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

Reviewed-on: #20

import_export_canalblog.py

@@ -3,12 +3,19 @@ from requests.auth import HTTPBasicAuth
 from getpass import getpass
 from urllib.parse import urlparse
 from concurrent import futures
-from concurrent.futures import as_completed, wait
+from concurrent.futures import as_completed, wait, ALL_COMPLETED
 
-import argparse, logging, threading
+import argparse, logging, threading, 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)
@@ -33,21 +40,21 @@ def remove(index, number, args, basic, logger, ssl_wordpress):
     del removeWp
 
 
-def download(name_thread, max_thread, url, logger, parser, directory, html, img, ssl_canalblog):
+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)    
-    webpage = exportWp.getUrlPage(name_thread, max_thread)
+    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(webpage[j][i])
+                exportWp.downloadHTML(j, i)
 
             if img is False:
-                exportWp.downloadImg(webpage[j][i])
+                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):
-
-def importUrl(name_thread, max_thread, canalblog, logger, parser, wordpress, basic, serial, ssl_wordpress, ssl_canalblog):
     canalblog = canalblog.split(",")
     wordpress = wordpress.split(",")
     name = "Thread-{0}".format(int(name_thread) + 1)
@@ -63,14 +70,15 @@ def importUrl(name_thread, max_thread, canalblog, logger, parser, wordpress, bas
             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)
+            exportWp = WPExport(name="Thread-{0}".format(int(name_thread) + 1), url=url, logger=logger, parser=parser, ssl_canalblog=ssl_canalblog, tmp=tmp)
-            webpage = exportWp.getUrlPage(name_thread, max_thread)
+            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)
+                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(webpage[l][k])
+                        importWp.fromUrl(l, k)
 
                 del importWp
     else:
@@ -86,9 +94,10 @@ def importUrl(name_thread, max_thread, canalblog, logger, parser, wordpress, bas
                 logger.error("parsing error : {0}".format(err))
                 exit(1)
             exportWp = WPExport(name=name, url=url, logger=logger, parser=parser, ssl_canalblog=ssl_canalblog)
-            webpage = exportWp.getUrlPage(name_thread, max_thread)
+            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)
+            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"]:
@@ -97,23 +106,23 @@ def importUrl(name_thread, max_thread, canalblog, logger, parser, wordpress, bas
             del importWp
     
 
-def importDirectory(name_thread, max_thread, directory, logger, parser, wordpress, basic, serial, ssl_wordpress):
+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)
+            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 differant than wordpress".format(name))
+            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)
+            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
 
@@ -127,6 +136,8 @@ if __name__ == '__main__':
     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")
 
@@ -143,6 +154,10 @@ if __name__ == '__main__':
     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")
@@ -168,6 +183,13 @@ 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()
@@ -204,6 +226,8 @@ if __name__ == '__main__':
         fileHandler.setFormatter(formatter)
         logger.addHandler(fileHandler)
 
+    os.makedirs(args.tmp, exist_ok=True)
+
     if args.command == "import" or args.command == "remove":
         password = args.password
         if len(args.password) == 0:
@@ -221,23 +245,36 @@ if __name__ == '__main__':
                 importWp.setUrl(i)
                 importWp.fromFile(files=args.file.split(","))
         if len(args.directory) > 0:
-            remove(args, basic, logger, ssl_wordpress)
             try:
                 with futures.ThreadPoolExecutor(max_workers=int(args.parallel)) as ex:
+                    wait_for = [ ex.submit(remove, i, args.parallel, args, basic, logger, ssl_wordpress) for i in range(0, int(args.parallel)) ]
+                    wait(wait_for, return_when=ALL_COMPLETED)
                     wait_for = [
-                    ex.submit(importDirectory, i, int(args.parallel), args.directory, logger, args.parser, args.wordpress, basic, args.serial, ssl_wordpress)
+                    ex.submit(importDirectory, i, int(args.parallel), args.directory, logger, args.parser, args.wordpress, basic, args.serial, ssl_wordpress, args.create, args.update, args.image)
                     for i in range(0, int(args.parallel))
                     ]
             except Exception as err:
                 logger.error("Threading error : {0}".format(err))
         if len(args.canalblog) > 0:
-            remove(args, basic, logger, ssl_wordpress)
             try:
                 with futures.ThreadPoolExecutor(max_workers=int(args.parallel)) as ex:
+                    wait_for = [ ex.submit(remove, i, args.parallel, args, basic, logger, ssl_wordpress) for i in range(0, int(args.parallel)) ]
+                    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)
+                        
                     wait_for = [
-                    ex.submit(importUrl, i, int(args.parallel), args.canalblog, logger, args.parser, args.wordpress, basic, args.serial, ssl_wordpress, ssl_canalblog)
+                    ex.submit(importUrl, i, int(args.parallel), args.canalblog, logger, args.parser, args.wordpress, basic, args.serial, ssl_wordpress, ssl_canalblog, args.create, args.update, args.image, args.revert, args.tmp)
                     for i in range(0, int(args.parallel))
                     ]
+
+
             except Exception as err:
                 logger.error("Threading error : {0}".format(err))
         exit(0)
@@ -271,7 +308,7 @@ if __name__ == '__main__':
                 try:
                     with futures.ThreadPoolExecutor(max_workers=int(args.parallel)) as ex:
                         wait_for = [
-                        ex.submit(download, i, int(args.parallel), url, logger, args.parser, args.directory, args.html, args.img, ssl_canalblog)
+                        ex.submit(download, i, int(args.parallel), url, logger, args.parser, args.directory, args.html, args.img, ssl_canalblog, args.revert, args.tmp)
                         for i in range(0, int(args.parallel))
                         ]
                 except Exception as err:
@@ -286,3 +323,16 @@ if __name__ == '__main__':
         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

