From 3744c4836bf52149823c05f455e3190bf1af583c Mon Sep 17 00:00:00 2001
From: cedricbonhomme <devnull@localhost>
Date: Mon, 19 Mar 2012 23:07:37 +0100
Subject: Added script to convert the SQLite database to a MongoDB database for
 the new version of pyAggr3g470r.

---
 mongodb.py      | 22 ++++++----------
 sqlite2mongo.py | 82 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 90 insertions(+), 14 deletions(-)
 create mode 100644 sqlite2mongo.py

diff --git a/mongodb.py b/mongodb.py
index 0eb4e002..a37b80c0 100644
--- a/mongodb.py
+++ b/mongodb.py
@@ -27,9 +27,8 @@ class Articles(object):
         """
         Creates a new collection for a new feed.
         """
-        #pymongo.collection.Collection(self.db, new_collection["feed_id"])
         collection = self.db[new_collection["feed_id"]]
-        collection.create_index([("article_link", pymongo.ASCENDING)], {"unique":True, "sparse":True})
+        #collection.create_index([("feed_link", pymongo.ASCENDING)], {"unique":True, "sparse":True})
         collection.insert(new_collection)
 
     def add_articles(self, articles, feed_id):
@@ -37,6 +36,10 @@ class Articles(object):
         Add article(s) in a collection.
         """
         collection = self.db[str(feed_id)]
+
+        collection.create_index([("article_link", pymongo.ASCENDING), ("article_date", pymongo.DESCENDING)], \
+                                        {"unique":False, "sparse":False})
+
         for article in articles:
             cursor = collection.find({"article_id":article["article_id"]})
             if cursor.count() == 0:
@@ -102,7 +105,8 @@ class Articles(object):
             cursor = collection.find({"type":1})
         else:
             cursor = collection.find({"type":1, condition[0]:condition[1]})
-        return cursor.sort([("article_date", pymongo.DESCENDING)])
+        #return cursor.sort([("article_date", pymongo.DESCENDING)])
+        return cursor
 
     def print_articles_from_collection(self, collection_id):
         """
@@ -245,16 +249,6 @@ if __name__ == "__main__":
     #print articles.get_all_articles()
 
 
-
-
-    for feed in articles.get_all_collections():
-        for article in articles.get_articles_from_collection(feed["feed_id"]):
-            try:
-                #print article["article_title"], article["article_date"]
-                pass
-            except:
-                pass
-
     
     # Drop the database
-    #articles.drop_database()
\ No newline at end of file
+    articles.drop_database()
\ No newline at end of file
diff --git a/sqlite2mongo.py b/sqlite2mongo.py
new file mode 100644
index 00000000..b129ebce
--- /dev/null
+++ b/sqlite2mongo.py
@@ -0,0 +1,82 @@
+#! /usr/bin/env python
+# -*- coding: utf-8 -*-
+
+import hashlib
+
+import sqlite3
+import mongodb
+
+SQLITE_BASE = "./var/feed.db"
+
+
+def load_feed():
+    """
+    Load feeds and articles in a dictionary.
+    """
+    mongo = mongodb.Articles()
+    list_of_feeds = []
+    list_of_articles = []
+
+    try:
+        conn = sqlite3.connect(SQLITE_BASE, isolation_level = None)
+        c = conn.cursor()
+        list_of_feeds = c.execute("SELECT * FROM feeds").fetchall()
+    except:
+        pass
+
+
+    if list_of_feeds != []:
+        # Walk through the list of feeds
+        for feed in list_of_feeds:
+            try:
+                list_of_articles = c.execute(\
+                        "SELECT * FROM articles WHERE feed_link='" + \
+                        feed[2] + "'").fetchall()
+            except:
+                continue
+            sha1_hash = hashlib.sha1()
+            sha1_hash.update(feed[2].encode('utf-8'))
+            feed_id = sha1_hash.hexdigest()
+
+
+            new_collection = {"feed_id" : feed_id.encode('utf-8'), \
+                                "type": 0, \
+                                "feed_image" : feed[3].encode('utf-8'), \
+                                "feed_title" : feed[0].encode('utf-8'), \
+                                "feed_link" : feed[2].encode('utf-8'), \
+                                "site_link" : feed[1].encode('utf-8'), \
+                                "mail" : feed[4]=="1"}
+            
+
+            mongo.add_collection(new_collection)
+
+            
+            if list_of_articles != []:
+                # Walk through the list of articles for the current feed.
+                articles = []
+                for article in list_of_articles:
+                    sha1_hash = hashlib.sha1()
+                    sha1_hash.update(article[2].encode('utf-8'))
+                    article_id = sha1_hash.hexdigest()
+
+
+                    article = {"article_id": article_id.encode('utf-8'), \
+                                "type":1, \
+                                "article_date": article[0].encode('utf-8'), \
+                                "article_link": article[2].encode('utf-8'), \
+                                "article_title": article[1].encode('utf-8'), \
+                                "article_content": article[3].encode('utf-8'), \
+                                "article_readed": article[4]=="1", \
+                                "article_like": article[6]=="1" \
+                                }
+
+                    articles.append(article)
+
+
+                mongo.add_articles(articles, feed_id)
+
+        c.close()
+
+
+if __name__ == "__main__":
+    load_feed()
\ No newline at end of file
-- 
cgit