fix: remove duplicate representation entries for absolute BaseURLs

src/inheritAttributes.js

@@ -30,7 +30,7 @@ export const buildBaseUrls = (references, baseUrlElements) => {
     return references;
   }
 
-  return flatten(references.map(function(reference) {
+  const resolvedUrls = flatten(references.map(function(reference) {
     return baseUrlElements.map(function(baseUrlElement) {
       const initialBaseUrl = getContent(baseUrlElement);
       const resolvedBaseUrl = resolveUrl(reference.baseUrl, initialBaseUrl);
@@ -49,6 +49,25 @@ export const buildBaseUrls = (references, baseUrlElements) => {
       return finalBaseUrl;
     });
   }));
+
+  // Absolute child BaseURLs do not resolve against the parent, so the product of
+  // N parent BaseURLs repeats that child URL N times. Keep one copy of each unique
+  // resolved BaseURL object so content-steering copies that differ by serviceLocation
+  // (or other attributes) are preserved.
+  // See: https://github.com/videojs/mpd-parser/issues/175
+  const seen = {};
+
+  return resolvedUrls.filter(function(url) {
+    const key = JSON.stringify(url);
+
+    if (seen[key]) {
+      return false;
+    }
+
+    seen[key] = true;
+
+    return true;
+  });
 };
 
 /**
@@ -630,13 +649,6 @@ export const inheritAttributes = (mpd, options = {}) => {
   return {
     locations: mpdAttributes.locations,
     contentSteeringInfo: generateContentSteeringInformation(contentSteeringNodes, eventHandler),
-    // TODO: There are occurences where this `representationInfo` array contains undesired
-    // duplicates. This generally occurs when there are multiple BaseURL nodes that are
-    // direct children of the MPD node. When we attempt to resolve URLs from a combination of the
-    // parent BaseURL and a child BaseURL, and the value does not resolve,
-    // we end up returning the child BaseURL multiple times.
-    // We need to determine a way to remove these duplicates in a safe way.
-    // See: https://github.com/videojs/mpd-parser/pull/17#discussion_r162750527
     representationInfo: flatten(periods.map(toAdaptationSets(mpdAttributes, mpdBaseUrls))),
     eventStream: flatten(periods.map(toEventStream))
   };

test/inheritAttributes.test.js

@@ -81,6 +81,67 @@ QUnit.test('absolute BaseURL overwrites reference', function(assert) {
   );
 });
 
+QUnit.test('absolute BaseURL is not duplicated across multiple references', function(assert) {
+  const reference = [
+    { baseUrl: 'https://cdn1.example.com/' },
+    { baseUrl: 'https://cdn2.example.com/' }
+  ];
+  const node = [{ textContent: 'https://example.com/en.vtt' }];
+  const expected = [{ baseUrl: 'https://example.com/en.vtt' }];
+
+  assert.deepEqual(
+    buildBaseUrls(reference, node), expected,
+    'absolute url is not duplicated'
+  );
+});
+
+QUnit.test('absolute BaseURL does not duplicate when mixed with relative BaseURLs', function(assert) {
+  const reference = [
+    { baseUrl: 'https://cdn1.example.com/' },
+    { baseUrl: 'https://cdn2.example.com/' }
+  ];
+  const nodes = [
+    { textContent: 'bar/' },
+    { textContent: 'https://example.com/en.vtt' }
+  ];
+  const expected = [
+    { baseUrl: 'https://cdn1.example.com/bar/' },
+    { baseUrl: 'https://example.com/en.vtt' },
+    { baseUrl: 'https://cdn2.example.com/bar/' }
+  ];
+
+  assert.deepEqual(
+    buildBaseUrls(reference, nodes), expected,
+    'relative urls still combine with each reference'
+  );
+});
+
+QUnit.test('absolute BaseURLs with different serviceLocation are both kept', function(assert) {
+  const reference = [
+    { baseUrl: 'https://cdn1.example.com/' },
+    { baseUrl: 'https://cdn2.example.com/' }
+  ];
+  const nodes = [
+    {
+      textContent: 'https://example.com/en.vtt',
+      attributes: [{ name: 'serviceLocation', value: 'alpha' }]
+    },
+    {
+      textContent: 'https://example.com/en.vtt',
+      attributes: [{ name: 'serviceLocation', value: 'beta' }]
+    }
+  ];
+  const expected = [
+    { baseUrl: 'https://example.com/en.vtt', serviceLocation: 'alpha' },
+    { baseUrl: 'https://example.com/en.vtt', serviceLocation: 'beta' }
+  ];
+
+  assert.deepEqual(
+    buildBaseUrls(reference, nodes), expected,
+    'distinct serviceLocation values are preserved'
+  );
+});
+
 QUnit.test('reference attributes are ignored when there is a BaseURL node', function(assert) {
   const reference = [{ baseUrl: 'https://example.com', attributes: [{ name: 'test', value: 'wow' }] }];
   const node = [{ textContent: 'https://foo.com/bar/' }];
@@ -869,22 +930,6 @@ QUnit.test('end to end - content steering - non resolvable base URLs', function(
           template: {}
         }
       },
-      {
-        attributes: {
-          NOW,
-          bandwidth: 256,
-          baseUrl: 'https://example.com/en.vtt',
-          clientOffset: 0,
-          id: 'en',
-          lang: 'en',
-          mimeType: 'text/vtt',
-          periodStart: 0,
-          role: {},
-          sourceDuration: 0,
-          type: 'dyanmic'
-        },
-        segmentInfo: {}
-      },
       {
         attributes: {
           NOW,
@@ -904,7 +949,7 @@ QUnit.test('end to end - content steering - non resolvable base URLs', function(
     ]
   };
 
-  assert.equal(actual.representationInfo.length, 4);
+  assert.equal(actual.representationInfo.length, 3);
   assert.deepEqual(actual, expected);
 });
 
@@ -1388,26 +1433,10 @@ QUnit.test('end to end - alternate BaseURLs', function(assert) {
         clientOffset: 0
       },
       segmentInfo: {}
-    }, {
-      attributes: {
-        bandwidth: 256,
-        baseUrl: 'https://example.com/en.vtt',
-        id: 'en',
-        lang: 'en',
-        mediaPresentationDuration: 30,
-        mimeType: 'text/vtt',
-        periodStart: 0,
-        role: {},
-        sourceDuration: 30,
-        type: 'static',
-        NOW,
-        clientOffset: 0
-      },
-      segmentInfo: {}
     }]
   };
 
-  assert.equal(actual.representationInfo.length, 6);
+  assert.equal(actual.representationInfo.length, 5);
   assert.deepEqual(actual, expected);
 });