Node module that is bridged with the Java Hapi HL7 library.
With this library, you can:
- Transform plain JavaScript objects into HL7 messages.
- Transform HL7 messages into plain JavaScript objects.
- Benefit from Java Hapi HL7 being used under the hood, with much coverage e.x.: ADT, BAR, DFT, MDM, MFN, ORM, ORU, QRY, RAS, RDE, RGV, SIU, etc...
This HL7 solution works by making calls to the Java Hapi HL7 library under the hood. Hapi is the gold-standard implementation of HL7 parsing.
When this project started, many folks were using Java 8. Unfortunately, the Java dependencies in this project relies on that version to function. While it is a major inconvenience to downgrade your Java version, you may find success with Jabba to manage your version (simiar to rvm, nvm, etc).
In addition, version 3.1.0 and above of this project expects your Node version to be greater than 12. Check out Node Version Manager to manage versions.
Docker Solution
This project also offers a Docker solution that will completely abstract away the Java 8 dependency. The container runs a server with 2 simple endpoints that match what would be used in the programmatic context. See "Running with Docker" a bit below.
var NodeHL7Complete = require('node-hl7-complete');
var nodeHL7Instance = new NodeHL7Complete();
// Stolen from http://www.dt7.com/cdc/sampmsgs.html
var hl7Message = '';
hl7Message += 'MSH|^~\\&|||||||VXU^V04|19970522MA53|P|2.3.1|||AL\r';
hl7Message += 'PID|||221345671^^^^SS||KENNEDY^JOHN^FITZGERALD^JR|BOUVIER^^^^^^M|19900607|M|||^^^^MA^^^BLD\r';
hl7Message += 'NK1|1|KENNEDY^JACQUELINE^LEE|32^MOTHER^HL70063\r';
hl7Message += 'RXA|0|1|19900607|19900607|08^HEPB-PEDIATRIC/ADOLESCENT^CVX|.5|ML^^ISO+||||||||MRK12345||MSD^MERCK^MVX\r';
nodeHL7Instance.hl7ToJs(hl7Message, function(error, data) {
if (error) { console.error(error); return; }
console.log(data);
});
var jsData = {
'VXU_V04': {
'MSH': {
'MSH.1': '|',
'MSH.2': '^~\\&',
'MSH.9': {
'MSG.1': 'VXU',
'MSG.2': 'V04'
},
'MSH.10': '19970522MA53',
'MSH.11': {
'PT.1': 'P'
},
'MSH.12': {
'VID.1': '2.3.1'
},
'MSH.15': 'AL'
},
'PID': {
'PID.3': {
'CX.1': 221345671,
'CX.5': 'SS'
},
'PID.5': {
'XPN.1': {
'FN.1': 'KENNEDY'
},
'XPN.2': 'JOHN',
'XPN.3': 'FITZGERALD',
'XPN.4': 'JR'
},
'PID.6': {
'XPN.1': {
'FN.1': 'BOUVIER'
},
'XPN.7': 'M'
},
'PID.7': {
'TS.1': 19900607
},
'PID.8': 'M',
'PID.11': {
'XAD.5': 'MA',
'XAD.8': 'BLD'
}
},
'NK1': {
'NK1.1': 1,
'NK1.2': {
'XPN.1': {
'FN.1': 'KENNEDY'
},
'XPN.2': 'JACQUELINE',
'XPN.3': 'LEE'
},
'NK1.3': {
'CE.1': 32,
'CE.2': 'MOTHER',
'CE.3': 'HL70063'
}
},
'VXU_V04.ORCRXARXROBXNTE': {
'RXA': {
'RXA.1': 0,
'RXA.2': 1,
'RXA.3': {
'TS.1': 19900607
},
'RXA.4': {
'TS.1': 19900607
},
'RXA.5': {
'CE.1': 08,
'CE.2': 'HEPB-PEDIATRIC/ADOLESCENT',
'CE.3': 'CVX'
},
'RXA.6': 0.5,
'RXA.7': {
'CE.1': 'ML',
'CE.3': 'ISO+'
},
'RXA.15': 'MRK12345',
'RXA.17': {
'CE.1': 'MSD',
'CE.2': 'MERCK',
'CE.3': 'MVX'
}
}
}
}
};
nodeHL7Instance.jsToHl7('VXU_V04', jsData, function(error, data) {
if (error) { console.error(error); return; }
console.log(data.split('\r').join('\n'));
});... these two console.logs yield the following:
MSH|^~\&|||||||VXU^V04|19970522MA53|P|2.3.1|||AL
PID|||221345671^^^^SS||KENNEDY^JOHN^FITZGERALD^JR|BOUVIER^^^^^^M|19900607|M|||^^^^MA^^^BLD
NK1|1|KENNEDY^JACQUELINE^LEE|32^MOTHER^HL70063
RXA|0|1|19900607|19900607|8^HEPB-PEDIATRIC/ADOLESCENT^CVX|0.5|ML^^ISO+||||||||MRK12345||MSD^MERCK^MVX
{ VXU_V04:
{ MSH: [ [Object] ],
PID: [ [Object] ],
NK1: [ [Object] ],
'VXU_V04.ORCRXARXROBXNTE': [ [Object] ] } }
As far as data-integrity goes, see the following graphic to see what the original HL7 message looked like and what the transformed JS->HL7 message looks like (very small diff):
By default, all HL7-to-JS messages will be validated for correctness. If you are receiving HL7 messages that are somewhat valid and wish to skip validation, you can use nodeHL7Instance.setStrictMode(false).
- clone this repo
cd dockerdocker build -t hl7_docker .docker run -dp 8000:8000 hl7_dockercd examples- HL7->JSON test call:
./hl7ToJson.sh - JSON->HL7 test call:
./jsonToHl7.sh
Note: Outside of PM2, you may wish to use a more formal process manager solution.
- JavaScript keys must be in quotes because 'PID.3', for example, cannot be used in dot-notation.
- The first argument to
NodeHL7Complete.jsToHl7must be the root XML node name. For exampleVXU_V04. MSHsegment must be escaped like so^~\\&- The Java layer is strictly speaking HL7 and XML... XML is converted to JavaScript objects in the
index.jsfile. Was getting a ton of exceptions when trying to bring inGSON. It appears that Hapi is really best used with plain XML for now. - Segments must have
\rat the end of them. - If an
erroris returned, it will be the Java exception. - The
node-javamodule is what embeds Java and does the bridge work. Really cool. - Functional test HL7 data was found by Googling around. Credit is applied by the name of the folder holding the test data files. Cerner/NIST/Realm test data was provided by https://github.com/ruby-hl7/ruby-hl7/tree/master/test_data.
workingDirmust be passed in when requiring the module because the java dependencies have to be absolutely referenced.
> npm run unit_tests> npm run functional_tests
...unit test coverage reports live in coverage/.
- Test on Windows.
- Add
.gitignorefor all of the Spring Tool Suite/Eclipse/Intellij files as well as NPM files. - Quiet down the warnings on the Java side (logging stuff).
- Implicitly determine the root XML node name instead of passing it in as the first argument to
NodeHL7Complete.jsToHl7. - Look into ways to get around the
^~\\&escaping. - Figure out how to make Travis CI run the functional tests even though it is specified to be a node environment and doesn't have Java installed at the moment.
MIT