mimesniff.bs

<pre class=metadata>
Group: WHATWG
H1: MIME Sniffing
Shortname: mimesniff
Text Macro: TWITTER mimesniff
Text Macro: LATESTRD 2026-01
Abstract: The MIME Sniffing standard defines sniffing resources.
Translation: ja https://triple-underscore.github.io/mimesniff-ja.html
Translation: zh-Hans https://htmlspecs.com/mimesniff/
Translation: ko https://ko.htmlspecs.com/mimesniff/
Markup Shorthands: css off
</pre>

<pre class=biblio>
{
 "FTP": {
  "aliasOf": "rfc0959"
 },
 "HTTP-SEMANTICS": {
  "aliasOf": "rfc9110"
 },
 "KEYWORDS": {
  "aliasOf": "rfc2119"
 },
 "MIMETYPE": {
  "aliasOf": "rfc2046"
 },
 "SECCONTSNIFF": {
  "authors": ["Adam Barth", "Juan Caballero", "Dawn Song"],
  "date": "May 2009",
  "href": "https://www.adambarth.com/papers/2009/barth-caballero-song.pdf",
  "title": "Secure Content Sniffing for Web Browsers, or How to Stop Papers from Reviewing Themselves"
 }
}
</pre>

<pre class=anchors>
spec: HTTP-SEMANTICS; urlPrefix: https://www.rfc-editor.org/rfc/rfc9110
    type: dfn
        text: media-type; url: #name-media-type
        text: quoted-string; url: #name-quoted-strings
        text: token; url: #name-tokens
</pre>

<pre class=link-defaults>
spec:html;
    type:element; text:script
    type:element-attr; text:type
</pre>



<h2 id=introduction>Introduction</h2>

<p>
 The HTTP <code>Content-Type</code> header field is intended to indicate the
 MIME type of an HTTP response.
 However, many HTTP servers supply a <code>Content-Type</code> header field
 value that does not match the actual contents of the response.
 Historically, web browsers have tolerated these servers by examining the
 content of HTTP responses in addition to the <code>Content-Type</code> header
 field in order to determine the effective MIME type of the response.

<p>
 Without a clear specification for how to "sniff" the MIME type, each user
 agent has been forced to reverse-engineer the algorithms of other user agents
 in order to maintain interoperability.
 Inevitably, these efforts have not been entirely successful, resulting in
 divergent behaviors among user agents.
 In some cases, these divergent behaviors have had security implications, as a
 user agent could interpret an HTTP response as a different MIME type than
 the server intended.

<p>
 These security issues are most severe when an "honest" server allows
 potentially malicious users to upload their own files and then serves the
 contents of those files with a low-privilege MIME type.
 For example, if a server believes that the client will treat a contributed
 file as an image (and thus treat it as benign), but a user agent believes the
 content to be HTML (and thus privileged to execute any scripts contained
 therein), an attacker might be able to steal the user's authentication
 credentials and mount other cross-site scripting attacks.
 (Malicious servers, of course, can specify an arbitrary MIME type in the
 <code>Content-Type</code> header field.)

<p>
 This document describes a content sniffing algorithm that carefully balances
 the compatibility needs of user agent with the security constraints imposed
 by existing web content.
 The algorithm originated from research conducted by Adam Barth, Juan
 Caballero, and Dawn Song, based on content sniffing algorithms present in
 popular user agents, an extensive database of existing web content, and
 metrics collected from implementations deployed to a sizable number of users.
 [[SECCONTSNIFF]]



<h2 id=conformance-requirements>Conformance requirements</h2>

<p>
 The keywords "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD",
 "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this
 document are to be interpreted as described in RFC 2119.
 For readability, these keywords will generally not appear in all uppercase
 letters.
 [[!KEYWORDS]]

<p>
 Requirements phrased in the imperative as part of algorithms (such as "strip
 any leading space characters" or "return false and abort these steps") are to
 be interpreted with the meaning of the keyword used in introducing the
 algorithm.

<p>
 Conformance requirements phrased as algorithms or specific steps can be
 implemented in any manner, so long as the end result is equivalent.
 In particular, note that the algorithms defined in this specification are
 intended to be easy to understand and are not intended to be performant.



<h2 id=terminology>Terminology</h2>

<p>
 This specification depends on the Infra Standard. [[!INFRA]]

<p>An <dfn>HTTP token code point</dfn> is U+0021 (!), U+0023 (#), U+0024 ($), U+0025 (%),
U+0026 (&amp;), U+0027 ('), U+002A (*), U+002B (+), U+002D (-), U+002E (.), U+005E (^), U+005F (_),
U+0060 (`), U+007C (|), U+007E (~), or an <a>ASCII alphanumeric</a>.</p>

<p class=note>This matches the value space of the <a>token</a> token production. [[HTTP-SEMANTICS]]

<p>An <dfn>HTTP quoted-string token code point</dfn> is U+0009 TAB, a <a>code point</a> in the range
U+0020 SPACE to U+007E (~), inclusive, or a <a>code point</a> in the range U+0080 through
U+00FF (ÿ), inclusive.

<p class=note>This matches the effective value space of the <a>quoted-string</a> token
production. By definition it is a superset of the <a>HTTP token code points</a>. [[HTTP-SEMANTICS]]

<p>
 A <dfn>binary data byte</dfn> is a <a>byte</a> in the range 0x00 to
 0x08 (NUL to BS), the <a>byte</a> 0x0B (VT), a <a>byte</a> in the
 range 0x0E to 0x1A (SO to SUB), or a <a>byte</a> in the range 0x1C to
 0x1F (FS to US).

<p>
 A <dfn>whitespace byte</dfn> (abbreviated
 <abbr lt="whitespace byte">0xWS</abbr>) is any one of the following
 <a>bytes</a>: 0x09 (HT), 0x0A (LF), 0x0C (FF), 0x0D (CR),
 0x20 (SP).

<p>
 A <dfn>tag-terminating byte</dfn> (abbreviated
 <abbr lt="tag terminating byte">0xTT</abbr>) is any one of the following
 <a>bytes</a>: 0x20 (SP), 0x3E ("<code>></code>").

<p>
  Equations are using the mathematical operators as defined in
  [[!ENCODING]]. In addition, the bitwise NOT is
  represented by ~.


<h2 id=understanding-mime-types>MIME types</h2>

<h3 id=mime-type-representation>MIME type representation</h3>

<p>A <dfn export lt="MIME type|MIME type record" id=mime-type>MIME type</dfn> represents an
<i>internet media type</i> as defined by
<cite>Multipurpose Internet Mail Extensions (MIME) Part Two: Media Types</cite>. It can also be
referred to as a <a>MIME type record</a>. [[!MIMETYPE]]

<p class=note>Standards are encouraged to consistently use the term <a>MIME type</a> to avoid
confusion with the use of <i>media type</i> as described in <cite>Media Queries</cite>.
[[MEDIAQUERIES]]

<p>A <a>MIME type</a>'s <dfn export for="MIME type" id=type>type</dfn> is a non-empty
<a>ASCII string</a>.

<p>A <a>MIME type</a>'s <dfn export for="MIME type" id=subtype>subtype</dfn> is a non-empty
<a>ASCII string</a>.

<p>A <a>MIME type</a>'s <dfn export for="MIME type" id=parameters>parameters</dfn> is an
<a>ordered map</a> whose <a for=map>keys</a> are <a>ASCII strings</a> and <a for=map>values</a> are
<a for=/>strings</a> limited to <a>HTTP quoted-string token code points</a>. It is initially empty.
<!-- Candidate for "isomorphic string" if we ever decide to add that. -->


<h3 id=mime-type-miscellaneous>MIME type miscellaneous</h3>

<p>The <dfn export for="MIME type">essence</dfn> of a <a>MIME type</a> <var>mimeType</var> is
<var>mimeType</var>'s <a for="MIME type">type</a>, followed by U+002F (/), followed by
<var>mimeType</var>'s <a for="MIME type">subtype</a>.

<p>A <a>MIME type</a> is <dfn export>supported by the user agent</dfn> if the user agent has the
capability to interpret a <a>resource</a> of that <a>MIME type</a> and present it to the user.

<p class=XXX>Ideally this would be more precise. See
<a href=https://github.com/w3c/preload/issues/113>w3c/preload #113</a>.

<div algorithm>
<p>To <dfn export>minimize a supported MIME type</dfn> given a <a>MIME type</a> <var>mimeType</var>,
run these steps. They return an <a>ASCII string</a>.

<ol>
 <li><p>If <var>mimeType</var> is a <a>JavaScript MIME type</a>, then return
 "<code>text/javascript</code>".

 <li><p>If <var>mimeType</var> is a <a>JSON MIME type</a>, then return
 "<code>application/json</code>".

 <li>
  <p>If <var>mimeType</var>'s <a for="MIME type">essence</a> is "<code>image/svg+xml</code>", then
  return "<code>image/svg+xml</code>".

  <p class=note>SVG is worth distinguishing from other <a>XML MIME types</a>.

 <li><p>If <var>mimeType</var> is an <a>XML MIME type</a>, then return
 "<code>application/xml</code>".

 <li><p>If <var>mimeType</var> is <a>supported by the user agent</a>, then return
 <var>mimeType</var>'s <a for="MIME type">essence</a>.

 <li><p>Return the empty string.
</ol>

<p class=note>The goal of this algorithm is to allow the caller to distinguish MIME types with
different processing models, such as those for GIF and PNG, but otherwise provide as little
information as possible.
</div>


<h3 id=mime-type-writing>MIME type writing</h3>

<p>A <dfn export id=valid-mime-type>valid MIME type string</dfn> is a string that matches the
<a spec="HTTP-SEMANTICS">media-type</a> token production. In particular, a <a>valid MIME type string</a> may
include <a for="MIME type">parameters</a>. [[!HTTP-SEMANTICS]]

<p class=note>A <a>valid MIME type string</a> is supposed to be used for conformance checkers only.

