Examples with HoodieBootstrapSchemaProvider org.apache.hudi.client.bootstrap.HoodieBootstrapSchemaProvider used on opensource projects

Example 1 with HoodieBootstrapSchemaProvider

use of org.apache.hudi.client.bootstrap.HoodieBootstrapSchemaProvider 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));
}