diff --git a/get_incidents.py b/get_incidents.py
new file mode 100644
index 0000000..f47936e
--- /dev/null
+++ b/get_incidents.py
@@ -0,0 +1,65 @@
+# This file contains the logic that periodically fetches all pages on the
+# Reachout Berlin homepage, checks if they're already in the database and inserts
+# them if needed.
+
+import sqlite3
+import hashlib
+from scraper.scraper import Scraper
+
+encoding = 'utf-8'
+
+scraper = Scraper()
+articles = scraper.scrape()
+
+conn = sqlite3.connect('violence.db')
+c = conn.cursor()
+
+# setup database schema
+c.execute('''
+    CREATE TABLE IF NOT EXISTS incidents (
+        incident_id INTEGER PRIMARY KEY,
+        date TEXT,
+        place TEXT,
+        additional_place TEXT,
+        description TEXT,
+        hash
+    );
+''')
+
+c.execute('''
+    CREATE INDEX IF NOT EXISTS incidents_date
+    ON incidents (date);
+''')
+
+c.execute('''
+    CREATE INDEX IF NOT EXISTS incidents_hash
+    ON incidents (hash);
+''')
+
+# insert articles if necessary
+select_query = 'SELECT * FROM incidents WHERE hash=?'
+insert_query = '''
+    INSERT INTO incidents (
+        date, place, additional_place, description, hash
+    ) VALUES (?)
+'''
+for article in articles:
+    # build a hash so we can more easily find out if we have an article already
+    h = h.sha256()
+    h.update(article.date.encode(encoding))
+    h.update(article.place.encode(encoding))
+    h.update(article.additional_place.encode(encoding))
+    h.update(article.description.encode(encoding))
+    digest = h.digest()
+
+    c.execute(select_query)
+
+    if (not c.fetchone()):
+        article_tuple = (
+            article.date,
+            article.place,
+            article.additional_place,
+            article.description,
+            digest
+        )
+        c.execute(insert_query, article_tuple)
diff --git a/scraper/scraper.py b/scraper/scraper.py
index 72ebf83..2a2da71 100644
--- a/scraper/scraper.py
+++ b/scraper/scraper.py
@@ -1,5 +1,6 @@
 from bs4 import BeautifulSoup
 from urllib import parse, request
+from datetime import date
 
 class Scraper():
 
@@ -27,7 +28,7 @@ class Scraper():
             # headlines are always YYYY-MM-DD Berlin-DISTRICT (+ sometimes additional info)
             headline = table.select('tr:first-child')[0].get_text()
 
-            date = headline[:headline.find(' ')]
+            year, month, day = headline[:headline.find(' ')].strip().split('-')
             places = headline[headline.find(' ') + 1:]
 
             if places.find(' ') == -1:
@@ -35,9 +36,9 @@ class Scraper():
                 additional = None
             else:
                 district = places[:places.find(' ')]
-                additional = places[places.find(' ') + 1:]
+                additional = places[places.find(' ') + 1:].strip()
 
-            text = table.select('tr:nth-of-type(3)')[0].select('td:nth-of-type(2)')[0].get_text()
+            text = table.select('tr')[2].select('td')[1].get_text()
 
             article = {
                 'date': date.strip(),