<div class=example id=example-valid-mime-type-string>
 <p>"<code>text/html</code>" is a <a>valid MIME type string</a>.

 <p>"<code>text/html;</code>" is not a <a>valid MIME type string</a>, though
 <a>parse a MIME type</a> returns a <a>MIME type record</a> for it identical to if the input had
 been "<code>text/html</code>".
</div>

<p>A
<dfn export id=valid-mime-type-with-no-parameters>valid MIME type string with no parameters</dfn> is
a <a>valid MIME type string</a> that does not contain U+003B (;).


<h3 id=parsing-a-mime-type>Parsing a MIME type</h3>

<p>To <dfn export>parse a MIME type</dfn>, given a string <var>input</var>, run these steps:

<ol>
 <li><p>Remove any leading and trailing <a>HTTP whitespace</a> from <var>input</var>.

 <li><p>Let <var>position</var> be a <a for=string>position variable</a> for <var>input</var>,
 initially pointing at the start of <var>input</var>.

 <li><p>Let <var>type</var> be the result of <a>collecting a sequence of code points</a> that are
 not U+002F (/) from <var>input</var>, given <var>position</var>.

 <li><p>If <var>type</var> is the empty string or does not solely contain
 <a>HTTP token code points</a>, then return failure.

 <li><p>If <var>position</var> is past the end of <var>input</var>, then return failure.

 <li><p>Advance <var>position</var> by 1. (This skips past U+002F (/).)

 <li><p>Let <var>subtype</var> be the result of <a>collecting a sequence of code points</a> that are
 not U+003B (;) from <var>input</var>, given <var>position</var>.

 <li><p>Remove any trailing <a>HTTP whitespace</a> from <var>subtype</var>.

 <li><p>If <var>subtype</var> is the empty string or does not solely contain
 <a>HTTP token code points</a>, then return failure.

 <li><p>Let <var>mimeType</var> be a new <a>MIME type record</a> whose <a for="MIME type">type</a>
 is <var>type</var>, in <a>ASCII lowercase</a>, and <a for="MIME type">subtype</a> is
 <var>subtype</var>, in <a>ASCII lowercase</a>.

 <li>
  <p>While <var>position</var> is not past the end of <var>input</var>:

  <ol>
   <li><p>Advance <var>position</var> by 1. (This skips past U+003B (;).)

   <li>
    <p><a>Collect a sequence of code points</a> that are <a>HTTP whitespace</a> from
    <var>input</var> given <var>position</var>.

    <p class=note>This is roughly equivalent to <a>skip ASCII whitespace</a>, except that
    <a>HTTP whitespace</a> is used rather than <a>ASCII whitespace</a>.

   <li><p>Let <var>parameterName</var> be the result of <a>collecting a sequence of code points</a>
   that are not U+003B (;) or U+003D (=) from <var>input</var>, given <var>position</var>.

   <li><p>Set <var>parameterName</var> to <var>parameterName</var>, in <a>ASCII lowercase</a>.

   <li>
    <p>If <var>position</var> is not past the end of <var>input</var>, then:

    <ol>
     <li><p>If the <a>code point</a> at <var>position</var> within <var>input</var> is U+003B (;),
     then <a for=iteration>continue</a>.

     <li><p>Advance <var>position</var> by 1. (This skips past U+003D (=).)
    </ol>

   <li><p>If <var>position</var> is past the end of <var>input</var>, then
   <a for=iteration>break</a>.

   <li><p>Let <var>parameterValue</var> be null.

   <li>
    <p>If the <a>code point</a> at <var>position</var> within <var>input</var> is U+0022 ("),
    then:

    <ol>
     <li><p>Set <var>parameterValue</var> to the result of <a>collecting an HTTP quoted string</a>
     from <var>input</var>, given <var>position</var> and true.

     <li>
      <p><a>Collect a sequence of code points</a> that are not U+003B (;) from <var>input</var>,
      given <var>position</var>.

      <p class=example id=example-mime-type-parser-trailing-garbage>Given
      <code>text/html;charset="shift_jis"iso-2022-jp</code> you end up with
      <code>text/html;charset=shift_jis</code>.
    </ol>

   <li>
    <p>Otherwise:

    <ol>
     <li><p>Set <var>parameterValue</var> to the result of
     <a>collecting a sequence of code points</a> that are not U+003B (;) from <var>input</var>,
     given <var>position</var>.

     <li><p>Remove any trailing <a>HTTP whitespace</a> from <var>parameterValue</var>.

     <li><p>If <var>parameterValue</var> is the empty string, then <a for=iteration>continue</a>.
    </ol>

   <li>
    <p>If all of the following are true

    <ul class=brief>
     <li><var>parameterName</var> is not the empty string

     <li><var>parameterName</var> solely contains <a>HTTP token code points</a>

     <li><var>parameterValue</var> solely contains <a>HTTP quoted-string token code points</a>

     <li><var>mimeType</var>'s <a for="MIME type">parameters</a>[<var>parameterName</var>]
     <a for=map lt=exist>does not exist</a>
    </ul>

    <p>then <a for=map>set</a> <var>mimeType</var>'s
    <a for="MIME type">parameters</a>[<var>parameterName</var>] to <var>parameterValue</var>.
  </ol>

 <li><p>Return <var>mimeType</var>.
</ol>

<hr>

<p>To <dfn export>parse a MIME type from bytes</dfn>, given a <a>byte sequence</a> <var>input</var>,
run these steps:

<ol>
 <li><p>Let <var>string</var> be <var>input</var>, <a>isomorphic decoded</a>.

 <li><p>Return the result of <a>parse a MIME type</a> with <var>string</var>.
</ol>


<h3 id=serializing-a-mime-type>Serializing a MIME type</h3>

<p>To <dfn export>serialize a MIME type</dfn>, given a <a>MIME type</a> <var>mimeType</var>, run
these steps:

<ol>
 <li><p>Let <var>serialization</var> be the concatenation of <var>mimeType</var>'s
 <a for="MIME type">type</a>, U+002F (/), and <var>mimeType</var>'s <a for="MIME type">subtype</a>.

 <li>
  <p><a for=map>For each</a> <var>name</var> → <var>value</var> of <var>mimeType</var>'s
  <a for="MIME type">parameters</a>:

  <ol>
   <li><p>Append U+003B (;) to <var>serialization</var>.

   <li><p>Append <var>name</var> to <var>serialization</var>.

   <li><p>Append U+003D (=) to <var>serialization</var>.

   <li>
    <p>If <var>value</var> does not solely contain <a>HTTP token code points</a> or <var>value</var>
    is the empty string, then:

    <ol>
     <li><p>Precede each occurrence of U+0022 (") or U+005C (\) in <var>value</var> with U+005C (\).

     <li><p>Prepend U+0022 (") to <var>value</var>.

     <li><p>Append U+0022 (") to <var>value</var>.
    </ol>

   <li><p>Append <var>value</var> to <var>serialization</var>.
  </ol>

 <li><p>Return <var>serialization</var>.
</ol>

<hr>

<p>To <dfn export>serialize a MIME type to bytes</dfn>, given a <a>MIME type</a>
<var>mimeType</var>, run these steps:

<ol>
 <li><p>Let <var>stringSerialization</var> be the result of <a>serialize a MIME type</a> with
 <var>mimeType</var>.

 <li><p>Return <var>stringSerialization</var>, <a>isomorphic encoded</a>.
</ol>


<h3 id=mime-type-groups>MIME type groups</h3>

<p>An <dfn export>image MIME type</dfn> is a <a>MIME type</a> whose <a for="MIME type">type</a> is
"<code>image</code>".

<p>An <dfn export>audio or video MIME type</dfn> is any <a>MIME type</a> whose
<a for="MIME type">type</a> is "<code>audio</code>" or "<code>video</code>", or whose
<a for="MIME type">essence</a> is "<code>application/ogg</code>".

<p>A <dfn export>font MIME type</dfn> is any <a>MIME type</a> whose <a for="MIME type">type</a> is
"<code>font</code>", or whose <a for="MIME type">essence</a> is one of the following: [[RFC8081]]

<ul class="brief">
 <li><code>application/font-cff</code>
 <li><code>application/font-otf</code>
 <li><code>application/font-sfnt</code>
 <li><code>application/font-ttf</code>
 <li><code>application/font-woff</code>
 <li><code>application/vnd.ms-fontobject</code>
 <li><code>application/vnd.ms-opentype</code>
</ul>

<p>A <dfn export>ZIP-based MIME type</dfn> is any <a>MIME type</a> whose
<a for="MIME type">subtype</a> ends in "<code>+zip</code>" or whose <a for="MIME type">essence</a>
is one of the following:

<ul class="XXX brief">
 <li><code>application/zip</code>
</ul>

<p>An <dfn export>archive MIME type</dfn> is any <a>MIME type</a> whose
<!--<span>type</span> is equal to "<code title>archive</code>" or-->
<a for="MIME type">essence</a> is one of the following:

<ul class="brief">
 <li><code>application/x-rar-compressed</code>
 <li><code>application/zip</code>
 <li><code>application/x-gzip</code>
</ul>

<p>An <dfn export>XML MIME type</dfn> is any <a>MIME type</a> whose <a for="MIME type">subtype</a>
ends in "<code>+xml</code>" or whose <a for="MIME type">essence</a> is "<code>text/xml</code>" or
"<code>application/xml</code>". [[RFC7303]]

<p>An <dfn export>HTML MIME type</dfn> is any <a>MIME type</a> whose <a for="MIME type">essence</a>
is "<code>text/html</code>".

<p>A <dfn export>scriptable MIME type</dfn> is an <a>XML MIME type</a>, <a>HTML MIME type</a>, or
any <a>MIME type</a> whose <a for="MIME type">essence</a> is "<code>application/pdf</code>".

