Process Preload link headers

Preload links based on link headers, some of them on document creation,
and the rest (viewport-dependant ones) when response chunks start arriving.

This can be used as an infra for early-hints and other link types.

Conforms to the current common denominator of implementations.

See #4224

Author: noamr

source

@@ -14239,20 +14239,72 @@ interface <dfn interface>HTMLLinkElement</dfn> : <span>HTMLElement</span> {
 
   <h5>Processing `<code data-x="http-link">Link</code>` headers</h5>
 
-  <p>HTTP `<code data-x="http-link">Link</code>` headers, if supported, must be assumed to come
-  before any links in the document, in the order that they were given in the HTTP message. These
-  headers are to be processed according to the rules given in the relevant specifications. <ref
-  spec=HTTP> <ref spec=WEBLINK></p> <!-- WEBLINK is ref for Link: header -->
-
-  <p class="note">Registration of relation types in HTTP `<code data-x="http-link">Link</code>`
-  headers is distinct from <a href="#linkTypes">HTML link types</a>, and thus their semantics can be
-  different from same-named HTML types.</p>
-
-  <p class="XXX">The processing of `<code data-x="http-link">Link</code>` headers, in particular
-  their influence on a <code>Document</code>'s <span>script-blocking style sheet counter</span>, is
-  not defined. See <a href="https://github.com/whatwg/html/issues/4224">issue #4224</a> for
-  discussion on integrating this into the spec.</p>
+  <p>To <dfn>process link headers</dfn> given <code>Document</code> <var>doc</var>,
+  <span data-x="concept-response">response</span>, and a "<code data-x="">pre-media</code>" or
+  "<code data-x="">media</code>" <var>phase</var>:</p>
+
+  <ol>
+    <li><p>Let <var>linkHeaders</var> be the result of <span>getting a structured field value</span>
+    given `<code>Link</code>` and "<code data-x="">list</code>" from
+    <var>response</var>'s <span data-x="concept-response-header-list">header list</span>.</p></li>
+
+    <li><p>If <var>linkHeaders</var> is null, then return.</p></li>
+
+    <li>
+     <p><span data-x="list iterate">For each</span> <var>rawLink</var> of
+     <var>linkHeaders</var>, run these steps:</p>
 
+     <ol>
+      <li><p>Let (<var>href</var>, <var>params</var>) be the result of parsing
+      <var>linkHeader</var> according to the `<code data-x="http-link">Link</code>` header
+      Serialization rules. <ref spec=HTTP> <ref spec=WEBLINK></p></li>
+      <!-- WEBLINK is ref for Link: header -->
+
+      <li>
+       <p>If <var>params</var>["<code data-x="attr-link-rel">rel</code>"] is
+       "<code data-x="rel-preload">preload</code>", then abort these steps.</p>
+
+       <p class=note>Additional support for link types can be added here in the future.</p>
+      </li>
+
+      <li><p>Let <var>expectedPhase</var> be "<code data-x="">media</code>" if either
+      "<code data-x="attr-img-srcset">srcset</code>",
+      "<code data-x="attr-link-imagesrcset">imagesrcset</code>", or <code
+      data-x="attr-link-media">media</code>" <span data-x="map exists">exist</span> in
+      <var>params</var>; Otherwise false.</p></li>
+
+      <li><p>If <var>expectedPhase</var> is not <var>phase</var>, then abort these
+      steps.</p></li>
+
+      <li><p>If <var>params</var> includes "<code data-x="attr-link-media">media</code>" and
+      <var>params</var>["<code data-x="attr-link-media">media</code>"] does not
+      <span data-x="matches the environment">match the environment</span>, then abort these
+      steps.</p></li>
+
+      <li><p>Let <var>element</var> be the result of
+      <span data-x="create an element">creating an element</span> given <var>document</var>,
+      <code>link</code>, and the <span>HTML namespace</span>.</p></li>
+
+      <li><p><span data-x="map iterate">For each</span> <var>entry</var> in <var>params</var>: If
+      <var>entry</var>[0] is "<code data-x="attr-link-as">as</code>",
+      "<code data-x="attr-img-srcset">rel</code>", "<code data-x="attr-link-type">type</code>",
+      "<code data-x="attr-link-crossorigin">crossorigin</code>",
+      "<code data-x="attr-link-imagesizes">imagesizes</code>",
+      "<code data-x="attr-link-imagesrcset">imagesrcset</code>", or
+      "<code data-x="attr-link-media">media</code>", then <span data-x="list append">append</span>
+      <var>entry</var> to <var>element</var>'s <span>attribute list</span>.</p></li>
+
+      <li><p><span data-x="fetch and process the linked resource">Fetch and process</span>
+      <var>el</var>.</p></li>
+     </ol>
+    </li>
+  </ol>
+
+  <p class="XXX">The processing of additional `<code data-x="http-link">Link</code>` headers, in
+  particular their influence on a <code>Document</code>'s
+  <span>script-blocking style sheet counter</span>, is not defined. See
+  <a href="https://github.com/whatwg/html/issues/4224">issue #4224</a> for
+  discussion on integrating this into the spec.</p>
 
   </div>
 
@@ -88778,7 +88830,7 @@ interface <dfn interface>Location</dfn> { // but see also <a href="#the-location
    data-x="navigation-params-response">response</span>'s <span
    data-x="concept-response-service-worker-timing-info">service worker timing info</span>.</p></li>
 
-   <li>
+     <li>
     <p>If <var>navigationParams</var>'s <span data-x="navigation-params-response">response</span>
     has a `<code>Refresh</code>` header, then:</p>
 
@@ -88795,6 +88847,10 @@ interface <dfn interface>Location</dfn> { // but see also <a href="#the-location
     href="https://github.com/whatwg/html/issues/2900">issue #2900</a>.</p>
    </li>
 
+   <li><p><span>Process link headers</span> given <var>document</var>, <var>navigationParams</var>'s
+   <span data-x="navigation-params-response">response</span>, and
+   "<code data-x="">pre-media</code>".</p></li>
+
    <li><p>Return <var>document</var>.</p></li>
   </ol>
 
@@ -89008,12 +89064,18 @@ new PaymentRequest(&hellip;); // Allowed to use
    <var>navigationParams</var>.</p></li>
 
    <li>
-    <p>Create an <span>HTML parser</span> and associate it with the <var>document</var>.  Each
+    <p>Create an <span>HTML parser</span> and associate it with the <var>document</var>. Each
     <span data-x="concept-task">task</span> that the <span>networking task source</span> places on
     the <span>task queue</span> while fetching runs must then fill the parser's <span>input byte
     stream</span> with the fetched bytes and cause the <span>HTML parser</span> to perform the
     appropriate processing of the input stream.</p>
 
+    <p>The first <span data-x="concept-task">task</span> that the
+    <span>networking task source</span> places on the <span>task queue</span> while fetching runs
+    must <li><p><span>process link headers</span> given <var>document</var>,
+    <var>navigationParams</var>'s <span data-x="navigation-params-response">response</span>, and
+    "<code data-x="">media</code>".</p>
+
     <p class="note">The <span>input byte stream</span> converts bytes into characters for use in the
     <span data-x="tokenization">tokenizer</span>. This process relies, in part, on character encoding
     information found in the real <span data-x="Content-Type">Content-Type metadata</span> of the