New Amazon scripts

create_core.sh

@@ -1,8 +1,14 @@
 #Simple script to create solr cores and reload/index with the right XML's
+if [ $# -eq 0 ]
+  then
+    echo "No arguments supplied"
+    echo "./create_core.sh <name of your keyspace>"
+    exit 1
+fi
 
 #add the path to dsetool if dsetool is not in your path
 echo 'Creating Solr cores...'
-dsetool create_core amazon.metadata schema=schema_metadata.xml solrconfig=solr_config_metadata.xml reindex=true
-dsetool create_core amazon.rank schema=schema_rank.xml solrconfig=solr_config_rank.xml reindex=true
-dsetool create_core amazon.clicks schema=schema_clicks.xml solrconfig=solr_config_clicks.xml reindex=true
+dsetool create_core $1.metadata schema=schema_metadata.xml solrconfig=solr_config_metadata.xml reindex=true
+dsetool create_core $1.rank schema=schema_rank.xml solrconfig=solr_config_rank.xml reindex=true
+dsetool create_core $1.clicks schema=schema_clicks.xml solrconfig=solr_config_clicks.xml reindex=true
 echo 'finished creating Solr cores!'

create_data.sh

@@ -0,0 +1,8 @@
+if [ $# -eq 0 ]
+  then
+    echo "No arguments supplied"
+    echo "./create_database <name of your keyspace>"
+    exit 1
+fi
+python ./solr_dataloader.py $1
+

solr_dataloader.py

@@ -15,7 +15,7 @@
 
 from cassandra.cluster import Cluster
 
-KEYSPACE = "amazon"
+KEYSPACE = sys.argv[1]
 META_COLUMN_FAMILY = "metadata"
 RANK_COLUMN_FAMILY = "rank"
 GEO_COLUMN_FAMILY = "clicks"
@@ -194,6 +194,7 @@ def load_geo_data(session, geopath):
 
     meta_prepared = session.prepare(META_INSERT_STATEMENT)
     rank_prepared = session.prepare(RANK_INSERT_STATEMENT)
+    print ('Loading into Keyspace',sys.argv[1])
     print ('loading geo')
     load_geo_data(session, geo_path)
     print ('loading meta')