Implement legacy collections using glob (#11976)

* feat: support pattern arrays with glob * wip * feat: emulate legacy content collections * Fixes * Lint * Correctly handle legacy data * Fix tests * Switch flag handling * Fix warnings * Add layout warning * Update fixtures * More tests! * Handle empty md files * Lockfile * Dedupe name * Handle data ID unslug * Fix e2e * Clean build * Clean builds in tests * Test fixes * Fix test * Fix typegen * Fix tests * Fixture updates * Test updates * Update changeset * Test * Remove wait in test * Handle race condition * Lock * chore: changes from review * Handle folders without config * lint * Fix test * Update wording for auto-collections * Delete legacyId * Sort another fixture * Rename flag to `legacy.collections` * Apply suggestions from code review Co-authored-by: Sarah Rainsberger <[email protected]> * Changes from review * Apply suggestions from code review Co-authored-by: Sarah Rainsberger <[email protected]> * lockfile * lock --------- Co-authored-by: Sarah Rainsberger <[email protected]>
withastro · Oct 4, 2024 · 4d590a2 · 4d590a2
1 parent 953e6e0
commit 4d590a2
Show file tree

Hide file tree

Showing 117 changed files with 2,167 additions and 506 deletions.
diff --git a/.changeset/quick-onions-leave.md b/.changeset/quick-onions-leave.md
@@ -0,0 +1,47 @@
+---
+'astro': patch
+---
+
+Refactors legacy `content` and `data` collections to use the Content Layer API `glob()` loader for better performance and to support backwards compatibility. Also introduces the `legacy.collections` flag for projects that are unable to update to the new behavior immediately.
+
+:warning: **BREAKING CHANGE FOR LEGACY CONTENT COLLECTIONS** :warning:
+
+By default, collections that use the old types (`content` or `data`) and do not define a `loader` are now implemented under the hood using the Content Layer API's built-in `glob()` loader, with extra backward-compatibility handling.
+
+In order to achieve backwards compatibility with existing `content` collections, the following have been implemented:
+
+- a `glob` loader collection is defined, with patterns that match the previous handling (matches `src/content/<collection name>/**/*.md` and other content extensions depending on installed integrations, with underscore-prefixed files and folders ignored)
+- When used in the runtime, the entries have an ID based on the filename in the same format as legacy collections
+- A `slug` field is added with the same format as before
+- A `render()` method is added to the entry, so they can be called using `entry.render()`
+- `getEntryBySlug` is supported
+
+In order to achieve backwards compatibility with existing `data` collections, the following have been implemented:
+
+- a `glob` loader collection is defined, with patterns that match the previous handling (matches `src/content/<collection name>/**/*{.json,.yaml}` and other data extensions, with underscore-prefixed files and folders ignored)
+- Entries have an ID that is not slugified
+- `getDataEntryById` is supported
+
+While this backwards compatibility implementation is able to emulate most of the features of legacy collections, **there are some differences and limitations that may cause breaking changes to existing collections**:
+
+- In previous versions of Astro, collections would be generated for all folders in `src/content/`, even if they were not defined in `src/content/config.ts`. This behavior is now deprecated, and collections should always be defined in `src/content/config.ts`. For existing collections, these can just be empty declarations (e.g. `const blog = defineCollection({})`) and Astro will implicitly define your legacy collection for you in a way that is compatible with the new loading behavior.
+- The special `layout` field is not supported in Markdown collection entries. This property is intended only for standalone page files located in `src/pages/` and not likely to be in your collection entries. However, if you were using this property, you must now create dynamic routes that include your page styling.
+- Sort order of generated collections is non-deterministic and platform-dependent. This means that if you are calling `getCollection()`, the order in which entries are returned may be different than before. If you need a specific order, you should sort the collection entries yourself.
+- `image().refine()` is not supported. If you need to validate the properties of an image you will need to do this at runtime in your page or component.
+- the `key` argument of `getEntry(collection, key)` is typed as `string`, rather than having types for every entry.
+
+A new legacy configuration flag `legacy.collections` is added for users that want to keep their current legacy (content and data) collections behavior (available in Astro v2 - v4), or who are not yet ready to update their projects:
+
+```js
+// astro.config.mjs
+import { defineConfig } from 'astro/config';
+
+export default defineConfig({
+  legacy: {
+    collections: true
+  }
+});
+```
+
+When set, no changes to your existing collections are necessary, and the restrictions on storing both new and old collections continue to exist: legacy collections (only) must continue to remain in `src/content/`, while new collections using a loader from the Content Layer API are forbidden in that folder.
+
diff --git a/examples/with-markdoc/src/content/config.ts b/examples/with-markdoc/src/content/config.ts
@@ -0,0 +1,5 @@
+import { defineCollection } from 'astro:content';
+
+export const collections = {
+	docs: defineCollection({})
+};
diff --git a/packages/astro/e2e/fixtures/content-collections/src/content/config.ts b/packages/astro/e2e/fixtures/content-collections/src/content/config.ts
@@ -0,0 +1,6 @@
+import { defineCollection } from "astro:content";
+
+
+const posts = defineCollection({});
+
+export const collections = { posts };
diff --git a/packages/astro/src/content/data-store.ts b/packages/astro/src/content/data-store.ts
@@ -34,6 +34,8 @@ export interface DataEntry<TData extends Record<string, unknown> = Record<string
 	 */
 	deferredRender?: boolean;
 	assetImports?: Array<string>;
+	/** @deprecated */
+	legacyId?: string;
 }
 
