Merge pull request 'parent-comment' (#8) from parent-comment into master

Reviewed-on: #8

lib/WPExport.py

@@ -69,6 +69,10 @@ class WPExport:
                 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 = []
         for i in page_url:
@@ -113,6 +117,10 @@ class WPExport:
                                             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
 
 
@@ -179,6 +187,10 @@ class WPExport:
                                 o = o._replace(netloc=u.netloc)
                                 o = o._replace(scheme=u.scheme)
                             page_url.append(o.geturl())        
+        else:
+            self._logger.error("JS or CSS did not get due status code : {0}".format(page.status_code))
+            self._logger.debug(page.content)
+
         return page_url
 
     # Get image
@@ -201,6 +213,10 @@ class WPExport:
                         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
 
 
@@ -232,4 +248,7 @@ class WPExport:
                     open(fileDownload, "wb").write(r.content)
                 except Exception as err:
                     self._logger.error("file error : {0}".format(err))
-                    exit(1)
+                    exit(1)
+            else:
+                self._logger.error("Not download due status code : {0}".format(r.status_code))
+                self._logger.debug(r.content)

lib/WPImport.py

@@ -13,6 +13,7 @@ class WPimport:
         self._wordpress = wordpress
         self._logger = logger
         self._parser = parser
+        self._headers_json = {'Content-Type': 'application/json', 'Accept':'application/json'}
 
         self._request = requests.Session()
 
@@ -41,6 +42,9 @@ class WPimport:
                     self._addOrUpdatePost(soup)
                 else:
                     self._addOrUpdateFeaturedMedia(soup)
+            else:
+                self._logger.error("Connection error with status code : {0}".format(r.status_code))
+                self._logger.debug(r.content)
 
 
     def fromDirectory(self, directory):
@@ -122,18 +126,33 @@ class WPimport:
                                     res = page.json()
                                     if len(res) > 0:
                                         id_media = res[0]["id"]
-                                        headers = {'Content-Type': 'application/json', 'Accept':'application/json'}
                                         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=headers, data=json.dumps(data))
+                                            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))
                                         except Exception as err:
                                             self._logger.error("Connection error : {0}".format(err))
                                             exit(1)
                                         if r.status_code == 200:
                                             self._logger.info("Ajout media featured : {0}".format(r.json()["title"]["raw"]))
+                                        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))
-                
+                                else:
+                                    self._logger.error("Connection error with status code : {0}".format(page.status_code))
+                                    self._logger.debug(page.content)
+
+                            else:
+                                self._logger.error("Connection error with status code : {0}".format(page.status_code))
+                                self._logger.debug(page.content)
+
+            else:
+                self._logger.error("Connection error with status code : {0}".format(page.status_code))
+                self._logger.debug(page.content)
+
+
     ## Association image to post
 
     def _linkImgPost(self, title, list_img, post_id):
@@ -146,6 +165,10 @@ class WPimport:
                 exit(1)
             if r.status_code == 200:
                 self._logger.info("Association d'une image à l'article {0}".format(title))
+            else:
+                self._logger.error("Connection error with status code : {0}".format(r.status_code))
+                self._logger.debug(r.content)
+
     
     ## Add or update img 
 
@@ -169,7 +192,11 @@ class WPimport:
                     self._logger.error("Connection error : {0}".format(err))
                     exit(1)
                 if r.status_code == 200:
-                    self._logger.info("Image supprimé {0}".format(img_name))
+                    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":
@@ -181,48 +208,129 @@ class WPimport:
                 self._logger.error("Connection error : {0}".format(err))
                 exit(1)
             if r.status_code == 201:
-                self._logger.info("Ajout d'image {0}".format(img_name))
+                self._logger.info("Image added {0}".format(img_name))
                 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
 
     def _addOrUpdateComment(self, post, comment, title):
-        params = {"post": post}
-        block = True
-        try:
-            page = self._request.get("http://{0}/wp-json/wp/v2/comments".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()
             for i in comment:
-                comment_exist = False
+
-                for j in result:
+                try:
-                    if i["author"] == j["author_name"] and i["date"] == j["date"]:
+                    params = {"post": post, "author_name":i["author"], "date":i["date"]}
-                        comment_exist = True
+                    page = self._request.get("http://{0}/wp-json/wp/v2/comments".format(self._wordpress), auth=self._basic, params=params)
-                        id_comment = j["id"]
+                except Exception as err:
-                data = {"post": post, "content": i["content"], "date": i["date"], "author_name": i["author"]}
+                    self._logger.error("Connection error : {0}".format(err))
-                if comment_exist is True:
+                    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:
+                    self._logger.error("Comment not listed for {0} due status code : {1}".format(title, page.status_code))
+                    self._logger.debug(page.content)
+   
+            for i in comment:           
+                data = {"post": post, "content": i["content"], "date": i["date"], "author_name": i["author"], "status": "approved"}
+    
+                if i["parent_id"] != -1:
+                    parent_id = int(i["parent_id"])
+                    params = {"post": post, "author_name":comment[parent_id]["author"], "date":comment[parent_id]["date"]}
                     try:
-                        page = page = self._request.post("http://{0}/wp-json/wp/v2/comments/{1}".format(self._wordpress, id_comment), auth=self._basic, data=data)
+                        page = self._request.get("http://{0}/wp-json/wp/v2/comments".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:
-                        self._logger.info("Commentaire mise à jour pour {0}".format(title))
+                        result = page.json()
+                        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)
+                except Exception as err:
+                    self._logger.error("Connection error : {0}".format(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:
-                    try:
+                    self._logger.error("Comment not added for {0} due status code : {1}".format(title, page.status_code))
-                        page = self._request.post("http://{0}/wp-json/wp/v2/comments".format(self._wordpress), auth=self._basic, data=data)
+                    self._logger.debug(page.content)
-                    except Exception as err:
-                        self._logger.error("Connection error : {0}".format(err))
-                        exit(1)
-                    if page.status_code == 201:
-                        self._logger.info("Commentaire ajoute pour {0}".format(title))
     