@@ -1,12 +1,12 @@
 #!/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, name = "Thread-0", url = "", logger = None, parser = "html.parser", directory = "backup", ssl_canalblog=True):
+    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
@@ -22,7 +22,7 @@ class WPExport:
                 status_forcelist=[429, 500, 502, 503, 504], backoff_factor=2)
 
         self._request.mount('{0}://'.format(self._protocol), HTTPAdapter(max_retries=retries))
-
+        self._tmp = tmp
 
     # Destructor 
     def __del__(self):
@@ -57,13 +57,13 @@ 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"))
 
@@ -161,7 +161,14 @@ class WPExport:
             except Exception as err:
                 self._logger.error("{0} : Exception error : {1}".format(self._name, err))
                 exit(1)
-        return webpage
+        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))
+
+
+
         
 
 

lib/WPImport.py

@@ -8,7 +8,7 @@ from requests.packages.urllib3.util.retry import Retry
 
 class WPimport:
     # Constructor
-    def __init__(self, name="Thread-0", basic=None, wordpress="", logger=None, parser="html.parser", ssl_wordpress=True):
+    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
@@ -16,6 +16,7 @@ class WPimport:
         self._parser = parser
         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()
@@ -24,7 +25,10 @@ class WPimport:
                 status_forcelist=[429, 500, 502, 503, 504], backoff_factor=2)
 
         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):
@@ -35,38 +39,56 @@ class WPimport:
     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())
-                if r.status_code == 200:
+                self._logger.debug("{0} : size of webpage : {1}".format(self._name, len(webpage_content)))
-                    self._logger.info("{0} : ({1}/{2}) : Page is importing : {3}".format(self._name, i+1, len(webpage), webpage[i]))
+                webpage = webpage_content[first][second]
-                    soup = BeautifulSoup(r.content, self._parser)
+                for i in range(0, len(webpage)):
-                    articlebody = soup.find_all("div", class_="articlebody")
+                    try:
-                    if len(articlebody) > 0:
+                        r = self._request.get(webpage[i])
-                        self._addOrUpdatePost(soup)
+                        if r.status_code == 200:
-                    else:
+                            self._logger.info("{0} : ({1}/{2}) : Page is importing : {3}".format(self._name, i+1, len(webpage), webpage[i]))
-                        self._addOrUpdateFeaturedMedia(soup)
+                            soup = BeautifulSoup(r.content, self._parser)
-                else:
+                            articlebody = soup.find_all("div", class_="articlebody")
-                    self._logger.error("{0} : Connection error for get url {1} with status code : {2}".format(self._name, webpage[i], r.status_code))
+                            if len(articlebody) > 0:
-                    self._logger.debug("{0} : {1}".format(self._name, r.content))
+                                self._addOrUpdatePost(soup)
-            except ConnectionError as err:
+                            else:
-                self._logger.error("{0} : Connection error for get url {1} : {2}".format(self._name, webpage[i], err))
+                                self._addOrUpdateFeaturedMedia(soup)
-                exit(1)
+                            del webpage_content[first][second][i]
-            except Exception as err:
+                            webpage_content = json.dumps(webpage_content)
-                self._logger.error("{0} : Exception error for get url {1} : {2}".format(self._name, webpage[i], err))
+                            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="", number_thread=1, max_thread=1):
+        self._directory = directory
         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)
+            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=[], number_thread=1, max_thread=1):
         divFiles = int(len(files) / max_thread)
         currentRangeFiles = int(divFiles * (number_thread+1))
@@ -435,48 +457,64 @@ class WPimport:
         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 = {}
-            self._logger.debug("{0} : Number of image's tag : {1}".format(self._name, len(img)))
+                img = i.find_all("img")
-            if len(img) > 0:
+                self._logger.debug("{0} : Number of image's tag : {1}".format(self._name, len(img)))
-                href_a = i.get("href")
+                if len(img) > 0:
-                href_img = img[0].get("src")
+                    href_a = i.get("href")
-                new_img["old_src"]=href_img
+                    href_img = img[0].get("src")
-                new_img["old_href"]=href_a
+                    href_a_o = urlparse(href_a)
-                try:
+                    href_img_o = urlparse(href_img)
-                    page_img = self._request.get(href_img)
+                    new_img["old_src"]=href_img
+                    new_img["old_href"]=href_a
+                    try:
+                        if len(href_img_o.netloc) > 0:
+                            img_ok = False
+                            page_img = self._request.get(href_img)
                 
