Mais refactorings

edwardoliveira · Sep 1, 2016 · 28bf901 · 28bf901
1 parent 72625a6
commit 28bf901
Show file tree

Hide file tree

Showing 3 changed files with 29 additions and 3 deletions.
diff --git a/src/main/java/br/unb/cassandra/etl/CassandraBulkLoader.java b/src/main/java/br/unb/cassandra/etl/CassandraBulkLoader.java
@@ -2,4 +2,22 @@
 
 public class CassandraBulkLoader {
 
+    public static final String SCHEMA = "CREATE TABLE %s.%s (ID TIMEUUID,\n" +
+                                        "                    UF TEXT,\n" +
+                                        "                    CODIGO_MUNICIPIO TEXT,\n" +
+                                        "                    NOME_MUNICIPIO TEXT,\n" +
+                                        "                    NOME_BENEFICIARIO TEXT,\n" +
+                                        "                    VALOR_PAGO FLOAT,\n" +
+                                        "                    MES_ANO TEXT,\n" +
+                                        "                    PRIMARY KEY (ID, VALOR_PAGO))";
+
+    public static final String INSERT = "INSERT INTO %s.%s (ID, UF, CODIGO_MUNICIPIO, NOME_MUNICIPIO, NOME_BENEFICIARIO, VALOR_PAGO, MES_ANO) " +
+            " VALUES (?, ?, ?, ?, ?, ?, ?)";
+
+
+    public static void main(String[] args) {
+
+    }
+
+
 }
diff --git a/src/main/java/br/unb/cassandra/etl/CassandraClientLoader.java b/src/main/java/br/unb/cassandra/etl/CassandraClientLoader.java
@@ -11,8 +11,8 @@ public class CassandraClientLoader {
     public static final String KEYSPACE = "bolsafamilia";
     public static final String TABLE = "bf";
 
-    private String insertCmd = "INSERT INTO %s.%s (ID, UF, CODIGO_MUNICIPIO, NOME_MUNICIPIO, NOME_BENEFICIARIO, VALOR_PAGO, MES_ANO) " +
-                               " VALUES (?, ?, ?, ?, ?, ?, ?)";
+    public static final String INSERT = "INSERT INTO %s.%s (ID, UF, CODIGO_MUNICIPIO, NOME_MUNICIPIO, NOME_BENEFICIARIO, VALOR_PAGO, MES_ANO) " +
+                                        " VALUES (?, ?, ?, ?, ?, ?, ?)";
 
     public boolean connect() {
         cluster = Cluster.builder().addContactPoint(host).build();
@@ -29,7 +29,7 @@ public void close() {
     // TODO: use BATCH insert to speedup?
     public void insert(Registro registro) {
 
-       PreparedStatement pstmt = session.prepare(String.format(insertCmd, KEYSPACE, TABLE));
+       PreparedStatement pstmt = session.prepare(String.format(INSERT, KEYSPACE, TABLE));
 
        BoundStatement bstmt = pstmt.bind(UUIDs.timeBased(),
                                          registro.getUf(),

diff --git a/src/main/resources/NOTES.txt b/src/main/resources/NOTES.txt
@@ -0,0 +1,8 @@
+wc -l ~/Downloads/201606_BolsaFamiliaFolhaPagamento.csv
+13849867 /home/eribeiro/Downloads/201606_BolsaFamiliaFolhaPagamento.csv
+
+
+* Adicionar indices de baixa cardinalidade para UF
+* Colocar VALOR_PAGO como CLUSTERING KEY da chave primária
+* Usar SASI indexes para NOME_BENEFICIARIO? NOME_MUNICIPIO? CODIGO_MUNICIPIO?
+