Chapter 61. HL7

HL7 Component

The HL7 component is used for working with the HL7 MLLP protocol and HL7 v2 messages using the HAPI library.
This component supports the following:
Maven users will need to add the following dependency to their pom.xml for this component:
<dependency>
    <groupId>org.apache.camel</groupId>
    <artifactId>camel-hl7</artifactId>
    <version>x.x.x</version>
    <!-- use the same version as your Camel core version -->
</dependency>

HL7 MLLP protocol

HL7 is often used with the HL7 MLLP protocol, which is a text based TCP socket based protocol. This component ships with a Mina and Netty4 Codec that conforms to the MLLP protocol so you can easily expose an HL7 listener accepting HL7 requests over the TCP transport layer.
To expose a HL7 listener service, the camel-mina2 or camel-netty4 component is used with the HL7MLLPCodec (mina2) or HL7MLLPNettyDecoder/HL7MLLPNettyEncoder (Netty4).
The HL7 MLLP codec has the following options:
Name Default Value Description
startByte 0x0b The start byte spanning the HL7 payload.
endByte1 0x1c The first end byte spanning the HL7 payload.
endByte2 0x0d The 2nd end byte spanning the HL7 payload.
charset JVM Default The encoding (a charset name) to use for the codec. If not provided, Camel will use the JVM default Charset.
produceString true Camel 2.14.1: If true, the codec creates a string using the defined charset. If false, the codec sends a plain byte array into the route, so that the HL7 Data Format can determine the actual charset from the HL7 message content.
convertLFtoCR false Will convert \n to \r (0x0d, 13 decimal) as HL7 stipulates \r as segment terminators. The HAPI library requires the use of \r.

Exposing a HL7 listener using Mina

In the Spring XML file, we configure a Mina2 endpoint to listen for HL7 requests using TCP on port 8888:
<endpoint id="hl7MinaListener" uri="mina2:tcp://localhost:8888?sync=true&amp;codec=#hl7codec"/>
sync=true indicates that this listener is synchronous and therefore will return a HL7 response to the caller. The HL7 codec is set up with codec=#hl7codec. Note that hl7codec is just a Spring bean ID, so it could be named mygreatcodecforhl7 or whatever you like. The codec is also set up in the Spring XML file:
    <bean id="hl7codec" class="org.apache.camel.component.hl7.HL7MLLPCodec">
        <property name="charset" value="iso-8859-1"/>
    </bean>
The endpoint hl7MinaListener can then be used in a route as a consumer, as this Java DSL example illustrates:
    from("hl7MinaListener").beanRef("patientLookupService");
This is a very simple route that will listen for HL7 and route it to a service named patientLookupService. This is also Spring bean ID, configured in the Spring XML as:
    <bean id="patientLookupService" class="com.mycompany.healthcare.service.PatientLookupService"/>
Another powerful feature of Camel is that we can have our business logic in POJO classes that is not tied to Camel as shown here:
import ca.uhn.hl7v2.HL7Exception;
import ca.uhn.hl7v2.model.Message;
import ca.uhn.hl7v2.model.v24.segment.QRD;

public class PatientLookupService {
    public Message lookupPatient(Message input) throws HL7Exception {
        QRD qrd = (QRD)input.get("QRD");
        String patientId = qrd.getWhoSubjectFilter(0).getIDNumber().getValue();

        // find patient data based on the patient id and create a HL7 model object with the response
        Message response = ... create and set response data
        return response
    }
Notice that this class uses just imports from the HAPI library and not from Camel.

Exposing an HL7 listener using Netty (available from Camel 2.15 onwards)

In the Spring XML file, we configure a Netty4 endpoint to listen for HL7 requests using TCP on port 8888:
<endpoint id="hl7NettyListener" uri="netty4:tcp://localhost:8888?sync=true&amp;encoder=#hl7encoder&amp;decoder=#hl7decoder"/>
sync=true indicates that this listener is synchronous and therefore will return a HL7 response to the caller. The HL7 codec is set up with encoder=#hl7encoder and decoder=#hl7decoder. Note that hl7encoder and hl7decoder are just bean IDs, so they could be named differently. The beans can be set in the Spring XML file:
<bean id="hl7decoder" class="org.apache.camel.component.hl7.HL7MLLPNettyDecoderFactory"/>
<bean id="hl7encoder" class="org.apache.camel.component.hl7.HL7MLLPNettyEncoderFactory"/>
The hl7NettyListener endpoint can then be used in a route as a consumer, as this Java DSL example illustrates:
from("hl7NettyListener").beanRef("patientLookupService");

