chore: Standardize files

src/lib/LDWorkbenchConfiguration.d.ts → src/LDWorkbenchConfiguration.d.ts

@@ -9,7 +9,7 @@
  * JSON Schema for LDWorkbench.
  * It helps with the writing of the configuration needed to run LDWorkbench pipelines.
  */
-export interface LDWorkbenchConfiguration {
+export interface configuration {
   /**
    * The name of your pipeline, it must be unique over all your configurations.
    */

src/cliArgs.ts → src/cli.ts

@@ -27,16 +27,16 @@ program
   )
   .version(version());
 program.parse();
-export const cliArgs: {
+export const cli: {
   config: string;
   pipeline?: string;
   stage?: string;
   silent?: boolean;
   init?: boolean;
 } = program.opts();
 
-if (cliArgs.init !== undefined) {
-  if (Object.values(cliArgs).length !== 2) {
+if (cli.init !== undefined) {
+  if (Object.values(cli).length !== 2) {
     error(
       'The --init flag can not be used in conjunction with other CLI arguments.'
     );

src/configuration.d.ts

@@ -0,0 +1,158 @@
+/* eslint-disable */
+/**
+ * This file was automatically generated by json-schema-to-typescript.
+ * DO NOT MODIFY IT BY HAND. Instead, modify the source JSONSchema file,
+ * and run json-schema-to-typescript to regenerate this file.
+ */
+
+/**
+ * JSON Schema for LDWorkbench.
+ * It helps with the writing of the configuration needed to run LDWorkbench pipelines.
+ */
+export interface Configuration {
+  /**
+   * The name of your pipeline. It must be unique over all your configurations.
+   */
+  name: string;
+  /**
+   * An optional description for your pipeline.
+   */
+  description?: string;
+  /**
+   * An optional base directory for files referenced by `file://...` paths. This defaults to the directory that contains the YAML configuration file.
+   */
+  baseDir?: string;
+  /**
+   * The file where the final result of your pipeline is saved.
+   */
+  destination?: string;
+  /**
+   * A pipeline stage consists of an iterator and one or more generators.
+   *
+   * @minItems 1
+   */
+  stages: [
+    {
+      /**
+       * The name of the stage. It must be unique within the pipeline.
+       */
+      name: string;
+      iterator: {
+        /**
+         * SPARQL SELECT query that returns a `$this` binding for each URI that will be passed to the generator(s). Either an inline string (`SELECT $this WHERE {...}`) or a reference to a file (`file://...`) that contains the query.
+         */
+        query: string;
+        /**
+         * SPARQL endpoint for the iterator. If it starts with `file://`, a local RDF file is queried. If omitted the result of the previous stage is used.
+         */
+        endpoint?: string;
+        /**
+         * Number of `$this` bindings retrieved per query. Defaults to the LIMIT value of your iterator query or 10 if no LIMIT is present.
+         */
+        batchSize?: number;
+        /**
+         * Human-readable time delay for requests to the the iterator’s SPARQL endpoint (e.g. `5ms`, `100 milliseconds`, `1s`).
+         */
+        delay?: string;
+      };
+      /**
+       * @minItems 1
+       */
+      generator: [
+        {
+          /**
+           * SPARQL CONSTRUCT query that takes a `$this` binding from the iterator and generates triples for it. Either an inline string (`CONSTRUCT $this schema:name ?name WHERE {$this ...}`) or a reference to a file (`file://...`) that contains the query.
+           */
+          query: string;
+          /**
+           * The SPARQL endpoint for the generator. If it starts with `file://`, a local RDF file is queried. If omitted, the endpoint of the iterator is used.
+           */
+          endpoint?: string;
+          /**
+           * Overrule the generator's behaviour of fetching results for 10 bindings of `$this` per request.
+           */
+          batchSize?: number;
+        },
+        ...{
+          /**
+           * SPARQL CONSTRUCT query that takes a `$this` binding from the iterator and generates triples for it. Either an inline string (`CONSTRUCT $this schema:name ?name WHERE {$this ...}`) or a reference to a file (`file://...`) that contains the query.
+           */
+          query: string;
+          /**
+           * The SPARQL endpoint for the generator. If it starts with `file://`, a local RDF file is queried. If omitted, the endpoint of the iterator is used.
+           */
+          endpoint?: string;
+          /**
+           * Overrule the generator's behaviour of fetching results for 10 bindings of `$this` per request.
+           */
+          batchSize?: number;
+        }[]
+      ];
+      /**
+       * The optional path where the results are saved. If omitted, a temporary file will be created.
+       */
+      destination?: string;
+    },
+    ...{
+      /**
+       * The name of the stage. It must be unique within the pipeline.
+       */
+      name: string;
+      iterator: {
+        /**
+         * SPARQL SELECT query that returns a `$this` binding for each URI that will be passed to the generator(s). Either an inline string (`SELECT $this WHERE {...}`) or a reference to a file (`file://...`) that contains the query.
+         */
+        query: string;
+        /**
+         * SPARQL endpoint for the iterator. If it starts with `file://`, a local RDF file is queried. If omitted the result of the previous stage is used.
+         */
+        endpoint?: string;
+        /**
+         * Number of `$this` bindings retrieved per query. Defaults to the LIMIT value of your iterator query or 10 if no LIMIT is present.
+         */
+        batchSize?: number;
+        /**
+         * Human-readable time delay for requests to the the iterator’s SPARQL endpoint (e.g. `5ms`, `100 milliseconds`, `1s`).
+         */
+        delay?: string;
+      };
+      /**
+       * @minItems 1
+       */
+      generator: [
+        {
+          /**
+           * SPARQL CONSTRUCT query that takes a `$this` binding from the iterator and generates triples for it. Either an inline string (`CONSTRUCT $this schema:name ?name WHERE {$this ...}`) or a reference to a file (`file://...`) that contains the query.
+           */
+          query: string;
+          /**
+           * The SPARQL endpoint for the generator. If it starts with `file://`, a local RDF file is queried. If omitted, the endpoint of the iterator is used.
+           */
+          endpoint?: string;
+          /**
+           * Overrule the generator's behaviour of fetching results for 10 bindings of `$this` per request.
+           */
+          batchSize?: number;
+        },
+        ...{
+          /**
+           * SPARQL CONSTRUCT query that takes a `$this` binding from the iterator and generates triples for it. Either an inline string (`CONSTRUCT $this schema:name ?name WHERE {$this ...}`) or a reference to a file (`file://...`) that contains the query.
+           */
+          query: string;
+          /**
+           * The SPARQL endpoint for the generator. If it starts with `file://`, a local RDF file is queried. If omitted, the endpoint of the iterator is used.
+           */
+          endpoint?: string;
+          /**
+           * Overrule the generator's behaviour of fetching results for 10 bindings of `$this` per request.
+           */
+          batchSize?: number;
+        }[]
+      ];
+      /**
+       * The optional path where the results are saved. If omitted, a temporary file will be created.
+       */
+      destination?: string;
+    }[]
+  ];
+}