 /**

diff --git a/packages/astro/src/content/loaders/glob.ts b/packages/astro/src/content/loaders/glob.ts
@@ -35,7 +35,7 @@ function generateIdDefault({ entry, base, data }: GenerateIdOptions): string {
 	if (data.slug) {
 		return data.slug as string;
 	}
-	const entryURL = new URL(entry, base);
+	const entryURL = new URL(encodeURI(entry), base);
 	const { slug } = getContentEntryIdAndSlug({
 		entry: entryURL,
 		contentDir: base,
@@ -55,6 +55,15 @@ function checkPrefix(pattern: string | Array<string>, prefix: string) {
  * Loads multiple entries, using a glob pattern to match files.
  * @param pattern A glob pattern to match files, relative to the content directory.
  */
+export function glob(globOptions: GlobOptions): Loader;
+/** @private */
+export function glob(
+	globOptions: GlobOptions & {
+		/** @deprecated */
+		_legacy?: true;
+	},
+): Loader;
+
 export function glob(globOptions: GlobOptions): Loader {
 	if (checkPrefix(globOptions.pattern, '../')) {
 		throw new Error(
@@ -80,19 +89,21 @@ export function glob(globOptions: GlobOptions): Loader {
 			>();
 
 			const untouchedEntries = new Set(store.keys());
-
+			const isLegacy = (globOptions as any)._legacy;
+			// If global legacy collection handling flag is *not* enabled then this loader is used to emulate them instead
+			const emulateLegacyCollections = !config.legacy.collections;
 			async function syncData(entry: string, base: URL, entryType?: ContentEntryType) {
 				if (!entryType) {
 					logger.warn(`No entry type found for ${entry}`);
 					return;
 				}
-				const fileUrl = new URL(entry, base);
+				const fileUrl = new URL(encodeURI(entry), base);
 				const contents = await fs.readFile(fileUrl, 'utf-8').catch((err) => {
 					logger.error(`Error reading ${entry}: ${err.message}`);
 					return;
 				});
 
-				if (!contents) {
+				if (!contents && contents !== '') {
 					logger.warn(`No contents found for ${entry}`);
 					return;
 				}
@@ -103,6 +114,17 @@ export function glob(globOptions: GlobOptions): Loader {
 				});
 
 				const id = generateId({ entry, base, data });
+				let legacyId: string | undefined;
+
+				if (isLegacy) {
+					const entryURL = new URL(encodeURI(entry), base);
+					const legacyOptions = getContentEntryIdAndSlug({
+						entry: entryURL,
+						contentDir: base,
+						collection: '',
+					});
+					legacyId = legacyOptions.id;
+				}
 				untouchedEntries.delete(id);
 
 				const existingEntry = store.get(id);
@@ -132,6 +154,12 @@ export function glob(globOptions: GlobOptions): Loader {
 					filePath,
 				});
 				if (entryType.getRenderFunction) {
+					if (isLegacy && data.layout) {
+						logger.error(
+							`The Markdown "layout" field is not supported in content collections in Astro 5. Ignoring layout for ${JSON.stringify(entry)}. Enable "legacy.collections" if you need to use the layout field.`,
+						);
+					}
+
 					let render = renderFunctionByContentType.get(entryType);
 					if (!render) {
 						render = await entryType.getRenderFunction(config);
@@ -160,6 +188,7 @@ export function glob(globOptions: GlobOptions): Loader {
 						digest,
 						rendered,
 						assetImports: rendered?.metadata?.imagePaths,
+						legacyId,
 					});
 
