fix(core): in isInternalUrl(), URI protocol scheme detection should implement the spec more strictly

packages/docusaurus/src/client/exports/__tests__/isInternalUrl.test.ts

@@ -28,7 +28,7 @@ describe('isInternalUrl', () => {
     expect(isInternalUrl('https://foo.com')).toBeFalsy();
   });
 
-  it('returns false for whatever protocol links', () => {
+  it('returns false for relative protocol links', () => {
     expect(isInternalUrl('//foo.com')).toBeFalsy();
   });
 
@@ -43,4 +43,50 @@ describe('isInternalUrl', () => {
   it('returns false for undefined links', () => {
     expect(isInternalUrl(undefined)).toBeFalsy();
   });
+
+  describe('custom scheme links', () => {
+    it('returns true for invalid protocol schemes', () => {
+      expect(isInternalUrl('+customScheme://')).toBeTruthy();
+      expect(isInternalUrl('+customScheme://whatever')).toBeTruthy();
+      expect(isInternalUrl('+customScheme:whatever')).toBeTruthy();
+
+      expect(isInternalUrl('.customScheme://')).toBeTruthy();
+      expect(isInternalUrl('.customScheme://whatever')).toBeTruthy();
+      expect(isInternalUrl('.customScheme:whatever')).toBeTruthy();
+
+      expect(isInternalUrl('-customScheme://')).toBeTruthy();
+      expect(isInternalUrl('-customScheme://whatever')).toBeTruthy();
+      expect(isInternalUrl('-customScheme:whatever')).toBeTruthy();
+
+      expect(isInternalUrl('custom_scheme://')).toBeTruthy();
+      expect(isInternalUrl('custom_scheme://whatever')).toBeTruthy();
+      expect(isInternalUrl('custom_scheme:whatever')).toBeTruthy();
+
+      expect(isInternalUrl('custom scheme://')).toBeTruthy();
+      expect(isInternalUrl('custom scheme://whatever')).toBeTruthy();
+      expect(isInternalUrl('custom scheme:whatever')).toBeTruthy();
+
+      expect(isInternalUrl('custom$scheme://')).toBeTruthy();
+      expect(isInternalUrl('custom$scheme://whatever')).toBeTruthy();
+      expect(isInternalUrl('custom$scheme:whatever')).toBeTruthy();
+    });
+
+    it('returns false valid protocol schemes', () => {
+      expect(isInternalUrl('customScheme://')).toBeFalsy();
+      expect(isInternalUrl('customScheme://whatever')).toBeFalsy();
+      expect(isInternalUrl('customScheme:whatever')).toBeFalsy();
+
+      expect(isInternalUrl('custom-scheme://')).toBeFalsy();
+      expect(isInternalUrl('custom-scheme://whatever')).toBeFalsy();
+      expect(isInternalUrl('custom-scheme:whatever')).toBeFalsy();
+
+      expect(isInternalUrl('custom.scheme://')).toBeFalsy();
+      expect(isInternalUrl('custom.scheme://whatever')).toBeFalsy();
+      expect(isInternalUrl('custom.scheme:whatever')).toBeFalsy();
+
+      expect(isInternalUrl('custom-sch.eme+-.://')).toBeFalsy();
+      expect(isInternalUrl('custom-sch.eme+-.://whatever')).toBeFalsy();
+      expect(isInternalUrl('custom-sch.eme+-.:whatever')).toBeFalsy();
+    });
+  });
 });

packages/docusaurus/src/client/exports/isInternalUrl.ts

@@ -5,8 +5,11 @@
  * LICENSE file in the root directory of this source tree.
  */
 
+// Poor man's protocol detection
+// Spec: https://datatracker.ietf.org/doc/html/rfc3986#section-3.1
+// In particular: scheme = ALPHA *( ALPHA / DIGIT / "+" / "-" / "." )
 export function hasProtocol(url: string): boolean {
-  return /^(?:\w*:|\/\/)/.test(url);
+  return /^(?:[A-Za-z][A-Za-z\d+.-]*:|\/\/)/.test(url);
 }
 
 export default function isInternalUrl(url?: string): boolean {

website/_dogfooding/_pages tests/markdown-tests-mdx.mdx

@@ -360,29 +360,27 @@ This is a test page to see if Docusaurus Markdown features are working properly
 See [#3337](https://github.com/facebook/docusaurus/issues/3337)
 
 - [/someFile.pdf](/someFile.pdf)
-
 - [/someFile.xyz](/someFile.xyz)
-
 - [/image with space.png](</image with spaces.png>)
-
 - [@site/\_dogfooding/\_asset-tests/someFile.pdf](@site/_dogfooding/_asset-tests/someFile.pdf)
-
 - [@site/\_dogfooding/\_asset-tests/someFile.xyz](@site/_dogfooding/_asset-tests/someFile.xyz)
-
 - [@site/\_dogfooding/\_asset-tests/image with space.png](<@site/_dogfooding/_asset-tests/image with spaces.png>)
 
 ### Linking to non-SPA page hosted within website
 
 See [#3309](https://github.com/facebook/docusaurus/issues/3309)
 
 - [pathname:///dogfooding/javadoc](pathname:///dogfooding/javadoc)
-
 - [pathname:///dogfooding/javadoc/index.html](pathname:///dogfooding/javadoc/index.html)
-
 - [pathname://../dogfooding/javadoc](pathname://../dogfooding/javadoc)
-
 - [pathname://../dogfooding/javadoc/index.html](pathname://../dogfooding/javadoc/index.html)
 
+### Linking to non-SPA paths with custom schemes
+
+- [customScheme://whatever](customScheme://whatever)
+- [custom-scheme://whatever](custom-scheme://whatever)
+- [custom-sch.eme+-.://whatever](custom-sch.eme+-.://whatever)
+
 ### Linking to non-SPA page with Link component
 
 See [#9758](https://github.com/facebook/docusaurus/issues/9758), these external urls should not be reported by the broken links checker: