Merge Chromium + Blink git repositories

[chromium-blink-merge.git] / chrome / common / extensions / docs / templates / intros / webRequest.html

<h2 id="manifest">Manifest</h2>
<p>You must declare the "webRequest" permission in the <a
  href="manifest">extension manifest</a> to use the web request
API, along with <a href="declare_permissions">host permissions</a>
for any hosts whose network requests you want to access. If you want to
use the web request API in a blocking fashion, you need to request
the "webRequestBlocking" permission in addition.
For example:</p>
<pre data-filename="manifest.json">
{
  "name": "My extension",
  ...
  <b>"permissions": [
    "webRequest",
    "*://*.google.com/"
  ]</b>,
  ...
}
</pre>

<h2 id="life_cycle">Life cycle of requests</h2>

<p>
The web request API defines a set of events that follow the life cycle of a web
request. You can use these events to observe and analyze traffic. Certain
synchronous events will allow you to intercept, block, or modify a request.
</p>

<p>
The event life cycle for successful requests is illustrated here, followed by
event definitions:<br/>
<img src="{{static}}/images/webrequestapi.png"
  width="385" height="503"
  alt="Life cycle of a web request from the perspective of the webrequest API"
  style="margin-left: auto; margin-right: auto; display: block"/>
</p>

<p>
<dl>
  <dt><code>onBeforeRequest</code> (optionally synchronous)</dt>
  <dd>Fires when a request is about to occur. This event is sent before any TCP
  connection is made and can be used to cancel or redirect requests.</dd>
  <dt><code>onBeforeSendHeaders</code> (optionally synchronous)</dt>
  <dd>Fires when a request is about to occur and the initial headers have been
  prepared. The event is intended to allow extensions to add, modify, and delete
  request headers <a href="#life_cycle_footnote">(*)</a>. The
  <code>onBeforeSendHeaders</code> event is passed to all subscribers, so
  different subscribers may attempt to modify the request; see the <a
    href="#implementation">Implementation details</a> section for how this is
  handled. This event can be used to cancel the request.</dd>
  <dt><code>onSendHeaders</code></dt>
  <dd>Fires after all extensions have had a chance to modify the request
  headers, and presents the final <a href="#life_cycle_footnote">(*)</a>
  version. The event is triggered before the headers are sent to the network.
  This event is informational and handled asynchronously. It does not allow
  modifying or cancelling the request.</dd>
  <dt><code>onHeadersReceived</code> (optionally synchronous)</dt>
  <dd>Fires each time that an HTTP(S) response header is received. Due
  to redirects and authentication requests this can happen multiple times per
  request. This event is intended to allow extensions to add, modify, and delete
  response headers, such as incoming Set-Cookie headers. The caching directives
  are processed before this event is triggered, so modifying headers such as
  Cache-Control has no influence on the browser's cache.
  It also allows you to redirect the request.</dd>
  <dt><code>onAuthRequired</code> (optionally synchronous)</dt>
  <dd>Fires when a request requires authentication of the user. This event can
  be handled synchronously to provide authentication credentials. Note that
  extensions may provide invalid credentials. Take care not to enter an infinite
  loop by repeatedly providing invalid credentials.</dd>
  <dt><code>onBeforeRedirect</code></dt>
  <dd>Fires when a redirect is about to be executed. A redirection can be
  triggered by an HTTP response code or by an extension. This event is
  informational and handled asynchronously. It does not allow you to modify or
  cancel the request. </dd>
  <dt><code>onResponseStarted</code></dt>
  <dd>Fires when the first byte of the response body is received. For HTTP
  requests, this means that the status line and response headers are
  available. This event is informational and handled asynchronously. It does not
  allow modifying or cancelling the request.</dd>
  <dt><code>onCompleted</code></dt>
  <dd>Fires when a request has been processed successfully.</dd>
  <dt><code>onErrorOccurred</code></dt>
  <dd>Fires when a request could not be processed successfully.</dd>
</dl>
The web request API guarantees that for each request either
<code>onCompleted</code> or <code>onErrorOccurred</code> is fired as the final
event with one exception: If a request is redirected to a <code>data://</code>
URL, <code>onBeforeRedirect</code> is the last reported event.
</p>

