Amazon Glue examples using SDK for C++

Code for the CMakeLists.txt CMake file.

# Set the minimum required version of CMake for this project.
cmake_minimum_required(VERSION 3.13)

# Set the AWS service components used by this project.
set(SERVICE_COMPONENTS glue)

# Set this project's name.
project("hello_glue")

# Set the C++ standard to use to build this target.
# At least C++ 11 is required for the AWS SDK for C++.
set(CMAKE_CXX_STANDARD 11)

# Use the MSVC variable to determine if this is a Windows build.
set(WINDOWS_BUILD ${MSVC})

if (WINDOWS_BUILD) # Set the location where CMake can find the installed libraries for the AWS SDK.
    string(REPLACE ";" "/aws-cpp-sdk-all;" SYSTEM_MODULE_PATH "${CMAKE_SYSTEM_PREFIX_PATH}/aws-cpp-sdk-all")
    list(APPEND CMAKE_PREFIX_PATH ${SYSTEM_MODULE_PATH})
endif ()

# Find the AWS SDK for C++ package.
find_package(AWSSDK REQUIRED COMPONENTS ${SERVICE_COMPONENTS})

if (WINDOWS_BUILD AND AWSSDK_INSTALL_AS_SHARED_LIBS)
     # Copy relevant AWS SDK for C++ libraries into the current binary directory for running and debugging.

     # set(BIN_SUB_DIR "/Debug") # if you are building from the command line you may need to uncomment this 
                                    # and set the proper subdirectory to the executables' location.

     AWSSDK_CPY_DYN_LIBS(SERVICE_COMPONENTS "" ${CMAKE_CURRENT_BINARY_DIR}${BIN_SUB_DIR})
endif ()

add_executable(${PROJECT_NAME}
        hello_glue.cpp)

target_link_libraries(${PROJECT_NAME}
        ${AWSSDK_LINK_LIBRARIES})

Code for the hello_glue.cpp source file.

#include <aws/core/Aws.h>
#include <aws/glue/GlueClient.h>
#include <aws/glue/model/ListJobsRequest.h>
#include <iostream>

/*
 *  A "Hello Glue" starter application which initializes an AWS Glue client and lists the
 *  AWS Glue job definitions.
 *
 *  main function
 *
 *  Usage: 'hello_glue'
 *
 */

int main(int argc, char **argv) {
    Aws::SDKOptions options;
    // Optionally change the log level for debugging.
//   options.loggingOptions.logLevel = Utils::Logging::LogLevel::Debug;
    Aws::InitAPI(options); // Should only be called once.
    int result = 0;
    {
        Aws::Client::ClientConfiguration clientConfig;
        // Optional: Set to the AWS Region (overrides config file).
        // clientConfig.region = "us-east-1";

        Aws::Glue::GlueClient glueClient(clientConfig);

        std::vector<Aws::String> jobs;

        Aws::String nextToken;  // Used for pagination.
        do {
            Aws::Glue::Model::ListJobsRequest listJobsRequest;
            if (!nextToken.empty()) {
                listJobsRequest.SetNextToken(nextToken);
            }

            Aws::Glue::Model::ListJobsOutcome listRunsOutcome = glueClient.ListJobs(
                    listJobsRequest);

            if (listRunsOutcome.IsSuccess()) {
                const std::vector<Aws::String> &jobNames = listRunsOutcome.GetResult().GetJobNames();
                jobs.insert(jobs.end(), jobNames.begin(), jobNames.end());

                nextToken = listRunsOutcome.GetResult().GetNextToken();
            } else {
                std::cerr << "Error listing jobs. "
                          << listRunsOutcome.GetError().GetMessage()
                          << std::endl;
                result = 1;
                break;
            }
        } while (!nextToken.empty());

        std::cout << "Your account has " << jobs.size() << " jobs."
                  << std::endl;
        for (size_t i = 0; i < jobs.size(); ++i) {
            std::cout << "   " << i + 1 << ". " << jobs[i] << std::endl;
        }
    }
    Aws::ShutdownAPI(options); // Should only be called once.
    return result;
}

        Aws::Client::ClientConfiguration clientConfig;
        // Optional: Set to the AWS Region in which the bucket was created (overrides config file).
        // clientConfig.region = "us-east-1";

    Aws::Glue::GlueClient client(clientConfig);

        Aws::Glue::Model::S3Target s3Target;
        s3Target.SetPath("s3://crawler-public-us-east-1/flight/2016/csv");
        Aws::Glue::Model::CrawlerTargets crawlerTargets;
        crawlerTargets.AddS3Targets(s3Target);

        Aws::Glue::Model::CreateCrawlerRequest request;
        request.SetTargets(crawlerTargets);
        request.SetName(CRAWLER_NAME);
        request.SetDatabaseName(CRAWLER_DATABASE_NAME);
        request.SetTablePrefix(CRAWLER_DATABASE_PREFIX);
        request.SetRole(roleArn);

        Aws::Glue::Model::CreateCrawlerOutcome outcome = client.CreateCrawler(request);

        if (outcome.IsSuccess()) {
            std::cout << "Successfully created the crawler." << std::endl;
        }
        else {
            std::cerr << "Error creating a crawler. " << outcome.GetError().GetMessage()
                      << std::endl;
            deleteAssets("", CRAWLER_DATABASE_NAME, "", bucketName, clientConfig);
            return false;
        }

        Aws::Client::ClientConfiguration clientConfig;
        // Optional: Set to the AWS Region in which the bucket was created (overrides config file).
        // clientConfig.region = "us-east-1";

    Aws::Glue::GlueClient client(clientConfig);

        Aws::Glue::Model::CreateJobRequest request;
        request.SetName(JOB_NAME);
        request.SetRole(roleArn);
        request.SetGlueVersion(GLUE_VERSION);

        Aws::Glue::Model::JobCommand command;
        command.SetName(JOB_COMMAND_NAME);
        command.SetPythonVersion(JOB_PYTHON_VERSION);
        command.SetScriptLocation(
                Aws::String("s3://") + bucketName + "/" + PYTHON_SCRIPT);
        request.SetCommand(command);

