perf: Reuse one shared context parser + document loader across all files

lib/loading/ComponentRegistry.ts

@@ -1,5 +1,6 @@
-import type { Readable } from 'stream';
+import type { Readable } from 'node:stream';
 import type * as RDF from '@rdfjs/types';
+import type { ContextParser } from 'jsonld-context-parser';
 import type { Resource, RdfObjectLoader } from 'rdf-object';
 import type { Logger } from 'winston';
 import { RdfParser } from '../rdf/RdfParser';
@@ -20,6 +21,7 @@ export class ComponentRegistry {
   private readonly componentResources: Record<string, Resource>;
   private readonly skipContextValidation: boolean;
   private readonly remoteContextLookups: boolean;
+  private readonly contextParser?: ContextParser;
 
   public constructor(options: IComponentLoaderRegistryOptions) {
     this.moduleState = options.moduleState;
@@ -28,6 +30,7 @@ export class ComponentRegistry {
     this.componentResources = options.componentResources;
     this.skipContextValidation = options.skipContextValidation;
     this.remoteContextLookups = options.remoteContextLookups;
+    this.contextParser = options.contextParser;
   }
 
   /**
@@ -37,7 +40,7 @@ export class ComponentRegistry {
    */
   public async registerAvailableModules(): Promise<void> {
     await Promise.all(Object.values(this.moduleState.componentModules)
-      .flatMap(Object.values)
+      .flatMap(x => Object.values(x))
       .map((moduleResourceUrl: string) => this.registerModule(moduleResourceUrl)));
   }
 
@@ -54,6 +57,7 @@ export class ComponentRegistry {
       logger: this.logger,
       skipContextValidation: this.skipContextValidation,
       remoteContextLookups: this.remoteContextLookups,
+      contextParser: this.contextParser,
     }));
   }
 
@@ -123,4 +127,8 @@ export interface IComponentLoaderRegistryOptions {
   componentResources: Record<string, Resource>;
   skipContextValidation: boolean;
   remoteContextLookups: boolean;
+  /**
+   * An optional shared, cache-bearing JSON-LD context parser reused across all module files.
+   */
+  contextParser?: ContextParser;
 }

lib/loading/ComponentsManagerBuilder.ts

@@ -2,6 +2,9 @@ import type { Resource } from 'rdf-object';
 import { RdfObjectLoader } from 'rdf-object';
 import type { Logger } from 'winston';
 import { createLogger, format, transports } from 'winston';
+
+// eslint-disable-next-line import/extensions
+import contextJson from '../../components/context.json';
 import { ComponentsManager } from '../ComponentsManager';
 import { ConfigConstructorPool } from '../construction/ConfigConstructorPool';
 import type { IConfigConstructorPool } from '../construction/IConfigConstructorPool';
@@ -11,6 +14,7 @@ import { ConfigPreprocessorComponent } from '../preprocess/ConfigPreprocessorCom
 import { ConfigPreprocessorComponentMapped } from '../preprocess/ConfigPreprocessorComponentMapped';
 import { ConfigPreprocessorOverride } from '../preprocess/ConfigPreprocessorOverride';
 import { ParameterHandler } from '../preprocess/ParameterHandler';
+import { RdfParser } from '../rdf/RdfParser';
 import type { LogLevel } from '../util/LogLevel';
 import { ComponentRegistry } from './ComponentRegistry';
 import { ComponentRegistryFinalizer } from './ComponentRegistryFinalizer';
@@ -21,25 +25,25 @@ import type { IModuleState } from './ModuleStateBuilder';
 /**
  * Builds {@link ComponentsManager}'s based on given options.
  */
-export class ComponentsManagerBuilder<Instance = any> {
+export class ComponentsManagerBuilder<TInstance = any> {
   private readonly mainModulePath: string;
   private readonly componentLoader: (registry: ComponentRegistry) => Promise<void>;
   private readonly configLoader: (registry: ConfigRegistry) => Promise<void>;
-  private readonly constructionStrategy: IConstructionStrategy<Instance>;
+  private readonly constructionStrategy: IConstructionStrategy<TInstance>;
   private readonly dumpErrorState: boolean;
   private readonly logger: Logger;
   private readonly moduleState?: IModuleState;
   private readonly skipContextValidation: boolean;
   private readonly typeChecking: boolean;
   private readonly remoteContextLookups: boolean;
 
