Add support for BYTES data type

Author: mziccard

gcloud-java-bigquery/src/main/java/com/google/cloud/bigquery/Field.java

@@ -73,7 +73,7 @@ public static class Type implements Serializable {
     private static final long serialVersionUID = 2841484762609576959L;
 
     public enum Value {
-      STRING, INTEGER, FLOAT, BOOLEAN, TIMESTAMP, RECORD
+      BYTES, STRING, INTEGER, FLOAT, BOOLEAN, TIMESTAMP, RECORD
     }
 
     private final Value value;
@@ -108,6 +108,13 @@ public List<Field> fields() {
       return fields;
     }
 
+    /**
+     * Returns a {@link Value#BYTES} field value.
+     */
+    public static Type bytes() {
+      return new Type(Value.BYTES);
+    }
+
     /**
      * Returns a {@link Value#STRING} field value.
      */

gcloud-java-bigquery/src/main/java/com/google/cloud/bigquery/FieldValue.java

@@ -23,6 +23,7 @@
 import com.google.common.base.Function;
 import com.google.common.base.MoreObjects;
 import com.google.common.collect.Lists;
+import com.google.common.io.BaseEncoding;
 
 import java.io.Serializable;
 import java.util.List;
@@ -54,7 +55,7 @@ public FieldValue apply(Object pb) {
   public enum Attribute {
     /**
      * A primitive field value. A {@code FieldValue} is primitive when the corresponding field has
-     * type {@link Field.Type#bool()}, {@link Field.Type#string()},
+     * type {@link Field.Type#bytes()}, {@link Field.Type#bool()}, {@link Field.Type#string()},
      * {@link Field.Type#floatingPoint()}, {@link Field.Type#integer()},
      * {@link Field.Type#timestamp()} or the value is set to {@code null}.
      */
@@ -80,7 +81,7 @@ public enum Attribute {
    * Returns the attribute of this Field Value.
    *
    * @return {@link Attribute#PRIMITIVE} if the field is a primitive type
-   *     ({@link Field.Type#bool()}, {@link Field.Type#string()},
+   *     ({@link Field.Type#bytes()}, {@link Field.Type#bool()}, {@link Field.Type#string()},
    *     {@link Field.Type#floatingPoint()}, {@link Field.Type#integer()},
    *     {@link Field.Type#timestamp()}) or is {@code null}. Returns {@link Attribute#REPEATED} if
    *     the corresponding field has ({@link Field.Mode#REPEATED}) mode. Returns
@@ -108,8 +109,8 @@ public Object value() {
 
   /**
    * Returns this field's value as a {@link String}. This method should only be used if the
-   * corresponding field has primitive type ({@link Field.Type#bool()}, {@link Field.Type#string()},
-   * {@link Field.Type#floatingPoint()}, {@link Field.Type#integer()},
+   * corresponding field has primitive type ({@link Field.Type#bytes()}, {@link Field.Type#bool()},
+   * {@link Field.Type#string()}, {@link Field.Type#floatingPoint()}, {@link Field.Type#integer()},
    * {@link Field.Type#timestamp()}).
    *
    * @throws ClassCastException if the field is not a primitive type
@@ -121,6 +122,22 @@ public String stringValue() {
     return (String) value;
   }
 
+  /**
+   * Returns this field's value as a byte array. This method should only be used if the
+   * corresponding field has primitive type ({@link Field.Type#bytes()}.
+   *
+   * @throws ClassCastException if the field is not a primitive type
+   * @throws NullPointerException if {@link #isNull()} returns {@code true}
+   * @throws IllegalStateException if the field value is not encoded in base64
+   */
+  public byte[] bytesValue() {
+    try {
+      return BaseEncoding.base64().decode(stringValue());
+    } catch (IllegalArgumentException ex) {
+      throw new IllegalStateException(ex);
+    }
+  }
+
   /**
    * Returns this field's value as a {@code long}. This method should only be used if the
    * corresponding field has {@link Field.Type#integer()} type.

gcloud-java-bigquery/src/main/java/com/google/cloud/bigquery/InsertAllRequest.java

@@ -49,7 +49,9 @@ public final class InsertAllRequest implements Serializable {
 
   /**
    * A Google Big Query row to be inserted into a table. Each {@code RowToInsert} has an associated
-   * id used by BigQuery to detect duplicate insertion requests on a best-effort basis.
+   * id used by BigQuery to detect duplicate insertion requests on a best-effort basis. Please
+   * notice that data for fields of type {@link Field.Type#bytes()} must be provided as a base64
+   * encoded string.
    *
    * <p>Example usage of creating a row to insert:
    * <pre> {@code
@@ -58,8 +60,9 @@ public final class InsertAllRequest implements Serializable {
    * recordContent.put("subfieldName1", "value");
    * recordContent.put("subfieldName2", repeatedFieldValue);
    * Map<String, Object> rowContent = new HashMap<String, Object>();
-   * rowContent.put("fieldName1", true);
-   * rowContent.put("fieldName2", recordContent);
+   * rowContent.put("booleanFieldName", true);
+   * rowContent.put("bytesFieldName", "DQ4KDQ==");
+   * rowContent.put("recordFieldName", recordContent);
    * RowToInsert row = new RowToInsert("rowId", rowContent);
    * }</pre>
    *
@@ -116,7 +119,8 @@ public boolean equals(Object obj) {
     }
 
     /**
-     * Creates a row to be inserted with associated id.
+     * Creates a row to be inserted with associated id. Please notice that data for fields of type
+     * {@link Field.Type#bytes()} must be provided as a base64 encoded string.
      *
      * @param id id of the row, used to identify duplicates
      * @param content the actual content of the row
@@ -126,7 +130,8 @@ public static RowToInsert of(String id, Map<String, Object> content) {
     }
 