        Aws::Glue::Model::CreateJobOutcome outcome = client.CreateJob(request);

        if (outcome.IsSuccess()) {
            std::cout << "Successfully created the job." << std::endl;
        }
        else {
            std::cerr << "Error creating the job. " << outcome.GetError().GetMessage()
                      << std::endl;
            deleteAssets(CRAWLER_NAME, CRAWLER_DATABASE_NAME, "", bucketName,
                         clientConfig);
            return false;
        }

        Aws::Client::ClientConfiguration clientConfig;
        // Optional: Set to the AWS Region in which the bucket was created (overrides config file).
        // clientConfig.region = "us-east-1";

    Aws::Glue::GlueClient client(clientConfig);

        Aws::Glue::Model::DeleteCrawlerRequest request;
        request.SetName(crawler);

        Aws::Glue::Model::DeleteCrawlerOutcome outcome = client.DeleteCrawler(request);

        if (outcome.IsSuccess()) {
            std::cout << "Successfully deleted the crawler." << std::endl;
        }
        else {
            std::cerr << "Error deleting the crawler. "
                      << outcome.GetError().GetMessage() << std::endl;
            result = false;
        }

        Aws::Client::ClientConfiguration clientConfig;
        // Optional: Set to the AWS Region in which the bucket was created (overrides config file).
        // clientConfig.region = "us-east-1";

    Aws::Glue::GlueClient client(clientConfig);

        Aws::Glue::Model::DeleteDatabaseRequest request;
        request.SetName(database);

        Aws::Glue::Model::DeleteDatabaseOutcome outcome = client.DeleteDatabase(
                request);

        if (outcome.IsSuccess()) {
            std::cout << "Successfully deleted the database." << std::endl;
        }
        else {
            std::cerr << "Error deleting database. " << outcome.GetError().GetMessage()
                      << std::endl;
            result = false;
        }

        Aws::Client::ClientConfiguration clientConfig;
        // Optional: Set to the AWS Region in which the bucket was created (overrides config file).
        // clientConfig.region = "us-east-1";

    Aws::Glue::GlueClient client(clientConfig);

        Aws::Glue::Model::DeleteJobRequest request;
        request.SetJobName(job);

        Aws::Glue::Model::DeleteJobOutcome outcome = client.DeleteJob(request);


        if (outcome.IsSuccess()) {
            std::cout << "Successfully deleted the job." << std::endl;
        }
        else {
            std::cerr << "Error deleting the job. " << outcome.GetError().GetMessage()
                      << std::endl;
            result = false;
        }

        Aws::Client::ClientConfiguration clientConfig;
        // Optional: Set to the AWS Region in which the bucket was created (overrides config file).
        // clientConfig.region = "us-east-1";

    Aws::Glue::GlueClient client(clientConfig);

        Aws::Glue::Model::GetCrawlerRequest request;
        request.SetName(CRAWLER_NAME);

        Aws::Glue::Model::GetCrawlerOutcome outcome = client.GetCrawler(request);

        if (outcome.IsSuccess()) {
            Aws::Glue::Model::CrawlerState crawlerState = outcome.GetResult().GetCrawler().GetState();
            std::cout << "Retrieved crawler with state " <<
                      Aws::Glue::Model::CrawlerStateMapper::GetNameForCrawlerState(
                              crawlerState)
                      << "." << std::endl;
        }
        else {
            std::cerr << "Error retrieving a crawler.  "
                      << outcome.GetError().GetMessage() << std::endl;
            deleteAssets(CRAWLER_NAME, CRAWLER_DATABASE_NAME, "", bucketName,
                         clientConfig);
            return false;
        }

        Aws::Client::ClientConfiguration clientConfig;
        // Optional: Set to the AWS Region in which the bucket was created (overrides config file).
        // clientConfig.region = "us-east-1";

    Aws::Glue::GlueClient client(clientConfig);

        Aws::Glue::Model::GetDatabaseRequest request;
        request.SetName(CRAWLER_DATABASE_NAME);

        Aws::Glue::Model::GetDatabaseOutcome outcome = client.GetDatabase(request);

        if (outcome.IsSuccess()) {
            const Aws::Glue::Model::Database &database = outcome.GetResult().GetDatabase();

            std::cout << "Successfully retrieve the database\n" <<
                      database.Jsonize().View().WriteReadable() << "'." << std::endl;
        }
        else {
            std::cerr << "Error getting the database.  "
                      << outcome.GetError().GetMessage() << std::endl;
            deleteAssets(CRAWLER_NAME, CRAWLER_DATABASE_NAME, "", bucketName,
                         clientConfig);
            return false;
        }

        Aws::Client::ClientConfiguration clientConfig;
        // Optional: Set to the AWS Region in which the bucket was created (overrides config file).
        // clientConfig.region = "us-east-1";

    Aws::Glue::GlueClient client(clientConfig);

        Aws::Glue::Model::GetJobRunRequest jobRunRequest;
        jobRunRequest.SetJobName(jobName);
        jobRunRequest.SetRunId(jobRunID);

        Aws::Glue::Model::GetJobRunOutcome jobRunOutcome = client.GetJobRun(
                jobRunRequest);

        if (jobRunOutcome.IsSuccess()) {
            std::cout << "Displaying the job run JSON description." << std::endl;
            std::cout
                    << jobRunOutcome.GetResult().GetJobRun().Jsonize().View().WriteReadable()
                    << std::endl;
        }
        else {
            std::cerr << "Error get a job run. "
                      << jobRunOutcome.GetError().GetMessage()
                      << std::endl;
        }

        Aws::Client::ClientConfiguration clientConfig;
        // Optional: Set to the AWS Region in which the bucket was created (overrides config file).
        // clientConfig.region = "us-east-1";

    Aws::Glue::GlueClient client(clientConfig);

        Aws::Glue::Model::GetJobRunsRequest getJobRunsRequest;
        getJobRunsRequest.SetJobName(jobName);

