Search in sources :

Example 16 with EventData

use of com.alibaba.otter.shared.etl.model.EventData in project otter by alibaba.

the class DbLoadActionTest method test_db_load_mysql.

@Test
public void test_db_load_mysql() {
    ArbitrateConfigRegistry.regist(configClientService);
    dbLoadAction = (DbLoadAction) TestedObject.getSpringBeanFactory().getBean("dbLoadAction");
    final Channel channel = new Channel();
    channel.setId(1L);
    final Pipeline pipeline = new Pipeline();
    pipeline.setId(100L);
    List<DataMediaPair> pairs = generatorDataMediaPairForMysql(20);
    pipeline.setPairs(pairs);
    pipeline.getParameters().merge(new SystemParameter());
    pipeline.getParameters().merge(new ChannelParameter());
    // pipeline.getParameters().setChannelInfo("LJH_DEMO");
    // final Pipeline oppositePipeline = new Pipeline();
    // oppositePipeline.setId(101L);
    channel.setPipelines(Arrays.asList(pipeline));
    final Node currentNode = new Node();
    currentNode.setId(1L);
    new NonStrictExpectations() {

        {
            configClientService.findChannel(anyLong);
            returns(channel);
            configClientService.findPipeline(anyLong);
            returns(pipeline);
            configClientService.currentNode();
            returns(currentNode);
        }
    };
    Identity identity = new Identity();
    identity.setChannelId(100L);
    identity.setPipelineId(100L);
    identity.setProcessId(100L);
    RowBatch rowBatch = new RowBatch();
    rowBatch.setIdentity(identity);
    List<EventData> eventDatas = generatorEventDataForMysql(0, 20, EventType.INSERT);
    for (EventData eventData : eventDatas) {
        rowBatch.merge(eventData);
    }
    eventDatas = generatorEventDataForMysql(10, 10, EventType.INSERT);
    for (EventData eventData : eventDatas) {
        rowBatch.merge(eventData);
    }
    eventDatas = generatorEventDataForMysql(19, 1, EventType.DELETE);
    for (EventData eventData : eventDatas) {
        rowBatch.merge(eventData);
    }
    WeightController controller = new WeightController(1);
    dbLoadAction.load(rowBatch, controller);
}
Also used : DataMediaPair(com.alibaba.otter.shared.common.model.config.data.DataMediaPair) ChannelParameter(com.alibaba.otter.shared.common.model.config.channel.ChannelParameter) Channel(com.alibaba.otter.shared.common.model.config.channel.Channel) Node(com.alibaba.otter.shared.common.model.config.node.Node) EventData(com.alibaba.otter.shared.etl.model.EventData) Pipeline(com.alibaba.otter.shared.common.model.config.pipeline.Pipeline) RowBatch(com.alibaba.otter.shared.etl.model.RowBatch) SystemParameter(com.alibaba.otter.shared.common.model.config.parameter.SystemParameter) WeightController(com.alibaba.otter.node.etl.load.loader.weight.WeightController) Identity(com.alibaba.otter.shared.etl.model.Identity) Test(org.testng.annotations.Test) BaseDbTest(com.alibaba.otter.node.etl.BaseDbTest)

Example 17 with EventData

use of com.alibaba.otter.shared.etl.model.EventData in project otter by alibaba.

the class DbLoadActionTest method generatorEventDataForOracle.

