Convert json-schema into AWS Glue struct schema (for automating JSON parsing in Glue using AWS CDK)

gistfile1.txt

import request from 'sync-request';
import * as glue from '@aws-cdk/aws-glue';

export function polytectJsonSchemaToGlue(): glue.Type {
    console.log("Getting Polytect reference schema...")
    const res = request('GET', 'https://raw.githubusercontent.com/polyverse/polytect/master/reference/schema.json');
    const body = res.getBody();
    const schema = JSON.parse(body.toString());
    return recursiveGlueColumns(schema, schema.definitions);
}

function recursiveGlueColumns(schema: any, definitions: any): glue.Type {
    if (typeof(schema.type) !== 'undefined') {
        if (schema.type == 'object') {
            let columns: glue.Column[] = [];
            for (let propertyName in schema.properties) {
                const property = schema.properties[propertyName];
                columns.push({
                    name: propertyName,
                    //comment: property.description,
                    type: recursiveGlueColumns(property, definitions),
                });
            }
            return glue.Schema.struct(columns);
        } else if (schema.type == 'string') {
            return glue.Schema.STRING;
        } else if (schema.type == 'boolean') {
            return glue.Schema.BOOLEAN;
        } else if (schema.type == 'integer') {
            return glue.Schema.INTEGER;
        } else if (Array.isArray(schema.type) && schema.type[0] == 'string' && schema.type[1] == 'null') {
            return glue.Schema.STRING;
        } else if (Array.isArray(schema.type) && schema.type[0] == 'integer' && schema.type[1] == 'null') {
            return glue.Schema.INTEGER;
        } else {
            console.log("Reached the typed end!! ", JSON.stringify(schema));
        }
    } else if (typeof(schema["$ref"]) !== 'undefined') {
        const ref = schema["$ref"];
        const refName = ref.split("/")[2];
        const refSchema = definitions[refName];
        return recursiveGlueColumns(refSchema, definitions);
    } else if (typeof(schema.allOf) !== 'undefined') {
        // get the schema elsewhere...
        if (schema.allOf.length > 1 && schema.allOf[1].type != "null") {
            console.log("Unable to handle multiple schema allOfs: ", schema);
            throw "Error!!!";
        }
        if (schema.allOf.length < 1) {
            console.log("Unable to handle zero schema allOfs: ", schema);
            throw "Error!!!";
        }
        return recursiveGlueColumns(schema.allOf[0], definitions);
    } else if (typeof(schema.anyOf) !== 'undefined') {
        if (schema.anyOf.length < 1) {
            console.log("Unable to handle zero schema anyOfs: ", schema);
            throw "Error!!!";
        }

        if (typeof(schema.anyOf[0]["$ref"]) !== 'undefined') {
            return recursiveGlueColumns(schema.anyOf[0], definitions);
        }

        // get the schema elsewhere...
        let columns: glue.Column[] = [];
        schema.anyOf.forEach((struct: any) => {
            if (typeof(struct.required) !== 'undefined') {
                const propertyName = struct.required[0];
                columns.push({
                    name: propertyName,
                    //comment: struct.description,
                    type: recursiveGlueColumns(struct.properties[propertyName], definitions),
                });
            } else if (typeof(struct.enum) !== 'undefined') {
                columns.push({
                    name: struct.enum[0],
                    //comment: struct.description,
                    type: glue.Schema.STRING,
                });
            } else if (typeof(struct.type) !== 'undefined' && struct.type == "null") {
                // do nothing
            } else {
                console.log("Reached the anyOf end!! ", JSON.stringify(schema));
                throw "Error!";
            }
        });
        return glue.Schema.struct(columns);
    } else if (typeof(schema.enum) !== 'undefined') {
        return glue.Schema.STRING;
    } else {
        console.log("Reached the end VERY !! ", JSON.stringify(schema));
        throw "Error!";
    }
    return glue.Schema.STRING;
}