        Aws::String nextToken; // Used for pagination.
        std::vector<Aws::Glue::Model::JobRun> allJobRuns;
        do {
            if (!nextToken.empty()) {
                getJobRunsRequest.SetNextToken(nextToken);
            }
            Aws::Glue::Model::GetJobRunsOutcome jobRunsOutcome = client.GetJobRuns(
                    getJobRunsRequest);

            if (jobRunsOutcome.IsSuccess()) {
                const std::vector<Aws::Glue::Model::JobRun> &jobRuns = jobRunsOutcome.GetResult().GetJobRuns();
                allJobRuns.insert(allJobRuns.end(), jobRuns.begin(), jobRuns.end());

                nextToken = jobRunsOutcome.GetResult().GetNextToken();
            }
            else {
                std::cerr << "Error getting job runs. "
                          << jobRunsOutcome.GetError().GetMessage()
                          << std::endl;
                break;
            }
        } while (!nextToken.empty());

        Aws::Client::ClientConfiguration clientConfig;
        // Optional: Set to the AWS Region in which the bucket was created (overrides config file).
        // clientConfig.region = "us-east-1";

    Aws::Glue::GlueClient client(clientConfig);

        Aws::Glue::Model::GetTablesRequest request;
        request.SetDatabaseName(CRAWLER_DATABASE_NAME);
        std::vector<Aws::Glue::Model::Table> all_tables;
        Aws::String nextToken; // Used for pagination.
        do {
            Aws::Glue::Model::GetTablesOutcome outcome = client.GetTables(request);

            if (outcome.IsSuccess()) {
                const std::vector<Aws::Glue::Model::Table> &tables = outcome.GetResult().GetTableList();
                all_tables.insert(all_tables.end(), tables.begin(), tables.end());
                nextToken = outcome.GetResult().GetNextToken();
            }
            else {
                std::cerr << "Error getting the tables. "
                          << outcome.GetError().GetMessage()
                          << std::endl;
                deleteAssets(CRAWLER_NAME, CRAWLER_DATABASE_NAME, "", bucketName,
                             clientConfig);
                return false;
            }
        } while (!nextToken.empty());

        std::cout << "The database contains " << all_tables.size()
                  << (all_tables.size() == 1 ?
                      " table." : "tables.") << std::endl;
        std::cout << "Here is a list of the tables in the database.";
        for (size_t index = 0; index < all_tables.size(); ++index) {
            std::cout << "    " << index + 1 << ":  " << all_tables[index].GetName()
                      << std::endl;
        }

        if (!all_tables.empty()) {
            int tableIndex = askQuestionForIntRange(
                    "Enter an index to display the database detail ",
                    1, static_cast<int>(all_tables.size()));
            std::cout << all_tables[tableIndex - 1].Jsonize().View().WriteReadable()
                      << std::endl;

            tableName = all_tables[tableIndex - 1].GetName();
        }

        Aws::Client::ClientConfiguration clientConfig;
        // Optional: Set to the AWS Region in which the bucket was created (overrides config file).
        // clientConfig.region = "us-east-1";

    Aws::Glue::GlueClient client(clientConfig);

        Aws::Glue::Model::ListJobsRequest listJobsRequest;

        Aws::String nextToken;
        std::vector<Aws::String> allJobNames;

        do {
            if (!nextToken.empty()) {
                listJobsRequest.SetNextToken(nextToken);
            }
            Aws::Glue::Model::ListJobsOutcome listRunsOutcome = client.ListJobs(
                    listJobsRequest);

            if (listRunsOutcome.IsSuccess()) {
                const std::vector<Aws::String> &jobNames = listRunsOutcome.GetResult().GetJobNames();
                allJobNames.insert(allJobNames.end(), jobNames.begin(), jobNames.end());
                nextToken = listRunsOutcome.GetResult().GetNextToken();
            }
            else {
                std::cerr << "Error listing jobs. "
                          << listRunsOutcome.GetError().GetMessage()
                          << std::endl;
            }
        } while (!nextToken.empty());

        Aws::Client::ClientConfiguration clientConfig;
        // Optional: Set to the AWS Region in which the bucket was created (overrides config file).
        // clientConfig.region = "us-east-1";

    Aws::Glue::GlueClient client(clientConfig);

        Aws::Glue::Model::StartCrawlerRequest request;
        request.SetName(CRAWLER_NAME);

        Aws::Glue::Model::StartCrawlerOutcome outcome = client.StartCrawler(request);


        if (outcome.IsSuccess() || (Aws::Glue::GlueErrors::CRAWLER_RUNNING ==
                                    outcome.GetError().GetErrorType())) {
            if (!outcome.IsSuccess()) {
                std::cout << "Crawler was already started." << std::endl;
            }
            else {
                std::cout << "Successfully started crawler." << std::endl;
            }

            std::cout << "This may take a while to run." << std::endl;

            Aws::Glue::Model::CrawlerState crawlerState = Aws::Glue::Model::CrawlerState::NOT_SET;
            int iterations = 0;
            while (Aws::Glue::Model::CrawlerState::READY != crawlerState) {
                std::this_thread::sleep_for(std::chrono::seconds(1));
                ++iterations;
                if ((iterations % 10) == 0) { // Log status every 10 seconds.
                    std::cout << "Crawler status " <<
                              Aws::Glue::Model::CrawlerStateMapper::GetNameForCrawlerState(
                                      crawlerState)
                              << ". After " << iterations
                              << " seconds elapsed."
                              << std::endl;
                }
                Aws::Glue::Model::GetCrawlerRequest getCrawlerRequest;
                getCrawlerRequest.SetName(CRAWLER_NAME);

                Aws::Glue::Model::GetCrawlerOutcome getCrawlerOutcome = client.GetCrawler(
                        getCrawlerRequest);

                if (getCrawlerOutcome.IsSuccess()) {
                    crawlerState = getCrawlerOutcome.GetResult().GetCrawler().GetState();
                }
                else {
                    std::cerr << "Error getting crawler.  "
                              << getCrawlerOutcome.GetError().GetMessage() << std::endl;
                    break;
                }
            }

            if (Aws::Glue::Model::CrawlerState::READY == crawlerState) {
                std::cout << "Crawler finished running after " << iterations
                          << " seconds."
                          << std::endl;
            }
        }
        else {
            std::cerr << "Error starting a crawler.  "
                      << outcome.GetError().GetMessage()
                      << std::endl;

            deleteAssets(CRAWLER_NAME, CRAWLER_DATABASE_NAME, "", bucketName,
                         clientConfig);
            return false;
        }

