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

Reviewed-on: #20

import_export_canalblog.py

@@ -54,7 +54,6 @@ def download(name_thread, max_thread, url, logger, parser, directory, html, img,
     del exportWp
 
 
-
 def importUrl(name_thread, max_thread, canalblog, logger, parser, wordpress, basic, serial, ssl_wordpress, ssl_canalblog, create, update, image, revert, tmp):
     canalblog = canalblog.split(",")
     wordpress = wordpress.split(",")
@@ -107,30 +106,29 @@ def importUrl(name_thread, max_thread, canalblog, logger, parser, wordpress, bas
             del importWp
     
 
-def importDirectory(name_thread, max_thread, directory, logger, parser, wordpress, basic, serial, ssl_wordpress, create, update, image, tmp, revert):
+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, tmp=tmp, revert=revert)
+            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, no_create=create, no_update=update, no_image=image, tmp=tmp, revert=revert)
+            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__':
-    TMP = "/tmp/import_export_canablog"
     parser = argparse.ArgumentParser()
     parser.add_argument("--debug", help="Verbosity",  action="store_true")
     parser.add_argument("--logfile", help="Log file",  default="")
@@ -139,6 +137,7 @@ if __name__ == '__main__':
     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")
 
@@ -192,9 +191,6 @@ if __name__ == '__main__':
                       default="",
                       help="File")
 
-
-    if not os.path.exists(TMP):
-        os.mkdir(TMP)
     
     args = parser.parse_args()
 
@@ -230,6 +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:
@@ -252,7 +250,7 @@ if __name__ == '__main__':
                     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, args.create, args.update, args.image, TMP, revert)
+                    ex.submit(importDirectory, i, int(args.parallel), args.directory, logger, args.parser, args.wordpress, basic, args.serial, ssl_wordpress, args.create, args.update, args.image)
                     for i in range(0, int(args.parallel))
                     ]
             except Exception as err:
@@ -262,8 +260,17 @@ if __name__ == '__main__':
                 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, args.create, args.update, args.image, args.revert, TMP)
+                    ex.submit(importUrl, i, int(args.parallel), args.canalblog, logger, args.parser, args.wordpress, basic, args.serial, ssl_wordpress, ssl_canalblog, args.create, args.update, args.image, args.revert, args.tmp)
                     for i in range(0, int(args.parallel))
                     ]
 
@@ -301,7 +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, args.revert, TMP)
+                        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:

lib/WPExport.py

@@ -1,7 +1,7 @@
 #!/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
 

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, no_create=False, no_update=False, no_image=False, tmp="/tmp/import_export_canablog", revert=False):
+    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
@@ -29,7 +29,6 @@ class WPimport:
         self._no_update = no_update
         self._no_image = no_image
         self._tmp = tmp
-        self._revert = revert
     
     # Destructor 
     def __del__(self):
@@ -42,55 +41,50 @@ class WPimport:
 
     def fromUrl(self, first, second):
         try:
-            content_file = open("{0}/{1}.json".format(self._name, self._tmp))
-            webpage_content = json.loads(content_file)
-            webpage = webpage_content[first][second]
-            for i in range(0, len(webpage)):
-                try:
-                    r = self._request.get(webpage[i])
-                    if r.status_code == 200:
-                        self._logger.info("{0} : ({1}/{2}) : Page is importing : {3}".format(self._name, i+1, len(webpage), webpage[i]))
-                        soup = BeautifulSoup(r.content, self._parser)
-                        articlebody = soup.find_all("div", class_="articlebody")
-                        if len(articlebody) > 0:
-                            self._addOrUpdatePost(soup)
+            with open("{0}/{1}.json".format(self._tmp, self._name)) as file:
+                webpage_content = json.loads(file.read())
+                self._logger.debug("{0} : size of webpage : {1}".format(self._name, len(webpage_content)))
+                webpage = webpage_content[first][second]
+                for i in range(0, len(webpage)):
+                    try:
+                        r = self._request.get(webpage[i])
+                        if r.status_code == 200:
+                            self._logger.info("{0} : ({1}/{2}) : Page is importing : {3}".format(self._name, i+1, len(webpage), webpage[i]))
+                            soup = BeautifulSoup(r.content, self._parser)
+                            articlebody = soup.find_all("div", class_="articlebody")
+                            if len(articlebody) > 0:
+                                self._addOrUpdatePost(soup)
+                            else:
+                                self._addOrUpdateFeaturedMedia(soup)
+                            del webpage_content[first][second][i]
+                            webpage_content = json.dumps(webpage_content)
+                            open("{0}/{1}.json".format(self._tmp, self._name), "wt").write(webpage_content)
                         else:
-                            self._addOrUpdateFeaturedMedia(soup)
-                        del webpage_content[first][second][i]
-                        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))
-
-
+                            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):
-        if self._revert:
-            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)
-            else:
-                self._logger.error("{0} : No files for {1}".format(self._name, directory))
+        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=files, number_thread=number_thread, max_thread=max_thread)
         else:
-            try:
-                files = open("{0}/{1}.json".format(self._name, self._tmp))
-                self.fromFile(files, number_thread, max_thread)
-            except Exception as ex:
-                self._logger.error("{0} : Read file json from tmp : {1}".format(self._name, ex))
+            self._logger.error("{0} : No files for {1}".format(self._name, directory))
+    
+