Examples with HoodieFileStatus org.apache.hudi.avro.model.HoodieFileStatus used on opensource projects

Example 11 with HoodieFileStatus

use of org.apache.hudi.avro.model.HoodieFileStatus in project hudi by apache.

the class SparkBootstrapCommitActionExecutor method runMetadataBootstrap.

private HoodieData<BootstrapWriteStatus> runMetadataBootstrap(List<Pair<String, List<HoodieFileStatus>>> partitions) {
    if (null == partitions || partitions.isEmpty()) {
        return context.emptyHoodieData();
    }
    TypedProperties properties = new TypedProperties();
    properties.putAll(config.getProps());
    KeyGeneratorInterface keyGenerator;
    try {
        keyGenerator = HoodieSparkKeyGeneratorFactory.createKeyGenerator(properties);
    } catch (IOException e) {
        throw new HoodieKeyGeneratorException("Init keyGenerator failed ", e);
    }
    BootstrapPartitionPathTranslator translator = (BootstrapPartitionPathTranslator) ReflectionUtils.loadClass(config.getBootstrapPartitionPathTranslatorClass(), properties);
    List<Pair<String, Pair<String, HoodieFileStatus>>> bootstrapPaths = partitions.stream().flatMap(p -> {
        String translatedPartitionPath = translator.getBootstrapTranslatedPath(p.getKey());
        return p.getValue().stream().map(f -> Pair.of(p.getKey(), Pair.of(translatedPartitionPath, f)));
    }).collect(Collectors.toList());
    context.setJobStatus(this.getClass().getSimpleName(), "Bootstrap metadata table.");
    return context.parallelize(bootstrapPaths, config.getBootstrapParallelism()).map(partitionFsPair -> getMetadataHandler(config, table, partitionFsPair.getRight().getRight()).runMetadataBootstrap(partitionFsPair.getLeft(), partitionFsPair.getRight().getLeft(), keyGenerator));
}

Example 12 with HoodieFileStatus

use of org.apache.hudi.avro.model.HoodieFileStatus in project hudi by apache.

the class SparkBootstrapCommitActionExecutor method listAndProcessSourcePartitions.

/**
 * Return Bootstrap Mode selections for partitions listed and figure out bootstrap Schema.
 * @return
 * @throws IOException
 */
private Map<BootstrapMode, List<Pair<String, List<HoodieFileStatus>>>> listAndProcessSourcePartitions() throws IOException {
    List<Pair<String, List<HoodieFileStatus>>> folders = BootstrapUtils.getAllLeafFoldersWithFiles(table.getMetaClient(), bootstrapSourceFileSystem, config.getBootstrapSourceBasePath(), context);
    LOG.info("Fetching Bootstrap Schema !!");
    HoodieBootstrapSchemaProvider sourceSchemaProvider = new HoodieSparkBootstrapSchemaProvider(config);
    bootstrapSchema = sourceSchemaProvider.getBootstrapSchema(context, folders).toString();
    LOG.info("Bootstrap Schema :" + bootstrapSchema);
    BootstrapModeSelector selector = (BootstrapModeSelector) ReflectionUtils.loadClass(config.getBootstrapModeSelectorClass(), config);
    Map<BootstrapMode, List<String>> result = selector.select(folders);
    Map<String, List<HoodieFileStatus>> partitionToFiles = folders.stream().collect(Collectors.toMap(Pair::getKey, Pair::getValue));
    // Ensure all partitions are accounted for
    ValidationUtils.checkArgument(partitionToFiles.keySet().equals(result.values().stream().flatMap(Collection::stream).collect(Collectors.toSet())));
    return result.entrySet().stream().map(e -> Pair.of(e.getKey(), e.getValue().stream().map(p -> Pair.of(p, partitionToFiles.get(p))).collect(Collectors.toList()))).collect(Collectors.toMap(Pair::getKey, Pair::getValue));
}

Example 13 with HoodieFileStatus

use of org.apache.hudi.avro.model.HoodieFileStatus in project hudi by apache.

the class SparkBootstrapCommitActionExecutor method metadataBootstrap.

/**
 * Perform Metadata Bootstrap.
 * @param partitionFilesList List of partitions and files within that partitions
 */
protected Option<HoodieWriteMetadata<HoodieData<WriteStatus>>> metadataBootstrap(List<Pair<String, List<HoodieFileStatus>>> partitionFilesList) {
    if (null == partitionFilesList || partitionFilesList.isEmpty()) {
        return Option.empty();
    }
    HoodieTableMetaClient metaClient = table.getMetaClient();
    metaClient.getActiveTimeline().createNewInstant(new HoodieInstant(State.REQUESTED, metaClient.getCommitActionType(), HoodieTimeline.METADATA_BOOTSTRAP_INSTANT_TS));
    table.getActiveTimeline().transitionRequestedToInflight(new HoodieInstant(State.REQUESTED, metaClient.getCommitActionType(), HoodieTimeline.METADATA_BOOTSTRAP_INSTANT_TS), Option.empty());
    HoodieData<BootstrapWriteStatus> bootstrapWriteStatuses = runMetadataBootstrap(partitionFilesList);
    HoodieWriteMetadata<HoodieData<WriteStatus>> result = new HoodieWriteMetadata<>();
    updateIndexAndCommitIfNeeded(bootstrapWriteStatuses.map(w -> w), result);
    return Option.of(result);
}