        Aws::Client::ClientConfiguration clientConfig;
        // Optional: Set to the AWS Region in which the bucket was created (overrides config file).
        // clientConfig.region = "us-east-1";

    Aws::Glue::GlueClient client(clientConfig);

        Aws::Glue::Model::StartJobRunRequest request;
        request.SetJobName(JOB_NAME);

        Aws::Map<Aws::String, Aws::String> arguments;
        arguments["--input_database"] = CRAWLER_DATABASE_NAME;
        arguments["--input_table"] = tableName;
        arguments["--output_bucket_url"] = Aws::String("s3://") + bucketName + "/";
        request.SetArguments(arguments);

        Aws::Glue::Model::StartJobRunOutcome outcome = client.StartJobRun(request);

        if (outcome.IsSuccess()) {
            std::cout << "Successfully started the job." << std::endl;

            Aws::String jobRunId = outcome.GetResult().GetJobRunId();

            int iterator = 0;
            bool done = false;
            while (!done) {
                ++iterator;
                std::this_thread::sleep_for(std::chrono::seconds(1));
                Aws::Glue::Model::GetJobRunRequest jobRunRequest;
                jobRunRequest.SetJobName(JOB_NAME);
                jobRunRequest.SetRunId(jobRunId);

                Aws::Glue::Model::GetJobRunOutcome jobRunOutcome = client.GetJobRun(
                        jobRunRequest);

                if (jobRunOutcome.IsSuccess()) {
                    const Aws::Glue::Model::JobRun &jobRun = jobRunOutcome.GetResult().GetJobRun();
                    Aws::Glue::Model::JobRunState jobRunState = jobRun.GetJobRunState();

                    if ((jobRunState == Aws::Glue::Model::JobRunState::STOPPED) ||
                        (jobRunState == Aws::Glue::Model::JobRunState::FAILED) ||
                        (jobRunState == Aws::Glue::Model::JobRunState::TIMEOUT)) {
                        std::cerr << "Error running job. "
                                  << jobRun.GetErrorMessage()
                                  << std::endl;
                        deleteAssets(CRAWLER_NAME, CRAWLER_DATABASE_NAME, JOB_NAME,
                                     bucketName,
                                     clientConfig);
                        return false;
                    }
                    else if (jobRunState ==
                             Aws::Glue::Model::JobRunState::SUCCEEDED) {
                        std::cout << "Job run succeeded after  " << iterator <<
                                  " seconds elapsed." << std::endl;
                        done = true;
                    }
                    else if ((iterator % 10) == 0) { // Log status every 10 seconds.
                        std::cout << "Job run status " <<
                                  Aws::Glue::Model::JobRunStateMapper::GetNameForJobRunState(
                                          jobRunState) <<
                                  ". " << iterator <<
                                  " seconds elapsed." << std::endl;
                    }
                }
                else {
                    std::cerr << "Error retrieving job run state. "
                              << jobRunOutcome.GetError().GetMessage()
                              << std::endl;
                    deleteAssets(CRAWLER_NAME, CRAWLER_DATABASE_NAME, JOB_NAME,
                                 bucketName, clientConfig);
                    return false;
                }
            }
        }
        else {
            std::cerr << "Error starting a job. " << outcome.GetError().GetMessage()
                      << std::endl;
            deleteAssets(CRAWLER_NAME, CRAWLER_DATABASE_NAME, JOB_NAME, bucketName,
                         clientConfig);
            return false;
        }

//! Scenario which demonstrates using AWS Glue to add a crawler and run a job.
/*!
 \\sa runGettingStartedWithGlueScenario()
 \param bucketName: An S3 bucket created in the setup.
 \param roleName: An AWS Identity and Access Management (IAM) role created in the setup.
 \param clientConfig: AWS client configuration.
 \return bool: Successful completion.
 */

bool AwsDoc::Glue::runGettingStartedWithGlueScenario(const Aws::String &bucketName,
                                                     const Aws::String &roleName,
                                                     const Aws::Client::ClientConfiguration &clientConfig) {
    Aws::Glue::GlueClient client(clientConfig);

    Aws::String roleArn;
    if (!getRoleArn(roleName, roleArn, clientConfig)) {
        std::cerr << "Error getting role ARN for role." << std::endl;
        return false;
    }

    // 1. Upload the job script to the S3 bucket.
    {
        std::cout << "Uploading the job script '"
                  << AwsDoc::Glue::PYTHON_SCRIPT
                  << "'." << std::endl;

        if (!AwsDoc::Glue::uploadFile(bucketName,
                                      AwsDoc::Glue::PYTHON_SCRIPT_PATH,
                                      AwsDoc::Glue::PYTHON_SCRIPT,
                                      clientConfig)) {
            std::cerr << "Error uploading the job file." << std::endl;
            return false;
        }
    }

    // 2. Create a crawler.
    {
        Aws::Glue::Model::S3Target s3Target;
        s3Target.SetPath("s3://crawler-public-us-east-1/flight/2016/csv");
        Aws::Glue::Model::CrawlerTargets crawlerTargets;
        crawlerTargets.AddS3Targets(s3Target);

        Aws::Glue::Model::CreateCrawlerRequest request;
        request.SetTargets(crawlerTargets);
        request.SetName(CRAWLER_NAME);
        request.SetDatabaseName(CRAWLER_DATABASE_NAME);
        request.SetTablePrefix(CRAWLER_DATABASE_PREFIX);
        request.SetRole(roleArn);

        Aws::Glue::Model::CreateCrawlerOutcome outcome = client.CreateCrawler(request);

        if (outcome.IsSuccess()) {
            std::cout << "Successfully created the crawler." << std::endl;
        }
        else {
            std::cerr << "Error creating a crawler. " << outcome.GetError().GetMessage()
                      << std::endl;
            deleteAssets("", CRAWLER_DATABASE_NAME, "", bucketName, clientConfig);
            return false;
        }
    }