-                    if page_img.status_code == 404:
+                            if page_img.status_code == 404:
-                        href_img = href_a
+                                href_img = href_a
-                        try:
+                                try:
-                            page_img = self._request.get(href_a)
+                                    page_img = self._request.get(href_a)
-                        except ConnectionError as err:
+                                    if page_img.status_code == 200:
-                            self._logger.error("{0} : Connection error for get image : {1}".format(self._name, err))
+                                        img_ok = True
-                            exit(1)
+                                except ConnectionError as err:
-                        except Exception as err:
+                                    self._logger.error("{0} : Connection error for get image : {1}".format(self._name, err))
-                            self._logger.error("{0} : Exception error for get image : {1}".format(self._name, err))
+                                    exit(1)
-                            exit(1)
+                                except Exception as err:
-                    self._logger.debug("{0} : Status code for image {1} : {2}".format(self._name, href_img, page_img.status_code))
+                                    self._logger.error("{0} : Exception error for get image : {1}".format(self._name, err))
-                    if page_img.status_code == 200:
+                                    exit(1)
-                        media=self._addOrUpdateMedia(href_img, page_img)
+                            
-                        new_img["id"]=media["id"]
+                        else:
-                        new_img["new_src"]=media["rendered"]
+                            if os.path.exists("{0}/..{1}".format(self._directory, href_img)):
-                        list_img.append(new_img)
+                                page_img = open("{0}/..{1}".format(self._directory, href_img), "r")
-                        if href_img != href_a:
+                                img_ok = True
-                            media=self._addOrUpdateMedia(href_a, page_img)
+                            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 page_img.status_code not in [200, 404]:
+                            if href_img != href_a:
-                        self._logger.error("{0} : Connection error with status code for get image : {1}".format(self._name, page_img.status_code))
+                                media=self._addOrUpdateMedia(href_a, page_img)
-                        self._logger.debug("{0} : {1}".format(self._name, page_img.content))
+                                new_img["id"]=media["id"]
-                except ConnectionError as err:
+                                new_img["new_src"]=media["rendered"]
-                    self._logger.error("{0} : Connection error for get image : {1}".format(self._name, err))
+                                list_img.append(new_img)
-                    exit(1)
+                        if page_img.status_code not in [200, 404]:
-                except Exception as err:
+                            self._logger.error("{0} : Connection error with status code for get image : {1}".format(self._name, page_img.status_code))
-                    self._logger.error("{0} : Exception error for get image : {1}".format(self._name, err))
+                            self._logger.debug("{0} : {1}".format(self._name, page_img.content))
-                    exit(1)
+                    except ConnectionError as err:
+                        self._logger.error("{0} : Connection error for get image : {1}".format(self._name, err))
+                        exit(1)
+                    except Exception as err:
+                        self._logger.error("{0} : Exception error for get image : {1}".format(self._name, err))
+                        exit(1)
 