private List<EventData> generatorEventDataForOracle(int start, int count, EventType type) {
    List<EventData> eventDatas = new ArrayList<EventData>();
    for (int i = 0; i < count; i++) {
        int index = i + 1 + start;
        EventData eventData = new EventData();
        eventData.setPairId(index);
        eventData.setTableId(1L);
        eventData.setSchemaName("srf");
        eventData.setTableName("columns");
        eventData.setEventType(type);
        eventData.setExecuteTime(100L);
        eventData.getKeys().add(buildColumn("id", Types.NUMERIC, "" + index, true, false));
        eventData.getKeys().add(buildColumn("name", Types.VARCHAR, "ljh_" + index, true, false));
        eventData.getOldKeys().add(buildColumn("id", Types.NUMERIC, "" + index, true, false));
        eventData.getOldKeys().add(buildColumn("name", Types.VARCHAR, "ljh_" + index, true, false));
        eventData.getColumns().add(buildColumn("alias_name", Types.CHAR, "hello_" + index, false, false));
        eventData.getColumns().add(buildColumn("amount", Types.NUMERIC, "100.01", false, false));
        eventData.getColumns().add(buildColumn("text_b", Types.BLOB, "[116,101,120,116,95,98]", false, false));
        eventData.getColumns().add(buildColumn("text_c", Types.CLOB, "中文", false, false));
        eventData.getColumns().add(buildColumn("curr_date", Types.DATE, "2011-01-01", false, false));
        eventData.getColumns().add(buildColumn("gmt_create", Types.DATE, "2011-01-01 11:11:11", false, false));
        eventData.getColumns().add(buildColumn("gmt_modify", Types.DATE, "2011-01-01 11:11:11", false, false));
        // OracleSqlTemplate sqlTemplate = new OracleSqlTemplate();
        // String sql = null;
        // if (type.isInsert()) {
        // sql = sqlTemplate.getMergeSql(eventData.getSchemaName(),
        // eventData.getTableName(),
        // buildColumnNames(eventData.getKeys()),
        // buildColumnNames(eventData.getColumns()), new String[] {});
        // } else if (type.isUpdate()) {
        // sql = sqlTemplate.getUpdateSql(eventData.getSchemaName(),
        // eventData.getTableName(),
        // buildColumnNames(eventData.getKeys()),
        // buildColumnNames(eventData.getColumns()));
        // } else if (type.isDelete()) {
        // sql = sqlTemplate.getDeleteSql(eventData.getSchemaName(),
        // eventData.getTableName(),
        // buildColumnNames(eventData.getKeys()));
        // }
        // eventData.setSql(sql);
        eventDatas.add(eventData);
    }
    return eventDatas;
}
Also used : ArrayList(java.util.ArrayList) EventData(com.alibaba.otter.shared.etl.model.EventData)

Example 18 with EventData

use of com.alibaba.otter.shared.etl.model.EventData in project otter by alibaba.

the class RowDataHttpPipe method saveDbBatch.

