(new_img)
-                    if href_img != href_a:
-                        media=self._addOrUpdateMedia(href_a, page_img)
-                        new_img["id"]=media["id"]
-                        new_img["new_src"]=media["rendered"]
-                        list_img.append(new_img)
-                if page_img.status_code not in [200, 404]:
-                    self._logger.error("Connection error with status code : {0}".format(page_img.status_code))
-                    self._logger.debug(page_img.content)
-
-                
         comment_post = self._getComment(comment)
 
         a = itemfooter[0].find_all("a", {"rel": True})
@@ -404,38 +531,62 @@ class WPimport:
         for i in liste:
             for j in element[i]:
                 element_exist = False
-                try:
+                title_element = self._removeSpace(j)
-                    params = {"params":j}
+                for index in range(1,10):
-                    page = self._request.get("http://{0}/wp-json/wp/v2/{1}".format(self._wordpress, i), auth=self._basic, params=params)
+                    self._logger.info("{0} : search {1} with index {2} : {3}".format(self._name, i, index, title_element))
-                except Exception as err:
-                    self._logger.error("Connection error : {0}".format(err))
-                    exit(1)
-                if page.status_code == 200:
-                    element_exist = True
-                    result = page.json()
-                    listelement[i].append(result[0]["id"])
-
-                else:
-                    self._logger.error("{0} not found due status code : {1}".format(i, page.status_code))
-                    self._logger.debug(page.content)
-
-
-                if element_exist is False:
-                    data = {"name": j}
-                    self._logger.debug("URL : {0} ".format("http://{0}/wp-json/wp/v2/{1}".format(self._wordpress, i)))
-                    self._logger.debug("data : {0}".format(data))
-                    self._logger.debug("headers : {0}".format(self._headers_form))
                     try:
-                        page = self._request.post("http://{0}/wp-json/wp/v2/{1}".format(self._wordpress, i), auth=self._basic, headers=self._headers_json, data=data)
+                        params = {"search":title_element, "per_page":"100", "page":index}
-                    except Exception as err:
+                        page = self._request.get("{2}://{0}/wp-json/wp/v2/{1}".format(self._wordpress, i, self._protocol), auth=self._basic, params=params)
-                        self._logger.error("Connection error : {0}".format(err))
+                    
+                        if page.status_code == 200:
+                            result = page.json()
+                            self._logger.debug("{0} : content {3} {2} : {1}".format(self._name, result, title_element, i))
+                            if len(result) > 0:
+                                for k in result:
+                                    title_rendered = k["name"]
+                                    self._logger.debug("{0} : content {2} : {1}".format(self._name, title_rendered, i))
+                                    self._logger.debug("{0} : size of content {3} : {2} - {1}".format(self._name, len(title_rendered), len(title_element), i))
+                                    if len(title_element) != len(title_rendered):
+                                        title_rendered = self._replaceCaracter(title_rendered)
+                        
+                                    if title_element == title_rendered:    
+                                        self._logger.info("{0} : {1} found : {2}".format(self._name, i, title_rendered))
+                                        element_exist = True
+                                        listelement[i].append(k["id"])
+                            else:
+                                break
+                        if page.status_code == 400:
+                            self._logger.error("{0} : {1} not found due status code : {2}".format(self._name, i, page.status_code))
+                            self._logger.debug("{0} : {1}".format(self._name, page.content))
+                            break
+                        else:
+                            self._logger.error("{0} : {1} not found due status code : {2}".format(self._name, i, page.status_code))
+                            self._logger.debug("{0} : {1}".format(self._name, page.content))
+                    except ConnectionError as err:
+                        self._logger.error("{0} : Connection error for {1} : {2}".format(self._name, i, err))
                         exit(1)
-                    if page.status_code == 201:
+                    except Exception as err:
-                        result = page.json()
+                        self._logger.error("{0} : Exception error for {1} : {2}".format(self._name, i, err))
-                        listelement[i].append(result["id"])
