[CAI-710] move sitemap generation (#1941)

* move sitemap sync to script

* add changeset

---------

Co-authored-by: marcobottaro <39835990+marcobottaro@users.noreply.github.com>

Author: MarBert

.changeset/loose-actors-retire.md

@@ -0,0 +1,5 @@
+---
+"gitbook-docs": patch
+---
+
+Move sitemap generation after the s3 sync

.github/workflows/sync_gitbook_docs.yaml

@@ -186,6 +186,13 @@ jobs:
             fi
           fi
 
+      - name: Sync sitemap  to S3
+        env:
+          ENVIRONMENT: ${{ inputs.environment }}
+          S3_BUCKET_NAME: devportal-${{ env.ENV_SHORT }}-website-static-content
+          SITEMAP_URL: ${{ env.SITEMAP_URL || 'https://developer.pagopa.it/sitemap.xml' }}
+        run: npm run sync-sitemap -w gitbook-docs
+
       - name: Invalidate CloudFront asset bucket cache
         run: |
           aws cloudfront create-invalidation \
@@ -311,6 +318,13 @@ jobs:
             done
           fi
 
+      - name: Sync sitemap  to S3
+        env:
+          ENVIRONMENT: 'prod'
+          S3_BUCKET_NAME: devportal-p-website-static-content
+          SITEMAP_URL: 'https://developer.pagopa.it/sitemap.xml'
+        run: npm run sync-sitemap -w gitbook-docs
+
       - name: Invalidate CloudFront asset bucket cache
         run: |
           aws cloudfront create-invalidation \

packages/gitbook-docs/package.json

@@ -28,6 +28,7 @@
     "parse-docs": "ts-node src/scripts/parseDocUrlsAndIncludes.ts",
     "replace-escaped-tokens-with-backticks": "ts-node src/scripts/replaceEscapedTokensWithBackticks.ts",
     "sync-all-metadata": "ts-node src/scripts/syncAllMetadata.ts",
+    "sync-sitemap": "ts-node src/scripts/syncSitemap.ts",
     "delete-unused-directories": "ts-node src/scripts/deleteUnusedDirectories.ts"
   },
   "dependencies": {

packages/gitbook-docs/src/scripts/syncAllMetadata.ts

@@ -14,7 +14,7 @@ import {
   makeS3Client,
   putS3File,
 } from '../helpers/s3Bucket.helper';
-import { PutObjectCommand, S3Client } from '@aws-sdk/client-s3';
+import { S3Client } from '@aws-sdk/client-s3';
 import { extractTitleFromMarkdown } from '../helpers/extractTitle.helper';
 import { fetchFromStrapi } from '../helpers/fetchFromStrapi';
 import { MetadataInfo, MetadataType } from '../helpers/guidesMetadataHelper';
@@ -27,7 +27,6 @@ import {
   getSyncedSolutionsResponseJsonPath,
 } from '../syncedResponses';
 import { DOCUMENTATION_PATH } from '../helpers/documentationParsing.helper';
-import { baseUrl } from 'nextjs-website/src/config';
 import {
   StrapiApiData,
   StrapiGuide,
@@ -95,9 +94,6 @@ const S3_RELEASE_NOTES_DIRNAMES_JSON_PATH =
   process.env.S3_RELEASE_NOTES_DIRNAMES_JSON_PATH ||
   'release-notes-dirNames.json';
 
-const SITEMAP_URL = process.env.SITEMAP_URL || `${baseUrl}/sitemap.xml`;
-const S3_SITEMAP_PATH = process.env.S3_SITEMAP_PATH || 'sitemap.xml';
-
 const DOCUMENTATION_ABSOLUTE_PATH = path.resolve(DOCUMENTATION_PATH);
 
 interface StrapiData {
@@ -199,19 +195,6 @@ async function fetchAllStrapiData(): Promise<StrapiData> {
   };
 }
 
-async function fetchSitemapXml(): Promise<string> {
-  console.log(`Fetching sitemap from ${SITEMAP_URL}...`);
-  const response = await fetch(SITEMAP_URL);
-  if (!response.ok) {
-    // eslint
-    // eslint-disable-next-line functional/no-throw-statements
-    throw new Error(
-      `Failed to fetch sitemap: ${response.status} ${response.statusText}`
-    );
-  }
-  return await response.text();
-}
-
 // Generate URL path helper functions
 function generateUrlPath(
   filePath: string,
@@ -647,15 +630,6 @@ async function main() {
         S3_BUCKET_NAME!,
         getS3Client()
       );
-      const sitemapXml = await fetchSitemapXml();
-
-      await getS3Client().send(
-        new PutObjectCommand({
-          Bucket: `${S3_BUCKET_NAME}`,
-          Key: S3_SITEMAP_PATH,
-          Body: sitemapXml,
-        })
-      );
     }
 
     // Save synced responses

packages/gitbook-docs/src/scripts/syncSitemap.ts

@@ -0,0 +1,53 @@
+/* eslint-disable functional/prefer-readonly-type */
+/* eslint-disable functional/no-loop-statements */
+/* eslint-disable functional/no-expression-statements */
+/* eslint-disable functional/immutable-data */
+/* eslint-disable functional/no-try-statements */
+/* eslint-disable functional/no-let */
+
+import { baseUrl } from 'nextjs-website/src/config';
+import { PutObjectCommand, S3Client } from '@aws-sdk/client-s3';
+import { makeS3Client } from '../helpers/s3Bucket.helper';
+import dotenv from 'dotenv';
+
+// Load environment variables
+dotenv.config();
+
+const S3_BUCKET_NAME = process.env.S3_BUCKET_NAME;
+const SITEMAP_URL = process.env.SITEMAP_URL || `${baseUrl}/sitemap.xml`;
+const S3_SITEMAP_PATH = process.env.S3_SITEMAP_PATH || 'sitemap.xml';
+
+let s3Client: S3Client | undefined;
+
+async function fetchSitemapXml(): Promise<string> {
+  console.log(`Fetching sitemap from ${SITEMAP_URL}...`);
+  const response = await fetch(SITEMAP_URL);
+  if (!response.ok) {
+    // eslint-disable-next-line functional/no-throw-statements
+    throw new Error(
+      `Failed to fetch sitemap: ${response.status} ${response.statusText}`
+    );
+  }
+  return await response.text();
+}
+
+function getS3Client(): S3Client {
+  if (!s3Client) {
+    s3Client = makeS3Client();
+  }
+  return s3Client;
+}
+
+async function main() {
+  const sitemapXml = await fetchSitemapXml();
+
+  await getS3Client().send(
+    new PutObjectCommand({
+      Bucket: `${S3_BUCKET_NAME}`,
+      Key: S3_SITEMAP_PATH,
+      Body: sitemapXml,
+    })
+  );
+}
+
+main();