-  public constructor(options: IComponentsManagerBuilderOptions<Instance>) {
+  public constructor(options: IComponentsManagerBuilderOptions<TInstance>) {
     this.mainModulePath = options.mainModulePath;
-    this.componentLoader = options.moduleLoader || (async registry => registry.registerAvailableModules());
-    this.configLoader = options.configLoader || (async() => {
+    this.componentLoader = options.moduleLoader ?? (async registry => registry.registerAvailableModules());
+    this.configLoader = options.configLoader ?? (async() => {
       // Do nothing
     });
-    this.constructionStrategy = options.constructionStrategy || new ConstructionStrategyCommonJs({ req: require });
+    this.constructionStrategy = options.constructionStrategy ?? new ConstructionStrategyCommonJs({ req: require });
     this.dumpErrorState = options.dumpErrorState === undefined ? true : Boolean(options.dumpErrorState);
     this.logger = ComponentsManagerBuilder.createLogger(options.logLevel);
     this.moduleState = options.moduleState;
@@ -73,14 +77,14 @@ export class ComponentsManagerBuilder<Instance = any> {
   public static createObjectLoader(): RdfObjectLoader {
     return new RdfObjectLoader({
       uniqueLiterals: true,
-      context: require('../../components/context.json'),
+      context: contextJson,
     });
   }
 
   /**
    * @return A new instance of {@link ComponentsManager}.
    */
-  public async build(): Promise<ComponentsManager<Instance>> {
+  public async build(): Promise<ComponentsManager<TInstance>> {
     // Initialize module state
     let moduleState: IModuleState;
     if (this.moduleState) {
@@ -95,6 +99,16 @@ export class ComponentsManagerBuilder<Instance = any> {
     // Initialize object loader with built-in context
     const objectLoader: RdfObjectLoader = ComponentsManagerBuilder.createObjectLoader();
 
+    // Create a single, cache-bearing JSON-LD context parser (with one shared prefetched document
+    // loader) that is reused across every component and config file. This avoids re-loading and
+    // re-normalizing the shared well-known @contexts once per file.
+    const contextParser = RdfParser.createSharedContextParser({
+      contexts: moduleState.contexts,
+      logger: this.logger,
+      remoteContextLookups: this.remoteContextLookups,
+      skipContextValidation: this.skipContextValidation,
+    });
+
     // Load modules
     this.logger.info(`Initiating component loading`);
     const componentResources: Record<string, Resource> = {};
@@ -105,6 +119,7 @@ export class ComponentsManagerBuilder<Instance = any> {
       componentResources,
       skipContextValidation: this.skipContextValidation,
       remoteContextLookups: this.remoteContextLookups,
+      contextParser,
     });
     await this.componentLoader(componentRegistry);
     const componentFinalizer = new ComponentRegistryFinalizer({
@@ -122,14 +137,15 @@ export class ComponentsManagerBuilder<Instance = any> {
       logger: this.logger,
       skipContextValidation: this.skipContextValidation,
       remoteContextLookups: this.remoteContextLookups,
+      contextParser,
     });
     await this.configLoader(configRegistry);
     this.logger.info(`Loaded configs`);
 
     // Build constructor pool
     const runTypeConfigs = {};
     const parameterHandler = new ParameterHandler({ objectLoader, typeChecking: this.typeChecking });
-    const configConstructorPool: IConfigConstructorPool<Instance> = new ConfigConstructorPool({
+    const configConstructorPool: IConfigConstructorPool<TInstance> = new ConfigConstructorPool({
       objectLoader,
       configPreprocessors: [
         new ConfigPreprocessorOverride({
@@ -156,7 +172,7 @@ export class ComponentsManagerBuilder<Instance = any> {
       moduleState,
     });
 
-    return new ComponentsManager<Instance>({
+    return new ComponentsManager<TInstance>({
       moduleState,
       objectLoader,
       componentResources,
@@ -168,7 +184,7 @@ export class ComponentsManagerBuilder<Instance = any> {
   }
 }
 
