| commit | 181a8b93d48e5570712ed9347a4ed8942b2e1b62 | [log] [tgz] |
|---|---|---|
| author | Alexander Afanasyev <alexander.afanasyev@ucla.edu> | Thu Feb 28 13:28:53 2013 -0800 |
| committer | Alexander Afanasyev <alexander.afanasyev@ucla.edu> | Sat Mar 02 11:40:03 2013 -0800 |
| tree | 103f9db9bed2a7e7b695ce9a5a34e482a2436079 | |
| parent | fad41d356871d4ae057e394b3a8d8a9a2cc68c23 [diff] |
Initial reorganization
NDN.JS is the first native version of the NDN protocol written in JavaScript. It is wire format compatible with PARC's CCNx.
The project by the UCLA NDN team - for more information on NDN, see http://named-data.net/ http://ndn.ucla.edu/
NDN.JS is open source under a license described in the file COPYING. While the license does not require it, we really would appreciate it if others would share their contributions to the library if they are willing to do so under the same license.
This is a young project, with minimal documentation that we are slowly enhancing. Please email Jeff Burke (jburke@remap.ucla.edu) with any questions.
The primary goal of NDN.JS is to provide a pure Javascript implementation of the NDN API that enables developers to create browser-based applications using Named Data Networking. The approach requires no native code or signed Java applets, and thus can be delivered over the current web to modern browsers with no hassle for the end user.
Additional goals for the project:
The library currently requires a remote NDN daemon, and has been tested with ccnd, from the's CCNx package: http://ccnx.org/
Currently, the library has two APIs for developers:
1. The Javascript API for asynchronous Interest/Data exchange. This uses WebSockets for transport and currently requires a proxy for communication with a remote ccnd daemon. 2. A Firefox plug-in, which implements an "ndn:/" url scheme following CCNx repository conventions for file retrieval.
By default, both parts of the library connect automatically to a set of proxies and hubs that are part of the NDN research project's testbed. http://named-data.net/testbed.html There are currently no restrictions on non-commercial, research-oriented data exchange on this testbed. (Contact jburke@remap.ucla.edu for more details.) The developer can also specify a local or remote ccnd as well, as an argument to the NDN constructor.
See files in js/ and examples in js/testing, js/examples
NDN.JS currently supports expressing Interests (and receiving data) and publishing Data (that answers Interests). This includes encoding and decoding data packets as well as signing and verifying them using RSA keys.
** NDN connectivity ** The only way (for now) to get connectivity to other NDN nodes is via ccnd. For the Javascript API, a Websockets proxy that can communicate the target ccnd is currently required. Code for such a proxy (using Node.js) is in the wsproxy directory. It currently listens on port 9696 and passes messages (using either TCP or UDP) to ccnd on the same host.
** Including the scripts in a web page ** To use NDN.JS in a web page, one of two scripts must be included using a tag:
ndn-js.js, a combined, compressed library designed for efficient distribution. It can be built using js/tools/build/make-js.sh This is used in examples/ndn-ping.html
Helper.js, which loads each component script independently. This is used in most of the tests in testing/
** Example to retrieve content ** A simple example of the current API to express an Interest and receive data:
var ndn = new NDN(); // connect to a default hub/proxy
var AsyncGetClosure = function AsyncGetClosure() { // Inherit from Closure. Closure.call(this); }; AsyncGetClosure.prototype.upcall = function(kind, upcallInfo) { if (kind == Closure.UPCALL_CONTENT) { console.log("Received " + upcallInfo.contentObject.name.to_uri()); console.log(upcallInfo.contentObject.content); } return Closure.RESULT_OK; };
ndn.expressInterest(new Name("/ndn/ucla.edu/apps/ndn-js-test/hello.txt"), new AsyncGetClosure());
** Example to publish content **
// Note that publishing content requires knowledge of a // routable prefix for your upstream ccnd. We are working // on a way to either obtain that prefix or use the /local // convention.
For now, see testing/test-publish-async.html
See files in js/ndnProtocol
NDN.JS includes a Firefox extension for the ndn protocol built using the Javascript library. It currently obtains NDN connectivity through the NDN testbed. (This is hard-coded.)
To install, either download https://github.com/remap/ndn-js/blob/master/js/ndnProtocol.xpi or use js/ndnProtocol.xpi in the distribution. In Firefox, open Tools > Add-ons. In the "gear" or "wrench" menu, click Install Add-on From File and open ndnProtocol.xpi. Restart Firefox.
Firefox uses the protocol extension to load any URI starting with ndn. See this test page for examples ndn:/ndn/ucla.edu/apps/ndn-js-test/NDNProtocolExamples.html?ndn.ChildSelector=1
When the page is loaded, Firefox updates the address bar with the full matched name from the retrieved content object including the version, but without the implicit digest or segment number (see below).
You can add interest selectors. For example, this uses 1 to select the "rightmost" child (latest version). ndn:/ndn/ucla.edu/apps/ndn-js-test/hello.txt?ndn.ChildSelector=1&key=value#ref
The browser loads the latest version and changes the address to: ndn:/ndn/ucla.edu/apps/ndn-js-test/hello.txt/%FD%05%0B%16z%22%D1?key=value#ref
The child selector was used and removed. Note that the other non-ndn query values and ref "?key=value#ref" are still present, in case they are needed by the web application.
The following selector keys are supported: ndn.MinSuffixComponent= non-negative int ndn.MaxSuffixComponents= non-negative int ndn.ChildSelector= non-negative int ndn.AnswerOriginKind= non-negative int ndn.Scope= non-negative int ndn.InterestLifetime= non-negative int (milliseconds) ndn.PublisherPublicKeyDigest= % escaped value ndn.Nonce= % escaped value ndn.Exclude= comma-separated list of % escaped values or * for ANY
A URI for content with multiple segments is handled as follows. If the URI has a segment number, just retrieve that segment and return the content to the browser.
Otherwise look at the name in the returned ContentObject. If the returned name has no segment number, just return the content to the browser. If the name has a segment number which isn't 0, store it and express an interest for segment 0. Also express an interest for the highest segment to try to determine the FinalBlockID early. Fetch multiple segments in order and return each content to the browser (in order) as the arrive until we get the segment for FinalBlockID.