feat: JSONSchema/Schema converter

all/index.ts

@@ -1,15 +1,21 @@
+export * from "@dpkit/arrow"
 export * from "@dpkit/camtrap"
-export * from "@dpkit/csv"
 export * from "@dpkit/ckan"
 export * from "@dpkit/core"
+export * from "@dpkit/csv"
+export * from "@dpkit/database"
 export * from "@dpkit/datahub"
 export * from "@dpkit/file"
 export * from "@dpkit/folder"
 export * from "@dpkit/github"
 export * from "@dpkit/inline"
+export * from "@dpkit/json"
+export * from "@dpkit/jsonschema"
+export * from "@dpkit/ods"
+export * from "@dpkit/parquet"
 export * from "@dpkit/table"
+export * from "@dpkit/xlsx"
 export * from "@dpkit/zenodo"
-
 // TODO: Enable after migrated away from yauzl-promise (uses native crc32)
 //export * from "@dpkit/zip"
 

all/package.json

@@ -32,6 +32,7 @@
     "@dpkit/github": "workspace:*",
     "@dpkit/inline": "workspace:*",
     "@dpkit/json": "workspace:*",
+    "@dpkit/jsonschema": "workspace:*",
     "@dpkit/ods": "workspace:*",
     "@dpkit/parquet": "workspace:*",
     "@dpkit/table": "workspace:*",

core/dialect/Dialect.ts

@@ -7,14 +7,14 @@ import type { Metadata } from "../general/index.ts"
  */
 export interface Dialect extends Metadata {
   /**
-   * The name of this dialect
+   * JSON schema profile URL for validation
    */
-  name?: string
+  $schema?: string
 
   /**
-   * JSON schema profile URL for validation
+   * The name of this dialect
    */
-  $schema?: string
+  name?: string
 
   /**
    * Whether the file includes a header row with field names

core/package/Package.ts

@@ -7,6 +7,11 @@ import type { Contributor } from "./Contributor.ts"
  * @see https://datapackage.org/standard/data-package/
  */
 export interface Package extends Metadata {
+  /**
+   * URL of profile (optional)
+   */
+  $schema?: string
+
   /**
    * Data resources in this package (required)
    */
@@ -18,11 +23,6 @@ export interface Package extends Metadata {
    */
   name?: string
 
-  /**
-   * Package schema URL for validation
-   */
-  $schema?: string
-
   /**
    * Human-readable title
    */

core/resource/Resource.ts

@@ -9,6 +9,11 @@ import type { Source } from "./Source.ts"
  * @see https://datapackage.org/standard/data-resource/
  */
 export interface Resource extends Metadata {
+  /**
+   * JSON schema profile URL for validation
+   */
+  $schema?: string
+
   /**
    * Unique resource identifier
    * Should use lowercase alphanumeric characters, periods, hyphens, and underscores

core/schema/Schema.ts

@@ -8,14 +8,29 @@ import type { ForeignKey } from "./ForeignKey.ts"
  */
 export interface Schema extends Metadata {
   /**
-   * Fields in this schema (required)
+   * URL of profile (optional)
    */
-  fields: Field[]
+  $schema?: string
 
   /**
-   * URL of schema (optional)
+   * Name of schema (optional)
    */
-  $schema?: string
+  name?: string
+
+  /**
+   * Title of schema (optional)
+   */
+  title?: string
+
+  /**
+   * Description of schema (optional)
+   */
+  description?: string
+
+  /**
+   * Fields in this schema (required)
+   */
+  fields: Field[]
 
   /**
    * Field matching rule (optional)

docs/astro.config.ts

@@ -19,6 +19,7 @@ const PACKAGES = {
   "@dpkit/github": "../github",
   "@dpkit/inline": "../inline",
   "@dpkit/json": "../json",
+  "@dpkit/jsonschema": "../jsonschema",
   "@dpkit/ods": "../ods",
   "@dpkit/parquet": "../parquet",
   "@dpkit/table": "../table",