+    ## Check class name 
+
+    def _hasClassName(self, tag, className):
+        for i in tag["class"]:
+            if i == className:
+                return True
+        return False
+    
+    ## Get class name 
+
+    def _getClassName(self, tag, className):
+        for i in tag["class"]:
+            if re.match(className, i):
+                return i
+        return ""
+    
+    ## Get all comments
+
+    def _getComment(self, comment):
+        comment_post = []
+        for i in range(0, len(comment)):
+            comment_div = comment[i].find("div", class_="comment_item")
+            comment_item = comment_div.text.split("\n")
+            footer = comment_div.find_all("div", class_="itemfooter")
+            comment_author = footer[0].text.split(",")[0].replace("Posté par ", "")
+            comment_date = footer[0].find_all("abbr")[0].get("title")
+            comment_content = "<p>"
+            for j in range(0, len(comment_item)-2):
+                if len(comment_item[j]) > 0:
+                    comment_content = comment_content + comment_item[j] + "<br />"
+            comment_content = comment_content + "</p>"
+            parent = -1
+            if self._hasClassName(comment[i], "level-1") is False:
+                block = False
+                className = self._getClassName(comment[i], "level-").split("-")
+                level = 1
+                if len(className) > 0:
+                    level = int(className[1])
+                for j in range(i-1, 0, -1):
+                    if block is False:
+                        levelName = "level-{0}".format(level - 1)
+                        if self._hasClassName(comment[j], levelName) is True:
+                            parent = j
+                            block = True
+
+            comment_post.append({"author": comment_author, "date": comment_date, "content": comment_content, "parent_id":parent})
+        return comment_post
     ## Add or Update post 
 
     def _addOrUpdatePost(self, soup):
@@ -234,15 +342,8 @@ class WPimport:
         listelement = {}
 
         for i in liste:
-            try:
+            element[i] = []
-                page = self._request.get("http://{0}/wp-json/wp/v2/{1}".format(self._wordpress,i))
+            listelement[i] = []
-            except Exception as err:
-                self._logger.error("Connection error : {0}".format(err))
-                exit(1)
-            if page.status_code == 200:
-                elements[i] = page.json()
-                element[i] = []
-                listelement[i] = []
 
         articletitle = soup.find_all("h2", class_="articletitle")
         articlebody = soup.find_all("div", class_="articlebody")
@@ -250,7 +351,7 @@ class WPimport:
         articleacreator = soup.find_all("span", class_="articlecreator")
         dateheader = soup.find_all("div", class_="dateheader")
         itemfooter = soup.find_all("div", class_="itemfooter")
-        comment = soup.find_all("div", class_="comment_item")
+        comment = soup.find_all("li", class_="comment")
         img_a = articlebody[0].find_all("a", {"target": "_blank"})
         list_img = []
         for i in img_a:
@@ -273,6 +374,7 @@ class WPimport:
                     except Exception as err:
                         self._logger.error("Connection error : {0}".format(err))
                         exit(1)
+                
                 if page_img.status_code == 200:
                     media=self._addOrUpdateMedia(href_img, page_img)
                     new_img["id"]=media["id"]
@@ -283,19 +385,13 @@ class WPimport:
                         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 = []
+        comment_post = self._getComment(comment)
-        for i in comment:
+
-            comment_item = i.text.split("\n")
-            footer = i.find_all("div", class_="itemfooter")
-            comment_author = footer[0].text.split(",")[0].replace("Posté par ", "")
-            comment_date = footer[0].find_all("abbr")[0].get("title")
-            comment_content = "<p>"
-            for j in range(0, len(comment_item)-2):
-                if len(comment_item[j]) > 0:
-                    comment_content = comment_content + comment_item[j] + "<br />"
-            comment_content = comment_content + "</p>"
-            comment_post.append({"author": comment_author, "date": comment_date, "content": comment_content})
         a = itemfooter[0].find_all("a", {"rel": True})
         for i in a:
             rel = i.get("rel")
@@ -308,20 +404,38 @@ class WPimport:
         for i in liste:
             for j in element[i]:
                 element_exist = False
-                for k in elements[i]:
+                try:
-                    if k["name"] == j:
+                    params = {"params":j}
-                        element_exist = True
+                    page = self._request.get("http://{0}/wp-json/wp/v2/{1}".format(self._wordpress, i), auth=self._basic, params=params)
-                        listelement[i].append(k["id"])
+                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, data=data)
+                        page = self._request.post("http://{0}/wp-json/wp/v2/{1}".format(self._wordpress, i), auth=self._basic, headers=self._headers_json, data=data)
                     except Exception as err:
                         self._logger.error("Connection error : {0}".format(err))
                         exit(1)
                     if page.status_code == 201:
                         result = page.json()
                         listelement[i].append(result["id"])
+                    else:
+                        self._logger.error("{0} not added due status code : {1}".format(i, page.status_code))
+                        self._logger.debug(page.content)
 
         title = articletitle[0].text
         author = articleacreator[0].text.lower()        
@@ -349,6 +463,10 @@ class WPimport:
         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":title}
         try:
@@ -375,6 +493,14 @@ class WPimport:
                     self._logger.info("Article mis à jour : {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 updated due status code : {0}".format(page.status_code))
+                    self._logger.debug(page.content)
+
+        else:
+            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:
@@ -386,4 +512,7 @@ class WPimport:
                 result = page.json()
                 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"])
+                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)