Java 写parquet格式数据到hdfs

场景： 将接口数据接入数仓 

hive 版本：2.1.1   

hadoop 版本： 3.0.0

引入maven 依赖：

 <dependency>
            <groupId>com.alibaba</groupId>
            <artifactId>fastjson</artifactId>
            <version>1.2.60</version>
        </dependency>
        <dependency>
            <groupId>org.apache.hadoop</groupId>
            <artifactId>hadoop-hdfs</artifactId>
            <version>${hadoop.version}</version>
        </dependency>
        <dependency>
            <groupId>org.apache.hadoop</groupId>
            <artifactId>hadoop-client</artifactId>
            <version>${hadoop.version}</version>
        </dependency>
        <dependency>
            <groupId>org.apache.parquet</groupId>
            <artifactId>parquet-hadoop</artifactId>
            <version>1.12.0</version>
        </dependency>
        <dependency>
            <groupId>org.apache.parquet</groupId>
            <artifactId>parquet-column</artifactId>
            <version>1.12.0</version>
        </dependency>
        <dependency>
            <groupId>org.apache.hive</groupId>
            <artifactId>hive-jdbc</artifactId>
            <version>${hive.version}</version>
   </dependency>

数据写入hdfs 关键代码段：

使用append  模式

 /**
     * @desc append 模式写parquet 文件
     * @param filePath
     * @param FILE_SCHEMA
     * @return
     * @throws IOException
     */
    private ParquetWriter<Group> getWriter(String filePath, MessageType FILE_SCHEMA) throws IOException {
        Path path = new Path(filePath);
        return ExampleParquetWriter.builder(path)
                .withWriteMode(ParquetFileWriter.Mode.CREATE)
                .withWriterVersion(ParquetProperties.WriterVersion.PARQUET_1_0)
                .withCompressionCodec(CompressionCodecName.UNCOMPRESSED)
                .withConf(conf)
                .withType(FILE_SCHEMA).build();

    }

将数据集写分区表的hdfs 文件中

 /**
     * @desc 将数据集写分区表
     * @param partition
     * @param tableId
     * @param rows
     */
    public void write(String partition, String tableId, List<Row> rows) {
        try {
            //定义表结构
            MessageType messageType = Types.buildMessage()
                    .optional(PrimitiveType.PrimitiveTypeName.BINARY)
                    .as(LogicalTypeAnnotation.stringType())
                    .named("column1")
                    .optional(PrimitiveType.PrimitiveTypeName.BINARY)
                    .as(LogicalTypeAnnotation.stringType())
                    .named("column2")
                    .optional(PrimitiveType.PrimitiveTypeName.BINARY)
                    .as(LogicalTypeAnnotation.stringType())
                    .named("column3")
                    .optional(PrimitiveType.PrimitiveTypeName.BINARY)
                    .as(LogicalTypeAnnotation.stringType())
                    .named("column4")
                    .optional(PrimitiveType.PrimitiveTypeName.BINARY)
                    .as(LogicalTypeAnnotation.stringType())
                    .named("column5")
                    .optional(PrimitiveType.PrimitiveTypeName.BINARY)
                    .as(LogicalTypeAnnotation.stringType())
                    .named("column6")
                    .optional(PrimitiveType.PrimitiveTypeName.BINARY)
                    .as(LogicalTypeAnnotation.stringType())
                    .named("column7")
                    .optional(PrimitiveType.PrimitiveTypeName.BINARY)
                    .as(LogicalTypeAnnotation.stringType())
                    .named("column8")
                    .optional(PrimitiveType.PrimitiveTypeName.BINARY)
                    .as(LogicalTypeAnnotation.stringType())
                    .named("column9")
                    .optional(PrimitiveType.PrimitiveTypeName.BINARY)
                    .as(LogicalTypeAnnotation.stringType())
                    .named("column10")
                    .optional(PrimitiveType.PrimitiveTypeName.BINARY)
                    .as(LogicalTypeAnnotation.stringType())
                    .named("column11").named(tableId);


            //查看多条数据是否可以一次性写入一个Parquet文件
            ParquetWriter<Group> writer = getWriter(FS_PATH + tableId + "/" + partition + "/" + uuid, messageType);
            SimpleGroupFactory simpleGroupFactory = new SimpleGroupFactory(messageType);
            //校验目录
            for (Row row : rows) {
                try {
                    Group group = simpleGroupFactory.newGroup();

                    group.append("column1", "value1");
                    group.append("column2", "value2");
                    group.append("column3", "value3");
                    writer.write(group);
                } catch (Exception e) {
                    logger.error(rows.toString());
                }
            }
            writer.close();


        } catch (Exception e) {
            logger.error(ExceptionUtils.getStackTrace(e));
        }
    }

最后将 hdfs 中的 parquet 数据load 到 hive 分区外表中