Merge pull request #148 from thegalactiks/improve-sitemap

feat: improve sitemap generation

examples/astro/src/assets/.gitkeep

@@ -0,0 +1 @@
+/var/data/workspace/thegalactiks/explorer/examples/contentlayer/assets/.gitkeep

examples/contentlayer/content/websites/example.com.mdx → examples/contentlayer/content/websites/galactiks.com.mdx

@@ -1,7 +1,7 @@
 ---
 name: Website name
 description: Website description.
-identifier: example.com
-url: https://example.com/
+identifier: galactiks.com
+url: https://www.galactiks.com/
 dateCreated: 1970-01-01
 ---

packages/adapters/astro/package.json

@@ -51,15 +51,15 @@
     "@astrojs/partytown": "~2.0",
     "@astrojs/react": "~3.0",
     "@astrojs/rss": "~4.0",
-    "@astrojs/sitemap": "~3.1",
+    "@galactiks/sitemap": "^0.1.0",
     "astro": "~4.4",
     "astro-robots-txt": "~1.0"
   },
   "devDependencies": {
     "@astrojs/partytown": "2.0.4",
     "@astrojs/react": "3.0.10",
     "@astrojs/rss": "4.0.5",
-    "@astrojs/sitemap": "3.1.1",
+    "@galactiks/sitemap": "workspace:^",
     "@types/debug": "4.1.12",
     "@types/react": "18.2.58",
     "@types/react-dom": "18.2.19",

packages/adapters/astro/src/index.ts

@@ -53,6 +53,13 @@ export default function createPlugin(
         updateConfig,
         addWatchFile,
       }) => {
+        let trailingSlash = config.trailingSlash;
+        if (trailingSlash === 'ignore') {
+          trailingSlash = config.build.format === 'directory' ? 'always' : 'never';
+        }
+
+        setConfig('trailingSlash', config.trailingSlash);
+
         assetsPath = join(fileURLToPath(config.srcDir), 'assets');
         galactiksConfig = setConfig('content.assets', assetsPath);
 
@@ -61,6 +68,7 @@ export default function createPlugin(
 
         updateConfig({
           site: galactiksConfig.webManifest.start_url,
+          trailingSlash,
           prefetch: true,
           vite: {
             resolve: {

packages/adapters/astro/src/pages.ts

@@ -6,11 +6,11 @@ import {
 
 export async function getStaticPaths() {
   return (await getPages({ inLanguages: getLanguages() }))
-    .filter((_p) => _p.path && _p.path !== '/')
     .map((page) => ({
-      params: { path: page.path.slice(1) },
+      params: { path: page.path.endsWith('/') ? page.path.slice(0, -1) : page.path },
       props: { page },
-    }));
+    }))
+    .filter(page => page.params.path && page.params.path !== '/');
 }
 
 export function getIndexPage() {

packages/adapters/astro/src/plugins/sitemap.ts

@@ -0,0 +1,112 @@
+import path from 'node:path';
+import { fileURLToPath } from 'node:url';
+import type { AstroConfig, AstroIntegration } from 'astro';
+import { getDefaultLanguage } from '@galactiks/config';
+import { type Content, getPageByURL } from '@galactiks/explorer';
+import { generateSitemaps } from '@galactiks/sitemap';
+
+const sitemapIndexOutput = 'sitemap-index.xml';
+
+const createPlugin = (): AstroIntegration => {
+  let config: AstroConfig;
+
+  return {
+    name: '@galactiks/astro-integration/sitemap',
+
+    hooks: {
+      'astro:config:done': async ({ config: cfg }) => {
+        config = cfg;
+      },
+
+      'astro:build:done': async ({ dir, routes, pages, logger }) => {
+        if (!config.site) {
+          logger.warn(
+            'The Sitemap integration requires the `site` astro.config option. Skipping.'
+          );
+          return;
+        }
+
+        let finalSiteUrl: URL;
+        if (config.site) {
+          finalSiteUrl = new URL(config.base, config.site);
+        } else {
+          console.warn(
+            'The Sitemap integration requires the `site` astro.config option. Skipping.'
+          );
+          return;
+        }
+
+        let pageUrls = pages
+          .map((p) => {
+            if (p.pathname !== '' && !finalSiteUrl.pathname.endsWith('/'))
+              finalSiteUrl.pathname += '/';
+            if (p.pathname.startsWith('/')) p.pathname = p.pathname.slice(1);
+            const fullPath = finalSiteUrl.pathname + p.pathname;
+            return new URL(fullPath, finalSiteUrl).href;
+          });
+
+        const routeUrls = routes.reduce<string[]>((urls, r) => {
+          if (r.type !== 'page') return urls;
+
+          /**
+           * Dynamic URLs have entries with `undefined` pathnames
+           */
+          if (r.pathname) {
+            // `finalSiteUrl` may end with a trailing slash
+            // or not because of base paths.
+            let fullPath = finalSiteUrl.pathname;
+            if (fullPath.endsWith('/')) fullPath += r.generate(r.pathname).substring(1);
+            else fullPath += r.generate(r.pathname);
+
+            const newUrl = new URL(fullPath, finalSiteUrl).href;
+
+            if (config.trailingSlash === 'never') {
+              urls.push(newUrl);
+            } else if (config.build.format === 'directory' && !newUrl.endsWith('/')) {
+              urls.push(newUrl + '/');
+            } else {
+              urls.push(newUrl);
+            }
+          }
+
+          return urls;
+        }, []);
+
+        pageUrls = Array.from(new Set([...pageUrls, ...routeUrls]));
+        logger.info(`Generating sitemap for ${pageUrls.length} pages`);
+        logger.info(pageUrls.join('\n'));
+
+        const contentPages = (
+          await Promise.all(
+            pageUrls.map(getPageByURL)
+          )
+        ).filter((page) => page !== undefined) as Content[];
+        if (contentPages.length === 0) {
+          logger.warn(`No pages found!\n\`${sitemapIndexOutput}\` not created.`);
+          return;
+        }
+
+        const destDir = fileURLToPath(dir);
+        await generateSitemaps({
+          destinationDir: destDir,
+          hostname: finalSiteUrl.href,
+          pages: contentPages,
+          defaultLanguage: getDefaultLanguage(),
+          publication: {
+            name: 'Galactiks',
+          },
+        });
+        // await simpleSitemapAndIndex({
+        //   hostname: finalSiteUrl.href,
+        //   destinationDir: destDir,
+        //   sourceData: urlData,
+        //   limit: entryLimit,
+        //   gzip: false,
+        // });
+        logger.info(`\`${sitemapIndexOutput}\` created at \`${path.relative(process.cwd(), destDir)}\``);
+      },
+    },
+  };
+};
+
+export default createPlugin;

