feat: Have ExtensionBasedMapper handle extensions correctly

src/storage/ExtensionBasedMapper.ts

@@ -1,11 +1,13 @@
+import { promises as fsPromises } from 'fs';
 import { posix } from 'path';
-import { types } from 'mime-types';
+import * as mime from 'mime-types';
 import type { ResourceIdentifier } from '../ldp/representation/ResourceIdentifier';
 import { APPLICATION_OCTET_STREAM, TEXT_TURTLE } from '../util/ContentTypes';
 import { ConflictHttpError } from '../util/errors/ConflictHttpError';
 import { NotFoundHttpError } from '../util/errors/NotFoundHttpError';
+import { UnsupportedHttpError } from '../util/errors/UnsupportedHttpError';
 import { trimTrailingSlashes } from '../util/Util';
-import type { FileIdentifierMapper } from './FileIdentifierMapper';
+import type { FileIdentifierMapper, ResourceLink } from './FileIdentifierMapper';
 
 const { join: joinPath, normalize: normalizePath } = posix;
 
@@ -22,52 +24,136 @@ export interface ResourcePath {
   documentName?: string;
 }
 
+/**
+ * A mapper that stores the content-type of resources in the file path extension.
+ * In case the extension of the identifier does not correspond to the correct content-type,
+ * a new extension will be appended (with a `$` in front of it).
+ * E.g. if the path is `input.ttl` with content-type `text/plain`, the path would actually be `input.ttl$.txt`.
+ * This new extension is stripped again when generating an identifier.
+ */
 export class ExtensionBasedMapper implements FileIdentifierMapper {
-  private readonly base: string;
-  private readonly prootFilepath: string;
+  private readonly baseRequestURI: string;
+  private readonly rootFilepath: string;
   private readonly types: Record<string, any>;
 
   public constructor(base: string, rootFilepath: string, overrideTypes = { acl: TEXT_TURTLE, metadata: TEXT_TURTLE }) {
-    this.base = base;
-    this.prootFilepath = rootFilepath;
-    this.types = { ...types, ...overrideTypes };
-  }
-
-  public get baseRequestURI(): string {
-    return trimTrailingSlashes(this.base);
-  }
-
-  public get rootFilepath(): string {
-    return trimTrailingSlashes(this.prootFilepath);
+    this.baseRequestURI = trimTrailingSlashes(base);
+    this.rootFilepath = trimTrailingSlashes(normalizePath(rootFilepath));
+    this.types = { ...mime.types, ...overrideTypes };
   }
 
