consumeRemoteDataModel.md

How to Consume a CDS Data Model from a Different Git Repository

When you wish to consume a CDS data model defined in some other repository, then the following steps provide an outline of the approach.

1) Referencing an External Git Repository

The Web IDE project stored in the public SAP repository cloud-sample-spaceflight is packaged as an NPM module. This means we can use NPM's normal module dependency mechanism to reference it.

1. Ensure that the repository you wish to access is available on a public Git server. In this particular example, we wish to consume the CDS data model found in the public SAP repository cloud-sample-spaceflight.

2. Look into the definition of the CDS data model you wish to consume and make a note of its namespace. In our case, the namespace is teched.flight.trip (see line 6 of flight-model.cds)

3. Edit the top-level package.json file of your project to include an NPM dependency to the external repository.

   {
     "name": "spaceflight-model-hana"
   , "version": "0.1.0"
   , "dependencies": {
       "@sap/cds": "2.x"
     , "spaceflight-model": "https://github.com/SAP/cloud-sample-spaceflight"
     }
   , "scripts": {
       "build": "cds build --clean && reuseTableData"
     }
   , "license": "ISC"
   }

   You are free to choose any dependency name that is appropriate. Here, we are using the name spaceflight-model followed by the URL of the public Git repository.

2) Copying the Instructions to Load Tables with Data

One important detail here concerns the fact that in the original repository, the data model contains additional instructions to populate the tables with data at the time the database is deployed.

These instructions are stored in the file db/src/csv/Data.hdbtabledata of the external repository, and the actual data to be loaded into the database is stored in the various .csv files found in the same folder.

At the moment however, when a CDS data model from a remote Git repository is referenced, the instructions to populate the database tables (in any .hdbtabledata files) and the corresponding .csv files are not copied automatically.

To solve this problem, in the scripts object of package.json shown above, you can see that the build property contains not only the normal instruction to start the CDS compiler (cds build --clean), but also a further instruction to run a script called reuseTableData.

The name reuseTableData is declared in the bin object of the top-level package.json file of the external repository. This means that the name reuseTableData is available in the path used by npm install and can now be referenced by any other dependant NPM module.

The result of running this script is that the .hdbtabledata and .csv files are copied from the node_modules/spaceflight-model/db folder into our db/src/gen folder.

3) Using the Referenced Data Model

We have now done three things:

1. We have declared an NPM dependency on an external repository that contains a CDS Data Model
2. We have given this dependency the name spaceflight-model
3. We have invoked a workaround script called reuseTableData to copy the instructions for loading our database tables into our local db/src/gen folder

What remains is now to reference the spaceflight-model in our own .cds file.

In the db folder of our project, there is a single .cds file. If you created the project manually, then this file will be called data-model.cds, otherwise if you cloned the template repository, it will be called index.cds.

Either way, this file only needs to contain the following single line:

using teched.flight.trip from 'spaceflight-model/db';

All we must ensure is that the name spaceflight-model used here, matches the dependency name defined in package.json. This is all that is needed to import the data model from the db folder inside the NPM dependency called spaceflight-model.

4) So What Happens When We Run the CDS Compiler?

When we now run the CDS compiler on our project, the following things will happen:

1. Selecting Build -> Build CDS from the project's top-level context menu invokes the npm install command
2. npm reads the top-level package.json file and discovers some dependencies
3. These dependencies are imported into the local project and appear as sub-folders underneath the node_modules folder. This is now where the spaceflight-model module appears
4. npm then invokes the scripts.build command found in package.json.
   1. Firstly, this invokes the CDS compiler that transforms all the copied .cds files in the db folder into files suitable for building HANA database tables. (These files then appear in the db/src/gen folder)
   2. Secondly, the reuseTableData script is invoked to copy all the .hdbtabledata and .csv files from the node_modules/spaceflight-model/db directory into the local db/src/gen folder.

All the necessary data model files have been copied into our project and it is now ready to be deployed to HANA.

Deployment to HANA is achieved by right-clicking on the db/ folder name, and selecting Build -> Build

5) By the Way...

As an aside, since our particular data model is contained within a single namespace, and we wish to consume everything in this data model, it is actually unnecessary to provide the name of the namespace in the index.cds file.

Instead, of writing:

using teched.flight.trip from 'spaceflight-model/db';

We could equally well have written:

using from 'spaceflight-model/db';

because this means use the entire data model found in the spaceflight-model/db folder.