<p>A <dfn export>JavaScript MIME type</dfn> is any <a>MIME type</a> whose
<a for="MIME type">essence</a> is one of the following:

<ul class="brief">
 <li><code>application/ecmascript</code>
 <li><code>application/javascript</code>
 <li><code>application/x-ecmascript</code>
 <li><code>application/x-javascript</code>
 <li><code>text/ecmascript</code>
 <li><code>text/javascript</code>
 <li><code>text/javascript1.0</code>
 <li><code>text/javascript1.1</code>
 <li><code>text/javascript1.2</code>
 <li><code>text/javascript1.3</code>
 <li><code>text/javascript1.4</code>
 <li><code>text/javascript1.5</code>
 <li><code>text/jscript</code>
 <li><code>text/livescript</code>
 <li><code>text/x-ecmascript</code>
 <li><code>text/x-javascript</code>
</ul>

<p>A <a>string</a> is a <dfn export>JavaScript MIME type essence match</dfn> if it is an
<a>ASCII case-insensitive</a> match for one of the <a>JavaScript MIME type</a> essence strings.

<p class="note">This hook is used by the <{script/type}> attribute of <{script}> elements. [[HTML]]

<p>A <dfn export>JSON MIME type</dfn> is any <a>MIME type</a> whose <a for="MIME type">subtype</a>
ends in "<code>+json</code>" or whose <a for="MIME type">essence</a> is
"<code>application/json</code>" or "<code>text/json</code>".


<h2 id=handling-a-resource>Handling a resource</h2>

<p class=XXX>
 A <dfn>resource</dfn> is ….

<p>
 For each <a>resource</a> it handles, the user agent must keep track of
 the following associated metadata:

 <ul>
  <li>
   A <dfn export>supplied MIME type</dfn>, the <a>MIME type</a> determined by
   the <a>supplied MIME type detection algorithm</a>.

  <li>
   A <dfn>check-for-apache-bug flag</dfn>, which defaults to unset.

  <li>
   A <dfn>no-sniff flag</dfn>, which defaults to set if the user agent does
   not wish to perform sniffing on the <a>resource</a> and unset
   otherwise.

   <p class=note>
    The user agent can choose to use outside information, such as previous
    experience with a site, to determine whether to opt out of sniffing for a
    particular <a>resource</a>. The user agent can also choose to opt
    out of sniffing for all <a>resources</a>. However,
    opting out of sniffing does not exempt the user agent from using the
    <a>MIME type sniffing algorithm</a>.

  <li>
   A <dfn export>computed MIME type</dfn>, the <a>MIME type</a>
   determined by the <a>MIME type sniffing algorithm</a>.
 </ul>



<h3 id=interpreting-the-resource-metadata>Interpreting the resource metadata</h3>

<p>
 The <a>supplied MIME type</a> of a <a>resource</a> is provided
 to the user agent by an external source associated with that
 <a>resource</a>.
 The method of obtaining this information varies depending upon how the
 <a>resource</a> is retrieved.

<p>
 To determine the <a>supplied MIME type</a> of a <a>resource</a>,
 user agents must use the following <dfn>supplied MIME type detection
 algorithm</dfn>:

 <ol>
  <li>
   Let <var>supplied-type</var> be null.

  <li>
   If the <a>resource</a> is retrieved via HTTP, execute the following
   steps:

   <ol>
    <li>
     If one or more <code>Content-Type</code> headers are associated with the
     <a>resource</a>, execute the following steps:

     <ol>
      <li>
       Set <var>supplied-type</var> to the value of the last
       <code>Content-Type</code> header associated with the
       <a>resource</a>.

       <p class=note>
        File extensions are not used to determine the <a>supplied MIME
        type</a> of a <a>resource</a> retrieved via HTTP because they are
        unreliable and easily spoofed.

      <li>
       Set the <a>check-for-apache-bug flag</a> if
       <var>supplied-type</var> is <strong>exactly</strong> equal to one of
       the values in the following table:

       <table>
        <thead>
         <tr>
          <th>Bytes in Hexadecimal
          <th>Bytes in ASCII


        <tbody>
         <tr>
          <td>
           74 65 78 74 2F 70 6C 61 69 6E

          <td>
           <code>text/plain</code>


         <tr>
          <td>
           74 65 78 74 2F 70 6C 61 69 6E<br>
           3B 20 63 68 61 72 73 65 74 3D<br>
           49 53 4F 2D 38 38 35 39 2D 31

          <td>
           <code>text/plain; charset=ISO-8859-1</code>


         <tr>
          <td>
           74 65 78 74 2F 70 6C 61 69 6E<br>
           3B 20 63 68 61 72 73 65 74 3D<br>
           69 73 6F 2D 38 38 35 39 2D 31

          <td>
           <code>text/plain; charset=iso-8859-1</code>


         <tr>
          <td>
           74 65 78 74 2F 70 6C 61 69 6E<br>
           3B 20 63 68 61 72 73 65 74 3D<br>
           55 54 46 2D 38

          <td>
           <code>text/plain; charset=UTF-8</code>



       </table>

       <p class=note>
        The <a>supplied MIME type detection algorithm</a> detects these
        exact <a>byte</a> sequences because some older installations of
        Apache contain
        <a href="https://issues.apache.org/bugzilla/show_bug.cgi?id=13986">a
        bug</a> that causes them to supply one of these Content-Type headers
        when serving files with unrecognized <a lt="MIME type">MIME
        types</a>.
     </ol>
   </ol>

   [[!HTTP-SEMANTICS]]

  <li>
   If the <a>resource</a> is retrieved directly from the file system,
   set <var>supplied-type</var> to the <a>MIME type</a> provided by the
   file system.

  <li>
   If the <a>resource</a> is retrieved via another protocol (such as
   FTP), set <var>supplied-type</var> to the <a>MIME type</a> as
   determined by that protocol, if any.

   [[FTP]]

  <li>
   If <var>supplied-type</var> is not a <a>MIME type</a>, the
   <a>supplied MIME type</a> is undefined.

   Abort these steps.

  <li>
   The <a>supplied MIME type</a> is <var>supplied-type</var>.
 </ol>



<h3 id=reading-the-resource-header>Reading the resource header</h3>

<p>
 A <dfn>resource header</dfn> is the <a>byte sequence</a> at the
 beginning of a <a>resource</a>, as determined by
 <a lt="read the resource header">reading the resource header</a>.

<p>
 To <dfn>read the resource header</dfn>, perform the following steps:

 <ol>
  <li>
   Let <var>buffer</var> be a <a>byte sequence</a>.

  <li>
   Read <a>bytes</a> of the <a>resource</a> into
   <var>buffer</var> until one of the following conditions is met:

   <ul>
    <li>
     the end of the <a>resource</a> is reached.

    <li>
     the number of <a>bytes</a> in <var>buffer</var> is
     greater than or equal to 1445.

    <li>
     a reasonable amount of time has elapsed, as determined by the user
     agent.
   </ul>

   <p class=note>
    If the number of <a>bytes</a> in <var>buffer</var> is
    greater than or equal to 1445, the <a>MIME type sniffing
    algorithm</a> will be deterministic for the majority of cases.

    However, certain factors (such as a slow connection) may prevent the
    user agent from reading 1445 <a>bytes</a> in a
    reasonable amount of time.

   <li>
    The <a>resource header</a> is <var>buffer</var>.
 </ol>

 <p class=note>
  The <a>resource header</a> need only be determined once per
  <a>resource</a>.



<h2 id=matching-a-mime-type-pattern>Matching a MIME type pattern</h2>

<p>
 A <dfn>byte pattern</dfn> is a <a>byte sequence</a> used as a template
 to be matched against in the <a>pattern matching algorithm</a>.

<p>
 A <dfn>pattern mask</dfn> is a <a>byte sequence</a> used to determine
 the significance of <a>bytes</a> being compared against a
 <a>byte pattern</a> in the <a>pattern matching algorithm</a>.

<p class=note>
 In a <a>pattern mask</a>, 0xFF indicates the <a>byte</a> is
 strictly significant, 0xDF indicates that the <a>byte</a> is
 significant in an ASCII case-insensitive way, and 0x00 indicates that the
 <a>byte</a> is not significant.

<p>To determine whether a <a>byte sequence</a> matches a particular <a>byte pattern</a>, use the
following <dfn>pattern matching algorithm</dfn>. It is given a <a>byte sequence</a>
<var>input</var>, a <a>byte pattern</a> <var>pattern</var>, a <a>pattern mask</a> <var>mask</var>,
and a <a for=/>set</a> of <a>bytes</a> to be ignored <var>ignored</var>, and returns true or false.

<ol>
 <li><p>Assert: <var>pattern</var>'s <a for="byte sequence">length</a> is equal to
 <var>mask</var>'s <a for="byte sequence">length</a>.

 <li><p>If <var>input</var>'s <a for="byte sequence">length</a> is less than <var>pattern</var>'s
 <a for="byte sequence">length</a>, return false.

 <li><p>Let <var>s</var> be 0.

 <li>
  <p>While <var>s</var> &lt; <var>input</var>'s <a for="byte sequence">length</a>:

  <ol>
   <li><p>If <var>ignored</var> does not <a for=set>contain</a> <var>input</var>[<var>s</var>],
   <a for=iteration>break</a>.

   <li><p>Set <var>s</var> to <var>s</var> + 1.
  </ol>

 <li><p>Let <var>p</var> be 0.

 <li>
  <p>While <var>p</var> &lt; <var>pattern</var>'s <a for="byte sequence">length</a>:

  <ol>
   <li><p>Let <var>maskedData</var> be the result of applying the bitwise AND operator to
   <var>input</var>[<var>s</var>] and <var>mask</var>[<var>p</var>].

   <li><p>If <var>maskedData</var> is not equal to <var>pattern</var>[<var>p</var>], return false.

   <li><p>Set <var>s</var> to <var>s</var> + 1.