     /**
-     * Creates a row to be inserted without associated id.
+     * Creates a row to be inserted without associated id. Please notice that data for fields of
+     * type {@link Field.Type#bytes()} must be provided as a base64 encoded string.
      *
      * @param content the actual content of the row
      */
@@ -174,7 +179,8 @@ public Builder addRow(RowToInsert rowToInsert) {
     }
 
     /**
-     * Adds a row to be inserted with associated id.
+     * Adds a row to be inserted with associated id. Please notice that data for fields of type
+     * {@link Field.Type#bytes()} must be provided as a base64 encoded string.
      *
      * <p>Example usage of adding a row with associated id:
      * <pre> {@code
@@ -184,8 +190,9 @@ public Builder addRow(RowToInsert rowToInsert) {
      * recordContent.put("subfieldName1", "value");
      * recordContent.put("subfieldName2", repeatedFieldValue);
      * Map<String, Object> rowContent = new HashMap<String, Object>();
-     * rowContent.put("fieldName1", true);
-     * rowContent.put("fieldName2", recordContent);
+     * rowContent.put("booleanFieldName", true);
+     * rowContent.put("bytesFieldName", "DQ4KDQ==");
+     * rowContent.put("recordFieldName", recordContent);
      * builder.addRow("rowId", rowContent);
      * }</pre>
      */
@@ -195,7 +202,8 @@ public Builder addRow(String id, Map<String, Object> content) {
     }
 
     /**
-     * Adds a row to be inserted without an associated id.
+     * Adds a row to be inserted without an associated id. Please notice that data for fields of
+     * type {@link Field.Type#bytes()} must be provided as a base64 encoded string.
      *
      * <p>Example usage of adding a row without an associated id:
      * <pre> {@code
@@ -205,8 +213,9 @@ public Builder addRow(String id, Map<String, Object> content) {
      * recordContent.put("subfieldName1", "value");
      * recordContent.put("subfieldName2", repeatedFieldValue);
      * Map<String, Object> rowContent = new HashMap<String, Object>();
-     * rowContent.put("fieldName1", true);
-     * rowContent.put("fieldName2", recordContent);
+     * rowContent.put("booleanFieldName", true);
+     * rowContent.put("bytesFieldName", "DQ4KDQ==");
+     * rowContent.put("recordFieldName", recordContent);
      * builder.addRow(rowContent);
      * }</pre>
      */

gcloud-java-bigquery/src/test/java/com/google/cloud/bigquery/FieldValueTest.java

@@ -16,6 +16,7 @@
 
 package com.google.cloud.bigquery;
 
+import static org.junit.Assert.assertArrayEquals;
 import static org.junit.Assert.assertEquals;
 import static org.junit.Assert.assertFalse;
 import static org.junit.Assert.assertNull;
@@ -24,18 +25,22 @@
 import com.google.api.services.bigquery.model.TableCell;
 import com.google.common.collect.ImmutableList;
 import com.google.common.collect.ImmutableMap;
+import com.google.common.io.BaseEncoding;
 
 import org.junit.Test;
 
 import java.util.Map;
 
 public class FieldValueTest {
 
+  private static final byte[] BYTES = {0xD, 0xE, 0xA, 0xD};
+  private static final String BYTES_BASE64 = BaseEncoding.base64().encode(BYTES);
   private static final TableCell BOOLEAN_FIELD = new TableCell().setV("false");
   private static final Map<String, String> INTEGER_FIELD = ImmutableMap.of("v", "1");
   private static final Map<String, String> FLOAT_FIELD = ImmutableMap.of("v", "1.5");
   private static final Map<String, String> STRING_FIELD = ImmutableMap.of("v", "string");
   private static final Map<String, String> TIMESTAMP_FIELD = ImmutableMap.of("v", "42");
+  private static final Map<String, String> BYTES_FIELD = ImmutableMap.of("v", BYTES_BASE64);
   private static final Map<String, Object> NULL_FIELD =
       ImmutableMap.of("v", Data.nullOf(String.class));
   private static final Map<String, Object> REPEATED_FIELD =
@@ -60,6 +65,9 @@ public void testFromPb() {
     value = FieldValue.fromPb(TIMESTAMP_FIELD);
     assertEquals(FieldValue.Attribute.PRIMITIVE, value.attribute());
     assertEquals(42000000, value.timestampValue());
+    value = FieldValue.fromPb(BYTES_FIELD);
+    assertEquals(FieldValue.Attribute.PRIMITIVE, value.attribute());
+    assertArrayEquals(BYTES, value.bytesValue());
     value = FieldValue.fromPb(NULL_FIELD);
     assertNull(value.value());
     value = FieldValue.fromPb(REPEATED_FIELD);
@@ -94,6 +102,10 @@ public void testEquals() {
     assertEquals(timestampValue, FieldValue.fromPb(TIMESTAMP_FIELD));
     assertEquals(timestampValue.hashCode(), FieldValue.fromPb(TIMESTAMP_FIELD).hashCode());
 
+    FieldValue bytesValue = new FieldValue(FieldValue.Attribute.PRIMITIVE, BYTES_BASE64);
+    assertEquals(bytesValue, FieldValue.fromPb(BYTES_FIELD));
+    assertEquals(bytesValue.hashCode(), FieldValue.fromPb(BYTES_FIELD).hashCode());
+
     FieldValue nullValue = new FieldValue(FieldValue.Attribute.PRIMITIVE, null);
     assertEquals(nullValue, FieldValue.fromPb(NULL_FIELD));
     assertEquals(nullValue.hashCode(), FieldValue.fromPb(NULL_FIELD).hashCode());