HL7 Model using java.lang.String or byte[]

The HL7 MLLP codec uses plain String as its data format. Camel uses its Type Converter to convert to/from strings to the HAPI HL7 model objects, but you can use the plain String objects if you prefer, for instance if you wish to parse the data yourself.
As of Camel 2.14.1 you can also let both the Mina and Netty codecs use a plain byte[] as its data format by setting the produceString property to false. The Type Converter is also capable of converting the byte[] to/from HAPI HL7 model objects.

HL7v2 Model using HAPI

The HL7v2 model uses Java objects from the HAPI library. Using this library, you can encode and decode from the EDI format (ER7) that is mostly used with HL7v2.
The sample below is a request to lookup a patient with the patient ID 0101701234.
MSH|^~\\&|MYSENDER|MYRECEIVER|MYAPPLICATION||200612211200||QRY^A19|1234|P|2.4
QRD|200612211200|R|I|GetPatient|||1^RD|0101701234|DEM||
Using the HL7 model, you can work with a ca.uhn.hl7v2.model.Message object, for example to retrieve a patient ID:
Message msg = exchange.getIn().getBody(Message.class);
QRD qrd = (QRD)msg.get("QRD");
String patientId = qrd.getWhoSubjectFilter(0).getIDNumber().getValue();  // 0101701234
This is powerful when combined with the HL7 listener, because you don't have to work with byte[], String or any other simple object formats. You can just use the HAPI HL7v2 model objects. If you know the message type in advance, you can be more type-safe:
QRY_A19 msg = exchange.getIn().getBody(QRY_A19.class);
String patientId = msg.getQRD().getWhoSubjectFilter(0).getIDNumber().getValue();

HL7 DataFormat

The HL7 component ships with a HL7 data format that can be used to marshal or unmarshal HL7 model objects.
  • marshal = from Message to byte stream (can be used when responding using the HL7 MLLP codec)
  • unmarshal = from byte stream to Message (can be used when receiving streamed data from the HL7 MLLP
To use the data format, simply instantiate an instance and invoke the marshal or unmarshal operation in the route builder:
DataFormat hl7 = new HL7DataFormat();
...
from("direct:hl7in").marshal(hl7).to("jms:queue:hl7out");
In the sample above, the HL7 is marshalled from a HAPI Message object to a byte stream and put on a JMS queue. The next example is the opposite:
DataFormat hl7 = new HL7DataFormat();
...
from("jms:queue:hl7out").unmarshal(hl7).to("patientLookupService");
Here we unmarshal the byte stream into a HAPI Message object that is passed to our patient lookup service.
Note
As of HAPI 2.0 (used by Camel 2.11), the HL7v2 model classes are fully serializable. So you can put HL7v2 messages directly into a JMS queue (i.e. without calling marshal() and read them again directly from the queue (i.e. without calling unmarshal()).
Important
As of Camel 2.11, unmarshal does not automatically fix segment separators any more by converting \n to \r. If you need this conversion, org.apache.camel.component.hl7.HL7#convertLFToCR provides a handy Expression for this purpose.
Important
As of Camel 2.14.1, both marshal and unmarshal evaluate the charset provided in the field MSH-18. If this field is empty, by default the charset contained in the corresponding Camel charset property/header is assumed. You can even change this default behaviour by overriding the guessCharsetName method when inheriting from the HL7DataFormat class.
There is a shorthand syntax in Camel for well-known data formats that are commonly used. Then you don't need to create an instance of the HL7DataFormat object:
from("direct:hl7in").marshal().hl7().to("jms:queue:hl7out");
from("jms:queue:hl7out").unmarshal().hl7().to("patientLookupService");

Message Headers

The unmarshal operation adds these fields from the MSH segment as headers on the Camel message:
Key MSH field Example
CamelHL7SendingApplication MSH-3 MYSERVER
CamelHL7SendingFacility MSH-4 MYSERVERAPP
CamelHL7ReceivingApplication MSH-5 MYCLIENT
CamelHL7ReceivingFacility MSH-6 MYCLIENTAPP
CamelHL7Timestamp MSH-7 20071231235900
CamelHL7Security MSH-8 null
CamelHL7MessageType MSH-9-1 ADT
CamelHL7TriggerEvent MSH-9-2 A01
CamelHL7MessageControl MSH-10 1234
CamelHL7ProcessingId MSH-11 P
CamelHL7VersionId MSH-12 2.4
CamelHL7Context -
Camel 2.14: contains the HapiContext that was used to parse the message
CamelHL7Charset MSH-18
Camel 2.14.1: Unicode UTF-8
All headers except CamelHL7Context are String types. If a header value is missing, its value is null.