// ======================== help method ===================
// 保存对应的dbBatch
private HttpPipeKey saveDbBatch(DbBatch dbBatch) {
    RowBatch rowBatch = dbBatch.getRowBatch();
    // 转化为proto对象
    BatchProto.RowBatch.Builder rowBatchBuilder = BatchProto.RowBatch.newBuilder();
    rowBatchBuilder.setIdentity(build(rowBatch.getIdentity()));
    // 处理具体的字段rowData
    for (EventData eventData : rowBatch.getDatas()) {
        BatchProto.RowData.Builder rowDataBuilder = BatchProto.RowData.newBuilder();
        rowDataBuilder.setPairId(eventData.getPairId());
        rowDataBuilder.setTableId(eventData.getTableId());
        if (eventData.getSchemaName() != null) {
            rowDataBuilder.setSchemaName(eventData.getSchemaName());
        }
        rowDataBuilder.setTableName(eventData.getTableName());
        rowDataBuilder.setEventType(eventData.getEventType().getValue());
        rowDataBuilder.setExecuteTime(eventData.getExecuteTime());
        // add by ljh at 2012-10-31
        if (eventData.getSyncMode() != null) {
            rowDataBuilder.setSyncMode(eventData.getSyncMode().getValue());
        }
        if (eventData.getSyncConsistency() != null) {
            rowDataBuilder.setSyncConsistency(eventData.getSyncConsistency().getValue());
        }
        // 构造key column
        for (EventColumn keyColumn : eventData.getKeys()) {
            rowDataBuilder.addKeys(buildColumn(keyColumn));
        }
        // 构造old key column
        if (CollectionUtils.isEmpty(eventData.getOldKeys()) == false) {
            for (EventColumn keyColumn : eventData.getOldKeys()) {
                rowDataBuilder.addOldKeys(buildColumn(keyColumn));
            }
        }
        // 构造其他 column
        for (EventColumn column : eventData.getColumns()) {
            rowDataBuilder.addColumns(buildColumn(column));
        }
        rowDataBuilder.setRemedy(eventData.isRemedy());
        rowDataBuilder.setSize(eventData.getSize());
        if (StringUtils.isNotEmpty(eventData.getSql())) {
            rowDataBuilder.setSql(eventData.getSql());
        }
        if (StringUtils.isNotEmpty(eventData.getDdlSchemaName())) {
            rowDataBuilder.setDdlSchemaName(eventData.getDdlSchemaName());
        }
        if (StringUtils.isNotEmpty(eventData.getHint())) {
            rowDataBuilder.setHint(eventData.getHint());
        }
        rowDataBuilder.setWithoutSchema(eventData.isWithoutSchema());
        // 添加一条rowData记录
        rowBatchBuilder.addRows(rowDataBuilder.build());
    }
    // 处理下FileBatch
    FileBatch fileBatch = dbBatch.getFileBatch();
    BatchProto.FileBatch.Builder fileBatchBuilder = null;
    fileBatchBuilder = BatchProto.FileBatch.newBuilder();
    fileBatchBuilder.setIdentity(build(fileBatch.getIdentity()));
    // 构造对应的proto对象
    for (FileData fileData : fileBatch.getFiles()) {
        BatchProto.FileData.Builder fileDataBuilder = BatchProto.FileData.newBuilder();
        fileDataBuilder.setPairId(fileData.getPairId());
        fileDataBuilder.setTableId(fileData.getTableId());
        if (fileData.getNameSpace() != null) {
            fileDataBuilder.setNamespace(fileData.getNameSpace());
        }
        if (fileData.getPath() != null) {
            fileDataBuilder.setPath(fileData.getPath());
        }
        fileDataBuilder.setEventType(fileData.getEventType().getValue());
        fileDataBuilder.setSize(fileData.getSize());
        fileDataBuilder.setLastModifiedTime(fileData.getLastModifiedTime());
        // 添加一条fileData记录
        fileBatchBuilder.addFiles(fileDataBuilder.build());
    }
    // 处理构造对应的文件url
    String filename = buildFileName(rowBatch.getIdentity(), ClassUtils.getShortClassName(dbBatch.getClass()));
    // 写入数据
    File file = new File(htdocsDir, filename);
    OutputStream output = null;
    try {
        output = new BufferedOutputStream(new FileOutputStream(file));
        com.alibaba.otter.node.etl.model.protobuf.BatchProto.RowBatch rowBatchProto = rowBatchBuilder.build();
        // 输出大小
        output.write(ByteUtils.int2bytes(rowBatchProto.getSerializedSize()));
        // 输出row batch
        rowBatchProto.writeTo(output);
        com.alibaba.otter.node.etl.model.protobuf.BatchProto.FileBatch fileBatchProto = fileBatchBuilder.build();
        // 输出大小
        output.write(ByteUtils.int2bytes(fileBatchProto.getSerializedSize()));
        // 输出file batch
        fileBatchProto.writeTo(output);
        output.flush();
    } catch (IOException e) {
        throw new PipeException("write_byte_error", e);
    } finally {
        IOUtils.closeQuietly(output);
    }
    HttpPipeKey key = new HttpPipeKey();
    key.setUrl(remoteUrlBuilder.getUrl(rowBatch.getIdentity().getPipelineId(), filename));
    key.setDataType(PipeDataType.DB_BATCH);
    key.setIdentity(rowBatch.getIdentity());
    Pipeline pipeline = configClientService.findPipeline(rowBatch.getIdentity().getPipelineId());
    if (pipeline.getParameters().getUseFileEncrypt()) {
        // 加密处理
        EncryptedData encryptedData = encryptFile(file);
        key.setKey(encryptedData.getKey());
        key.setCrc(encryptedData.getCrc());
    }
    return key;
}
Also used : EventColumn(com.alibaba.otter.shared.etl.model.EventColumn) BufferedOutputStream(java.io.BufferedOutputStream) OutputStream(java.io.OutputStream) FileOutputStream(java.io.FileOutputStream) EventData(com.alibaba.otter.shared.etl.model.EventData) EncryptedData(com.alibaba.otter.node.etl.common.io.EncryptedData) FileData(com.alibaba.otter.shared.etl.model.FileData) BufferedOutputStream(java.io.BufferedOutputStream) FileBatch(com.alibaba.otter.shared.etl.model.FileBatch) IOException(java.io.IOException) BatchProto(com.alibaba.otter.node.etl.model.protobuf.BatchProto) Pipeline(com.alibaba.otter.shared.common.model.config.pipeline.Pipeline) RowBatch(com.alibaba.otter.shared.etl.model.RowBatch) FileOutputStream(java.io.FileOutputStream) PipeException(com.alibaba.otter.node.etl.common.pipe.exception.PipeException) File(java.io.File)

Example 19 with EventData

use of com.alibaba.otter.shared.etl.model.EventData in project otter by alibaba.

the class DataBatchLoader method split.

