Merge pull request 'add-parameter' (#17) from add-parameter into master

Reviewed-on: #17

import_export_canalblog.py

@@ -47,7 +47,7 @@ def download(name_thread, max_thread, url, logger, parser, directory, html, img,
 
 
 
-def importUrl(name_thread, max_thread, canalblog, logger, parser, wordpress, basic, serial, ssl_wordpress, ssl_canalblog):
+def importUrl(name_thread, max_thread, canalblog, logger, parser, wordpress, basic, serial, ssl_wordpress, ssl_canalblog, create, update, image):
     canalblog = canalblog.split(",")
     wordpress = wordpress.split(",")
     name = "Thread-{0}".format(int(name_thread) + 1)
@@ -67,7 +67,7 @@ def importUrl(name_thread, max_thread, canalblog, logger, parser, wordpress, bas
             webpage = exportWp.getUrlPage(name_thread, max_thread)
             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)
                 for k in ["article", "page"]:
                     for l in ["publications", "principal"]:
                         importWp.fromUrl(webpage[l][k])
@@ -88,7 +88,7 @@ def importUrl(name_thread, max_thread, canalblog, logger, parser, wordpress, bas
             exportWp = WPExport(name=name, url=url, logger=logger, parser=parser, ssl_canalblog=ssl_canalblog)
             webpage = 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)
 
             for k in ["article", "page"]:
                 for l in ["publications", "principal"]:
@@ -97,13 +97,13 @@ def importUrl(name_thread, max_thread, canalblog, logger, parser, wordpress, bas
             del importWp
     
 
-def importDirectory(name_thread, max_thread, directory, logger, parser, wordpress, basic, serial, 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
@@ -113,7 +113,7 @@ def importDirectory(name_thread, max_thread, directory, logger, parser, wordpres
             logger.error("{0} : Error : Number directory is differant than wordpress".format(name))
             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
 
@@ -143,6 +143,10 @@ if __name__ == '__main__':
     import_parser.add_argument("--remove-categories", help="Remove all categories", dest="categories", action="store_true")
     import_parser.add_argument("--remove-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")
@@ -226,7 +230,7 @@ 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)
+                    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:
@@ -237,7 +241,7 @@ 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(importUrl, i, int(args.parallel), args.canalblog, logger, args.parser, args.wordpress, basic, args.serial, ssl_wordpress, ssl_canalblog)
+                    ex.submit(importUrl, i, int(args.parallel), args.canalblog, logger, args.parser, args.wordpress, basic, args.serial, ssl_wordpress, ssl_canalblog, args.create, args.update, args.image)
                     for i in range(0, int(args.parallel))
                     ]
 

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):
         self._name = name
         self._basic = basic
         self._wordpress = wordpress
@@ -24,7 +24,9 @@ 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
     
     # Destructor 
     def __del__(self):
@@ -435,6 +437,7 @@ 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 = []
+        if self._no_image is False:
             for i in img_a:
                 new_img = {}
                 img = i.find_all("img")
@@ -615,6 +618,7 @@ class WPimport:
                         self._logger.debug("{0} : Search title posts for |{2}| : |{1}|".format(self._name, title_rendered, title))
                         self._logger.debug("{0} : 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
@@ -667,7 +671,7 @@ 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))