fix: handle base correctly (#114)

JustinBeckwith · Nov 19, 2019 · 9813a61 · 9813a61
1 parent d653f2c
commit 9813a61
Show file tree

Hide file tree

Showing 8 changed files with 170 additions and 1 deletion.
diff --git a/src/links.ts b/src/links.ts
@@ -43,12 +43,43 @@ export function getLinks(source: string, baseUrl: string): ParsedUrl[] {
       links.push(...values);
     });
   });
+
+  let realBaseUrl = baseUrl;
+  const base = $('base[href]');
+  if (base.length) {
+    // only first <base by specification
+    const htmlBaseUrl = base.first().attr('href');
+
+    realBaseUrl = getBaseUrl(htmlBaseUrl, baseUrl);
+  }
+
   const sanitized = links
     .filter(link => !!link)
-    .map(link => parseLink(link, baseUrl));
+    .map(link => parseLink(link, realBaseUrl));
   return sanitized;
 }
 
+function getBaseUrl(htmlBaseUrl: string, oldBaseUrl: string): string {
+  if (isAbsoluteUrl(htmlBaseUrl)) {
+    return htmlBaseUrl;
+  }
+
+  const url = new URL(htmlBaseUrl, oldBaseUrl);
+  url.hash = '';
+  return url.href;
+}
+
+function isAbsoluteUrl(url: string): boolean {
+  // Don't match Windows paths
+  if (/^[a-zA-Z]:\\/.test(url)) {
+    return false;
+  }
+
+  // Scheme: https://tools.ietf.org/html/rfc3986#section-3.1
+  // Absolute URL: https://tools.ietf.org/html/rfc3986#section-4.3
+  return /^[a-zA-Z][a-zA-Z\d+\-.]*:/.test(url);
+}
+
 function parseAttr(name: string, value: string): string[] {
   switch (name) {
     case 'srcset':

diff --git a/test/fixtures/basetag/absolute.html b/test/fixtures/basetag/absolute.html
@@ -0,0 +1,12 @@
+<html>
+  <head>
+    <base href="http://another.fake.local/" />
+    <base href="./unused" />
+  </head>
+  <body>
+    <a href="ok">relative link</a>
+    <a href="broken">relative link</a>
+    <a href="./ok">relative link</a>
+    <a href="./broken">relative link</a>
+  </body>
+</html>
diff --git a/test/fixtures/basetag/empty-base.html b/test/fixtures/basetag/empty-base.html
@@ -0,0 +1,11 @@
+<html>
+  <head>
+    <base href="" />
+  </head>
+  <body>
+    <a href="ok">relative link</a>
+    <a href="broken">relative link</a>
+    <a href="./ok">relative link</a>
+    <a href="./broken">relative link</a>
+  </body>
+</html>
diff --git a/test/fixtures/basetag/relative-dot-folder.html b/test/fixtures/basetag/relative-dot-folder.html
@@ -0,0 +1,11 @@
+<html>
+  <head>
+    <base href="./anotherBase/" />
+  </head>
+  <body>
+    <a href="ok">relative link</a>
+    <a href="broken">relative link</a>
+    <a href="./ok">relative link</a>
+    <a href="./broken">relative link</a>
+  </body>
+</html>
diff --git a/test/fixtures/basetag/relative-folder.html b/test/fixtures/basetag/relative-folder.html
@@ -0,0 +1,11 @@
+<html>
+  <head>
+    <base href="anotherBase/" />
+  </head>
+  <body>
+    <a href="ok">relative link</a>
+    <a href="broken">relative link</a>
+    <a href="./ok">relative link</a>
+    <a href="./broken">relative link</a>
+  </body>
+</html>
diff --git a/test/fixtures/basetag/relative-page.html b/test/fixtures/basetag/relative-page.html
@@ -0,0 +1,11 @@
+<html>
+  <head>
+    <base href="anotherBase" />
+  </head>
+  <body>
+    <a href="ok">relative link</a>
+    <a href="broken">relative link</a>
+    <a href="./ok">relative link</a>
+    <a href="./broken">relative link</a>
+  </body>
+</html>
diff --git a/test/fixtures/basetag/relative-to-root.html b/test/fixtures/basetag/relative-to-root.html
@@ -0,0 +1,11 @@
+<html>
+  <head>
+    <base href="/anotherBase/" />
+  </head>
+  <body>
+    <a href="ok">relative link</a>
+    <a href="broken">relative link</a>
+    <a href="./ok">relative link</a>
+    <a href="./broken">relative link</a>
+  </body>
+</html>
diff --git a/test/test.ts b/test/test.ts
@@ -88,6 +88,77 @@ describe('linkinator', () => {
     );
   });
 
+  it('should detect relative urls with relative base', async () => {
+    const cases = [
+      {
+        fixture: 'test/fixtures/basetag/relative-to-root.html',
+        nonBrokenUrl: '/anotherBase/ok',
+      },
+      {
+        fixture: 'test/fixtures/basetag/relative-folder.html',
+        nonBrokenUrl: '/pageBase/anotherBase/ok',
+      },
+      {
+        fixture: 'test/fixtures/basetag/relative-dot-folder.html',
+        nonBrokenUrl: '/pageBase/anotherBase/ok',
+      },
+      {
+        fixture: 'test/fixtures/basetag/relative-page.html',
+        nonBrokenUrl: '/pageBase/ok',
+      },
+      {
+        fixture: 'test/fixtures/basetag/empty-base.html',
+        nonBrokenUrl: '/pageBase/ok',
+      },
+    ];
+
+    for (let i = 0; i < cases.length; i++) {
+      const { fixture, nonBrokenUrl } = cases[i];
+      const scope = nock('http://fake.local')
+        .get('/pageBase/index')
+        .replyWithFile(200, fixture, {
+          'Content-Type': 'text/html; charset=UTF-8',
+        })
+        .head(nonBrokenUrl)
+        .reply(200);
+
+      const results = await check({
+        path: 'http://fake.local/pageBase/index',
+      });
+
+      assert.strictEqual(results.links.length, 3);
+      assert.strictEqual(
+        results.links.filter(x => x.state === LinkState.BROKEN).length,
+        1
+      );
+      scope.done();
+    }
+  });
+
+  it('should detect relative urls with absolute base', async () => {
+    const scope = nock('http://fake.local')
+      .get('/pageBase/index')
+      .replyWithFile(200, 'test/fixtures/basetag/absolute.html', {
+        'Content-Type': 'text/html; charset=UTF-8',
+      });
+
+    const anotherScope = nock('http://another.fake.local')
+      .head('/ok')
+      .reply(200);
+
+    const results = await check({
+      path: 'http://fake.local/pageBase/index',
+    });
+
+    assert.strictEqual(results.links.length, 3);
+    assert.strictEqual(
+      results.links.filter(x => x.state === LinkState.BROKEN).length,
+      1
+    );
+    scope.done();
+    anotherScope.done();
+  });
+
   it('should detect broken image links', async () => {
     const results = await check({ path: 'test/fixtures/image' });
     assert.strictEqual(