         self._logger.debug("{0} : Number of image : {1}".format(self._name, len(list_img)))
         comment_post = self._getComment(comment)
@@ -615,45 +653,46 @@ class WPimport:
                         self._logger.debug("{0} : Search title posts for |{2}| : |{1}|".format(self._name, title_rendered, title))
                         self._logger.debug("{0} : SIze of title : {1} - {2}".format(self._name, len(title), len(title_rendered)))
                         if title_rendered == title:
-                            page_is_exist = True
+                            if self._no_update is False:
-                            post_id = i["id"]
+                                page_is_exist = True
-                            count = count + 1
+                                post_id = i["id"]
-                            if count > 1:
+                                count = count + 1
-                                self._logger.info("{0} : Page {1} is double and going to delete".format(self._name, title))
+                                if count > 1:
-                                try:
+                                    self._logger.info("{0} : Page {1} is double and going to delete".format(self._name, title))
-                                    params = {"force":1}
+                                    try:
-                                    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)
+                                        params = {"force":1}
-                                    if page.status_code == 200:
+                                        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)
-                                        self._logger.info("{0} : Post deleted : {1}".format(self._name, title))
+                                        if page.status_code == 200:
-                                    else:
+                                            self._logger.info("{0} : Post deleted : {1}".format(self._name, title))
-                                        self._logger.error("{0} : Post not updated due status code : {1}".format(self._name, page.status_code))
+                                        else:
-                                        self._logger.debug("{0} : {1}".format(self._name, page.content))
+                                            self._logger.error("{0} : Post not updated due status code : {1}".format(self._name, page.status_code))
-                                except ConnectionError as err:
+                                            self._logger.debug("{0} : {1}".format(self._name, page.content))
-                                    self._logger.error("{0} : Connection error for deleted post : {1}".format(self._name, err))
+                                    except ConnectionError as err:
-                                    exit(1)
+                                        self._logger.error("{0} : Connection error for deleted post : {1}".format(self._name, err))
-                                except Exception as err:
+                                        exit(1)
-                                    self._logger.error("{0} : Exception error for deleted post : {1}".format(self._name, err))
+                                    except Exception as err:
+                                        self._logger.error("{0} : Exception error for deleted post : {1}".format(self._name, err))
                     
-                            else:
+                                else:
-                                self._logger.debug("{0} : Data for post to update : {1}".format(self._name, i))
+                                    self._logger.debug("{0} : Data for post to update : {1}".format(self._name, i))
-                                self._logger.info("{0} : Page {1} already exist and going to update".format(self._name, title))
+                                    self._logger.info("{0} : Page {1} already exist and going to update".format(self._name, title))
                 
-                                try:
+                                    try:
-                                    page = self._request.post("{2}://{0}/wp-json/wp/v2/posts/{1}".format(self._wordpress, post_id, self._protocol), auth=self._basic, headers=self._headers_json, data=json.dumps(data))
+                                        page = self._request.post("{2}://{0}/wp-json/wp/v2/posts/{1}".format(self._wordpress, post_id, self._protocol), auth=self._basic, headers=self._headers_json, data=json.dumps(data))
                                 
-                                    if page.status_code == 200:
+                                        if page.status_code == 200:
-                                        result = page.json()
+                                            result = page.json()
-                                        self._logger.info("{0} : Post updated : {1}".format(self._name, title))
+                                            self._logger.info("{0} : Post updated : {1}".format(self._name, title))
-                                        self._addOrUpdateComment(result["id"], comment_post, result["title"]["raw"])
+                                            self._addOrUpdateComment(result["id"], comment_post, result["title"]["raw"])
-                                        self._linkImgPost(result["title"]["raw"], list_img, result["id"])
+                                            self._linkImgPost(result["title"]["raw"], list_img, result["id"])
-                                    else:
+                                        else:
-                                        self._logger.error("{0} : Post not updated due status code : {1}".format(self._name, page.status_code))
+                                            self._logger.error("{0} : Post not updated due status code : {1}".format(self._name, page.status_code))
-                                        self._logger.debug("{0} : {1}".format(self._name, page.content))
+                                            self._logger.debug("{0} : {1}".format(self._name, page.content))
-                                except ConnectionError as err:
+                                    except ConnectionError as err:
-                                    self._logger.error("{0} : Connection error for update post : {1}".format(self._name, err))
+                                        self._logger.error("{0} : Connection error for update post : {1}".format(self._name, err))
-                                    exit(1)
+                                        exit(1)
-                                except Exception as err:
+                                    except Exception as err:
-                                    self._logger.error("{0} : Exception error for update post : {1}".format(self._name, 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))
@@ -667,7 +706,7 @@ class WPimport:
             except Exception as err:
                 self._logger.error("{0} : Exception error for search post : {1}".format(self._name, err))
 
-        if page_is_exist is False:
+        if page_is_exist is False and self._no_create is False:
             try:
                 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))