src/lib/File.class.ts → src/file.ts

@@ -6,11 +6,11 @@ import {
   mkdirSync,
   createReadStream,
 } from 'fs';
-import {isFile, isFilePathString} from '../utils/guards.js';
+import {isFile, isFilePathString} from './utils/guards.js';
 import {dirname} from 'path';
 import chalk from 'chalk';
 import {type Ora} from 'ora';
-import type Pipeline from './Pipeline.class.js';
+import type Pipeline from './pipeline.js';
 import {pipeline as streamPipeline} from 'stream/promises';
 
 export default class File {

src/lib/Generator.class.ts → src/generator.ts

@@ -1,12 +1,12 @@
 import type {ConstructQuery} from 'sparqljs';
-import type Stage from './Stage.class.js';
-import getSPARQLQuery from '../utils/getSPARQLQuery.js';
+import type Stage from './stage.js';
+import getSPARQLQuery from './utils/getSPARQLQuery.js';
 import type {Quad, NamedNode} from '@rdfjs/types';
-import getSPARQLQueryString from '../utils/getSPARQLQueryString.js';
-import getEndpoint from '../utils/getEndpoint.js';
+import getSPARQLQueryString from './utils/getSPARQLQueryString.js';
+import getEndpoint from './utils/getEndpoint.js';
 import type {Endpoint, QueryEngine, QuerySource} from './types.js';
-import getEngine from '../utils/getEngine.js';
-import getEngineSource from '../utils/getEngineSource.js';
+import getEngine from './utils/getEngine.js';
+import getEngineSource from './utils/getEngineSource.js';
 import EventEmitter from 'node:events';
 
 const DEFAULT_BATCH_SIZE = 10;

