com.marklogic.flux.impl.export.ExportArchiveFilesCommand Maven / Gradle / Ivy

Go to download
Show more of this group Show more artifacts with this name
Show all versions of flux-api Show documentation
Flux API for data movement with MarkLogic
There is a newer version: 1.0.0.ea1
/*
 * Copyright © 2024 MarkLogic Corporation. All Rights Reserved.
 */
package com.marklogic.flux.impl.export;

import com.marklogic.flux.api.ArchiveFilesExporter;
import com.marklogic.flux.impl.AbstractCommand;
import com.marklogic.flux.impl.OptionsUtil;
import com.marklogic.spark.Options;
import org.apache.spark.sql.*;
import picocli.CommandLine;

import java.util.Map;
import java.util.function.Consumer;
import java.util.stream.Collectors;
import java.util.stream.Stream;

@CommandLine.Command(
    name = "export-archive-files",
    description = "Read documents and their metadata from MarkLogic and write them to ZIP files on a local filesystem, HDFS, or S3."
)
public class ExportArchiveFilesCommand extends AbstractCommand implements ArchiveFilesExporter {

    @CommandLine.Mixin
    private ReadArchiveDocumentsParams readParams = new ReadArchiveDocumentsParams();

    @CommandLine.Mixin
    protected WriteArchiveFilesParams writeParams = new WriteArchiveFilesParams();

    @Override
    protected void validateDuringApiUsage() {
        writeParams.validatePath();
        readParams.verifyAtLeastOneQueryOptionIsSet("export");
    }

    @Override
    public void validateCommandLineOptions(CommandLine.ParseResult parseResult) {
        super.validateCommandLineOptions(parseResult);
        OptionsUtil.verifyHasAtLeastOneOption(parseResult, ReadDocumentParams.REQUIRED_QUERY_OPTIONS);
    }

    @Override
    protected Dataset loadDataset(SparkSession session, DataFrameReader reader) {
        final int fileCount = writeParams.getFileCount();
        if (fileCount > 0) {
            getCommonParams().setRepartition(fileCount);
        }
        return reader.format(MARKLOGIC_CONNECTOR)
            .options(getConnectionParams().makeOptions())
            .options(readParams.makeOptions())
            .load();
    }

    @Override
    protected void applyWriter(SparkSession session, DataFrameWriter writer) {
        writeParams.getS3Params().addToHadoopConfiguration(session.sparkContext().hadoopConfiguration());
        writer.format(MARKLOGIC_CONNECTOR)
            .options(writeParams.get())
            // The connector only supports "Append" in terms of how Spark defines it, but it will always overwrite files.
            .mode(SaveMode.Append)
            .save(writeParams.getPath());
    }

    public static class WriteArchiveFilesParams extends WriteFilesParams implements WriteArchiveFilesOptions {

        @CommandLine.Option(names = "--encoding", description = "Specify an encoding for writing files.")
        private String encoding;

        @Override
        public WriteArchiveFilesOptions encoding(String encoding) {
            this.encoding = encoding;
            return this;
        }

        @Override
        public Map get() {
            return OptionsUtil.makeOptions(
                Options.WRITE_FILES_COMPRESSION, "zip",
                Options.WRITE_FILES_ENCODING, encoding
            );
        }
    }

    public static class ReadArchiveDocumentsParams extends ReadDocumentParams implements ReadArchiveDocumentOptions {

        @CommandLine.Option(names = "--categories", description = "Comma-delimited sequence of categories of data to include. " +
            "Valid choices are: collections, permissions, quality, properties, and metadatavalues.")
        private String categories;

        @Override
        public Map makeOptions() {
            return OptionsUtil.addOptions(super.makeOptions(),
                Options.READ_DOCUMENTS_CATEGORIES, determineCategories()
            );
        }

        @Override
        public ReadArchiveDocumentOptions categories(String... categories) {
            this.categories = Stream.of(categories).collect(Collectors.joining(","));
            return this;
        }

        /**
         * While the "read documents" operation allows for only reading metadata, that isn't valid for an archive - we
         * always need content to be returned as well.
         *
         * @return
         */
        private String determineCategories() {
            if (categories != null && categories.trim().length() > 0) {
                return "content," + categories;
            }
            return "content,metadata";
        }
    }

    @Override
    public ArchiveFilesExporter from(Consumer consumer) {
        consumer.accept(readParams);
        return this;
    }

    @Override
    public ArchiveFilesExporter to(Consumer consumer) {
        consumer.accept(writeParams);
        return this;
    }

    @Override
    public ArchiveFilesExporter to(String path) {
        writeParams.path(path);
        return this;
    }
}