importing updates

.gitignore

@@ -8,3 +8,6 @@ logs/*
 #*
 Gemfile.lock
 temp.py
+import.xml
+import.json
+import.csv

bookie.py

@@ -1,5 +1,5 @@
 import os
-from flask import Flask, render_template, redirect, url_for, request
+from flask import Flask, render_template, redirect, url_for, request, Response
 from flask.ext.mongoengine import MongoEngine
 from flask.ext.security import Security, UserMixin, RoleMixin, login_required, MongoEngineUserDatastore
 from flask.ext.login import current_user
@@ -8,6 +8,7 @@ import base64
 import urllib
 from subprocess import call
 from bs4 import BeautifulSoup as BS
+from xml.sax.saxutils import escape
 
 app = Flask(__name__)
 
@@ -314,14 +315,16 @@ def list(count=100, format="HTML"):
             out = out + b.url + ',' + b.title + ',' + b.note + ',' + b.created_at.isoformat() + ',' + tags + ',' + str(b.unread) + ',bookie\n'
         return out
     elif format == "xml":
+        c = 0
         blist = Bookmark.objects(deleted=False).order_by("-created_at").only("url","title","short","note","created_at","tags","unread").limit(count)
         out = "<xml>\n"
         out += "\t<title>Bookmark Export "+datetime.datetime.now().strftime("%Y-%m-%d")+"</title>\n"
         out += "\t<created_at>"+datetime.datetime.now().isoformat()+"</created_at>\n"
         for b in blist:
             out += "\t<bookmark>\n"
-            out += "\t\t<title>"+b.title+"</title>\n"
+            out += "\t\t<index>"+str(c)+"</index>\n"
-            out += "\t\t<short>"+b.short+"</short>\n"
+            out += "\t\t<title>"+escape(b.title)+"</title>\n"
+            out += "\t\t<short>"+escape(b.short)+"</short>\n"
             out += "\t\t<created_at>"+b.created_at.isoformat()+"</created_at>\n"
             out += "\t\t<unread>"
             if b.unread:
@@ -329,17 +332,22 @@ def list(count=100, format="HTML"):
             else:
                 out += "False"
             out += "</unread>\n"
-            out += "\t\t<url>"+b.url+"</url>\n"
+            out += "\t\t<url>"+escape(b.url)+"</url>\n"
             out += "\t\t<tags>"
             for t in b.tags:
-                out += t.name + " "
+                out += escape(t.name) + " "
             out += "</tags>\n"
-            out += "\t\t<note>\n"
+            out += "\t\t<note><![CDATA[\n"
             out += "\t\t\t"+b.note+"\n"
-            out += "\t\t</note>\n"
+            out += "\t\t]]></note>\n"
             out += "\t</bookmark>\n"
+            c += 1
         out += "</xml>\n"
-        return out
+        
+        return Response(out, mimetype='application/xml')
+        
+        
+        
     elif format == "json":
         blist = Bookmark.objects(deleted=False).order_by("-created_at").only("url","title","short","note","created_at","tags","unread").limit(count)
         out = ""

csv_import.py

@@ -10,7 +10,7 @@ from bookie import *
 app.config["MONGODB_DB"] = "bookie"
 db = MongoEngine(app)
 
-file = open('bookmarks.csv')
+file = open('import.csv')
 
 date_marker = datetime.datetime.now()
 count = 0

json_import.py

@@ -14,7 +14,7 @@ db = MongoEngine(app)
 ###
 # File name is hard coded
 ###
-file = open('bookmarks.json')
+file = open('import.json')
 
 ###
 # Kill everything in the db

xml_import.py

@@ -0,0 +1,49 @@
+#!/usr/bin/env python3
+
+import os, re
+from flask.ext.mongoengine import MongoEngine
+import datetime
+import base64
+from bs4 import BeautifulSoup as BS
+from xml.sax.saxutils import unescape
+
+
+from bookie import *
+
+app.config["MONGODB_DB"] = "bookie"
+db = MongoEngine(app)
+
+date_marker = datetime.datetime.now()
+count = 0
+
+Bookmark.objects.all().delete()
+Tag.objects.all().delete()
+ArchivedText.objects.all().delete()
+ArchivedImage.objects.all().delete()
+
+with open("import.xml", "r") as importfile:
+    data = importfile.read()
+
+soup = BS(data)
+count = 0
+for i in soup(["bookmark"]):
+    b = Bookmark()
+    b.title = unescape(i.find("title").get_text())
+    b.short = unescape(i.find("short").get_text())
+    b.url = unescape(i.find("url").get_text())
+    b.note = i.find("note").get_text()
+    b.created_at = datetime.datetime(*map(int, re.split('[^\d]', unescape(i.find("created_at").get_text()))[:-1]))
+    if i.find("unread").get_text() == "True":
+        b.unread = True
+    else:
+        b.unread = False
+    b.tags = []
+    for t in i.find("tags").get_text().split(" "):
+        tag = Tag.objects.get_or_create(name=t)[0].save()
+        b.tags.append(tag)
+    count += 1
+    print(str(count)+" - "+b.short)
+    print(b.title)
+    print(b.url)
+    
+