packages/adapters/astro/src/preset.ts

@@ -1,29 +1,15 @@
 import type { AstroIntegration } from 'astro';
 import partytown from '@astrojs/partytown';
 import react from '@astrojs/react';
-import sitemap from '@astrojs/sitemap';
-import { getDefaultLanguage, getLanguages } from '@galactiks/config';
 import robotsTxt from 'astro-robots-txt';
 
-import { sitemapSerialize } from './sitemap.js';
+import sitemap from './plugins/sitemap.js';
 
 export const integrationsPreset = (): AstroIntegration[] => {
-  const defaultLanguage = getDefaultLanguage();
-
   return [
     react(),
     partytown(),
-    sitemap({
-      i18n: defaultLanguage
-        ? {
-            defaultLocale: defaultLanguage,
-            locales: Object.fromEntries(
-              getLanguages().map((lang) => [lang, lang])
-            ),
-          }
-        : undefined,
-      serialize: sitemapSerialize(),
-    }),
+    sitemap(),
     robotsTxt(),
   ];
 };

packages/config/src/config.ts

@@ -29,7 +29,7 @@ const galactiksConfigFileSchema = z.object({
   locales: localesSchema.optional(),
   template: z.string(),
   analytics: analyticsConfigSchema.optional(),
-  trailingSlash: z.enum(['always', 'never']).optional(),
+  trailingSlash: z.enum(['ignore', 'always', 'never']).optional(),
   pages: z
     .object({
       articles: pagesObjectItemSchema,

packages/explorer/src/index.ts

@@ -1 +1,3 @@
 export * from './content/index.js';
+
+export { Content } from './types/index.js';

packages/sitemap/README.md

@@ -0,0 +1,28 @@
+# @galactiks/config
+
+This package allows reading [Galactiks](https://www.galactiks.com) configurations files and get config during website generation.
+
+## Installation
+
+Install the `@galactiks/config` package using your preferred package manager:
+
+```sh
+# npm
+npm i @galactiks/config
+
+# yarn
+yarn add @galactiks/config
+
+# pnpm
+pnpm i @galactiks/config
+```
+
+If you run into any issues, [feel free to report them to us on GitHub](https://github.com/thegalactiks/explorer/issues).
+
+## Getting started
+
+Coming soon
+
+## License
+
+MIT © [Galactiks](https://www.galactiks.com)

packages/sitemap/package.json

@@ -0,0 +1,43 @@
+{
+  "name": "@galactiks/sitemap",
+  "version": "0.0.1",
+  "description": "A simple sitemap generator for Galactiks",
+  "author": "thegalactiks",
+  "type": "module",
+  "types": "./dist/index.d.ts",
+  "license": "MIT",
+  "engines": {
+    "node": ">=18.0.0"
+  },
+  "bugs": "https://github.com/thegalactiks/explorer/issues",
+  "homepage": "https://www.galactiks.com",
+  "repository": {
+    "type": "git",
+    "url": "https://github.com/thegalactiks/explorer.git",
+    "directory": "packages/sitemap"
+  },
+  "main": "dist/index.js",
+  "scripts": {
+    "build": "tsc -p ./tsconfig.json",
+    "build:ci": "tsc -p ./tsconfig.json"
+  },
+  "exports": {
+    "./package.json": "./package.json",
+    ".": "./dist/index.js"
+  },
+  "files": [
+    "dist"
+  ],
+  "keywords": [
+    "galactiks",
+    "sitemap",
+    "seo"
+  ],
+  "dependencies": {
+    "date-fns": "3.3.1",
+    "sitemap": "7.1.1"
+  },
+  "devDependencies": {
+    "@galactiks/explorer": "workspace:^"
+  }
+}