+                self._logger.debug("{0} : Element {3} {2} is {1}".format(self._name, element_exist, title_element, i))
-                    else:
+                if element_exist == False:
-                        self._logger.error("{0} not added due status code : {1}".format(i, page.status_code))
+                    data = {"name": title_element}
-                        self._logger.debug(page.content)
+                    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("{2}://{0}/wp-json/wp/v2/{1}".format(self._wordpress, i, self._protocol), auth=self._basic, headers=self._headers_json, data=json.dumps(data))
+                    
+                        if page.status_code == 201:
+                            self._logger.info("{0} : {1} created : {2}".format(self._name, i, j))
+                            result = page.json()
+                            listelement[i].append(result["id"])
+                        else:
+                            self._logger.error("{0} : {1} not added due status code : {2}".format(self._name, i, page.status_code))
+                            self._logger.debug("{0} : {1}".format(self._name, page.content))
+                    except ConnectionError as err:
+                        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()        
@@ -453,66 +604,123 @@ class WPimport:
             bodyhtml = bodyhtml.replace(i["old_src"], o.path)
         hour = articledate[0].text
         time = dateheader[0].text.split(" ")
+        self._logger.debug("{0} : Title post : |{1}|".format(self._name, title))
+        title = self._removeSpace(title)
+        self._logger.debug("{0} : Rendered Title post : |{1}|".format(self._name, title))
+
         data = {"title":title, "content":bodyhtml, "status":"publish", "date": "{0}-{1}-{2}T{3}:00".format(time[2],month[time[1]],time[0], hour), "tags": listelement["tags"], "categories": listelement["categories"]}
-        params = {"search":author}
+        self._logger.debug("{0} : Data for post : |{1}| : {2}" .format(self._name, title, data))
-        try:
-            page = self._request.get("http://{0}/wp-json/wp/v2/users".format(self._wordpress), auth=self._basic, params=params)
-        except Exception as err:
-            self._logger.error("Connection error : {0}".format(err))
-            exit(1)
-        if page.status_code == 200:
-            result = page.json()
-            data["author"] = result[0]["id"]
-        else:
-            self._logger.error("Connection error with status code : {0}".format(page.status_code))
-            self._logger.debug(page.content)
 
-
+        params = {"search":author, "per_page":100}
-        params = {"search":title}
         try:
-            page = self._request.get("http://{0}/wp-json/wp/v2/posts".format(self._wordpress), auth=self._basic, params=params)
+            self._logger.info("{0} : Search author : {1}".format(self._name, author))
-        except Exception as err:
+            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.error("Connection error : {0}".format(err))
+            self._logger.debug("{0} : End Search author : {1}".format(self._name, author))
-            exit(1)
+            self._logger.debug("{0} : Debug requests : {1}".format(self._name, page.content))
-        page_exist = True
+            if page.status_code == 200:
-        headers = {'Content-Type': 'application/json', 'Accept':'application/json'}
+                self._logger.info("{0} : Get author id : {1}".format(self._name, result))
-        if page.status_code == 200:
+                result = page.json()
-            result = page.json()
+                for a in result:
-            if len(result) == 0:
+                    data["author"] = a["id"]
-                page_exist = False
             else:
-                self._logger.info("La page {0} existe deja et mis à jour".format(title))
+                self._logger.error("{0} : Connection error with status code for get author : {1}".format(self._name, page.status_code))
-                post_id = result[0]["id"]
+                self._logger.debug("{0} : {1}".format(page.content))
-                try:
+        except ConnectionError as err:
-                    page = self._request.post("http://{0}/wp-json/wp/v2/posts/{1}".format(self._wordpress, post_id), auth=self._basic, headers=headers, data=json.dumps(data))
+            self._logger.error("{0} : Connection error for get author : {1}".format(self._name, err))
-                except Exception as err:
+            exit(1)
-                    self._logger.error("Connection error : {0}".format(err))
+        except Exception as err:
-                    exit(1)
+            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("{1}://{0}/wp-json/wp/v2/posts".format(self._wordpress, self._protocol), auth=self._basic, params=params, headers=self._headers_json)
                 if page.status_code == 200:
+                    self._logger.debug("{0} : Encoding : {1}".format(self._name, page.encoding))
+                    page.encoding = "utf-8"
                     result = page.json()