<p id="life_cycle_footnote">(*) Note that the web request API presents an
abstraction of the network stack to the extension. Internally, one URL request
can be split into several HTTP requests (for example to fetch individual byte
ranges from a large file) or can be handled by the network stack without
communicating with the network. For this reason, the API does not provide the
final HTTP headers that are sent to the network. For example, all headers that
are related to caching are invisible to the extension.</p>

<p>The following headers are currently <b>not provided</b> to the
<code>onBeforeSendHeaders</code> event. This list is not guaranteed to be
complete nor stable.
<ul>
  <li>Authorization</li>
  <li>Cache-Control</li>
  <li>Connection</li>
  <li>Content-Length</li>
  <li>Host</li>
  <li>If-Modified-Since</li>
  <li>If-None-Match</li>
  <li>If-Range</li>
  <li>Partial-Data</li>
  <li>Pragma</li>
  <li>Proxy-Authorization</li>
  <li>Proxy-Connection</li>
  <li>Transfer-Encoding</li>
</ul>
</p>

<p>
The webRequest API only exposes requests that the extension has
permission to see, given its
<a href="declare_permissions">host permissions</a>.
Moreover, only the following schemes are accessible:
<code>http://</code>,
<code>https://</code>,
<code>ftp://</code>,
<code>file://</code>, or
<code>chrome-extension://</code>.
In addition, even certain requests with URLs using one of the above schemes
are hidden, e.g.,
<code>chrome-extension://other_extension_id</code> where
<code>other_extension_id</code> is not the ID of the extension to handle
the request,
<code>https://www.google.com/chrome</code>,
and others (this list is not complete). Also synchronous XMLHttpRequests from
your extension are hidden from blocking event handlers in order to prevent
deadlocks.
Note that for some of the supported schemes the set of available events might be
limited due to the nature of the corresponding protocol.
For example, for the <q>file:</q> scheme, only <code>onBeforeRequest</code>,
<code>onResponseStarted</code>, <code>onCompleted</code>, and
<code>onErrorOccurred</code> may be dispatched.
</p>

<h2 id="concepts">Concepts</h2>

<p>As the following sections explain, events in the web request API use request
IDs, and you can optionally specify filters and extra information when you
register event listeners.</p>

<h3 id="Request IDs">Request IDs</h3>

<p>Each request is identified by a request ID. This ID is unique within a
browser session and the context of an extension. It remains constant during the
the life cycle of a request and can be used to match events for the same
request. Note that several HTTP requests are mapped to one web request in case
of HTTP redirection or HTTP authentication.</p>

<h3 id="subscription">Registering event listeners</h3>

<p>To register an event listener for a web request, you use a variation on the
<a href="events">usual <code>addListener()</code> function</a>.
In addition to specifying a callback function,
you have to specify a filter argument and you may specify an optional extra info
argument.</p>

<p>The three arguments to the web request API's <code>addListener()</code> have
the following definitions:</p>
<pre>
var callback = function(details) {...};
var filter = {...};
var opt_extraInfoSpec = [...];
</pre>

<p>Here's an example of listening for the <code>onBeforeRequest</code>
event:</p>
<pre>
chrome.webRequest.onBeforeRequest.addListener(
  callback, filter, opt_extraInfoSpec);
</pre>

<p>Each <code>addListener()</code> call takes a mandatory callback function as
the first parameter. This callback function is passed a dictionary containing
information about the current URL request. The information in this dictionary
depends on the specific event type as well as the content of
<code>opt_extraInfoSpec</code>.</p>

<p>If the optional <code>opt_extraInfoSpec</code> array contains the string
<code>'blocking'</code> (only allowed for specific events), the callback
function is handled synchronously. That means that the request is blocked until
the callback function returns. In this case, the callback can return a
  $(ref:webRequest.BlockingResponse) that determines the further
life cycle of the request. Depending on the context, this response allows
cancelling or redirecting a request (<code>onBeforeRequest</code>), cancelling a
request or modifying headers (<code>onBeforeSendHeaders</code>,
<code>onHeadersReceived</code>), or providing authentication credentials
(<code>onAuthRequired</code>).</p>