 					// todo: add an explicit way to opt in to deferred rendering
@@ -171,9 +200,10 @@ export function glob(globOptions: GlobOptions): Loader {
 						filePath: relativePath,
 						digest,
 						deferredRender: true,
+						legacyId,
 					});
 				} else {
-					store.set({ id, data: parsedData, body, filePath: relativePath, digest });
+					store.set({ id, data: parsedData, body, filePath: relativePath, digest, legacyId });
 				}
 
 				fileToIdMap.set(filePath, id);
@@ -222,7 +252,7 @@ export function glob(globOptions: GlobOptions): Loader {
 					if (isConfigFile(entry)) {
 						return;
 					}
-					if (isInContentDir(entry)) {
+					if (!emulateLegacyCollections && isInContentDir(entry)) {
 						skippedFiles.push(entry);
 						return;
 					}
@@ -240,7 +270,9 @@ export function glob(globOptions: GlobOptions): Loader {
 					? globOptions.pattern.join(', ')
 					: globOptions.pattern;
 
-				logger.warn(`The glob() loader cannot be used for files in ${bold('src/content')}.`);
+				logger.warn(
+					`The glob() loader cannot be used for files in ${bold('src/content')} when legacy mode is enabled.`,
+				);
 				if (skipCount > 10) {
 					logger.warn(
 						`Skipped ${green(skippedFiles.length)} files that matched ${green(patternList)}.`,

diff --git a/packages/astro/src/content/mutable-data-store.ts b/packages/astro/src/content/mutable-data-store.ts
@@ -4,7 +4,7 @@ import { Traverse } from 'neotraverse/modern';
 import { imageSrcToImportId, importIdToSymbolName } from '../assets/utils/resolveImports.js';
 import { AstroError, AstroErrorData } from '../core/errors/index.js';
 import { IMAGE_IMPORT_PREFIX } from './consts.js';
-import { type DataEntry, ImmutableDataStore, type RenderedContent } from './data-store.js';
+import { type DataEntry, ImmutableDataStore } from './data-store.js';
 import { contentModuleToId } from './utils.js';
 
 const SAVE_DEBOUNCE_MS = 500;
@@ -197,7 +197,17 @@ export default new Map([\n${lines.join(',\n')}]);
 			entries: () => this.entries(collectionName),
 			values: () => this.values(collectionName),
 			keys: () => this.keys(collectionName),
-			set: ({ id: key, data, body, filePath, deferredRender, digest, rendered, assetImports }) => {
+			set: ({
+				id: key,
+				data,
+				body,
+				filePath,
+				deferredRender,
+				digest,
+				rendered,
+				assetImports,
+				legacyId,
+			}) => {
 				if (!key) {
 					throw new Error(`ID must be a non-empty string`);
 				}
@@ -244,6 +254,9 @@ export default new Map([\n${lines.join(',\n')}]);
 				if (rendered) {
 					entry.rendered = rendered;
 				}
+				if (legacyId) {
+					entry.legacyId = legacyId;
+				}
 				if (deferredRender) {
 					entry.deferredRender = deferredRender;
 					if (filePath) {
@@ -335,30 +348,7 @@ export interface DataStore {
 		key: string,
 	) => DataEntry<TData> | undefined;
 	entries: () => Array<[id: string, DataEntry]>;
-	set: <TData extends Record<string, unknown>>(opts: {
-		/** The ID of the entry. Must be unique per collection. */
-		id: string;
-		/** The data to store. */
-		data: TData;
-		/** The raw body of the content, if applicable. */
-		body?: string;
-		/** The file path of the content, if applicable. Relative to the site root. */
-		filePath?: string;
-		/** A content digest, to check if the content has changed. */
-		digest?: number | string;
-		/** The rendered content, if applicable. */
-		rendered?: RenderedContent;
-		/**
-		 * If an entry is a deferred, its rendering phase is delegated to a virtual module during the runtime phase.
-		 */
-		deferredRender?: boolean;
-		/**
-		 * Assets such as images to process during the build. These should be files on disk, with a path relative to filePath.
-		 * Any values that use image() in the schema will already be added automatically.
-		 * @internal
-		 */
-		assetImports?: Array<string>;
-	}) => boolean;
+	set: <TData extends Record<string, unknown>>(opts: DataEntry<TData>) => boolean;
 	values: () => Array<DataEntry>;
 	keys: () => Array<string>;
 	delete: (key: string) => void;