   /**
-   * Strips the baseRequestURI from the identifier and checks if the stripped base URI matches the store's one.
-   * @param identifier - Incoming identifier.
+   * Maps the given resource identifier / URL to a file path.
+   * Determines the content-type if no content-type was provided.
+   * For containers the content-type input gets ignored.
+   * @param identifier - The input identifier.
+   * @param contentType - The (optional) content-type of the resource.
    *
-   * @throws {@link NotFoundHttpError}
-   * If the identifier does not match the baseRequestURI path of the store.
-   *
-   * @returns Absolute path of the file.
+   * @returns A ResourceLink with all the necessary metadata.
    */
-  public mapUrlToFilePath(identifier: ResourceIdentifier, id = ''): string {
-    return this.getAbsolutePath(this.getRelativePath(identifier), id);
+  public async mapUrlToFilePath(identifier: ResourceIdentifier, contentType?: string): Promise<ResourceLink> {
+    let path = this.getRelativePath(identifier);
+
+    if (!path.startsWith('/')) {
+      throw new UnsupportedHttpError('URL needs a / after the base.');
+    }
+
+    if (path.includes('/..')) {
+      throw new UnsupportedHttpError('Disallowed /.. segment in URL.');
+    }
+
+    path = this.getAbsolutePath(path);
+
+    // Container
+    if (identifier.path.endsWith('/')) {
+      return {
+        identifier,
+        filePath: path,
+      };
+    }
+
+    // Would conflict with how new extensions get stored
+    if (/\$\.\w+$/u.test(path)) {
+      throw new UnsupportedHttpError('Identifiers cannot contain a dollar sign before their extension.');
+    }
+
+    // Existing file
+    if (!contentType) {
+      const [ , folder, documentName ] = /^(.*\/)(.*)$/u.exec(path)!;
+
+      let fileName: string | undefined;
+      try {
+        const files = await fsPromises.readdir(folder);
+        fileName = files.find(
+          (file): boolean =>
+            file.startsWith(documentName) && /^(?:\$\..+)?$/u.test(file.slice(documentName.length)),
+        );
+      } catch {
+        // Parent folder does not exist (or is not a folder)
+        throw new NotFoundHttpError();
+      }
+
+      // File doesn't exist
+      if (!fileName) {
+        throw new NotFoundHttpError();
+      }
+
+      return {
+        identifier,
+        filePath: joinPath(folder, fileName),
+        contentType: this.getContentTypeFromExtension(fileName),
+      };
+    }
+
+    // If the extension of the identifier matches a different content-type than the one that is given,
+    // we need to add a new extension to match the correct type.
+    if (contentType !== this.getContentTypeFromExtension(path)) {
+      const extension = mime.extension(contentType);
+      if (!extension) {
+        throw new UnsupportedHttpError(`Unsupported content-type ${contentType}.`);
+      }
+      path = `${path}$.${extension}`;
+    }
+
+    return {
+      identifier,
+      filePath: path,
+      contentType,
+    };
   }
 
   /**
-   * Strips the rootFilepath path from the filepath and adds the baseRequestURI in front of it.
-   * @param path - The file path.
+   * Maps the given file path to an URL and determines the content-type
+   * @param filePath - The input file path.
+   * @param isContainer - If the path corresponds to a file.
    *
-   * @throws {@Link Error}
-   * If the file path does not match the rootFilepath path of the store.
-   *
-   * @returns Url of the file.
+   * @returns A ResourceLink with all the necessary metadata.
    */
-  public mapFilePathToUrl(path: string): string {
-    if (!path.startsWith(this.rootFilepath)) {
-      throw new Error(`File ${path} is not part of the file storage at ${this.rootFilepath}.`);
+  public async mapFilePathToUrl(filePath: string, isContainer: boolean): Promise<ResourceLink> {
+    if (!filePath.startsWith(this.rootFilepath)) {
+      throw new Error(`File ${filePath} is not part of the file storage at ${this.rootFilepath}.`);
     }
-    return this.baseRequestURI + path.slice(this.rootFilepath.length);
+
+    let relative = filePath.slice(this.rootFilepath.length);
+    if (isContainer) {
+      return {
+        identifier: { path: encodeURI(this.baseRequestURI + relative) },
+        filePath,
+      };
+    }
+
+    // Files
+    const extension = this.getExtension(relative);
+    const contentType = this.getContentTypeFromExtension(relative);
+    if (extension && relative.endsWith(`$.${extension}`)) {
+      relative = relative.slice(0, -(extension.length + 2));
+    }
+
+    return {
+      identifier: { path: encodeURI(this.baseRequestURI + relative) },
+      filePath,
+      contentType,
+    };
   }
 
   /**
@@ -76,9 +162,19 @@ export class ExtensionBasedMapper implements FileIdentifierMapper {
    *
    * @returns Content type of the file.
    */
-  public getContentTypeFromExtension(path: string): string {
+  private getContentTypeFromExtension(path: string): string {
+    const extension = this.getExtension(path);
+    return (extension && this.types[extension.toLowerCase()]) || APPLICATION_OCTET_STREAM;
+  }
+
+  /**
+   * Extracts the extension (without dot) from a path.
+   * Custom functin since `path.extname` does not work on all cases (e.g. ".acl")
+   * @param path - Input path to parse.
+   */
+  private getExtension(path: string): string | null {
     const extension = /\.([^./]+)$/u.exec(path);
-    return (extension && this.types[extension[1].toLowerCase()]) || APPLICATION_OCTET_STREAM;
+    return extension && extension[1];
   }
 