   <li><p>Set <var>p</var> to <var>p</var> + 1.
  </ol>

 <li><p>Return true.
</ol>



<h3 id=matching-an-image-type-pattern>Matching an image type pattern</h3>

<p>To determine which <a>image MIME type</a> <a>byte pattern</a> a <a>byte sequence</a>
<var>input</var> matches, if any, use the following
<dfn export>image type pattern matching algorithm</dfn>:

 <ol>
  <li><p>Execute the following steps for each row <var>row</var> in the following table:

   <ol>
    <li><p>Let <var>patternMatched</var> be the result of the <a>pattern matching algorithm</a>
    given <var>input</var>, the value in the first column of <var>row</var>, the value in the second
    column of <var>row</var>, and the value in the third column of <var>row</var>.

    <li><p>If <var>patternMatched</var> is true, return the value in the fourth column of
    <var>row</var>.
   </ol>

   <table>
    <thead>
     <tr>
      <th>
       <a>Byte Pattern</a>

      <th>
       <a>Pattern Mask</a>

      <th>
       Leading <a lt=byte>Bytes</a> to Be Ignored

      <th>
       <a>Image MIME Type</a>

      <th>
       Note



    <tbody>
     <!-- https://www.iana.org/assignments/media-types/image/vnd.microsoft.icon -->
     <!-- https://msdn.microsoft.com/en-us/library/ms997538.aspx -->
     <tr>
      <td>
       00 00 01 00

      <td>
       FF FF FF FF

      <td>
       None.

      <td>
       <code>image/x-icon</code>

      <td>
       A Windows Icon signature.


     <!-- https://msdn.microsoft.com/en-us/library/ms997538.aspx -->
     <tr>
      <td>
       00 00 02 00

      <td>
       FF FF FF FF

      <td>
       None.

      <td>
       <code>image/x-icon</code>

      <td>
       A Windows Cursor signature.


     <tr>
      <td>
       42 4D

      <td>
       FF FF

      <td>
       None.

      <td>
       <code>image/bmp</code>

      <td>
       The string "<code>BM</code>", a BMP signature.


     <!-- https://www.w3.org/Graphics/GIF/spec-gif87.txt -->
     <tr>
      <td>
       47 49 46 38 37 61

      <td>
       FF FF FF FF FF FF

      <td>
       None.

      <td>
       <code>image/gif</code>

      <td>
       The string "<code>GIF87a</code>", a GIF signature.


     <!-- https://www.w3.org/Graphics/GIF/spec-gif89a.txt -->
     <tr>
      <td>
       47 49 46 38 39 61

      <td>
       FF FF FF FF FF FF

      <td>
       None.

      <td>
       <code>image/gif</code>

      <td>
       The string "<code>GIF89a</code>", a GIF signature.


     <!-- https://developers.google.com/speed/webp/docs/riff_container#webp-file-header -->
     <tr>
      <td>
       52 49 46 46 00 00 00 00 57 45 42 50 56 50

      <td>
       FF FF FF FF 00 00 00 00 FF FF FF FF FF FF

      <td>
       None.

      <td>
       <code>image/webp</code>

      <td>
       The string "<code>RIFF</code>" followed by four
       <a>bytes</a> followed by the string
       "<code>WEBPVP</code>".


     <!-- https://www.w3.org/TR/PNG/#5PNG-file-signature -->
     <tr>
      <td>
       89 50 4E 47 0D 0A 1A 0A

      <td>
       FF FF FF FF FF FF FF FF

      <td>
       None.

      <td>
       <code>image/png</code>

      <td>
       An error-checking <a>byte</a> followed by the string
       "<code>PNG</code>" followed by CR LF SUB LF, the PNG signature.


     <!-- https://www.digicamsoft.com/itu/itu-t81-36.html -->
     <tr>
      <td>
       FF D8 FF

      <td>
       FF FF FF

      <td>
       None.

      <td>
       <code>image/jpeg</code>

      <td>
       The JPEG Start of Image marker followed by the indicator
       <a>byte</a> of another marker.



   </table>

  <li><p>Return undefined.
 </ol>



<h3 id=matching-an-audio-or-video-type-pattern>Matching an audio or video type pattern</h3>

<p>To determine which <a>audio or video MIME type</a> <a>byte pattern</a> a <a>byte sequence</a>
<var>input</var> matches, if any, use the following <dfn export>audio or video type pattern matching
algorithm</dfn>:

 <ol>
  <li><p>Execute the following steps for each row <var>row</var> in the following table:

   <ol>
    <li><p>Let <var>patternMatched</var> be the result of the <a>pattern matching algorithm</a>
    given <var>input</var>, the value in the first column of <var>row</var>, the value in the second
    column of <var>row</var>, and the value in the third column of <var>row</var>.

    <li><p>If <var>patternMatched</var> is true, return the value in the fourth column of
    <var>row</var>.
   </ol>

   <table>
    <thead>
     <tr>
      <th>
       <a>Byte Pattern</a>

      <th>
       <a>Pattern Mask</a>

      <th>
       Leading <a lt=byte>Bytes</a> to Be Ignored

      <th>
       <a>Audio or Video MIME Type</a>

      <th>
       Note



    <tbody>
     <!-- https://multimedia.cx/mirror/AudioIFF1_2_1.htm -->
     <tr>
      <td>
       46 4F 52 4D 00 00 00 00 41 49 46 46

      <td>
       FF FF FF FF 00 00 00 00 FF FF FF FF

      <td>
       None.

      <td>
       <code>audio/aiff</code>

      <td>
       The string "<code>FORM</code>" followed by four
       <a>bytes</a> followed by the string
       "<code>AIFF</code>", the AIFF signature.


     <!-- http://id3.org/id3v2.4.0-structure -->
     <tr>
      <td>
       49 44 33

      <td>
       FF FF FF

      <td>
       None.

      <td>
       <code>audio/mpeg</code>

      <td>
       The string "<code>ID3</code>", the ID3v2-tagged MP3 signature.


     <tr>
      <td>
       4F 67 67 53 00

      <td>
       FF FF FF FF FF

      <td>
       None.

      <td>
       <code>application/ogg</code>

      <td>
       The string "<code>OggS</code>" followed by NUL, the Ogg container
       signature.


     <!-- https://hostr.co/LJUEw0K -->
     <tr>
      <td>
       4D 54 68 64 00 00 00 06

      <td>
       FF FF FF FF FF FF FF FF

      <td>
       None.

      <td>
       <code>audio/midi</code>

      <td>
       The string "<code>MThd</code>" followed by four
       <a>bytes</a> representing the number 6 in 32 bits
       (big-endian), the MIDI signature.


     <!-- http://www.jmcgowan.com/avitech.html#Format -->
     <!-- http://www.the-labs.com/Video/odmlff2-avidef.pdf -->
     <tr>
      <td>
       52 49 46 46 00 00 00 00 41 56 49 20

      <td>
       FF FF FF FF 00 00 00 00 FF FF FF FF

      <td>
       None.

      <td>
       <code>video/avi</code>

      <td>
       The string "<code>RIFF</code>" followed by four
       <a>bytes</a> followed by the string
       "<code>AVI </code>", the AVI signature.


     <!-- http://www-mmsp.ece.mcgill.ca/Documents/AudioFormats/WAVE/WAVE.html -->
     <tr>
      <td>
       52 49 46 46 00 00 00 00 57 41 56 45

      <td>
       FF FF FF FF 00 00 00 00 FF FF FF FF

      <td>
       None.

      <td>
       <code>audio/wave</code>

      <td>
       The string "<code>RIFF</code>" followed by four
       <a>bytes</a> followed by the string
       "<code>WAVE</code>", the WAVE signature.



   </table>

  <li><p>If <var>input</var> <a>matches the signature for MP4</a>, return "<code>video/mp4</code>".

  <li><p>If <var>input</var> <a>matches the signature for WebM</a>, return
  "<code>video/webm</code>".

  <li><p>If <var>input</var> <a>matches the signature for MP3 without ID3</a>, return
  "<code>audio/mpeg</code>".

  <li><p>Return undefined.
 </ol>



<h4 id=signature-for-mp4>Signature for MP4</h4>

<p>
 To determine whether a <a>byte sequence</a> <dfn>matches the signature
 for MP4</dfn>, use the following steps:

 <ol>
  <li>
   Let <var>sequence</var> be the <a>byte sequence</a> to be matched,
   where <var>sequence</var>[<var>s</var>] is <a>byte</a> <var>s</var>
   in <var>sequence</var> and <var>sequence</var>[0] is the first
   <a>byte</a> in <var>sequence</var>.

  <li>
   Let <var>length</var> be the number of <a>bytes</a> in
   <var>sequence</var>.

  <li>
   If <var>length</var> is less than 12, return false.

  <li>
   Let <var>box-size</var> be the four <a>bytes</a> from
   <var>sequence</var>[0] to <var>sequence</var>[3], interpreted as a 32-bit
   unsigned big-endian integer.

  <li>
   If <var>length</var> is less than <var>box-size</var> or if
   <var>box-size</var> modulo 4 is not equal to 0, return false.

  <li>
   If the four <a>bytes</a> from <var>sequence</var>[4] to
   <var>sequence</var>[7] are not equal to 0x66 0x74 0x79 0x70
   ("<code>ftyp</code>"), return false.

  <li>
   If the three <a>bytes</a> from <var>sequence</var>[8] to
   <var>sequence</var>[10] are equal to 0x6D 0x70 0x34
   ("<code>mp4</code>"), return true.

  <li>
   Let <var>bytes-read</var> be 16.

   <p class=note>
    This ignores the four <a>bytes</a> that correspond to the
    version number of the "major brand".

  <li>
   While <var>bytes-read</var> is less than <var>box-size</var>, continuously
   loop through these steps:

   <ol>
    <li>
     If the three <a>bytes</a> from
     <var>sequence</var>[<var>bytes-read</var>] to
     <var>sequence</var>[<var>bytes-read</var> + 2] are equal to 0x6D 0x70
     0x34 ("<code>mp4</code>"), return true.

