Merge pull request #26 from nightt5879/nightt5879/fix-site-url-normalization-1.3.5

fix: normalize production site URLs for 1.3.5

Author: nightt5879

CHANGELOG.md

@@ -4,6 +4,16 @@ All notable changes to this project will be documented in this file.
 
 ## [Unreleased]
 
+## [1.3.5] - 2026-05-03
+
+### Changed
+- Bumped the package version from `1.3.4` to `1.3.5` for production Git build URL hardening.
+- Normalized site URL inputs so missing `https://` is treated as HTTPS instead of breaking static route generation.
+
+### Fixed
+- Prevented invalid `PUBLIC_SITE_URL` values from crashing Open Graph and canonical URL generation during Cloudflare Pages production builds.
+- Kept CMS config and smoke checks aligned with the same URL normalization rules.
+
 ## [1.3.4] - 2026-05-03
 
 ### Added

README_EN.md

@@ -228,6 +228,7 @@ venue: "Conference Name"
 - `1.3.2`: CMS stability patch that adds stable legacy news slugs, removes CMS test residue, and tightens news slug validation
 - `1.3.3`: build-log cleanup that clears Astro content cache before builds to remove duplicate news id warnings
 - `1.3.4`: Cloudflare Pages Git build stability patch that pins Node.js and records the Pages output directory in repo config
+- `1.3.5`: production Git build hardening that normalizes malformed site URL variables before static rendering
 
 ## License
 

docs/cloudflare-pages-build-settings.md

@@ -18,7 +18,7 @@ The repository pins the Node.js version in `.node-version`, and the Pages projec
 - `CMS_OAUTH_BASE_URL=https://doubleducklab-cms-oauth.<account>.workers.dev`
 - `PUBLIC_SITE_URL=https://doubleducklab.pages.dev`
 
-`NODE_VERSION` is a build-environment pin, not a CMS secret. It is listed with Pages secrets only because Wrangler manages Pages variables through `wrangler pages secret`.
+`PUBLIC_SITE_URL` must be an absolute site URL. The build now tolerates a missing `https://` prefix, but the stored Pages value should still include the full `https://` URL. `NODE_VERSION` is a build-environment pin, not a CMS secret. It is listed with Pages secrets only because Wrangler manages Pages variables through `wrangler pages secret`.
 
 ## Verification
 

package-lock.json

@@ -1,7 +1,7 @@
 {
   "name": "double-duck-lab",
   "type": "module",
-  "version": "1.3.4",
+  "version": "1.3.5",
   "private": true,
   "scripts": {
     "dev": "astro dev",

package.json

@@ -3,8 +3,22 @@ import { loadEnv } from 'vite';
 export const DEFAULT_SITE_URL = 'https://doubleducklab.com/';
 
 export function normalizeSiteUrl(value) {
-  const site = (value || DEFAULT_SITE_URL).trim();
-  return site.endsWith('/') ? site : `${site}/`;
+  const raw = `${value || DEFAULT_SITE_URL}`.trim().replace(/^['"]|['"]$/g, '');
+  const candidate = /^[a-z][a-z\d+\-.]*:\/\//i.test(raw) ? raw : `https://${raw}`;
+
+  try {
+    const url = new URL(candidate);
+    if (!['http:', 'https:'].includes(url.protocol)) {
+      throw new Error(`Unsupported site URL protocol: ${url.protocol}`);
+    }
+
+    url.search = '';
+    url.hash = '';
+    const site = url.toString();
+    return site.endsWith('/') ? site : `${site}/`;
+  } catch {
+    return DEFAULT_SITE_URL;
+  }
 }
 
 export function resolveConfiguredSiteUrl(mode = 'production') {

scripts/site-url.mjs

@@ -8,11 +8,24 @@ const publicSiteUrl = (process.env.PUBLIC_SITE_URL || '').trim();
 const cmsBranch = (process.env.CMS_BRANCH || 'main').trim() || 'main';
 
 function normalizeUrl(value) {
-  if (!value) {
+  const raw = (value || '').trim().replace(/^['"]|['"]$/g, '');
+  if (!raw) {
     return '';
   }
 
-  return value.replace(/\/$/, '');
+  const candidate = /^[a-z][a-z\d+\-.]*:\/\//i.test(raw) ? raw : `https://${raw}`;
+  try {
+    const url = new URL(candidate);
+    if (!['http:', 'https:'].includes(url.protocol)) {
+      return '';
+    }
+
+    url.search = '';
+    url.hash = '';
+    return url.toString().replace(/\/$/, '');
+  } catch {
+    return '';
+  }
 }
 
 function getUrlHostname(value) {
@@ -21,7 +34,7 @@ function getUrlHostname(value) {
   }
 
   try {
-    return new URL(`${value}/`).hostname;
+    return new URL(value).hostname;
   } catch {
     return '';
   }

scripts/smoke-build.mjs

@@ -16,17 +16,29 @@ export type CmsRuntimeConfig = {
 };
 
 function normalizeUrl(value: string | undefined, fallback = '') {
-  const normalized = (value || fallback).trim();
-  if (!normalized) {
+  const raw = (value || fallback).trim().replace(/^['"]|['"]$/g, '');
+  if (!raw) {
     return '';
   }
 
-  return normalized.endsWith('/') ? normalized.slice(0, -1) : normalized;
+  const candidate = /^[a-z][a-z\d+\-.]*:\/\//i.test(raw) ? raw : `https://${raw}`;
+  try {
+    const url = new URL(candidate);
+    if (!['http:', 'https:'].includes(url.protocol)) {
+      return '';
+    }
+
+    url.search = '';
+    url.hash = '';
+    return url.toString().replace(/\/$/, '');
+  } catch {
+    return '';
+  }
 }
 
 function getUrlHostname(value: string) {
   try {
-    return new URL(`${value}/`).hostname;
+    return new URL(value).hostname;
   } catch {
     return '';
   }

src/utils/cms-config.ts

@@ -53,10 +53,32 @@ export function buildDefaultLocaleAlternates(pathname: string): LocaleAlternates
 }
 
 export function toAbsoluteUrl(pathname: string, site: URL | string | undefined) {
-  const siteUrl = site instanceof URL ? site : new URL(site || DEFAULT_SITE_URL);
+  const siteUrl = normalizeSiteUrl(site);
   return new URL(pathname, siteUrl).toString();
 }
 
+export function normalizeSiteUrl(site: URL | string | undefined) {
+  if (site instanceof URL) {
+    return site;
+  }
+
+  const raw = `${site || DEFAULT_SITE_URL}`.trim().replace(/^['"]|['"]$/g, '');
+  const candidate = /^[a-z][a-z\d+\-.]*:\/\//i.test(raw) ? raw : `https://${raw}`;
+
+  try {
+    const url = new URL(candidate);
+    if (!['http:', 'https:'].includes(url.protocol)) {
+      throw new Error(`Unsupported site URL protocol: ${url.protocol}`);
+    }
+
+    url.search = '';
+    url.hash = '';
+    return url;
+  } catch {
+    return new URL(DEFAULT_SITE_URL);
+  }
+}
+
 export function summarizeText(value: string | undefined, maxLength = 160) {
   const plain = (value || '')
     .replace(/```[\s\S]*?```/g, ' ')