   /**
@@ -88,7 +184,7 @@ export class ExtensionBasedMapper implements FileIdentifierMapper {
    *
    * @returns Absolute path of the file.
    */
-  public getAbsolutePath(path: string, identifier = ''): string {
+  private getAbsolutePath(path: string, identifier = ''): string {
     return joinPath(this.rootFilepath, path, identifier);
   }
 
@@ -105,7 +201,7 @@ export class ExtensionBasedMapper implements FileIdentifierMapper {
     if (!identifier.path.startsWith(this.baseRequestURI)) {
       throw new NotFoundHttpError();
     }
-    return identifier.path.slice(this.baseRequestURI.length);
+    return decodeURI(identifier.path).slice(this.baseRequestURI.length);
   }
 
   /**
@@ -116,7 +212,7 @@ export class ExtensionBasedMapper implements FileIdentifierMapper {
    * @throws {@link ConflictHttpError}
    * If the root identifier is passed.
    *
-   * @returns A ResourcePath object containing path and (optional) slug fields.
+   * @returns A ResourcePath object containing (absolute) path and (optional) slug fields.
    */
   public extractDocumentName(identifier: ResourceIdentifier): ResourcePath {
     const [ , containerPath, documentName ] = /^(.*\/)([^/]+\/?)?$/u.exec(this.getRelativePath(identifier)) ?? [];
@@ -125,9 +221,9 @@ export class ExtensionBasedMapper implements FileIdentifierMapper {
       throw new ConflictHttpError('Container with that identifier already exists (root).');
     }
     return {
-      containerPath: normalizePath(containerPath),
+      containerPath: this.getAbsolutePath(normalizePath(containerPath)),
 
-      // If documentName is not undefined, return normalized documentName
+      // If documentName is defined, return normalized documentName
       documentName: typeof documentName === 'string' ? normalizePath(documentName) : undefined,
     };
   }

src/storage/FileIdentifierMapper.ts

@@ -1,21 +1,40 @@
 import type { ResourceIdentifier } from '../ldp/representation/ResourceIdentifier';
 
+export interface ResourceLink {
+  /**
+   * Identifier of a resource.
+   */
+  identifier: ResourceIdentifier;
+  /**
+   * File path of a resource.
+   */
+  filePath: string;
+  /**
+   * Content-type for a data resource (not defined for containers).
+   */
+  contentType?: string;
+}
+
 /**
  * Supports mapping a file to an URL and back.
  */
 export interface FileIdentifierMapper {
   /**
-   * Maps the given file path to an URL.
-   * @param file - The input file path.
+   * Maps the given file path to an URL and determines the content-type
+   * @param filePath - The input file path.
+   * @param isContainer - If the path corresponds to a file.
    *
-   * @returns The URL as a string.
+   * @returns A ResourceLink with all the necessary metadata.
    */
-  mapFilePathToUrl: (filePath: string) => string;
+  mapFilePathToUrl: (filePath: string, isContainer: boolean) => Promise<ResourceLink>;
   /**
    * Maps the given resource identifier / URL to a file path.
-   * @param url - The input URL.
+   * Determines the content-type if no content-type was provided.
+   * For containers the content-type input gets ignored.
+   * @param identifier - The input identifier.
+   * @param contentType - The (optional) content-type of the resource.
    *
-   * @returns The file path as a string.
+   * @returns A ResourceLink with all the necessary metadata.
    */
-  mapUrlToFilePath: (identifier: ResourceIdentifier) => string;
+  mapUrlToFilePath: (identifier: ResourceIdentifier, contentType?: string) => Promise<ResourceLink>;
 }