-                    self._logger.info("Article mis à jour : {0}".format(result["title"]["raw"]))
+                    if len(result) == 0:
+                        break
+                    self._logger.info("{0} : Number result posts : {1}".format(self._name, len(result)))
+                    count = 0
+                    for i in result:
+                        title_rendered = i["title"]["rendered"]
+                        self._logger.info("{0} : Search title posts for |{2}| : |{1}|".format(self._name, title_rendered, title))
+                        if len(title_rendered) != len(title):
+                            title_rendered = self._replaceCaracter(title_rendered)
+                        self._logger.debug("{0} : Search title posts for |{2}| : |{1}|".format(self._name, title_rendered, title))
+                        self._logger.debug("{0} : SIze of title : {1} - {2}".format(self._name, len(title), len(title_rendered)))
+                        if title_rendered == title:
+                            if self._no_update is False:
+                                page_is_exist = True
+                                post_id = i["id"]
+                                count = count + 1
+                                if count > 1:
+                                    self._logger.info("{0} : Page {1} is double and going to delete".format(self._name, title))
+                                    try:
+                                        params = {"force":1}
+                                        page = self._request.delete("{2}://{0}/wp-json/wp/v2/posts/{1}".format(self._wordpress, post_id, self._protocol), auth=self._basic, headers=self._headers_json, params=params)
+                                        if page.status_code == 200:
+                                            self._logger.info("{0} : Post deleted : {1}".format(self._name, title))
+                                        else:
+                                            self._logger.error("{0} : Post not updated due status code : {1}".format(self._name, page.status_code))
+                                            self._logger.debug("{0} : {1}".format(self._name, page.content))
+                                    except ConnectionError as err:
+                                        self._logger.error("{0} : Connection error for deleted post : {1}".format(self._name, err))
+                                        exit(1)
+                                    except Exception as err:
+                                        self._logger.error("{0} : Exception error for deleted post : {1}".format(self._name, err))
+                    
+                                else:
+                                    self._logger.debug("{0} : Data for post to update : {1}".format(self._name, i))
+                                    self._logger.info("{0} : Page {1} already exist and going to update".format(self._name, title))
+                
+                                    try:
+                                        page = self._request.post("{2}://{0}/wp-json/wp/v2/posts/{1}".format(self._wordpress, post_id, self._protocol), auth=self._basic, headers=self._headers_json, data=json.dumps(data))
+                                
+                                        if page.status_code == 200:
+                                            result = page.json()
+                                            self._logger.info("{0} : Post updated : {1}".format(self._name, title))
+                                            self._addOrUpdateComment(result["id"], comment_post, result["title"]["raw"])
+                                            self._linkImgPost(result["title"]["raw"], list_img, result["id"])
+                                        else:
+                                            self._logger.error("{0} : Post not updated due status code : {1}".format(self._name, page.status_code))
+                                            self._logger.debug("{0} : {1}".format(self._name, page.content))
+                                    except ConnectionError as err:
+                                        self._logger.error("{0} : Connection error for update post : {1}".format(self._name, err))
+                                        exit(1)
+                                    except Exception as err:
+                                        self._logger.error("{0} : Exception error for update post : {1}".format(self._name, err))
+                if page.status_code == 400:
+                    self._logger.error("{0} : Connection for update post unauthorized : {1}".format(self._name, page.status_code))
+                    self._logger.debug("{0} : {1}".format(self._name, page.content))
+                    break
+                else:
+                    self._logger.error("{0} : Connection for update post error with status code : {1}".format(self._name, page.status_code))
+                    self._logger.debug("{0} : {1}".format(self._name, page.content))
+            except ConnectionError as err:
+                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 and self._no_create is False:
+            try:
+                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))
+            
+                if page.status_code == 201:
+                    result = page.json()
+                    self._logger.info("{0} : Post added : {1}".format(self._name, result["title"]["raw"]))
                     self._addOrUpdateComment(result["id"], comment_post, result["title"]["raw"])
                     self._linkImgPost(result["title"]["raw"], list_img, result["id"])
                 else:
-                    self._logger.error("Post not updated due status code : {0}".format(page.status_code))
+                    self._logger.error("{0} : Post not added due status code : {1}".format(self._name, r.status_code))
-                    self._logger.debug(page.content)
+                    self._logger.debug("{0} : {1}".format(self._name, r.content))
-
+            except ConnectionError as err:
-        else:
+                self._logger.error("{0} : Connection error for create post : {1}".format(self._name, err))
-            self._logger.error("Connection for update post error with status code : {0}".format(page.status_code))
-            self._logger.debug(page.content)
-
-                    
-        if page_exist == False:
-            try:
-                page = self._request.post("http://{0}/wp-json/wp/v2/posts".format(self._wordpress), auth=self._basic, headers=headers, data=json.dumps(data))
-            except Exception as err:
-                self._logger.error("Connection error : {0}".format(err))
                 exit(1)
-            if page.status_code == 201:
+            except Exception as err:
-                result = page.json()
+                self._logger.error("{0} : Exception error for create post : {1}".format(self._name, err))
-                self._logger.info("Article ajoute : {0}".format(result["title"]["raw"]))
-                self._addOrUpdateComment(result["id"], comment_post, result["title"]["raw"])
-                self._linkImgPost(result["title"]["raw"], list_img, result["id"])
-            else:
-                self._logger.error("Post not added due status code : {0}".format(r.status_code))
-                self._logger.debug(r.content)

lib/WPRemove.py

@@ -0,0 +1,128 @@
+#!/usr/bin/python3
+
+from bs4 import BeautifulSoup
+from urllib.parse import urlparse
+import requests, os, logging, re, json
+from requests.adapters import HTTPAdapter
+from requests.packages.urllib3.util.retry import Retry
+
+class WPRemove:
+    # Constructor
+    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('{0}://'.format(self._protocol), HTTPAdapter(max_retries=retries))
+
+    
+    # Destructor 
+    def __del__(self):
+        print("{0} : Import finished for {1}".format(self._name, self._wordpress))
+
+    # 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
+    
+    def cleanPosts(self):
+        self._removeAll("posts")
+
+    def cleanTags(self):
+        self._removeAll("tags")
+
+    def cleanCategories(self):
+        self._removeAll("categories")
+    
+    def cleanMedia(self):
+        self._removeAll("media")
+
+    # Private method
+    
+    def _removeAll(self, composant):
+        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("{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:
+                            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("{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, title))
+                                    else:
+                                        self._logger.error("{0} : Connection error for post {1} {2} {3} with status code {4}".format(self._name, self._wordpress, composant, title, r.status_code))
+                                except ConnectionError as err:
+                                    self._logger.error("{0} : Connection error for {1} remove : {2}".format(self._name, composant, err))
+                                    exit(1)
+                                except Exception as err:
+                                    self._logger.error("{0} : Exception error for {1} remove : {2}".format(self._name, composant, err))
+                                self._removeAll(composant)
+                if r.status_code == 400:
+                    self._logger.error("{0} : No content for {1} to remove : {2}".format(self._name, composant, r.status_code))
+                else:
+                    self._logger.error("{0} : Error for list to remove {1} due status code {2}".format(self._name, composant, r.status_code))
+                    self._logger.debug("{0} : Content error for {1} : {2}".format(self._name, composant, r.content))
+            except ConnectionError as err:
+                self._logger.error("{0} : Connection error for list {1} to remove : {2}".format(self._name, composant, err))
+                exit(1)
+            except Exception as err:
+                self._logger.error("{0} : Exception error for list {1} to remove : {2}".format(self._name, composant, err))