src/lib/Iterator.class.ts → src/iterator.ts

@@ -1,14 +1,14 @@
 import EventEmitter from 'node:events';
 import type {SelectQuery} from 'sparqljs';
-import type Stage from './Stage.class.js';
+import type Stage from './stage.js';
 import type {NamedNode} from '@rdfjs/types';
-import getSPARQLQuery from '../utils/getSPARQLQuery.js';
+import getSPARQLQuery from './utils/getSPARQLQuery.js';
 import {type Bindings} from '@comunica/types';
-import getSPARQLQueryString from '../utils/getSPARQLQueryString.js';
-import getEndpoint from '../utils/getEndpoint.js';
+import getSPARQLQueryString from './utils/getSPARQLQueryString.js';
+import getEndpoint from './utils/getEndpoint.js';
 import type {Endpoint, QueryEngine, QuerySource} from './types.js';
-import getEngine from '../utils/getEngine.js';
-import getEngineSource from '../utils/getEngineSource.js';
+import getEngine from './utils/getEngine.js';
+import getEngineSource from './utils/getEngineSource.js';
 import parse from 'parse-duration';
 
 const DEFAULT_LIMIT = 10;

src/main.ts

@@ -3,25 +3,25 @@ import inquirer from 'inquirer';
 import chalk from 'chalk';
 import {error} from './utils/error.js';
 import version from './utils/version.js';
-import type {LDWorkbenchConfiguration} from './lib/LDWorkbenchConfiguration.js';
+import type {Configuration} from './configuration.js';
 import loadPipelines from './utils/loadPipelines.js';
-import Pipeline from './lib/Pipeline.class.js';
-import {cliArgs} from './cliArgs.js';
+import Pipeline from './pipeline.js';
+import {cli} from './cli.js';
 
 console.info(
   chalk.bold(`Welcome to LD Workbench version ${chalk.cyan(version())}`)
 );
 
 async function main(): Promise<void> {
-  const pipelines = loadPipelines(cliArgs.config ?? './pipelines/');
+  const pipelines = loadPipelines(cli.config ?? './pipelines/');
   const names = [...pipelines.keys()];
-  let configuration: LDWorkbenchConfiguration | undefined;
+  let configuration: Configuration | undefined;
 
-  if (cliArgs.pipeline !== undefined) {
-    configuration = pipelines.get(cliArgs.pipeline);
+  if (cli.pipeline !== undefined) {
+    configuration = pipelines.get(cli.pipeline);
     if (configuration === undefined) {
       error(
-        `No pipeline named “${cliArgs.pipeline}” was found.`,
+        `No pipeline named “${cli.pipeline}” was found.`,
         2,
         `Valid pipeline names are: ${names.map(name => `"${name}"`).join(', ')}`
       );
@@ -57,8 +57,8 @@ async function main(): Promise<void> {
 
   try {
     const pipeline = new Pipeline(configuration, {
-      startFromStageName: cliArgs.stage,
-      silent: cliArgs.silent,
+      startFromStageName: cli.stage,
+      silent: cli.silent,
     });
     await pipeline.run();
   } catch (e) {

src/lib/Pipeline.class.ts → src/pipeline.ts

@@ -1,17 +1,17 @@
 import ora, {Ora} from 'ora';
 import kebabcase from 'lodash.kebabcase';
-import type {LDWorkbenchConfiguration} from './LDWorkbenchConfiguration.js';
+import type {Configuration} from './configuration.js';
 import chalk from 'chalk';
-import Stage from './Stage.class.js';
-import formatDuration from '../utils/formatDuration.js';
+import Stage from './stage.js';
+import formatDuration from './utils/formatDuration.js';
 import {millify} from 'millify';
-import File from './File.class.js';
+import File from './file.js';
 import path from 'node:path';
 import * as fs from 'node:fs';
-import {isFilePathString, isTriplyDBPathString} from '../utils/guards.js';
-import TriplyDB from './TriplyDB.class.js';
+import {isFilePathString, isTriplyDBPathString} from './utils/guards.js';
+import TriplyDB from './triply-db.js';
 import prettyMilliseconds from 'pretty-ms';
-import {memoryConsumption} from '../utils/memory.js';
+import {memoryConsumption} from './utils/memory.js';
 interface PipelineOptions {
   startFromStageName?: string;
   silent?: boolean;
@@ -28,7 +28,7 @@ class Pipeline {
   private readonly opts?: PipelineOptions;
 
   public constructor(
-    private readonly $configuration: LDWorkbenchConfiguration,
+    private readonly $configuration: Configuration,
     pipelineOptions?: PipelineOptions
   ) {
     //  create data folder:
@@ -111,7 +111,7 @@ class Pipeline {
     }
   }
 
-  public get configuration(): LDWorkbenchConfiguration {
+  public get configuration(): Configuration {
     return this.$configuration;
   }