    <li>
     Increment <var>bytes-read</var> by 4.
   </ol>

  <li>
   Return false.
 </ol>

<h4 id=signature-for-webm>Signature for WebM</h4>

<!-- https://matroska-org.github.io/libebml/specs.html -->
<!-- https://www.matroska.org/technical/specs/index.html -->
<!-- https://www.webmproject.org/docs/container/ -->

<p>
 To determine whether a <a>byte sequence</a> <dfn>matches the signature
 for WebM</dfn>, use the following steps:

 <ol>
  <li>
   Let <var>sequence</var> be the <a>byte sequence</a> to be matched,
   where <var>sequence</var>[<var>s</var>] is <a>byte</a> <var>s</var>
   in <var>sequence</var> and <var>sequence</var>[0] is the first
   <a>byte</a> in <var>sequence</var>.

  <li>
   Let <var>length</var> be the number of <a>bytes</a> in
   <var>sequence</var>.

  <li>
   If <var>length</var> is less than 4, return false.

  <li>
   If the four <a>bytes</a> from
   <var>sequence</var>[0] to <var>sequence</var>[3], are not equal to 0x1A
   0x45 0xDF 0xA3, return false.

  <li>
   Let <var>iter</var> be 4.

  <li>
  While <var>iter</var> is less than <var>length</var> and <var>iter</var> is
  less than 38, continuously loop through these steps:

  <ol>
    <li> If the two bytes from <var>sequence</var>[<var>iter</var>] to
    <var>sequence</var>[<var>iter</var> + 1] are equal to 0x42 0x82,
    <ol>
      <li> Increment <var>iter</var> by 2.
      <li> If <var>iter</var> is greater or equal than <var>length</var>, abort
      these steps.
      <li> Let <var>number size</var> be the result
      of <a href=#parse-a-vint>parsing a <code>vint</code></a> starting at <var>sequence</var>[<var>iter</var>].
      <li> Increment <var>iter</var> by <var>number size</var>.
      <li> If <var>iter</var> is greater than or equal to <var>length</var> - 4, abort these
      steps.
      <li> Let <var>matched</var> be the result of <a href=#matching-a-padded-sequence>matching a padded
        sequence</a> 0x77 0x65 0x62 0x6D
      ("<code>webm</code>") on <var>sequence</var> at offset
      <var>iter</var>.
      <li> If <var>matched</var> is true, abort these steps and return true.
    </ol>
    <li> Increment <var>iter</var> by 1.
  </ol>
  <li> Return false.
  </ol>

To <dfn>parse a <code>vint</code></dfn> on a <a>byte sequence</a>
<var>sequence</var> of size <var>length</var>, starting at index <var>iter</var>
use the following steps:
<ol>
 <li>Let <var>mask</var> be 128.
 <li>Let <var>max vint length</var> be 8.
 <li>Let <var>number size</var> be 1.
 <li>Let <var>index</var> be 0.
 <li>
  While <var>number size</var> is less than <var>max vint length</var>, and less than
  <var>length</var>, continuously loop through these steps:

  <ol>
   <li>If the <var>sequence</var>[<var>index</var>] &amp; <var>mask</var> is
   not zero, abort these steps.
   <li>Let <var>mask</var> be the value of <var>mask</var> >> 1.
   <li>Increment <var>number size</var> by one.
  </ol>
 <li>Let <var>parsed number</var> be <var>sequence</var>[<var>index</var>] &amp; ~<var>mask</var>.
 <li>Increment <var>index</var> by one.
 <li>Let <var>bytes remaining</var> be the value of <var>number size</var> - 1.
 <li>
  While <var>bytes remaining</var> is not zero, execute there steps:
  <ol>
   <li>Let <var>parsed number</var> be <var>parsed number</var> &lt;&lt; 8.
   <li>Let <var>parsed number</var> be <var>parsed number</var> |
   <var>sequence</var>[<var>index</var>].
   <li>Increment <var>index</var> by one.
   <li>If <var>index</var> is greater or equal than <var>length</var>, abort these steps.
   <li>Decrement <var>bytes remaining</var> by one.
  </ol>
 <li>Return <var>parsed number</var> and <var>number size</var>
</ol>

<p>
<dfn>Matching a padded sequence</dfn> <var>pattern</var> on a sequence
<var>sequence</var> at starting at byte <var>offset</var> and ending at by
<var>end</var> means returning true if <var>sequence</var> has a length greater
than <var>end</var>, and contains exactly, in the range [<var>offset</var>,
<var>end</var>], the bytes in <var>pattern</var>, in the same order, eventually
preceded by bytes with a value of 0x00, false otherwise.


<h4 id=signature-for-mp3-without-id3>Signature for MP3 without ID3</h4>

To determine whether a <a>byte sequence</a> <dfn>matches the signature
for MP3 without ID3</dfn>, use the following steps:

<ol>
  <li>
  Let <var>sequence</var> be the <a>byte sequence</a> to be matched,
  where <var>sequence</var>[<var>s</var>] is <a>byte</a> <var>s</var> in
  <var>sequence</var> and <var>sequence</var>[0] is the first <a>byte</a>
  in <var>sequence</var>.
  <li> Let <var>length</var> be the number of <a>bytes</a> in
  <var>sequence</var>.
  <li>Initialize s to 0.
  <li>If the result of the operation <a href=#match-an-mp3-header>match mp3
  header</a> is false, return false.
  <li><a href=#parse-an-mp3-frame>Parse an mp3 frame</a> on
  <var>sequence</var> at offset <var>s</var>
  <li>Let <var>skipped-bytes</var> the return value of the execution of <a href=#compute-an-mp3-frame-size>mp3 framesize computation</a>
  <li>If <var>skipped-bytes</var> is less than 4, or <var>skipped-bytes</var> is
  greater than <var>s</var> - <var>length</var>, return false.
  <li>Increment s by <var>skipped-bytes</var>.
  <li>If the result of the operation <a href=#match-an-mp3-header>match mp3
  header</a> operation is false, return false, else, return true.
</ol>

To <dfn>match an mp3 header</dfn>, using a <a>byte sequence</a>
<var>sequence</var> of length <var>length</var> at offset <var>s</var>
execute these steps:
<ol>
  <li>If <var>length</var> is less than 4, return false.
  <li>If <var>sequence</var>[<var>s</var>] is not equal to 0xff and
  <var>sequence</var>[<var>s</var> + 1] &amp; 0xe0 is not equal to 0xe0, return
  false.
  <li>Let <var>layer</var> be the result of <var>sequence</var>[<var>s</var> +
  1] &amp; 0x06 >> 1.
  <li>If <var>layer</var> is 0, return false.
  <li>Let <var>bit-rate</var> be <var>sequence</var>[<var>s</var> + 2] &amp;
  0xf0 >> 4.
  <li>If <var>bit-rate</var> is 15, return false.
  <li>Let <var>sample-rate</var> be <var>sequence</var>[<var>s</var> + 2] &amp;
  0x0c >> 2.
  <li>If <var>sample-rate</var> is 3, return false.
  <li>Let <var>freq</var> be the value given by <var>sample-rate</var>
  in the table sample-rate.
  <li>Let <var>final-layer</var> be the result of 4 - ((<var>sequence</var>[<var>s</var> + 1] &amp; 0x06) >> 1).
  <li>If <var>final-layer</var> is not 3, return false.
  <li>Return true.
</ol>

To <dfn>compute an mp3 frame size</dfn>, execute these steps:
<ol>
  <li>If <var>version</var> is 1, let <var>scale</var> be 72, else, let
  <var>scale</var> be 144.
  <li>Let <var>size</var> be <var>bitrate</var> * <var>scale</var> /
  <var>freq</var>.
  <li>If <var>pad</var> is not zero, increment <var>size</var> by 1.
  <li>Return <var>size</var>.
</ol>

To <dfn>parse an mp3 frame</dfn>, execute these steps:
<ol>
 <li>Let <var>version</var> be <var>sequence</var>[<var>s</var> + 1] &amp; 0x18 >> 3.
 <li>Let <var>bitrate-index</var> be sequence[s + 2] &amp; 0xf0 >> 4.
 <li>If the <var>version</var> &amp; 0x01 is non-zero, let <var>bitrate</var> be the value given by
 <var>bitrate-index</var> in the table mp2.5-rates
 <li>If <var>version</var> &amp; 0x01 is zero, let <var>bitrate</var> be the value given by
 <var>bitrate-index</var> in the table mp3-rates
 <li>Let <var>samplerate-index</var> be <var>sequence</var>[<var>s</var> + 2] &amp; 0x0c >> 2.
 <li>Let <var ignore>samplerate</var> be the value given by <var>samplerate-index</var> in the
 <var>sample-rate</var> table.
 <li>If <var>version</var> is 2, let <var>samplerate</var> be the result of <var>samplerate</var> / 2.
 <li>If <var>version</var> is 0, let <var>samplerate</var> be the result of <var>samplerate</var> / 4.
 <li>Let <var>pad</var> be <var>sequence</var>[<var>s</var> + 2] &amp; 0x02 >> 1.
</ol>

<figure>
<figcaption>mp3-rates table</figcaption>
<table>
<thead>
<tr><th>index
<th>mp3-rates


<tbody>
<tr><td>0
<td>0

<tr><td>1
<td>32000

<tr><td>2
<td>40000

<tr><td>3
<td>48000

<tr><td>4
<td>56000

<tr><td>5
<td>64000

<tr><td>6
<td>80000

<tr><td>7
<td>96000

<tr><td>8
<td>112000

<tr><td>9
<td>128000

<tr><td>10
<td>160000

<tr><td>11
<td>192000

<tr><td>12
<td>224000

<tr><td>13
<td>256000

<tr><td>14
<td>320000


</table>
</figure>

<figure>
<figcaption>mp2.5-rates</figcaption>
<table>
<thead>
<tr><th>index
<th>mp2.5-rates


<tbody>
<tr><td>0
<td>0

<tr><td>1
<td>8000

<tr><td>2
<td>16000

<tr><td>3
<td>24000

<tr><td>4
<td>32000

<tr><td>5
<td>40000

<tr><td>6
<td>48000

<tr><td>7
<td>56000

<tr><td>8
<td>64000

<tr><td>9
<td>80000

<tr><td>10
<td>96000

<tr><td>11
<td>112000

<tr><td>12
<td>128000

<tr><td>13
<td>144000

<tr><td>14
<td>160000


</table>
</figure>

<figure>
<figcaption>sample-rate table</figcaption>
<table>
<thead>
<tr><th>index
<th>samplerate


<tbody>
<tr><td>0
<td>44100

<tr><td>1
<td>48000

<tr><td>2
<td>32000


</table>
</figure>

<h3 id=matching-a-font-type-pattern>Matching a font type pattern</h3>

<p>To determine which <a>font MIME type</a> <a>byte pattern</a> a <a>byte sequence</a>
<var>input</var> matches, if any, use the following <dfn>font type pattern matching algorithm</dfn>:

 <ol>
  <li><p>Execute the following steps for each row <var>row</var> in the following table:

   <ol>
    <li><p>Let <var>patternMatched</var> be the result of the <a>pattern matching algorithm</a>
    given <var>input</var>, the value in the first column of <var>row</var>, the value in the second
    column of <var>row</var>, and the value in the third column of <var>row</var>.

    <li><p>If <var>patternMatched</var> is true, return the value in the fourth column of
    <var>row</var>.
   </ol>

   <table>
    <thead>
     <tr>
      <th>
       <a>Byte Pattern</a>

      <th>
       <a>Pattern Mask</a>

      <th>
       Leading <a lt=byte>Bytes</a> to Be Ignored

      <th>
       <a>Font MIME Type</a>

      <th>
       Note



    <tbody>
     <!-- https://www.w3.org/Submission/EOT/ -->
     <tr>
      <td>
       00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
       00 00 00 00 00 00 00 00 00 00 4C 50

      <td>
       00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
       00 00 00 00 00 00 00 00 00 00 FF FF

      <td>
       None.

      <td>
       <code>application/vnd.ms-fontobject</code>

      <td>
       34 <a>bytes</a> followed by the string
       "<code>LP</code>", the Embedded OpenType signature.


     <!-- https://www.microsoft.com/typography/otspec/recom.htm -->
     <!-- https://www.ietf.org/mail-archive/web/websec/current/msg00235.html -->
     <tr>
      <td>
       00 01 00 00

      <td>
       FF FF FF FF

      <td>
       None.

      <td>
       <code>font/ttf</code>

      <td>
       4 <a>bytes</a> representing the version number 1.0, a
       TrueType signature.


     <!-- https://www.microsoft.com/typography/otspec/recom.htm -->
     <!-- https://www.ietf.org/mail-archive/web/websec/current/msg00235.html -->
     <tr>
      <td>
       4F 54 54 4F

      <td>
       FF FF FF FF

      <td>
       None.

      <td>
       <code>font/otf</code>

      <td>
       The string "<code>OTTO</code>", the OpenType signature.


     <!-- https://www.microsoft.com/typography/otspec/otff.htm -->
     <!-- https://www.ietf.org/mail-archive/web/websec/current/msg00235.html -->
     <tr>
      <td>
       74 74 63 66

      <td>
       FF FF FF FF

      <td>
       None.

      <td>
       <code>font/collection</code>

      <td>
       The string "<code>ttcf</code>", the TrueType Collection
       signature.


     <!-- https://www.ietf.org/mail-archive/web/websec/current/msg00235.html -->
     <tr>
      <td>
       77 4F 46 46

      <td>
       FF FF FF FF

      <td>
       None.

      <td>
       <code>font/woff</code>

      <td>
       The string "<code>wOFF</code>", the Web Open Font Format 1.0
       signature.

     <!-- https://www.w3.org/TR/WOFF2/ -->
     <tr>
      <td>
       77 4F 46 32

      <td>
       FF FF FF FF

      <td>
       None.

      <td>
       <code>font/woff2</code>

      <td>
       The string "<code>wOF2</code>", the Web Open Font Format 2.0
       signature.


   </table>

  <li><p>Return undefined.
 </ol>



<h3 id=matching-an-archive-type-pattern>Matching an archive type pattern</h3>

<p>To determine which <a>archive MIME type</a> <a>byte pattern</a> a <a>byte sequence</a>
<var>input</var> matches, if any, use the following <dfn>archive type pattern matching
algorithm</dfn>:

 <ol>
  <li><p>Execute the following steps for each row <var>row</var> in the following table:

   <ol>
    <li><p>Let <var>patternMatched</var> be the result of the <a>pattern matching algorithm</a>
    given <var>input</var>, the value in the first column of <var>row</var>, the value in the second
    column of <var>row</var>, and the value in the third column of <var>row</var>.

    <li><p>If <var>patternMatched</var> is true, return the value in the fourth column of
    <var>row</var>.
   </ol>

   <table>
    <thead>
     <tr>
      <th>
       <a>Byte Pattern</a>

      <th>
       <a>Pattern Mask</a>

      <th>
       Leading <a lt=byte>Bytes</a> to Be Ignored

      <th>
       <a>Archive MIME Type</a>

      <th>
       Note



    <tbody>
     <!-- https://tools.ietf.org/html/rfc1952 -->
     <tr>
      <td>
       <!-- 0x1F 0x8B = gzip format, 0x08 = deflate compression -->
       1F 8B 08

      <td>
       FF FF FF

      <td>
       None.

      <td>
       <code>application/x-gzip</code>

      <td>
       The GZIP archive signature.


     <tr>
      <td>
       50 4B 03 04

      <td>
       FF FF FF FF

      <td>
       None.

      <td>
       <code>application/zip</code>

      <td>
       The string "<code>PK</code>" followed by ETX EOT, the ZIP archive
       signature.


     <tr>
      <!-- RAR Legacy (up to 4.x) signature -->
      <!-- https://www.rarlab.com/technote.htm#rarsign -->
      <td>
       52 61 72 21 1A 07 00

      <td>
       FF FF FF FF FF FF FF

      <td>
       None.

      <td>
       <code>application/x-rar-compressed</code>

      <td>
       The string "<code>Rar!</code>" followed by SUB BEL NUL, the RAR
       4.x archive signature.



   </table>

  <li><p>Return undefined.
 </ol>



<h2 id=determining-the-computed-mime-type-of-a-resource>Determining the computed MIME type of a resource</h2>

<p>
 To determine the <a>computed MIME type</a> of a <a>resource</a>,
 user agents must use the following <dfn>MIME type sniffing algorithm</dfn>:

 <ol>
  <li>
   If the <a>supplied MIME type</a> is an <a>XML MIME type</a> or <a>HTML MIME type</a>, the
   <a>computed MIME type</a> is the <a>supplied MIME type</a>.

   Abort these steps.

  <li>
   If the <a>supplied MIME type</a> is undefined or if the
   <a>supplied MIME type</a>'s <a for="MIME type">essence</a> is
   "<code>unknown/unknown</code>",
   "<code>application/unknown</code>", or "<code>*/*</code>",
   execute the <a>rules for identifying an unknown MIME type</a> with
   the <a>sniff-scriptable flag</a> equal to the inverse of the
   <a>no-sniff flag</a> and abort these steps.

  <li>
   If the <a>no-sniff flag</a> is set, the <a>computed MIME
   type</a> is the <a>supplied MIME type</a>.

   Abort these steps.

  <li>
   If the <a>check-for-apache-bug flag</a> is set, execute the
   <a>rules for distinguishing if a resource is text or binary</a> and
   abort these steps.

  <li>
   If the <a>supplied MIME type</a> is an <a>image MIME type</a>
   <a>supported by the user agent</a>, let <var>matched-type</var> be
   the result of executing the <a>image type pattern matching
   algorithm</a> with the <a>resource header</a> as the <a>byte
   sequence</a> to be matched.

  <li>
   If <var>matched-type</var> is not undefined, the <a>computed MIME
   type</a> is <var>matched-type</var>.

   Abort these steps.

  <li>
   If the <a>supplied MIME type</a> is an <a>audio or video MIME type</a>
   <a>supported by the user agent</a>, let
   <var>matched-type</var> be the result of executing the <a>audio or video
   type pattern matching algorithm</a> with the <a>resource
   header</a> as the <a>byte sequence</a> to be matched.

  <li>
   If <var>matched-type</var> is not undefined, the <a>computed MIME
   type</a> is <var>matched-type</var>.

   Abort these steps.

  <li>
   The <a>computed MIME type</a> is the <a>supplied MIME
   type</a>.
 </ol>



<h3 id=identifying-a-resource-with-an-unknown-mime-type>Identifying a resource with an unknown MIME type</h3>

<p>
 The <dfn>sniff-scriptable flag</dfn> is used by the <a>rules for
 identifying an unknown MIME type</a> to determine whether to sniff for
 <a>scriptable MIME types</a>.

 If the setting of the <a>sniff-scriptable flag</a> is not specified
 when calling the <a>rules for identifying an unknown MIME type</a>,
 the <a>sniff-scriptable flag</a> must default to unset.

<p>To determine the <a>computed MIME type</a> of a <a>resource</a> <var>resource</var> with an
unknown <a>MIME type</a>, execute the following <dfn>rules for identifying an unknown MIME
type</dfn>:

 <ol>
  <li>
   <p>If the <a>sniff-scriptable flag</a> is set, execute the following steps for each row
   <var>row</var> in the following table:

   <ol>
    <li><p>Let <var>patternMatched</var> be the result of the <a>pattern matching algorithm</a>
    given <var>resource</var>'s <a>resource header</a>, the value in the first column of
    <var>row</var>, the value in the second column of <var>row</var>, and the value in the third
    column of <var>row</var>.

    <li><p>If <var>patternMatched</var> is true, return the value in the fourth column of
    <var>row</var>.
   </ol>

   <table>
    <thead>
     <tr>
      <th>
       <a>Byte Pattern</a>

      <th>
       <a>Pattern Mask</a>

      <th>
       Leading <a lt=byte>Bytes</a> to Be Ignored

      <th>
       <a>Computed MIME Type</a>

      <th>
       Note



    <tbody>
     <tr>
      <td>
       3C 21 44 4F 43 54 59 50 45 20 48 54 4D 4C <abbr lt="tag terminating byte">TT</abbr>

      <td>
       FF FF DF DF DF DF DF DF DF FF DF DF DF DF FF

      <td>
       <a lt="whitespace byte">Whitespace bytes</a>.

      <td>
       <code>text/html</code>

      <td>
       The case-insensitive string "<code>&lt;!DOCTYPE HTML</code>"
       followed by a <a>tag-terminating byte</a>.


     <tr>
      <td>
       3C 48 54 4D 4C <abbr lt="tag terminating byte">TT</abbr>

      <td>
       FF DF DF DF DF FF

      <td>
       <a lt="whitespace byte">Whitespace bytes</a>.

      <td>
       <code>text/html</code>

      <td>
       The case-insensitive string "<code>&lt;HTML</code>" followed by a
       <a>tag-terminating byte</a>.


     <tr>
      <td>
       3C 48 45 41 44 <abbr lt="tag terminating byte">TT</abbr>

      <td>
       FF DF DF DF DF FF

      <td>
       <a lt="whitespace byte">Whitespace bytes</a>.

      <td>
       <code>text/html</code>

      <td>
       The case-insensitive string "<code>&lt;HEAD</code>" followed by a
       <a>tag-terminating byte</a>.


     <tr>
      <td>
       3C 53 43 52 49 50 54 <abbr lt="tag terminating byte">TT</abbr>

      <td>
       FF DF DF DF DF DF DF FF

      <td>
       <a lt="whitespace byte">Whitespace bytes</a>.

      <td>
       <code>text/html</code>

      <td>
       The case-insensitive string "<code>&lt;SCRIPT</code>" followed by
       a <a>tag-terminating byte</a>.


     <tr>
      <td>
       3C 49 46 52 41 4D 45 <abbr lt="tag terminating byte">TT</abbr>

      <td>
       FF DF DF DF DF DF DF FF

      <td>
       <a lt="whitespace byte">Whitespace bytes</a>.

      <td>
       <code>text/html</code>

      <td>
       The case-insensitive string "<code>&lt;IFRAME</code>" followed by
       a <a>tag-terminating byte</a>.


     <tr>
      <td>
       3C 48 31 <abbr lt="tag terminating byte">TT</abbr>

      <td>
       FF DF FF FF

      <td>
       <a lt="whitespace byte">Whitespace bytes</a>.

      <td>
       <code>text/html</code>

      <td>
       The case-insensitive string "<code>&lt;H1</code>" followed by a
       <a>tag-terminating byte</a>.


     <tr>
      <td>
       3C 44 49 56 <abbr lt="tag terminating byte">TT</abbr>

      <td>
       FF DF DF DF FF

      <td>
       <a lt="whitespace byte">Whitespace bytes</a>.

      <td>
       <code>text/html</code>

      <td>
       The case-insensitive string "<code>&lt;DIV</code>" followed by a
       <a>tag-terminating byte</a>.


     <tr>
      <td>
       3C 46 4F 4E 54 <abbr lt="tag terminating byte">TT</abbr>

      <td>
       FF DF DF DF DF FF

      <td>
       <a lt="whitespace byte">Whitespace bytes</a>.

      <td>
       <code>text/html</code>

      <td>
       The case-insensitive string "<code>&lt;FONT</code>" followed by a
       <a>tag-terminating byte</a>.


     <tr>
      <td>
       3C 54 41 42 4C 45 <abbr lt="tag terminating byte">TT</abbr>

      <td>
       FF DF DF DF DF DF FF

      <td>
       <a lt="whitespace byte">Whitespace bytes</a>.

      <td>
       <code>text/html</code>

      <td>
       The case-insensitive string "<code>&lt;TABLE</code>" followed by
       a <a>tag-terminating byte</a>.


     <tr>
      <td>
       3C 41 <abbr lt="tag terminating byte">TT</abbr>

      <td>
       FF DF FF

      <td>
       <a lt="whitespace byte">Whitespace bytes</a>.

      <td>
       <code>text/html</code>

      <td>
       The case-insensitive string "<code>&lt;A</code>" followed by a
       <a>tag-terminating byte</a>.


     <tr>
      <td>
       3C 53 54 59 4C 45 <abbr lt="tag terminating byte">TT</abbr>

      <td>
       FF DF DF DF DF DF FF

      <td>
       <a lt="whitespace byte">Whitespace bytes</a>.

      <td>
       <code>text/html</code>

      <td>
       The case-insensitive string "<code>&lt;STYLE</code>" followed by
       a <a>tag-terminating byte</a>.


     <tr>
      <td>
       3C 54 49 54 4C 45 <abbr lt="tag terminating byte">TT</abbr>

      <td>
       FF DF DF DF DF DF FF

      <td>
       <a lt="whitespace byte">Whitespace bytes</a>.

      <td>
       <code>text/html</code>

      <td>
       The case-insensitive string "<code>&lt;TITLE</code>" followed by
       a <a>tag-terminating byte</a>.


     <tr>
      <td>
       3C 42 <abbr lt="tag terminating byte">TT</abbr>

      <td>
       FF DF FF

      <td>
       <a lt="whitespace byte">Whitespace bytes</a>.

      <td>
       <code>text/html</code>

      <td>
       The case-insensitive string "<code>&lt;B</code>" followed by a
       <a>tag-terminating byte</a>.


     <tr>
      <td>
       3C 42 4F 44 59 <abbr lt="tag terminating byte">TT</abbr>

      <td>
       FF DF DF DF DF FF

      <td>
       <a lt="whitespace byte">Whitespace bytes</a>.

      <td>
       <code>text/html</code>

      <td>
       The case-insensitive string "<code>&lt;BODY</code>" followed by a
       <a>tag-terminating byte</a>.


     <tr>
      <td>
       3C 42 52 <abbr lt="tag terminating byte">TT</abbr>

      <td>
       FF DF DF FF

      <td>
       <a lt="whitespace byte">Whitespace bytes</a>.

      <td>
       <code>text/html</code>

      <td>
       The case-insensitive string "<code>&lt;BR</code>" followed by a
       <a>tag-terminating byte</a>.


     <tr>
      <td>
       3C 50 <abbr lt="tag terminating byte">TT</abbr>

      <td>
       FF DF FF

      <td>
       <a lt="whitespace byte">Whitespace bytes</a>.

      <td>
       <code>text/html</code>

      <td>
       The case-insensitive string "<code>&lt;P</code>" followed by a
       <a>tag-terminating byte</a>.


     <tr>
      <td>
       3C 21 2D 2D <abbr lt="tag terminating byte">TT</abbr>

      <td>
       FF FF FF FF FF

      <td>
       <a lt="whitespace byte">Whitespace bytes</a>.

      <td>
       <code>text/html</code>

      <td>
       The string "<code>&lt;!--</code>" followed by a
       <a>tag-terminating byte</a>.


     <tr>
      <td>
       3C 3F 78 6D 6C

      <td>
       FF FF FF FF FF

      <td>
       <a lt="whitespace byte">Whitespace bytes</a>.

      <td>
       <code>text/xml</code>

      <td>
       The string "<code>&lt;?xml</code>".


     <tr>
      <td>
       25 50 44 46 2D

      <td>
       FF FF FF FF FF

      <td>
       None.

      <td>
       <code>application/pdf</code>

      <td>
       The string "<code>%PDF-</code>", the PDF signature.



   </table>

  <li>
   <p>Execute the following steps for each row <var>row</var> in the following table:

   <ol>
    <li><p>Let <var>patternMatched</var> be the result of the <a>pattern matching algorithm</a>
    given <var>resource</var>'s <a>resource header</a>, the value in the first column of
    <var>row</var>, the value in the second column of <var>row</var>, and the value in the third
    column of <var>row</var>.

    <li><p>If <var>patternMatched</var> is true, return the value in the fourth column of
    <var>row</var>.
   </ol>

   <table>
    <thead>
     <tr>
      <th>
       <a>Byte Pattern</a>

      <th>
       <a>Pattern Mask</a>

      <th>
       Leading <a lt=byte>Bytes</a> to Be Ignored

      <th>
       <a>Computed MIME Type</a>

      <th>
       Note



    <tbody>
     <tr>
      <td>
       25 21 50 53 2D 41 64 6F 62 65 2D

      <td>
       FF FF FF FF FF FF FF FF FF FF FF

      <td>
       None.

      <td>
       <code>application/postscript</code>

      <td>
       The string "<code>%!PS-Adobe-</code>", the PostScript signature.


     <tr>
      <td>
       FE FF 00 00

      <td>
       FF FF 00 00

      <td>
       None.

      <td>
       <code>text/plain</code>

      <td>
       UTF-16BE BOM


     <tr>
      <td>
       FF FE 00 00

      <td>
       FF FF 00 00

      <td>
       None.

      <td>
       <code>text/plain</code>

      <td>
       UTF-16LE BOM


     <tr>
      <td>
       EF BB BF 00

      <td>
       FF FF FF 00

      <td>
       None.

      <td>
       <code>text/plain</code>

      <td>
       UTF-8 BOM



   </table>

   <p>User agents may implicitly extend this table to support additional <a>MIME types</a>.

   <p>However, user agents should not implicitly extend this table to include additional
   <a>byte patterns</a> for any <a>computed MIME type</a> already present in this table, as doing so
   could introduce privilege escalation vulnerabilities.

