Skip to content

A compiler extension to extract abstract representation of Ballerina connector actions and their associated types

License

Notifications You must be signed in to change notification settings

ballerina-platform/module-ballerinax-datamapper

Repository files navigation

Build Daily build codecov

Ballerina Data Mapper Extension

The Ballerina Data Mapper extension is a compiler extension, which extracts an abstract representation of Ballerina connector actions and the associated types. This compiler extension gets executed when compiling a Ballerina connector project. The extension generates two types of files by default. Note that these two types of files get automatically extracted from connectors. Hence, the developer does not need to write them.

Functions File

These files are named as <CLIENT_NAME>_functions.json in which the CLIENT_NAME corresponds to the connector client name. When there are n number of clients existing in a connector, a n number of functions files will get created. Functions files get created only if there are client objects with remote function calls. These files can be used to identify the list of connector actions available in a particular connector as well as to identify what types of records are used as input parameters and return types.

Schema File

These files are named as <RECORD_NAME>_schema.json in which the RECORD_NAME corresponds to a record type defined within the connector module. In order for a schema file to be generated for a record type defined within a connector module, the record has to be one of the following,

  1. An input parameter or a return type from a connector action.
  2. A record type referred by a record, which falls into the first category.

In addition to the above two types of JSON file categories, which get automatically extracted from a connector by default, the connector developer can also specify sample data in the <RECORD_NAME>_data.json files. Note that the use of the term schema here does not correspond to JSON Schema.

Data File

Data files have to follow the structure given below.

{
    "<ORG_NAME>/<MODULE_NAME>:<VERSION_NUMBER>:<RECORD_NAME>": [
        <SAMPLE_JSON_INSTANCE_1>,
        <SAMPLE_JSON_INSTANCE_2>,
        <SAMPLE_JSON_INSTANCE_3>,
        ...,
        <SAMPLE_JSON_INSTANCE_N>
    ]
}

The <SAMPLE_JSON_INSTANCE_i> should have a structure as follows in which the JSON instance has M number of attributes.

{
    "<ATTRIBUTE_NAME_1>" : "<ATTRIBUTE_VALUE_1>",
    "<ATTRIBUTE_NAME_2>" : "<ATTRIBUTE_VALUE_2>",
    "<ATTRIBUTE_NAME_3>" : "<ATTRIBUTE_VALUE_3>",
    ...,
    "<ATTRIBUTE_NAME_M>" : "<ATTRIBUTE_VALUE_M>"
}

An <ATTRIBUTE_VALUE_j> can be a simple literal (string/numeric value), <SAMPLE_JSON_INSTANCE_k>, an array, or null.

Unlike functions files and schema files, which automatically get extracted from the Ballerina Abstract Syntax Tree (AST), data files are user-specified JSON files. Hence, the Data Mapper compiler extension conducts a validation of the content of the data JSON files. The validation process checks whether the data JSON file's attribute names are equivalent to the attribute names used in their corresponding types. Furthermore, it checks whether the same number of attributes exist between the record type and its corresponding data JSON file. In addition to these validations, the compiler extension also extracts sample data nested within a particular data JSON to separate top-level data JSON files.

The extracted JSON files get written to the resources directory within the connector module's folder. However, the functions and schema JSON files are not required to be committed to a Github repository since they get generated automatically from the connector's Ballerina code. However, the data files need to be version controlled.