-export interface IComponentsManagerBuilderOptions<Instance> {
+export interface IComponentsManagerBuilderOptions<TInstance> {
   /* ----- REQUIRED FIELDS ----- */
   /**
    * Absolute path to the package root from which module resolution should start.
@@ -192,7 +208,7 @@ export interface IComponentsManagerBuilderOptions<Instance> {
    * A strategy for constructing instances.
    * Defaults to {@link ConstructionStrategyCommonJs}.
    */
-  constructionStrategy?: IConstructionStrategy<Instance>;
+  constructionStrategy?: IConstructionStrategy<TInstance>;
   /**
    * If the error state should be dumped into `componentsjs-error-state.json`
    * after failed instantiations.

lib/loading/ConfigRegistry.ts

@@ -1,5 +1,6 @@
-import type { Readable } from 'stream';
+import type { Readable } from 'node:stream';
 import type * as RDF from '@rdfjs/types';
+import type { ContextParser } from 'jsonld-context-parser';
 import type { RdfObjectLoader, Resource } from 'rdf-object';
 import { termToString } from 'rdf-string';
 import type { Logger } from 'winston';
@@ -15,13 +16,15 @@ export class ConfigRegistry {
   private readonly logger: Logger;
   private readonly skipContextValidation: boolean;
   private readonly remoteContextLookups: boolean;
+  private readonly contextParser?: ContextParser;
 
   public constructor(options: IConfigLoaderRegistryOptions) {
     this.moduleState = options.moduleState;
     this.objectLoader = options.objectLoader;
     this.logger = options.logger;
     this.skipContextValidation = options.skipContextValidation;
     this.remoteContextLookups = options.remoteContextLookups;
+    this.contextParser = options.contextParser;
   }
 
   /**
@@ -38,6 +41,7 @@ export class ConfigRegistry {
       logger: this.logger,
       skipContextValidation: this.skipContextValidation,
       remoteContextLookups: this.remoteContextLookups,
+      contextParser: this.contextParser,
     }));
   }
 
@@ -62,6 +66,7 @@ export class ConfigRegistry {
   ): Promise<void> {
     // Create ad-hoc resource
     const configResource = this.objectLoader.createCompactedResource({
+      // eslint-disable-next-line ts/naming-convention
       '@id': configId,
       types: componentTypeIri,
     });
@@ -85,4 +90,8 @@ export interface IConfigLoaderRegistryOptions {
   logger: Logger;
   skipContextValidation: boolean;
   remoteContextLookups: boolean;
+  /**
+   * An optional shared, cache-bearing JSON-LD context parser reused across all config files.
+   */
+  contextParser?: ContextParser;
 }

lib/rdf/RdfParser.ts

@@ -1,26 +1,61 @@
-import { createReadStream } from 'fs';
-import type { Readable } from 'stream';
+import { createReadStream, promises as fs } from 'node:fs';
+import type { Readable } from 'node:stream';
 import type * as RDF from '@rdfjs/types';
+import * as JsonLdContextParser from 'jsonld-context-parser';
+import { ContextParser } from 'jsonld-context-parser';
 import type { ParseOptions } from 'rdf-parse';
-import rdfParser from 'rdf-parse';
+import { rdfParser } from 'rdf-parse';
 import type { Logger } from 'winston';
 import { PrefetchedDocumentLoader } from './PrefetchedDocumentLoader';
 import { RdfStreamIncluder } from './RdfStreamIncluder';
-// Import syntax only works in Node > 12
-const fs = require('fs').promises;
 