   <p>User agents must not introduce any privilege escalation vulnerabilities when extending this
   table.

  <li><p>Let <var>matchedType</var> be the result of executing the
  <a>image type pattern matching algorithm</a> given <var>resource</var>'s <a>resource header</a>.

  <li><p>If <var>matchedType</var> is not undefined, return <var>matchedType</var>.

  <li><p>Set <var>matchedType</var> to the result of executing the
  <a>audio or video type pattern matching algorithm</a> given <var>resource</var>'s
  <a>resource header</a>.

  <li><p>If <var>matchedType</var> is not undefined, return <var>matchedType</var>.

  <!-- Major browsers do not implement font sniffing in the regular browsing context.
  <li>
   Let <var>matched-type</var> be the result of executing the <span>font type
   pattern matching algorithm</span> with the <span>resource header</span> as
   the <span>byte sequence</span> to be matched.

  <li>
   If <var>matched-type</var> is not undefined, the <span>computed MIME
   type</span> is <var>matched-type</var>.

   Abort these steps.
  -->

  <li><p>Set <var>matchedType</var> to the result of executing the
  <a>archive type pattern matching algorithm</a> given <var>resource</var>'s
  <a>resource header</a>.

  <li><p>If <var>matchedType</var> is not undefined, return <var>matchedType</var>.

  <li><p>If <var>resource</var>'s <a>resource header</a> contains no <a>binary data bytes</a>,
  return  "<code>text/plain</code>".

  <li><p>Return "<code>application/octet-stream</code>".
 </ol>



<h3 id=sniffing-a-mislabeled-binary-resource>Sniffing a mislabeled binary resource</h3>

<p>
 To determine whether a binary <a>resource</a> has been mislabeled as
 plain text, execute the following <dfn id=rules-for-text-or-binary>rules for
 distinguishing if a resource is text or binary</dfn>:

<ol>
 <li>
  Let <var>length</var> be the number of <a>bytes</a> in the
  <a>resource header</a>.

 <li>
  If <var>length</var> is greater than or equal to 2 and the first 2 bytes of
  the <a>resource header</a> are equal to 0xFE 0xFF (UTF-16BE BOM) or
  0xFF 0xFE (UTF-16LE BOM), the <a>computed MIME type</a> is
  "<code>text/plain</code>".

  Abort these steps.

 <li>
  If <var>length</var> is greater than or equal to 3 and the first 3 bytes of
  the <a>resource header</a> are equal to 0xEF 0xBB 0xBF (UTF-8 BOM),
  the <a>computed MIME type</a> is "<code>text/plain</code>".

  Abort these steps.

 <li>
  If the <a>resource header</a> contains no
  <a>binary data bytes</a>, the <a>computed
  MIME type</a> is "<code>text/plain</code>".

  Abort these steps.

 <li>
  The <a>computed MIME type</a> is
  "<code>application/octet-stream</code>".

  <p class=warning>
   It is critical that the <a>rules for distinguishing if a resource is
   text or binary</a> never determine the <a>computed MIME type</a>
   to be a <a>scriptable MIME type</a>, as this could allow a privilege
   escalation attack.
</ol>



<h2 id=context-specific-sniffing>Context-specific sniffing</h2>

<p class=XXX>
 A <dfn>context</dfn> is ….

<p>
 In certain <a>contexts</a>, it is only useful to identify
 <a>resources</a> that belong to a certain subset of
 <a>MIME types</a>.

 In such <a>contexts</a>, it is appropriate to use a
 <a>context-specific sniffing algorithm</a> in place of the <a>MIME
 type sniffing algorithm</a> in order to determine the <a>computed MIME
 type</a> of a <a>resource</a>.

<p>
 A <dfn>context-specific sniffing algorithm</dfn> determines the
 <a>computed MIME type</a> of a <a>resource</a> only if the
 <a>resource</a> is a <a>MIME type</a> relevant to a particular
 <a>context</a>.



<h3 id=sniffing-in-a-browsing-context>Sniffing in a browsing context</h3>

<p class=XXX>
 Use the <a>MIME type sniffing algorithm</a>.



<h3 id=sniffing-in-an-image-context>Sniffing in an image context</h3>

<p>
 To determine the <a>computed MIME type</a> of a <a>resource</a>
 with an <a>image MIME type</a>, execute the following <dfn export>rules for
 sniffing images specifically</dfn>:

 <ol>
  <li>
   If the <a>supplied MIME type</a> is an <a>XML MIME type</a>, the
   <a>computed MIME type</a> is the <a>supplied MIME type</a>.

   Abort these steps.

  <li>
   Let <var>image-type-matched</var> be the result of executing the
   <a>image type pattern matching algorithm</a> with the <a>resource
   header</a> as the <a>byte sequence</a> to be matched.

  <li>
   If <var>image-type-matched</var> is not undefined, the <a>computed MIME
   type</a> is <var>image-type-matched</var>.

   Abort these steps.

  <li>
   The <a>computed MIME type</a> is the <a>supplied MIME
   type</a>.
 </ol>



<h3 id=sniffing-in-an-audio-or-video-context>Sniffing in an audio or video context</h3>

<p>
 To determine the <a>computed MIME type</a> of a <a>resource</a>
 with an <a>audio or video MIME type</a>, execute the following <dfn export>rules
 for sniffing audio and video specifically</dfn>:

 <ol>
  <li>
   If the <a>supplied MIME type</a> is an <a>XML MIME type</a>, the
   <a>computed MIME type</a> is the <a>supplied MIME type</a>.

   Abort these steps.

  <li>
   Let <var>audio-or-video-type-matched</var> be the result of executing the
   <a>audio or video type pattern matching algorithm</a> with the
   <a>resource header</a> as the <a>byte sequence</a> to be
   matched.

  <li>
   If <var>audio-or-video-type-matched</var> is not undefined, the
   <a>computed MIME type</a> is <var>audio-or-video-type-matched</var>.

   Abort these steps.

  <li>
   The <a>computed MIME type</a> is the <a>supplied MIME
   type</a>.
 </ol>



<h3 id=sniffing-in-a-plugin-context>Sniffing in a plugin context</h3>

<p>
 To determine the <a>computed MIME type</a> of a <a>resource</a>
 <a lt=fetch for=/>fetched</a> in a plugin context, execute
 the following <dfn export>rules for sniffing in a plugin context</dfn>:

 <ol>
  <li class=XXX>
   If the <a>supplied MIME type</a> is undefined, the <a>computed
   MIME type</a> is "<code>application/octet-stream</code>".

  <li>
   The <a>computed MIME type</a> is the <a>supplied MIME type</a>.
 </ol>



<h3 id=sniffing-in-a-style-context>Sniffing in a style context</h3>

<p>
 To determine the <a>computed MIME type</a> of a <a>resource</a>
 <a lt=fetch for=/>fetched</a> in a style context, execute the
 following <dfn export>rules for sniffing in a style context</dfn>:

 <ol>
  <li class=XXX>
   If the <a>supplied MIME type</a> is undefined, ….

  <li>
   The <a>computed MIME type</a> is the <a>supplied MIME type</a>.
 </ol>



<h3 id=sniffing-in-a-script-context>Sniffing in a script context</h3>

<p>
 To determine the <a>computed MIME type</a> of a <a>resource</a>
 <a lt=fetch for=/>fetched</a> in a script context, execute
 the following <dfn export>rules for sniffing in a script context</dfn>:

 <ol>
  <li class=XXX>
   If the <a>supplied MIME type</a> is undefined, ….

  <li>
   The <a>computed MIME type</a> is the <a>supplied MIME type</a>.
 </ol>



<h3 id=sniffing-in-a-font-context>Sniffing in a font context</h3>

<p>
 To determine the <a>computed MIME type</a> of a <a>resource</a>
 with a <a>font MIME type</a>, execute the following <dfn export>rules for sniffing
 fonts specifically</dfn>:

 <ol>
  <li>
   If the <a>supplied MIME type</a> is an <a>XML MIME type</a>, the
   <a>computed MIME type</a> is the <a>supplied MIME type</a>.

   Abort these steps.

  <li>
   Let <var>font-type-matched</var> be the result of executing the <a>font
   type pattern matching algorithm</a> with the <a>resource
   header</a> as the <a>byte sequence</a> to be matched.

  <li>
   If <var>font-type-matched</var> is not undefined, the <a>computed MIME
   type</a> is <var>font-type-matched</var>.

   Abort these steps.

  <li>
   The <a>computed MIME type</a> is the <a>supplied MIME
   type</a>.
 </ol>



<h3 id=sniffing-in-a-text-track-context>Sniffing in a text track context</h3>

<p class=XXX>
 The <a>computed MIME type</a> is "<code>text/vtt</code>".



<h3 id=sniffing-in-a-cache-manifest-context>Sniffing in a cache manifest context</h3>

<p class=XXX>
 The <a>computed MIME type</a> is
 "<code>text/cache-manifest</code>".



<h2 class=no-num id=acknowledgements>Acknowledgments</h2>

<p>
 Special thanks to Adam Barth and Ian Hickson for maintaining previous
 incarnations of this document.

<p>
 Thanks also to
 Alfred Hönes,
 Andreu Botella,
 Anne van Kesteren,
 Ben Eidson,
 Boris Zbarsky,
 Darien Maillet Valentine,
 David Singer,
 Domenic Denicola,
 Henri Sivonen,
 Jean-Yves Avenard,
 Jonathan Neal,
 Joshua Cranmer,
 Larry Masinter,
 罗泽轩,
 Mariko Kosaka,
 Mark Pilgrim,
 Paul Adenot,
 Peter Occil,
 Rob Buis,
 Russ Cox,
 Simon Pieters, and
 triple-underscore
 for their invaluable contributions.

<p>This standard is written by <a href=https://gphemsley.org/>Gordon P. Hemsley</a>
(<a href=mailto:me@gphemsley.org>me@gphemsley.org</a>).