dpkit/commands/schema/convert.tsx

@@ -0,0 +1,71 @@
+import { denormalizeJsonSchema, normalizeJsonSchema } from "@dpkit/all"
+import { loadDescriptor, saveDescriptor } from "@dpkit/all"
+import { Command, Option } from "commander"
+import { helpConfiguration } from "../../helpers/help.ts"
+import { Session } from "../../helpers/session.ts"
+import * as params from "../../params/index.ts"
+
+const format = new Option("--format <format>", "source schema format").choices([
+  "schema",
+  "jsonschema",
+])
+
+const toFormat = new Option(
+  "--to-format <format>",
+  "target schema format",
+).choices(["schema", "jsonschema"])
+
+export const convertSchemaCommand = new Command("convert")
+  .configureHelp(helpConfiguration)
+  .description("Convert schema between different formats")
+
+  .addArgument(params.positionalDescriptorPath)
+  .addOption(format)
+  .addOption(toFormat)
+  .addOption(params.toPath)
+  .addOption(params.json)
+  .addOption(params.debug)
+
+  .action(async (path, options) => {
+    const session = Session.create({
+      title: "Convert schema",
+      json: options.json || !options.toPath,
+      debug: options.debug,
+    })
+
+    if (!options.format && !options.toFormat) {
+      session.terminate("Either --format or --to-format must be specified")
+      process.exit(1)
+    }
+
+    if (options.format === options.toFormat) {
+      session.terminate("Source and target formats must be different")
+      process.exit(1)
+    }
+
+    const converter =
+      options.format === "schema" || options.toFormat === "jsonschema"
+        ? denormalizeJsonSchema
+        : normalizeJsonSchema
+
+    const source = await session.task("Loading schema", loadDescriptor(path))
+    const target = await session.task(
+      "Converting schema",
+      // @ts-ignore
+      converter(source.descriptor),
+    )
+
+    if (!options.toPath) {
+      session.render(target)
+      return
+    }
+
+    await session.task(
+      "Saving schema",
+      saveDescriptor(target as any, {
+        path: options.toPath,
+      }),
+    )
+
+    session.success(`Converted schema from ${path} to ${options.toPath}`)
+  })

dpkit/commands/schema/index.ts

@@ -1,5 +1,6 @@
 import { Command } from "commander"
 import { helpConfiguration } from "../../helpers/help.ts"
+import { convertSchemaCommand } from "./convert.tsx"
 import { exploreSchemaCommand } from "./explore.tsx"
 import { inferSchemaCommand } from "./infer.tsx"
 import { scriptSchemaCommand } from "./script.tsx"
@@ -10,6 +11,7 @@ export const schemaCommand = new Command("schema")
   .description("Table Schema related commands")
 
   .addCommand(inferSchemaCommand)
+  .addCommand(convertSchemaCommand)
   .addCommand(exploreSchemaCommand)
   .addCommand(scriptSchemaCommand)
   .addCommand(validateSchemaCommand)

jsonschema/README.md

@@ -0,0 +1,3 @@
+# @dpkit/jsonschema
+
+dpkit is a fast data management framework built on top of the Data Package standard and Polars DataFrames. It supports various formats like CSV, JSON, and Parquet and integrates with data platforms such as CKAN, Zenodo, and GitHub. For more information, please visit the [documentation portal](https://dpkit.datist.io).

jsonschema/index.ts

@@ -0,0 +1 @@
+export * from "./schema/index.ts"

jsonschema/package.json

@@ -0,0 +1,33 @@
+{
+  "name": "@dpkit/jsonschema",
+  "type": "module",
+  "version": "0.0.0-development",
+  "exports": "./build/index.js",
+  "license": "MIT",
+  "author": "Evgeny Karev",
+  "repository": "https://github.com/datisthq/dpkit",
+  "description": "Fast TypeScript data management framework built on top of the Data Package standard and Polars DataFrames",
+  "keywords": [
+    "data",
+    "polars",
+    "dataframe",
+    "datapackage",
+    "tableschema",
+    "typescript",
+    "validation",
+    "quality",
+    "fair",
+    "jsonschema"
+  ],
+  "scripts": {
+    "build": "tsc"
+  },
+  "dependencies": {
+    "@dpkit/core": "workspace:*",
+    "@types/json-schema": "^7.0.15",
+    "json-schema": "^0.4.0"
+  },
+  "devDependencies": {
+    "@dpkit/test": "workspace:*"
+  }
+}