    // 3. Get a crawler.
    {
        Aws::Glue::Model::GetCrawlerRequest request;
        request.SetName(CRAWLER_NAME);

        Aws::Glue::Model::GetCrawlerOutcome outcome = client.GetCrawler(request);

        if (outcome.IsSuccess()) {
            Aws::Glue::Model::CrawlerState crawlerState = outcome.GetResult().GetCrawler().GetState();
            std::cout << "Retrieved crawler with state " <<
                      Aws::Glue::Model::CrawlerStateMapper::GetNameForCrawlerState(
                              crawlerState)
                      << "." << std::endl;
        }
        else {
            std::cerr << "Error retrieving a crawler.  "
                      << outcome.GetError().GetMessage() << std::endl;
            deleteAssets(CRAWLER_NAME, CRAWLER_DATABASE_NAME, "", bucketName,
                         clientConfig);
            return false;
        }
    }

    // 4. Start a crawler.
    {
        Aws::Glue::Model::StartCrawlerRequest request;
        request.SetName(CRAWLER_NAME);

        Aws::Glue::Model::StartCrawlerOutcome outcome = client.StartCrawler(request);


        if (outcome.IsSuccess() || (Aws::Glue::GlueErrors::CRAWLER_RUNNING ==
                                    outcome.GetError().GetErrorType())) {
            if (!outcome.IsSuccess()) {
                std::cout << "Crawler was already started." << std::endl;
            }
            else {
                std::cout << "Successfully started crawler." << std::endl;
            }

            std::cout << "This may take a while to run." << std::endl;

            Aws::Glue::Model::CrawlerState crawlerState = Aws::Glue::Model::CrawlerState::NOT_SET;
            int iterations = 0;
            while (Aws::Glue::Model::CrawlerState::READY != crawlerState) {
                std::this_thread::sleep_for(std::chrono::seconds(1));
                ++iterations;
                if ((iterations % 10) == 0) { // Log status every 10 seconds.
                    std::cout << "Crawler status " <<
                              Aws::Glue::Model::CrawlerStateMapper::GetNameForCrawlerState(
                                      crawlerState)
                              << ". After " << iterations
                              << " seconds elapsed."
                              << std::endl;
                }
                Aws::Glue::Model::GetCrawlerRequest getCrawlerRequest;
                getCrawlerRequest.SetName(CRAWLER_NAME);

                Aws::Glue::Model::GetCrawlerOutcome getCrawlerOutcome = client.GetCrawler(
                        getCrawlerRequest);

                if (getCrawlerOutcome.IsSuccess()) {
                    crawlerState = getCrawlerOutcome.GetResult().GetCrawler().GetState();
                }
                else {
                    std::cerr << "Error getting crawler.  "
                              << getCrawlerOutcome.GetError().GetMessage() << std::endl;
                    break;
                }
            }

            if (Aws::Glue::Model::CrawlerState::READY == crawlerState) {
                std::cout << "Crawler finished running after " << iterations
                          << " seconds."
                          << std::endl;
            }
        }
        else {
            std::cerr << "Error starting a crawler.  "
                      << outcome.GetError().GetMessage()
                      << std::endl;

            deleteAssets(CRAWLER_NAME, CRAWLER_DATABASE_NAME, "", bucketName,
                         clientConfig);
            return false;
        }
    }

    // 5. Get a database.
    {
        Aws::Glue::Model::GetDatabaseRequest request;
        request.SetName(CRAWLER_DATABASE_NAME);

        Aws::Glue::Model::GetDatabaseOutcome outcome = client.GetDatabase(request);

        if (outcome.IsSuccess()) {
            const Aws::Glue::Model::Database &database = outcome.GetResult().GetDatabase();

            std::cout << "Successfully retrieve the database\n" <<
                      database.Jsonize().View().WriteReadable() << "'." << std::endl;
        }
        else {
            std::cerr << "Error getting the database.  "
                      << outcome.GetError().GetMessage() << std::endl;
            deleteAssets(CRAWLER_NAME, CRAWLER_DATABASE_NAME, "", bucketName,
                         clientConfig);
            return false;
        }
    }

    // 6. Get tables.
    Aws::String tableName;
    {
        Aws::Glue::Model::GetTablesRequest request;
        request.SetDatabaseName(CRAWLER_DATABASE_NAME);
        std::vector<Aws::Glue::Model::Table> all_tables;
        Aws::String nextToken; // Used for pagination.
        do {
            Aws::Glue::Model::GetTablesOutcome outcome = client.GetTables(request);

            if (outcome.IsSuccess()) {
                const std::vector<Aws::Glue::Model::Table> &tables = outcome.GetResult().GetTableList();
                all_tables.insert(all_tables.end(), tables.begin(), tables.end());
                nextToken = outcome.GetResult().GetNextToken();
            }
            else {
                std::cerr << "Error getting the tables. "
                          << outcome.GetError().GetMessage()
                          << std::endl;
                deleteAssets(CRAWLER_NAME, CRAWLER_DATABASE_NAME, "", bucketName,
                             clientConfig);
                return false;
            }
        } while (!nextToken.empty());

        std::cout << "The database contains " << all_tables.size()
                  << (all_tables.size() == 1 ?
                      " table." : "tables.") << std::endl;
        std::cout << "Here is a list of the tables in the database.";
        for (size_t index = 0; index < all_tables.size(); ++index) {
            std::cout << "    " << index + 1 << ":  " << all_tables[index].GetName()
                      << std::endl;
        }

        if (!all_tables.empty()) {
            int tableIndex = askQuestionForIntRange(
                    "Enter an index to display the database detail ",
                    1, static_cast<int>(all_tables.size()));
            std::cout << all_tables[tableIndex - 1].Jsonize().View().WriteReadable()
                      << std::endl;

            tableName = all_tables[tableIndex - 1].GetName();
        }
    }

    // 7. Create a job.
    {
        Aws::Glue::Model::CreateJobRequest request;
        request.SetName(JOB_NAME);
        request.SetRole(roleArn);
        request.SetGlueVersion(GLUE_VERSION);

