Back to EIP

PatternClaim Check will store messages in a persistent storage and send a claim to another application to access the origin stored message.
How Ballerina helps

Ballerina supports the persistent storage of messages. Popular storage technologies such as SQL, AWS S3, and Redis are available as Ballerina packages.

Claim CheckContent EnricherContent Filter
import ballerina/http;
import ballerina/io;
import ballerinax/kafka;

type ScanResult [string, string];

final http:Client awsS3Client = check new (",io");
final http:Client firebaseClient = check new ("");
final kafka:Consumer kafkaConsumer = check new (kafka:DEFAULT_URL, {
    groupId: "mri-scan-group",
    topics: ["topic-mri-scan"]

public function main() returns error? {
    while true {
        ScanResult[] mriScanResults = check kafkaConsumer->pollPayload(1);
        foreach var [scanId, patientId] in mriScanResults {
            http:Response s3Response = check awsS3Client->/mri\-scans/[scanId].get();
            string mriScanResult = analyzeMriScan(check s3Response.getByteStream());
            _ = check firebaseClient->/mri/[scanId]/reports\.json.put({mriScanResult, patientId}, targetType = json);

isolated function analyzeMriScan(stream<byte[], io:Error?> fileByteStream) returns string {
    // logic to analyze the MRI scan
    return "No Abnormalities Detected";
import ballerina/http;
import ballerina/io;
import ballerina/mime;
import ballerina/uuid;
import ballerinax/kafka;

final http:Client awsS3Client = check new (",io");
final kafka:Producer kafkaProducer = check new (kafka:DEFAULT_URL);

type ScanRequest record {|
    string patientId;
    stream<byte[], io:Error?> fileByteStream;

service /api/v1 on new http:Listener(8080) {

    isolated resource function post scans/mri(http:Request request) returns error? {
        ScanRequest {patientId, fileByteStream} = check scanRequestFromMultipart(request);

        string claimCheckId = uuid:createType1AsString();
        string fileName = string `${claimCheckId}.dicom`;

        http:Request s3ObjectCreationRequest = new;
        _ = check awsS3Client->/mri\-scans/[fileName].put(s3ObjectCreationRequest, targetType = http:Response);

        _ = check kafkaProducer->send({
            topic: "topic-mri-scan",
            value: [fileName, patientId]

isolated function scanRequestFromMultipart(http:Request request) returns ScanRequest|error {
    mime:Entity[] bodyParts = check request.getBodyParts();
    string? patientId = ();
    stream<byte[], io:Error?>? fileByteStream = ();
    foreach mime:Entity bodyPart in bodyParts {
        string partName = bodyPart.getContentDisposition().name;
        if partName == "patientId" {
            patientId = check bodyPart.getText();
        } else if partName == "file" {
            fileByteStream = check bodyPart.getByteStream();
    if patientId == () || fileByteStream == () {
        return error("Multipart request should contains both patientId and file parts");
    return {fileByteStream, patientId};