/**
     * 将rowBatch中的记录,按找载入的目标数据源进行分类
     */
private List<RowBatch> split(RowBatch rowBatch) {
    final Identity identity = rowBatch.getIdentity();
    Map<DataMediaSource, RowBatch> result = new MapMaker().makeComputingMap(new Function<DataMediaSource, RowBatch>() {

        public RowBatch apply(DataMediaSource input) {
            RowBatch rowBatch = new RowBatch();
            rowBatch.setIdentity(identity);
            return rowBatch;
        }
    });
    for (EventData eventData : rowBatch.getDatas()) {
        // 获取介质信息
        DataMedia media = ConfigHelper.findDataMedia(configClientService.findPipeline(identity.getPipelineId()), eventData.getTableId());
        // 归类
        result.get(media.getSource()).merge(eventData);
    }
    return new ArrayList<RowBatch>(result.values());
}
Also used : RowBatch(com.alibaba.otter.shared.etl.model.RowBatch) MapMaker(com.google.common.collect.MapMaker) ArrayList(java.util.ArrayList) DataMediaSource(com.alibaba.otter.shared.common.model.config.data.DataMediaSource) Identity(com.alibaba.otter.shared.etl.model.Identity) EventData(com.alibaba.otter.shared.etl.model.EventData) DataMedia(com.alibaba.otter.shared.common.model.config.data.DataMedia)

Example 20 with EventData

use of com.alibaba.otter.shared.etl.model.EventData in project otter by alibaba.

the class DbLoadAction method buildWeightBuckets.

/**
     * 构建基于weight权重分组的item集合列表
     */
private WeightBuckets<EventData> buildWeightBuckets(DbLoadContext context, List<EventData> datas) {
    WeightBuckets<EventData> buckets = new WeightBuckets<EventData>();
    for (EventData data : datas) {
        // 获取对应的weight
        DataMediaPair pair = ConfigHelper.findDataMediaPair(context.getPipeline(), data.getPairId());
        buckets.addItem(pair.getPushWeight(), data);
    }
    return buckets;
}
Also used : DataMediaPair(com.alibaba.otter.shared.common.model.config.data.DataMediaPair) WeightBuckets(com.alibaba.otter.node.etl.load.loader.weight.WeightBuckets) EventData(com.alibaba.otter.shared.etl.model.EventData)

Aggregations

EventData (com.alibaba.otter.shared.etl.model.EventData)48 ArrayList (java.util.ArrayList)20 Pipeline (com.alibaba.otter.shared.common.model.config.pipeline.Pipeline)19 EventColumn (com.alibaba.otter.shared.etl.model.EventColumn)18 DataMediaPair (com.alibaba.otter.shared.common.model.config.data.DataMediaPair)16 Test (org.testng.annotations.Test)16 BaseDbTest (com.alibaba.otter.node.etl.BaseDbTest)15 RowBatch (com.alibaba.otter.shared.etl.model.RowBatch)14 Identity (com.alibaba.otter.shared.etl.model.Identity)9 MapMaker (com.google.common.collect.MapMaker)9 RowKey (com.alibaba.otter.node.etl.load.loader.db.DbLoadMerger.RowKey)8 DataMedia (com.alibaba.otter.shared.common.model.config.data.DataMedia)8 DbBatch (com.alibaba.otter.shared.etl.model.DbBatch)7 DbDialect (com.alibaba.otter.node.etl.common.db.dialect.DbDialect)5 ExtractException (com.alibaba.otter.node.etl.extract.exceptions.ExtractException)5 LoadException (com.alibaba.otter.node.etl.load.exception.LoadException)4 FileData (com.alibaba.otter.shared.etl.model.FileData)4 SelectException (com.alibaba.otter.node.etl.select.exceptions.SelectException)3 Channel (com.alibaba.otter.shared.common.model.config.channel.Channel)3 EventType (com.alibaba.otter.shared.etl.model.EventType)3