add method getAuthor

import_export_canalblog.py

@@ -67,7 +67,7 @@ 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):
+def importUrl(name_thread, max_thread, canalblog, logger, parser, wordpress, basic, serial, ssl_wordpress, ssl_canalblog, create, update, image, revert, tmp, author):
     canalblog = canalblog.split(",")
     wordpress = wordpress.split(",")
     name = "Thread-{0}".format(int(name_thread) + 1)
@@ -88,7 +88,7 @@ def importUrl(name_thread, max_thread, canalblog, logger, parser, wordpress, bas
                 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)
+                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, author=author)
                 for k in ["article", "page"]:
                     for l in ["publications", "principal"]:
                         importWp.fromUrl(l, k)
@@ -110,7 +110,7 @@ def importUrl(name_thread, max_thread, canalblog, logger, parser, wordpress, bas
             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)
+            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, author=author)
 
             for k in ["article", "page"]:
                 for l in ["publications", "principal"]:
@@ -119,13 +119,13 @@ 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, create, update, image, revert):
+def importDirectory(name_thread, max_thread, directory, logger, parser, wordpress, basic, serial, ssl_wordpress, create, update, image, revert, author):
     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)
+            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, author=author)
             for j in directory:
                 importWp.fromDirectory(j, name_thread, max_thread, revert)
                 del importWp
@@ -135,7 +135,7 @@ def importDirectory(name_thread, max_thread, directory, logger, parser, wordpres
             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 = 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, author=author)
             importWp.fromDirectory(directory[i], name_thread, max_thread, revert)
             del importWp
 
@@ -170,6 +170,7 @@ if __name__ == '__main__':
     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")
+    import_parser.add_argument("--author", help="Define author for page album", default="")
 
 
 
@@ -258,7 +259,7 @@ if __name__ == '__main__':
         basic = HTTPBasicAuth(args.user, password)
     if args.command == "import":
         wordpress = args.wordpress.split(",")
-        importWp = WPimport(basic=basic, wordpress="", logger=logger, parser=args.parser, ssl_wordpress=ssl_wordpress)
+        importWp = WPimport(basic=basic, wordpress="", logger=logger, parser=args.parser, ssl_wordpress=ssl_wordpress, author=args.author)
         if len(args.file) > 0:
             for i in wordpress:
                 importWp.setUrl(i)
@@ -273,7 +274,7 @@ if __name__ == '__main__':
                     wait(wait_for, return_when=ALL_COMPLETED)
                     errorRevert(logger, args.revert, args.tmp)
                     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, args.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, args.revert, args.author)
                     for i in range(0, int(args.parallel))
                     ]
                     wait(wait_for, return_when=ALL_COMPLETED)
@@ -288,7 +289,7 @@ if __name__ == '__main__':
                     wait(wait_for, return_when=ALL_COMPLETED)
                     errorRevert(logger, args.revert, args.tmp)
                     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, args.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, args.author)
                     for i in range(0, int(args.parallel))
                     ]
                     wait(wait_for, return_when=ALL_COMPLETED)

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"):
+    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", author=""):
         self._name = name
         self._basic = basic
         self._wordpress = wordpress
@@ -29,6 +29,7 @@ class WPimport:
         self._no_update = no_update
         self._no_image = no_image
         self._tmp = tmp
+        self._author = author
     
     # Destructor 
     def __del__(self):
@@ -114,6 +115,30 @@ class WPimport:
     
     # Private method
 
+
+    def _getAuthor(self, author):
+        params = {"search":author, "per_page":100}
+        author = 0
+        try:
+            self._logger.info("{0} : Search author : {1}".format(self._name, author))
+            page = self._request.get("{1}://{0}/wp-json/wp/v2/users".format(self._wordpress, self._protocol), auth=self._basic, headers=self._headers_json, params=params)
+            self._logger.debug("{0} : End Search author : {1}".format(self._name, author))
+            self._logger.debug("{0} : Debug requests : {1}".format(self._name, page.content))
+            if page.status_code == 200:
+                self._logger.info("{0} : Get author  : {1}".format(self._name, author))
+                result = page.json()
+                for a in result:
+                    author = a["id"]
+            else:
+                self._logger.error("{0} : Connection error with status code for get author : {1}".format(self._name, page.status_code))
+                self._logger.debug("{0} : {1}".format(page.content))
+        except ConnectionError as err:
+            self._logger.error("{0} : Connection error for get author : {1}".format(self._name, err))
+            exit(1)
+        except Exception as err:
+            self._logger.error("{0} : Exception error for get author : {1}".format(self._name, err))
+        return author
+
     def _addOrUpdateAlbum(self, soup):
         self._logger.info("{0} : Add/Update Album".format(self._name))
         albumbody = soup.find("div", class_="albumbody")
@@ -131,10 +156,10 @@ class WPimport:
                 new_img = {}
                 href_img = i.get("src")
                 href_img_o = urlparse(href_img)
-                new_img["old_src"]=href_img
                 try:
                     if len(href_img_o.netloc) > 0:
                         img_ok = False
+                        href_img = href_img.replace("_q", "_o")
                         page_img = self._request.get(href_img)
                         if page_img.status_code == 200:
                             img_ok = True
@@ -158,19 +183,21 @@ class WPimport:
             content_html = ""
             if len(list_img) > 0:
 