Options

The HL7 Data Format supports the following options:
Option Default Description
validate true Whether the HAPI Parser should validate the message using the default validation rules. It is recommended to use the parser or hapiContext option and initialize it with the desired HAPI ValidationContext.
parser ca.uhn.hl7v2.parser.GenericParser Custom parser to be used. Must be of type ca.uhn.hl7v2.parser.Parser. Note that GenericParser also allows to parse XML-encoded HL7v2 messages
hapiContext ca.uhn.hl7v2.DefaultHapiContext
Camel 2.14: Custom HAPI context that can define a custom parser, custom ValidationContext etc. This gives you full control over the HL7 parsing and rendering process.

Dependencies

To use HL7 in your Camel routes you'll need to add a dependency on camel-hl7 listed above, which implements this data format.
The HAPI library is been split into a base library and several structure libraries, one for each HL7v2 message version:
By default camel-hl7 only references the HAPI base library. Applications are responsible for including structure libraries themselves. For example, if an application works with HL7v2 message versions 2.4 and 2.5 then the following dependencies must be added:
<dependency>
    <groupId>ca.uhn.hapi</groupId>
    <artifactId>hapi-structures-v24</artifactId>
    <version>2.2</version>
    <!-- use the same version as your hapi-base version -->
</dependency>
<dependency>
    <groupId>ca.uhn.hapi</groupId>
    <artifactId>hapi-structures-v25</artifactId>
    <version>2.2</version>
    <!-- use the same version as your hapi-base version -->
</dependency>
Alternatively, an OSGi bundle containing the base library, all structures libraries and required dependencies (on the bundle classpath) can be downloaded from the central Maven repository.
<dependency>
    <groupId>ca.uhn.hapi</groupId>
    <artifactId>hapi-osgi-base</artifactId>
    <version>2.2</version>
</dependency>

Terser language

HAPI provides a Terser class that provides access to fields using a commonly used terse location specification syntax. The Terser language allows to use this syntax to extract values from messages and to use them as expressions and predicates for filtering, content-based routing etc.
Sample:
import static org.apache.camel.component.hl7.HL7.terser;
...

   // extract patient ID from field QRD-8 in the QRY_A19 message above and put into message header
   from("direct:test1")
      .setHeader("PATIENT_ID",terser("QRD-8(0)-1"))
      .to("mock:test1");
   // continue processing if extracted field equals a message header
   from("direct:test2")
      .filter(terser("QRD-8(0)-1").isEqualTo(header("PATIENT_ID"))
      .to("mock:test2");

HL7 Validation predicate

Often it is preferable first to parse a HL7v2 message and in a separate step validate it against a HAPI ValidationContext.
Sample:
import static org.apache.camel.component.hl7.HL7.messageConformsTo;
import ca.uhn.hl7v2.validation.impl.DefaultValidation;
...

  // Use standard or define your own validation rules
   ValidationContext defaultContext = new DefaultValidation(); 

   // Throws PredicateValidationException if message does not validate
   from("direct:test1").validate(messageConformsTo(defaultContext)).to("mock:test1");

HL7 Validation predicate using the HapiContext (Camel 2.14)

The HAPI Context is always configured with a ValidationContext (or a ValidationRuleBuilder), so you can access the validation rules indirectly. Furthermore, when unmarshalling the HL7DataFormat forwards the configured HAPI context in the CamelHL7Context header, and the validation rules of this context can be easily reused:
import static org.apache.camel.component.hl7.HL7.messageConformsTo;
import static org.apache.camel.component.hl7.HL7.messageConforms
...

