fix requirements, app_config

geocoder/app_config.py.example

@@ -6,11 +6,14 @@ DB_USER = 'datamade'
 DB_PW = ''
 DB_HOST = 'localhost'
 DB_PORT = '5432'
-DB_NAME = 'sunshine'
+DB_NAME = 'geocoder'
 
 DB_CONN='postgresql+psycopg2://{0}:{1}@{2}:{3}/{4}'\
         .format(DB_USER, DB_PW, DB_HOST, DB_PORT, DB_NAME)
 
+# if no password, the conn_string should be changed to just
+# 'postgresql+psychopg2://{0}'.format(DB_NAME)
+
 SECRET_KEY = 'super secret key'
 
 # See: https://pythonhosted.org/Flask-Cache/#configuring-flask-cache

linkNewData.py

@@ -1,6 +1,6 @@
 import os
 import sqlalchemy as sa
-
+from geocoder.app_config import DB_CONN
 
 def checkForTable(engine, table_name):
     try:
@@ -22,7 +22,7 @@ def trainIncoming(name):
     import simplejson as json
     import dedupe
 
-    engine = create_engine('postgresql://localhost:5432/geocoder')
+    engine = create_engine(DB_CONN)
 
     deduper = DatabaseGazetteer([{'field': 'complete_address', 'type': 'Address'}],
                                 engine=engine)
@@ -71,7 +71,7 @@ def trainIncoming(name):
 def blockIncoming(name, train):
     from geocoder.deduper import StaticDatabaseGazetteer
 
-    engine = create_engine('postgresql://localhost:5432/geocoder')
+    engine = create_engine(DB_CONN)
 
     with open('geocoder/data/dedupe.settings', 'rb') as sf:
         deduper = StaticDatabaseGazetteer(sf, engine=engine)
@@ -170,7 +170,7 @@ def blockIncoming(name, train):
     if args.link:
         from geocoder.deduper import AddressLinkGazetteer
 
-        engine = create_engine('postgresql://localhost:5432/geocoder')
+        engine = create_engine(DB_CONN)
 
         sql_table = checkForTable(engine, args.name)
 

loadAddresses.py

@@ -5,6 +5,7 @@
 import os
 import sqlalchemy as sa
 from geocoder.data_loader import ETLThing
+from geocoder.app_config import DB_CONN
 
 class CookCountyETL(ETLThing):
 
@@ -194,7 +195,7 @@ class SuburbsETL(CookCountyETL):
     cook_county_data_portal = 'https://datacatalog.cookcountyil.gov/api/geospatial/%s?method=export&format=Original'
 
     if args.load_data:
-        engine = create_engine('postgresql://localhost:5432/geocoder')
+        engine = create_engine(DB_CONN)
         connection = engine.connect()
 
         chicago = ChicagoETL(connection, 'chicago_addresses')
@@ -223,7 +224,8 @@ class SuburbsETL(CookCountyETL):
         import simplejson as json
         import dedupe
 
-        engine = create_engine('postgresql://localhost:5432/geocoder')
+        from geocoder.app_config import DB_CONN
+        engine = create_engine(DB_CONN) 
 
         deduper = DatabaseGazetteer([{'field': 'complete_address', 'type': 'Address'}],
                                     engine=engine)
@@ -255,7 +257,7 @@ class SuburbsETL(CookCountyETL):
     if args.block:
         from geocoder.deduper import StaticDatabaseGazetteer
 
-        engine = create_engine('postgresql://localhost:5432/geocoder')
+        engine = create_engine(DB_CONN)
 
         with open('geocoder/data/dedupe.settings', 'rb') as sf:
             deduper = StaticDatabaseGazetteer(sf, engine=engine)

requirements.txt

@@ -1,7 +1,9 @@
-numpy==1.9.2
-dedupe==0.8.0.1.7
-dedupe-variable-address==0.0.4
+dedupe==1.4.15
+dedupe-variable-address==0.0.7
 Flask==0.10.1
 psycopg2==2.6.1
 csvkit==0.9.1
 requests==2.7.0
+simpledbf==0.2.6
+pytz==2016.6.1
+numpy==1.11.1