 /**
  * Parses a data stream to a triple stream.
  */
 export class RdfParser {
+  /**
+   * Create a single, cache-bearing JSON-LD context parser that can be reused across all files
+   * that are parsed during a component/config load.
+   *
+   * By default, Components.js constructs a fresh {@link PrefetchedDocumentLoader} and (indirectly,
+   * via jsonld-streaming-parser) a fresh `ContextParser` for every `.jsonld` file. As a result the
+   * well-known `@context`s (e.g. the shared `componentsjs` and `@solid/community-server` contexts)
+   * are re-loaded and re-normalized once per file - during a Community Solid Server boot the
+   * `componentsjs` context alone is normalized ~979 times.
+   *
+   * Passing the returned parser to every parse (see {@link RdfParserOptions#contextParser}) shares a
+   * single {@link PrefetchedDocumentLoader}, a single raw-document cache, and a single normalized
+   * {@link ContextCache} across all files, so the well-known contexts are loaded once and (once the
+   * shared cache can be hit) normalized once instead of once per file.
+   *
+   * @param options Options describing the prefetched contexts and validation behaviour.
+   */
+  public static createSharedContextParser(options: ISharedContextParserOptions): ContextParser {
+    const documentLoader = new PrefetchedDocumentLoader({
+      contexts: options.contexts ?? {},
+      logger: options.logger,
+      remoteContextLookups: options.remoteContextLookups,
+    });
+    const parserOptions: Record<string, any> = {
+      documentLoader,
+      skipValidation: options.skipContextValidation,
+    };
+    // Attach a shared normalized-context cache when the installed jsonld-context-parser exposes one
+    // (feature-detected so this remains compatible with versions that predate the cache).
+    if ((<any> JsonLdContextParser).ContextCache) {
+      parserOptions.contextCache = new (<any> JsonLdContextParser).ContextCache();
+    }
+    return new ContextParser(parserOptions);
+  }
+
   /**
    * Parses the given stream into RDF quads.
    * @param textStream A text stream.
    * @param options Parsing options.
    */
   public parse(textStream: NodeJS.ReadableStream, options: RdfParserOptions): RDF.Stream & Readable {
     // Parsing libraries don't work as expected if path contains backslashes
-    options.path = options.path.replace(/\\+/gu, '/');
+    options.path = options.path.replaceAll(/\\+/gu, '/');
 
     if (!options.baseIRI) {
       // Try converting path to URL using defined import paths
@@ -44,10 +79,12 @@ export class RdfParser {
     }
 
     // Set JSON-LD parser options
-    (<any> options)['@comunica/actor-rdf-parse-jsonld:parserOptions'] = {
-      // Override the JSON-LD document loader
+    const jsonLdParserOptions: Record<string, any> = {
+      // Override the JSON-LD document loader.
+      // This is always provided so that a jsonld-streaming-parser that does not (yet) support an
+      // injected context parser still uses the prefetched contexts and remote-lookup protection.
       documentLoader: new PrefetchedDocumentLoader({
-        contexts: options.contexts || {},
+        contexts: options.contexts ?? {},
         logger: options.logger,
         path: options.path,
         remoteContextLookups: options.remoteContextLookups,
@@ -57,6 +94,15 @@ export class RdfParser {
       // If JSON-LD context validation should be skipped
       skipContextValidation: options.skipContextValidation,
     };
+    if (options.contextParser) {
+      // Reuse one shared, cache-bearing context parser (which carries its own single prefetched
+      // document loader) across all files, so the shared @contexts are only loaded/normalized once
+      // instead of once per file. This is honoured by a jsonld-streaming-parser that supports the
+      // `contextParser` option; on older versions the option is ignored and the per-file
+      // `documentLoader` above is used, preserving the previous behaviour.
+      jsonLdParserOptions.contextParser = options.contextParser;
+    }
+    (<any> options)['@comunica/actor-rdf-parse-jsonld:parserOptions'] = jsonLdParserOptions;
 
     // Execute parsing
     const quadStream = rdfParser.parse(textStream, options);
@@ -131,4 +177,30 @@ export type RdfParserOptions = ParseOptions & {
    * If allowed, only a warning is emitted.
    */
   remoteContextLookups?: boolean;
+  /**
+   * An optional shared JSON-LD context parser to reuse across all files.
+   * When provided, this (cache-bearing) parser and its prefetched document loader are used for all
+   * context resolution instead of constructing a new document loader (and context parser) per file.
+   * Create one with {@link RdfParser#createSharedContextParser}.
+   */
+  contextParser?: ContextParser;
 };
+
+export interface ISharedContextParserOptions {
+  /**
+   * The cached JSON-LD contexts (id -> context document).
+   */
+  contexts?: Record<string, any>;
+  /**
+   * An optional logger, used to warn on remote context lookups.
+   */
+  logger?: Logger;
+  /**
+   * If remote context lookups are allowed.
+   */
+  remoteContextLookups?: boolean;
+  /**
+   * If JSON-LD context validation should be skipped.
+   */
+  skipContextValidation?: boolean;
+}