<abstract>This specification defines a protocol to request permissions from another entity to upload a file to a specific path on an HTTP server and at the same time receive a URL from which that file can later be downloaded again.</abstract>
<li>Remove statement about access control from Requirements. (Statement about long randomized file names had been moved to Security Considerations earlier.)</li>
<li>Add section about IP address leak to Security Considerations.</li>
<p>XMPP protocol extensions already define methods for peer-to-peer file transfer such as &xep0096; or &xep0234; however due to their very nature of being peer-to-peer they don't work very well in scenarios where it is requeried to send a file to multiple recipients or multiple resources of the same recipient at once. They also don’t work alongside offline storage, MUC history and &xep0313;.</p>
<p>Uploading files manually to an HTTP server and sharing the link has been a workaround for this for a long time now. While users have a variety of services to choose from the downside of this manual approach is that an XMPP client can not automate this process on behalf of the user since these services don’t share a common API. Furthermore using a third party service would probably require the user to enter additional credentials into their XMPP client specifically for the file upload.</p>
<p>This XEP defines an approach to request permissions from another entity to upload a file to a specific path on an HTTP server and at the same time receive an URL from which that file can later be downloaded again. These tuples consisting of a PUT and a GET-URL are called slots.</p>
</section1>
<section1topic='Requirements'anchor='reqs'>
<ul>
<li>Be as easy to implement as possible. This is grounded on the idea that most programming languages already have HTTP libraries available.</li>
<li>Be agnostic toward the distribution of the actual URL. Users can choose to send the URL in the body of a message stanza, utilize &xep0066;, &xep0370;, or even use it as their avatar in &xep0084;.</li>
<li>Anyone who knows the URL SHOULD be able to access it.</li>
<p>An entity advertises support for this protocol by including the "urn:xmpp:http:upload:0" in its service discovery information features as specified in &xep0030; or section 6.3 of &xep0115;. To avoid unnecessary round trips an entity SHOULD also include the maximum file size as specified in &xep0128; if such a limitation exists. The field name MUST be "max-file-size" and the value MUST be in bytes.</p>
<p>A user’s server SHOULD include any known entities that provide such services into its service discovery items.</p>
<p>A client requests a new upload slot by sending an IQ-get to the upload service containing a <request> child element qualified by the urn:xmpp:http:upload:0 namespace. This element MUST include the attributes filename and size containing the file name and size (in bytes) respectively.</p>
<p>The upload service responds with both a PUT and a GET URL wrapped by a <slot> element. The service SHOULD keep the file name and especially the file ending intact. Using the same hostname for PUT and GET is OPTIONAL. The host MUST provide Transport Layer Security (&rfc5246;). Both HTTPS URLs MUST adhere to &rfc3986;. Non ASCII characters MUST be percent-encoded.</p>
<p>The <put> element MAY also contain a number of <header> elements which correspond to HTTP header fields. Each <header> element MUST have a name-attribute and a content with the value of the header. Only the following header names are allowed: Authorization, Cookie, Expires. These headers MUST be included in the HTTP PUT request. Other header names MUST be ignored by the requesting entity and MUST NOT be included in the HTTP request. The requesting entity MUST strip any newline characters from the header name and value before performing the HTTP request.</p>
<p>Instead of providing the client with a slot the service MAY respond with an error if the requested file size is too large. In addition the entity MAY inform the requester about the maximum file size.</p>
<p>For temporary errors such as exceeding a personal quota the service MAY include a <retry/> element qualified by the urn:xmpp:http:upload:0 namespace as a child of the <error/> element. The retry element MUST include an attribute 'stamp' which indicates the time at which the requesting entity may try again. The format of the timestamp MUST adhere to the date-time format specified in &xep0082; and MUST be expressed in UTC.</p>
<p>The actual upload of the file happens via HTTP-PUT and is out of scope of this document. The upload service MUST reject the file upload if the Content-Length does not match the size of the slot request. The service SHOULD reject the file if the Content-Type has been specified beforehand and does not match. The service MAY assume application/octet-stream as a Content-Type if it the client did not specify a Content-Type at all.</p>
<p>There is no further XMPP communication required between the upload service and the client. A HTTP status Code of 201 means that the server is now ready to serve the file via the provided GET URL. If the upload fails for whatever reasons the client MAY request a new slot.</p>
<p>The upload service SHOULD choose an appropriate timeout for the validity of the PUT URL. Since there is no reason for a client to wait between requesting the slot and starting the upload, relatively low timeout values of around 300s are RECOMMENDED.</p>
<p>To make HTTP Upload work in web clients (including those hosted on a different domain) the upload service SHOULD set appropriate <linkurl="https://www.w3.org/TR/cors/">CORS</link>-Headers. The exact headers and values are out of scope of this document but may include: <em>Access-Control-Allow-Origin</em>, <em>Access-Control-Allow-Methods</em> and <em>Access-Control-Allow-Headers</em>. For HTTP upload services that use custom <em>Authorization</em> or <em>Cookie</em> request header the CORS-Header <em>Access-Control-Allow-Credentials</em> might also be of importance.</p>
<p><strong>Note: This section is not normative; it may be updated when general web security recommendations change in the future.</strong></p>
<p>It is recommended to run the HTTP upload domain used for GET requests in appropriate isolation from other HTTP based services to avoid user-generated, malicious scripts to be executed in the context of said services. Isolation techniques can include, but are not limited to, setting the <em>Content-Security-Policy</em>.</p>
<codecaption="HTTP Upload Server Headers for better Content Security Policy"><