        Aws::Glue::Model::JobCommand command;
        command.SetName(JOB_COMMAND_NAME);
        command.SetPythonVersion(JOB_PYTHON_VERSION);
        command.SetScriptLocation(
                Aws::String("s3://") + bucketName + "/" + PYTHON_SCRIPT);
        request.SetCommand(command);

        Aws::Glue::Model::CreateJobOutcome outcome = client.CreateJob(request);

        if (outcome.IsSuccess()) {
            std::cout << "Successfully created the job." << std::endl;
        }
        else {
            std::cerr << "Error creating the job. " << outcome.GetError().GetMessage()
                      << std::endl;
            deleteAssets(CRAWLER_NAME, CRAWLER_DATABASE_NAME, "", bucketName,
                         clientConfig);
            return false;
        }
    }

    // 8. Start a job run.
    {
        Aws::Glue::Model::StartJobRunRequest request;
        request.SetJobName(JOB_NAME);

        Aws::Map<Aws::String, Aws::String> arguments;
        arguments["--input_database"] = CRAWLER_DATABASE_NAME;
        arguments["--input_table"] = tableName;
        arguments["--output_bucket_url"] = Aws::String("s3://") + bucketName + "/";
        request.SetArguments(arguments);

        Aws::Glue::Model::StartJobRunOutcome outcome = client.StartJobRun(request);

        if (outcome.IsSuccess()) {
            std::cout << "Successfully started the job." << std::endl;

            Aws::String jobRunId = outcome.GetResult().GetJobRunId();

            int iterator = 0;
            bool done = false;
            while (!done) {
                ++iterator;
                std::this_thread::sleep_for(std::chrono::seconds(1));
                Aws::Glue::Model::GetJobRunRequest jobRunRequest;
                jobRunRequest.SetJobName(JOB_NAME);
                jobRunRequest.SetRunId(jobRunId);

                Aws::Glue::Model::GetJobRunOutcome jobRunOutcome = client.GetJobRun(
                        jobRunRequest);

                if (jobRunOutcome.IsSuccess()) {
                    const Aws::Glue::Model::JobRun &jobRun = jobRunOutcome.GetResult().GetJobRun();
                    Aws::Glue::Model::JobRunState jobRunState = jobRun.GetJobRunState();

                    if ((jobRunState == Aws::Glue::Model::JobRunState::STOPPED) ||
                        (jobRunState == Aws::Glue::Model::JobRunState::FAILED) ||
                        (jobRunState == Aws::Glue::Model::JobRunState::TIMEOUT)) {
                        std::cerr << "Error running job. "
                                  << jobRun.GetErrorMessage()
                                  << std::endl;
                        deleteAssets(CRAWLER_NAME, CRAWLER_DATABASE_NAME, JOB_NAME,
                                     bucketName,
                                     clientConfig);
                        return false;
                    }
                    else if (jobRunState ==
                             Aws::Glue::Model::JobRunState::SUCCEEDED) {
                        std::cout << "Job run succeeded after  " << iterator <<
                                  " seconds elapsed." << std::endl;
                        done = true;
                    }
                    else if ((iterator % 10) == 0) { // Log status every 10 seconds.
                        std::cout << "Job run status " <<
                                  Aws::Glue::Model::JobRunStateMapper::GetNameForJobRunState(
                                          jobRunState) <<
                                  ". " << iterator <<
                                  " seconds elapsed." << std::endl;
                    }
                }
                else {
                    std::cerr << "Error retrieving job run state. "
                              << jobRunOutcome.GetError().GetMessage()
                              << std::endl;
                    deleteAssets(CRAWLER_NAME, CRAWLER_DATABASE_NAME, JOB_NAME,
                                 bucketName, clientConfig);
                    return false;
                }
            }
        }
        else {
            std::cerr << "Error starting a job. " << outcome.GetError().GetMessage()
                      << std::endl;
            deleteAssets(CRAWLER_NAME, CRAWLER_DATABASE_NAME, JOB_NAME, bucketName,
                         clientConfig);
            return false;
        }
    }

    // 9. List the output data stored in the S3 bucket.
    {
        Aws::S3::S3Client s3Client;
        Aws::S3::Model::ListObjectsV2Request request;
        request.SetBucket(bucketName);
        request.SetPrefix(OUTPUT_FILE_PREFIX);

        Aws::String continuationToken; // Used for pagination.
        std::vector<Aws::S3::Model::Object> allObjects;
        do {
            if (!continuationToken.empty()) {
                request.SetContinuationToken(continuationToken);
            }
            Aws::S3::Model::ListObjectsV2Outcome outcome = s3Client.ListObjectsV2(
                    request);

            if (outcome.IsSuccess()) {
                const std::vector<Aws::S3::Model::Object> &objects =
                        outcome.GetResult().GetContents();
                allObjects.insert(allObjects.end(), objects.begin(), objects.end());
                continuationToken = outcome.GetResult().GetNextContinuationToken();
            }
            else {
                std::cerr << "Error listing objects. "
                          << outcome.GetError().GetMessage()
                          << std::endl;
                break;
            }
        } while (!continuationToken.empty());

        std::cout << "Data from your job is in " << allObjects.size() <<
                  " files in the S3 bucket, " << bucketName << "." << std::endl;

        for (size_t i = 0; i < allObjects.size(); ++i) {
            std::cout << "    " << i + 1 << ". " << allObjects[i].GetKey()
                      << std::endl;
        }

        int objectIndex = askQuestionForIntRange(
                std::string(
                        "Enter the number of a block to download it and see the first ") +
                std::to_string(LINES_OF_RUN_FILE_TO_DISPLAY) +
                " lines of JSON output in the block: ", 1,
                static_cast<int>(allObjects.size()));

        Aws::String objectKey = allObjects[objectIndex - 1].GetKey();

        std::stringstream stringStream;
        if (getObjectFromBucket(bucketName, objectKey, stringStream,
                                clientConfig)) {
            for (int i = 0; i < LINES_OF_RUN_FILE_TO_DISPLAY && stringStream; ++i) {
                std::string line;
                std::getline(stringStream, line);
                std::cout << "    " << line << std::endl;
            }
        }
        else {
            deleteAssets(CRAWLER_NAME, CRAWLER_DATABASE_NAME, JOB_NAME, bucketName,
                         clientConfig);
            return false;
        }
    }