  HapiContext hapiContext = new DefaultHapiContext();
  hapiContext.getParserConfiguration().setValidating(false); // don't validate during parsing
  
  // customize HapiContext some more ... e.g. enforce that PID-8 in ADT_A01 messages of version 2.4 is not empty
  ValidationRuleBuilder builder = new ValidationRuleBuilder() {
      @Override
      protected void configure() {
         forVersion(Version.V24)
              .message("ADT", "A01")
              .terser("PID-8", not(empty()));
         }
      };
  hapiContext.setValidationRuleBuilder(builder);

  HL7DataFormat hl7 = new HL7DataFormat();
  hl7.setHapiContext(hapiContext);

  from("direct:test1")
     .unmarshal(hl7)                // uses the GenericParser returned from the HapiContext
     .validate(messageConforms())   // uses the validation rules returned from the HapiContext
                                    // equivalent with .validate(messageConformsTo(hapiContext))
     // route continues from here

HL7 Acknowledgement expression

A common task in HL7v2 processing is to generate an acknowledgement message as response to an incoming HL7v2 message, e.g. based on a validation result. The ack expression lets us accomplish this very elegantly:
import static org.apache.camel.component.hl7.HL7.messageConformsTo;
import static org.apache.camel.component.hl7.HL7.ack;
import ca.uhn.hl7v2.validation.impl.DefaultValidation;
...

  // Use standard or define your own validation rules
   ValidationContext defaultContext = new DefaultValidation(); 

   from("direct:test1")
      .onException(Exception.class)
         .handled(true)
         .transform(ack()) // auto-generates negative ack because of exception in Exchange
         .end()
      .validate(messageConformsTo(defaultContext))
      // do something meaningful here
      ...
      // acknowledgement
      .transform(ack())

More Samples

In the following example, a plain String HL7 request is sent to an HL7 listener that sends back a response:
String line1 = "MSH|^~\\&|MYSENDER|MYRECEIVER|MYAPPLICATION||200612211200||QRY^A19|1234|P|2.4";
String line2 = "QRD|200612211200|R|I|GetPatient|||1^RD|0101701234|DEM||";

StringBuilder in = new StringBuilder();
in.append(line1);
in.append("\n");
in.append(line2);

String out = (String)template.requestBody("mina2:tcp://127.0.0.1:8888?sync=true&codec=#hl7codec", in.toString());
In the next sample, HL7 requests from the HL7 listener are routed to the business logic, which is implemented as plain POJO registered in the registry as hl7service:
public class MyHL7BusinessLogic {

    // This is a plain POJO that has NO imports whatsoever on Apache Camel.
    // its a plain POJO only importing the HAPI library so we can much easier work with the HL7 format.

    public Message handleA19(Message msg) throws Exception {
        // here you can have your business logic for A19 messages
        assertTrue(msg instanceof QRY_A19);
        // just return the same dummy response
        return createADR19Message();
    }

    public Message handleA01(Message msg) throws Exception {
        // here you can have your business logic for A01 messages
        assertTrue(msg instanceof ADT_A01);
        // just return the same dummy response
        return createADT01Message();
    }
}
Then the Camel routes using the RouteBuilder are as follows:
DataFormat hl7 = new HL7DataFormat();
// we setup or HL7 listener on port 8888 (using the hl7codec) and in sync mode so we can return a response
from("mina2:tcp://127.0.0.1:8888?sync=true&codec=#hl7codec")
    // we use the HL7 data format to unmarshal from HL7 stream to the HAPI Message model
    // this ensures that the camel message has been enriched with hl7 specific headers to
    // make the routing much easier (see below)
    .unmarshal(hl7)
    // using choice as the content base router
    .choice()
        // where we choose that A19 queries invoke the handleA19 method on our hl7service bean
        .when(header("CamelHL7TriggerEvent").isEqualTo("A19"))
            .beanRef("hl7service", "handleA19")
            .to("mock:a19")
        // and A01 should invoke the handleA01 method on our hl7service bean
        .when(header("CamelHL7TriggerEvent").isEqualTo("A01")).to("mock:a01")
            .beanRef("hl7service", "handleA01")
            .to("mock:a19")
        // other types should go to mock:unknown
        .otherwise()
            .to("mock:unknown")
    // end choice block
    .end()
    // marshal response back
    .marshal(hl7);
Note that by using the HL7 DataFormat the Camel message headers are populated with the fields from the MSH segment. The headers are particularly useful for filtering or content-based routing as shown in the example above.