UNPKG

ts-to-mongo-schema

Version:

Using the Typescript Compiler API, this program parses through Typescript, to generate a JSON Schema that conforms to MongoDB's bson format. Useful when generating large schemas, and being unable to use a javascript or typescript library like Moongoose.

137 lines (117 loc) 3.97 kB
## Introduction This program allows the user to convert Typescript defined interfaces or type alias declarations, to MongoDB bson type schemas. This is achieved through the steps below ## Program Logic <details> <summary>Click To Expand!</summary> <br> The program traverses all nodes that the interface or type depends on, according to the steps below. 1. Generate a file's AST through the [Typescript Compiler API](https://github.com/microsoft/TypeScript/wiki/Using-the-Compiler-API) 2. Recursively parse through each AST node to either: - Extract the property types - Search another file for an imported module 4. Map the extracted AST property types to supported MongoDB bson types 5. Combine the property types into one large MongoDB bson schema </details> ## Setup <details> <summary>Click To Expand!</summary> <br> ```typescript type GenerateSchema = { //Path to the tsconfig.json of the target project configPath: string; //Exact string name of type or interface declaration identifier: string; //Path to the file the identifier is located filePath: string; //typescript file extension extension: ".tsx" | ".ts"; /* Provide custom map that contains the name of the custom generic as a string, and a function that returns the bson schema value of that generic */ resolveCustomGenerics?: { [key: string]: (params: any) => any }; }; ``` Example: ```typescript import { generateSchema, ResolveCustomParams } from "ts-to-mongo-schema"; import * as fs from "fs"; //paths const projectPath = "../../testProject"; const configPath = projectPath + "/tsconfig.json"; const filePath = projectPath + "/src/types/testFile.tsx"; const bsonSchema = generateSchema({ configPath: configPath, identifier: "Person", filePath: filePath, extension: '.tsx', }); ``` </details> ## Error Management <details> <summary>Click To Expand!</summary> <br> If the program cannot parse the property type, an empty object will be returned in the property type's place. The user can then modify this manually, in the generated schema. Example Typescript: ```typescript type ArrayOneOrMore<T> = { 0: T; } & Array<T>; interface Person{ name: string; interests: ArrayOneOrMore<string> } ``` BSON Schema Result: ```typescript { bsonType: 'object' properties: { name: {bsonType: string} //empty object interests: {} } required: [ name, interests ] } ``` #### Note: This will usually occur if the interface or type depends on a custom generic, or an imported type from a third-party library. If this is the case, please use the resolveCustomGenerics function to provide a custom value. This outlined below. </details> ## Providing Custom Values for Custom Generics <details> <summary>Click To Expand!</summary> <br> When the program returns too many empty objects for property values, there could be an unsupported custom generic that does not allow for the extraction of properties. However, that does not mean all hope is lost. Find the offending generics and pass in a map to help the program identify them, and parse them according to custom logic ```typescript type ResolveCustomParams = { propertiesPerArg?: any[]; combinedProperties?: { [key: string]: any }; }; const bsonSchema = generateSchema({ configPath: configPath, identifier: "Person", filePath: filePath, extension: '.tsx', resolveCustomGenerics: { //the test file contains a custom generic declared as ArrayOneOrMore. //Therefore, the key is the name of the generic, and attached function //returns the custom value for that generic ArrayOneOrMore: (props: ResolveCustomParams) => { return { bsonType: "array", items: props.combinedProperties, minItems: 1, }; }, }, }); ``` </details>