insert_wordpress.py

@@ -68,12 +68,27 @@ if __name__ == '__main__':
       
 
         title = articletitle[0].text
-        body = articlebody[0]
+        body = articlebody[0].find_all("p")
+        bodyhtml = "<p>"
+        for i in body:
+            if len(i.text) == 1:
+                bodyhtml = bodyhtml + "<br />"
+            else:
+                bodyhtml = bodyhtml + str(i).replace("<p>", "").replace("</p>", "").replace("<br>", "<br />") + "<br />"
+        bodyhtml = bodyhtml + "</p>"
+        print(bodyhtml)
         hour = articledate[0].text
         time = dateheader[0].text.split(" ")
-        data = {"title":title, "content":body, "status":"publish", "date": "{0}-{1}-{2}T{3}:00".format(time[2],month[time[1]],time[0], hour), "tags": listelement["tags"], "categories": listelement["categories"]}
+        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"]}
+    params = {"search":title}
+    page = requests.get("http://localhost:8080/wp-json/wp/v2/posts", auth=basic, params=params)
+    page_exist = True
+    if page.status_code:
+        result= page.json()
+        if len(result) == 0:
+            page_exist = False
 
+    if page_exist == False:
         page = requests.post("http://localhost:8080/wp-json/wp/v2/posts", auth=basic, data=data)
-    print(page.status_code)
         if page.status_code == 201:
             print(page.content)