<p>The $(ref:webRequest.RequestFilter)
<code>filter</code> allows limiting the requests for which events are
triggered in various dimensions:
<dl>
  <dt>URLs</dt>
  <dd><a href="match_patterns">URL patterns</a> such as
  <code>*://www.google.com/foo*bar</code>.</dd>
  <dt>Types</dt>
  <dd>Request types such as <code>main_frame</code> (a document that is loaded
  for a top-level frame), <code>sub_frame</code> (a document that is loaded for
  an embedded frame), and <code>image</code> (an image on a web site).
  See $(ref:webRequest.RequestFilter).</dd>
  <dt>Tab ID</dt>
  <dd>The identifier for one tab.</dd>
  <dt>Window ID</dt>
  <dd>The identifier for a window.</dd>
</p>

<p>Depending on the event type, you can specify strings in
<code>opt_extraInfoSpec</code> to ask for additional information about the
request. This is used to provide detailed information on request's data only
if explicitly requested.</p>

<h2 id="implementation">Implementation details</h2>

<p>Several implementation details can be important to understand when developing
an extension that uses the web request API:</p>

<h3 id="conflict_resolution">Conflict resolution</h3>
<p>In the current implementation of the web request API, a request is considered
as cancelled if at least one extension instructs to cancel the request. If
an extension cancels a request, all extensions are notified by an
<code>onErrorOccurred</code> event. Only one extension is allowed to redirect a
request or modify a header at a time. If more than one extension attempts to
modify the request, the most recently installed extension wins and all others
are ignored. An extension is not notified if its instruction to modify or
redirect has been ignored.</p>

<h3 id="caching">Caching</h3>
<p>
Chrome employs two caches &mdash; an on-disk cache and a very fast in-memory
cache.  The lifetime of an in-memory cache is attached to the lifetime of a
render process, which roughly corresponds to a tab. Requests that are answered
from the in-memory cache are invisible to the web request API. If a request
handler changes its behavior (for example, the behavior according to which
requests are blocked), a simple page refresh might not respect this changed
behavior.  To make sure the behavior change goes through, call
<code>handlerBehaviorChanged()</code> to flush the in-memory cache. But don't do
it often; flushing the cache is a very expensive operation. You don't need to
call <code>handlerBehaviorChanged()</code> after registering or unregistering an
event listener.</p>

<h3 id="timestamps">Timestamps</h3>
<p>
The <code>timestamp</code> property of web request events is only guaranteed to
be <i>internally</i> consistent. Comparing one event to another event will give
you the correct offset between them, but comparing them to the current time
inside the extension (via <code>(new Date()).getTime()</code>, for instance)
might give unexpected results.
</p>

<h3 id="error_handling">Error handling</h3>
<p>
If you try to register an event with invalid arguments, then a JavaScript error
will be thrown, and the event handler will not be registered.

If an error is thrown while an event is handled, or if an event handler returns
an invalid blocking response, an error message is logged to your extension's
console and the handler is ignored for that request.
</p>

<h2 id="examples">Examples</h2>

<p>The following example illustrates how to block all requests to
<code>www.evil.com</code>:</p>
<pre>
chrome.webRequest.onBeforeRequest.addListener(
  function(details) {
    return {cancel: details.url.indexOf("://www.evil.com/") != -1};
  },
  {urls: ["&lt;all_urls&gt;"]},
  ["blocking"]);
</pre>

<p>As this function uses a blocking event handler, it requires the "webRequest"
as well as the "webRequestBlocking" permission in the manifest file.</p>

<p>The following example achieves the same goal in a more efficient way because
requests that are not targeted to <code>www.evil.com</code> do not need to be
passed to the extension:</p>
<pre>
chrome.webRequest.onBeforeRequest.addListener(
  function(details) { return {cancel: true}; },
  {urls: ["*://www.evil.com/*"]},
  ["blocking"]);
</pre>

<p>The following example illustrates how to delete the User-Agent header from
all requests:</p>
<pre>
chrome.webRequest.onBeforeSendHeaders.addListener(
  function(details) {
    for (var i = 0; i < details.requestHeaders.length; ++i) {
      if (details.requestHeaders[i].name === 'User-Agent') {
        details.requestHeaders.splice(i, 1);
        break;
      }
    }
    return {requestHeaders: details.requestHeaders};
  },
  {urls: ["&lt;all_urls&gt;"]},
  ["blocking", "requestHeaders"]);
</pre>

<p> For more example code, see the <a href="samples#search:webrequest">web request
samples</a>.</p>