-                content_html = "<!-- wp:paragraph --><p>{0}</p><!-- /wp:paragraph -->\n\n".format(albumdesc)
+                content_html = "<!-- wp:paragraph -->\n{0}\n<!-- /wp:paragraph -->\n\n".format(albumdesc)
 
 
-                content_html = content_html + "<figure class='wp-block-gallery has-nested-images columns-default is-cropped'><!-- wp:image {\"id\":id-image,'sizeSlug':'large','linkDestination':'none'} --><figure class='wp-block-image size-large'><img src='src-image' alt='' class=wp-image-id-image/></figure><!-- /wp:image --> \n\n".replace("id-image", str(list_img[0]["id"])).replace("src-image", list_img[0]["new_src"])
 
-                for i in range(1, len(list_img)):
-                    content_html = content_html + "<!-- wp:image {\"id\":id-image,\"sizeSlug\":\"large\",\"linkDestination\":\"none\"} --><figure class=\"wp-block-image size-large\"><img src='src-image' alt=\"\" class='wp-image-id-image'/></figure><!-- /wp:image -->\n\n".replace("id-image", str(list_img[0]["id"])).replace("src-image", list_img[0]["new_src"])
-                content_html = content_html + "<!-- /wp:gallery -->"
+                for i in range(0, len(list_img)):
+                    content_html = content_html + "<!-- wp:image {\"id\":id-image,\"sizeSlug\":\"large\",\"linkDestination\":\"none\"} --><figure class=\"wp-block-image size-large\"><img src=\"src-image\" alt=\"\" class=\"wp-image-id-image\"/></figure><!-- /wp:image -->\n\n".replace("id-image", str(list_img[i]["id"])).replace("src-image", list_img[i]["new_src"])
 
             self._logger.info("{0} : content html : {1}".format(self._name, content_html))
             if len(content_html) > 0:
                 data = {"title":albumtitle, "content":content_html, "status":"publish"}
-
+                if len(self._author) > 0:
+                    author = self._getAuthor(self._author)
+                    if author != 0:
+                        data = {"title":albumtitle, "content":content_html, "status":"publish", "author":author}
+                self._logger.debug("{0} : data for album page : {1}".format(self._name, data))
                 for index in range(1,10):
                     params = {"search": albumtitle, "per_page":100, "page": index}
                     try:
@@ -801,25 +828,7 @@ class WPimport:
         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"]}
         self._logger.debug("{0} : Data for post : |{1}| : {2}" .format(self._name, title, data))
 
-        params = {"search":author, "per_page":100}
-        try:
-            self._logger.info("{0} : Search author : {1}".format(self._name, author))
-            page = self._request.get("{1}://{0}/wp-json/wp/v2/users".format(self._wordpress, self._protocol), auth=self._basic, headers=self._headers_json, params=params)
-            self._logger.debug("{0} : End Search author : {1}".format(self._name, author))
-            self._logger.debug("{0} : Debug requests : {1}".format(self._name, page.content))
-            if page.status_code == 200:
-                self._logger.info("{0} : Get author id : {1}".format(self._name, result))
-                result = page.json()
-                for a in result:
-                    data["author"] = a["id"]
-            else:
-                self._logger.error("{0} : Connection error with status code for get author : {1}".format(self._name, page.status_code))
-                self._logger.debug("{0} : {1}".format(page.content))
-        except ConnectionError as err:
-            self._logger.error("{0} : Connection error for get author : {1}".format(self._name, err))
-            exit(1)
-        except Exception as err:
-            self._logger.error("{0} : Exception error for get author : {1}".format(self._name, err))
+        data["author"] = self._getAuthor(author)
         page_is_exist = False
 
         for index in range(1,10):