    // 10. List all the jobs.
    Aws::String jobName;
    {
        Aws::Glue::Model::ListJobsRequest listJobsRequest;

        Aws::String nextToken;
        std::vector<Aws::String> allJobNames;

        do {
            if (!nextToken.empty()) {
                listJobsRequest.SetNextToken(nextToken);
            }
            Aws::Glue::Model::ListJobsOutcome listRunsOutcome = client.ListJobs(
                    listJobsRequest);

            if (listRunsOutcome.IsSuccess()) {
                const std::vector<Aws::String> &jobNames = listRunsOutcome.GetResult().GetJobNames();
                allJobNames.insert(allJobNames.end(), jobNames.begin(), jobNames.end());
                nextToken = listRunsOutcome.GetResult().GetNextToken();
            }
            else {
                std::cerr << "Error listing jobs. "
                          << listRunsOutcome.GetError().GetMessage()
                          << std::endl;
            }
        } while (!nextToken.empty());
        std::cout << "Your account has " << allJobNames.size() << " jobs."
                  << std::endl;
        for (size_t i = 0; i < allJobNames.size(); ++i) {
            std::cout << "   " << i + 1 << ". " << allJobNames[i] << std::endl;
        }
        int jobIndex = askQuestionForIntRange(
                Aws::String("Enter a number between 1 and ") +
                std::to_string(allJobNames.size()) +
                " to see the list of runs for a job: ",
                1, static_cast<int>(allJobNames.size()));

        jobName = allJobNames[jobIndex - 1];
    }

    // 11. Get the job runs for a job.
    Aws::String jobRunID;
    if (!jobName.empty()) {
        Aws::Glue::Model::GetJobRunsRequest getJobRunsRequest;
        getJobRunsRequest.SetJobName(jobName);

        Aws::String nextToken; // Used for pagination.
        std::vector<Aws::Glue::Model::JobRun> allJobRuns;
        do {
            if (!nextToken.empty()) {
                getJobRunsRequest.SetNextToken(nextToken);
            }
            Aws::Glue::Model::GetJobRunsOutcome jobRunsOutcome = client.GetJobRuns(
                    getJobRunsRequest);

            if (jobRunsOutcome.IsSuccess()) {
                const std::vector<Aws::Glue::Model::JobRun> &jobRuns = jobRunsOutcome.GetResult().GetJobRuns();
                allJobRuns.insert(allJobRuns.end(), jobRuns.begin(), jobRuns.end());

                nextToken = jobRunsOutcome.GetResult().GetNextToken();
            }
            else {
                std::cerr << "Error getting job runs. "
                          << jobRunsOutcome.GetError().GetMessage()
                          << std::endl;
                break;
            }
        } while (!nextToken.empty());

        std::cout << "There are " << allJobRuns.size() << " runs in the job '"
                  <<
                  jobName << "'." << std::endl;

        for (size_t i = 0; i < allJobRuns.size(); ++i) {
            std::cout << "   " << i + 1 << ". " << allJobRuns[i].GetJobName()
                      << std::endl;
        }

        int runIndex = askQuestionForIntRange(
                Aws::String("Enter a number between 1 and ") +
                std::to_string(allJobRuns.size()) +
                " to see details for a run: ",
                1, static_cast<int>(allJobRuns.size()));
        jobRunID = allJobRuns[runIndex - 1].GetId();
    }

    // 12. Get a single job run.
    if (!jobRunID.empty()) {
        Aws::Glue::Model::GetJobRunRequest jobRunRequest;
        jobRunRequest.SetJobName(jobName);
        jobRunRequest.SetRunId(jobRunID);

        Aws::Glue::Model::GetJobRunOutcome jobRunOutcome = client.GetJobRun(
                jobRunRequest);

        if (jobRunOutcome.IsSuccess()) {
            std::cout << "Displaying the job run JSON description." << std::endl;
            std::cout
                    << jobRunOutcome.GetResult().GetJobRun().Jsonize().View().WriteReadable()
                    << std::endl;
        }
        else {
            std::cerr << "Error get a job run. "
                      << jobRunOutcome.GetError().GetMessage()
                      << std::endl;
        }
    }

    return deleteAssets(CRAWLER_NAME, CRAWLER_DATABASE_NAME, JOB_NAME, bucketName,
                        clientConfig);
}

//! Cleanup routine to delete created assets.
/*!
 \\sa deleteAssets()
 \param crawler: Name of an AWS Glue crawler.
 \param database: The name of an AWS Glue database.
 \param job: The name of an AWS Glue job.
 \param bucketName: The name of an S3 bucket.
 \param clientConfig: AWS client configuration.
 \return bool: Successful completion.
 */
bool AwsDoc::Glue::deleteAssets(const Aws::String &crawler, const Aws::String &database,
                                const Aws::String &job, const Aws::String &bucketName,
                                const Aws::Client::ClientConfiguration &clientConfig) {
    const Aws::Glue::GlueClient client(clientConfig);
    bool result = true;

    // 13. Delete a job.
    if (!job.empty()) {
        Aws::Glue::Model::DeleteJobRequest request;
        request.SetJobName(job);

        Aws::Glue::Model::DeleteJobOutcome outcome = client.DeleteJob(request);


        if (outcome.IsSuccess()) {
            std::cout << "Successfully deleted the job." << std::endl;
        }
        else {
            std::cerr << "Error deleting the job. " << outcome.GetError().GetMessage()
                      << std::endl;
            result = false;
        }
    }

    // 14. Delete a database.
    if (!database.empty()) {
        Aws::Glue::Model::DeleteDatabaseRequest request;
        request.SetName(database);

        Aws::Glue::Model::DeleteDatabaseOutcome outcome = client.DeleteDatabase(
                request);

        if (outcome.IsSuccess()) {
            std::cout << "Successfully deleted the database." << std::endl;
        }
        else {
            std::cerr << "Error deleting database. " << outcome.GetError().GetMessage()
                      << std::endl;
            result = false;
        }
    }

    // 15. Delete a crawler.
    if (!crawler.empty()) {
        Aws::Glue::Model::DeleteCrawlerRequest request;
        request.SetName(crawler);

        Aws::Glue::Model::DeleteCrawlerOutcome outcome = client.DeleteCrawler(request);

        if (outcome.IsSuccess()) {
            std::cout << "Successfully deleted the crawler." << std::endl;
        }
        else {
            std::cerr << "Error deleting the crawler. "
                      << outcome.GetError().GetMessage() << std::endl;
            result = false;
        }
    }

    // 16. Delete the job script and run data from the S3 bucket.
    result &= AwsDoc::Glue::deleteAllObjectsInS3Bucket(bucketName,
                                                       clientConfig);
    return result;
}

//! Routine which uploads a file to an S3 bucket.
/*!
 \\sa uploadFile()
 \param bucketName: An S3 bucket created in the setup.
 \param filePath: The path of the file to upload.
 \param fileName The name for the uploaded file.
 \param clientConfig: AWS client configuration.
 \return bool: Successful completion.
 */
bool
AwsDoc::Glue::uploadFile(const Aws::String &bucketName,
                         const Aws::String &filePath,
                         const Aws::String &fileName,
                         const Aws::Client::ClientConfiguration &clientConfig) {
    Aws::S3::S3Client s3_client(clientConfig);

    Aws::S3::Model::PutObjectRequest request;
    request.SetBucket(bucketName);
    request.SetKey(fileName);

    std::shared_ptr<Aws::IOStream> inputData =
            Aws::MakeShared<Aws::FStream>("SampleAllocationTag",
                                          filePath.c_str(),
                                          std::ios_base::in | std::ios_base::binary);

    if (!*inputData) {
        std::cerr << "Error unable to read file " << filePath << std::endl;
        return false;
    }

    request.SetBody(inputData);

    Aws::S3::Model::PutObjectOutcome outcome =
            s3_client.PutObject(request);

    if (!outcome.IsSuccess()) {
        std::cerr << "Error: PutObject: " <<
                  outcome.GetError().GetMessage() << std::endl;
    }
    else {
        std::cout << "Added object '" << filePath << "' to bucket '"
                  << bucketName << "'." << std::endl;
    }

    return outcome.IsSuccess();
}

//! Routine which deletes all objects in an S3 bucket.
/*!
 \\sa deleteAllObjectsInS3Bucket()
 \param bucketName: The S3 bucket name.
 \param clientConfig: AWS client configuration.
 \return bool: Successful completion.
 */
bool AwsDoc::Glue::deleteAllObjectsInS3Bucket(const Aws::String &bucketName,
                                              const Aws::Client::ClientConfiguration &clientConfig) {
    Aws::S3::S3Client client(clientConfig);
    Aws::S3::Model::ListObjectsV2Request listObjectsRequest;
    listObjectsRequest.SetBucket(bucketName);

    Aws::String continuationToken; // Used for pagination.
    bool result = true;
    do {
        if (!continuationToken.empty()) {
            listObjectsRequest.SetContinuationToken(continuationToken);
        }

        Aws::S3::Model::ListObjectsV2Outcome listObjectsOutcome = client.ListObjectsV2(
                listObjectsRequest);

        if (listObjectsOutcome.IsSuccess()) {
            const std::vector<Aws::S3::Model::Object> &objects = listObjectsOutcome.GetResult().GetContents();
            if (!objects.empty()) {
                Aws::S3::Model::DeleteObjectsRequest deleteObjectsRequest;
                deleteObjectsRequest.SetBucket(bucketName);

                std::vector<Aws::S3::Model::ObjectIdentifier> objectIdentifiers;
                for (const Aws::S3::Model::Object &object: objects) {
                    objectIdentifiers.push_back(
                            Aws::S3::Model::ObjectIdentifier().WithKey(
                                    object.GetKey()));
                }
                Aws::S3::Model::Delete objectsDelete;
                objectsDelete.SetObjects(objectIdentifiers);
                objectsDelete.SetQuiet(true);
                deleteObjectsRequest.SetDelete(objectsDelete);

                Aws::S3::Model::DeleteObjectsOutcome deleteObjectsOutcome =
                        client.DeleteObjects(deleteObjectsRequest);

                if (!deleteObjectsOutcome.IsSuccess()) {
                    std::cerr << "Error deleting objects. " <<
                              deleteObjectsOutcome.GetError().GetMessage() << std::endl;
                    result = false;
                    break;
                }
                else {
                    std::cout << "Successfully deleted the objects." << std::endl;

                }
            }
            else {
                std::cout << "No objects to delete in '" << bucketName << "'."
                          << std::endl;
            }

            continuationToken = listObjectsOutcome.GetResult().GetNextContinuationToken();
        }
        else {
            std::cerr << "Error listing objects. "
                      << listObjectsOutcome.GetError().GetMessage() << std::endl;
            result = false;
            break;
        }
    } while (!continuationToken.empty());

    return result;
}

//! Routine which retrieves an object from an S3 bucket.
/*!
 \\sa getObjectFromBucket()
 \param bucketName: The S3 bucket name.
 \param objectKey: The object's name.
 \param objectStream: A stream to receive the retrieved data.
 \param clientConfig: AWS client configuration.
 \return bool: Successful completion.
 */
bool AwsDoc::Glue::getObjectFromBucket(const Aws::String &bucketName,
                                       const Aws::String &objectKey,
                                       std::ostream &objectStream,
                                       const Aws::Client::ClientConfiguration &clientConfig) {
    Aws::S3::S3Client client(clientConfig);
    Aws::S3::Model::GetObjectRequest request;
    request.SetBucket(bucketName);
    request.SetKey(objectKey);

    Aws::S3::Model::GetObjectOutcome outcome = client.GetObject(request);


    if (outcome.IsSuccess()) {
        std::cout << "Successfully retrieved '" << objectKey << "'." << std::endl;
        auto &body = outcome.GetResult().GetBody();
        objectStream << body.rdbuf();
    }
    else {
        std::cerr << "Error retrieving object. " << outcome.GetError().GetMessage()
                  << std::endl;
    }

    return outcome.IsSuccess();
}