All over the Web you can find rectangular data in CSV files. Typically, this data is treated in its entirety, meaning you retrieve the entire CSV file and then you deal with it in your application. But that's not very webby.

Wouldn't it be nicer to be able to address parts of a table, like certain columns or rows? Now you can, with Addrables!

Addrable is short for Addressable Table - essentially making parts of a table addressable via URIs. OK, sounds nice in theory, but how does it work? Well, take a look at this example table:

city    person   visits
----------------------
Berlin  Richard  20
London  Richard  2
Rom     Richard  1
Berlin  Michael  4
London  Michael  10
Rom     Michael  2

Now, imagine you're only interested in the part of the table where the city column has the value Berlin. With Addrables, you'd state this as follows:

table1.csv#where:city=Berlin

This would yield the following part of the table (called slice, here):

person   visits
---------------
Richard  20
Michael  4

But you can also go a step further, for example addressing a single value, if you specify all but one of the columns (called dimension, here) of a table, like so:

table1.csv#where:city=Berlin,person=Richard

Which would result, not very surprisingly, in the value 20. You're now equipped with the Addrables basics, but if you want to dive into the details, you might want to have a look at the Addrable addressing scheme and syntax.

Addrable is a 100% JavaScript library for either client-side or server-side processing of tabular data. The result of the selection via an Addrable depends on the mode: on the client-side (in a browser) the selected slices are rendered visually, on the server-side, the slices are delivered as JSON. Both client and server implementations share a common core Addrable functionality. So, the Addrable library essentially contains:

• addrable-core.js, providing generic methods to parse the slice's addressing and filtering of the data.
• addrable-client.js, a jQuery-based implementation that renders slices in various ways.
• addrable-server.js, a node.js-based implementation that returns slices in JSON.

To better understand how Addrable works, you might want run the following examples either using the client demo (in your browser) or the server demo (using the command-line).

As already mentioned, the client is implemented using jQuery and a jQuery plug-in, see the lib/ directory for details. To play around with the client demo, simply grab the content of the repository via git clone or download it and point your browser to index.html.

Try the following Addrables:

• data/table2.csv#col:*
• data/table2.csv#col:temperature
• data/table2.csv#row:*
• data/table2.csv#row:10
• data/table2.csv#where:city=Galway,date=2011-03-01,reporter=Richard
• data/table2.csv#where:city=Galway,reporter=Richard
• data/table2.csv#where:city=Galway

I've tested it with jQuery 1.4.2 and the js-tables plug-in (which you can find in the lib/ directory) under MacOS X 10.5.8/Firefox 3.6.13. Note for Chrome users: to use the Addrable client demo under Chrome you must enable access from local files due to a known issue:

$ cd "/Applications/Google Chrome.app/Contents/MacOS/"
$ sudo mv "Google Chrome" Google.real
$ sudo printf '#!/bin/bash\ncd "/Applications/Google Chrome.app/Contents/MacOS"\n"/Applications/Google Chrome.app/Contents/MacOS/Google.real"  --allow-file-access-from-files "$@"\n' > Google\ Chrome
$ chmod 755 "Google Chrome"

On the server, you need to have node.js installed. You can then run the Addrable server demo:

$ node addrable-node.js 
Addrable v0.1 running on server 127.0.0.1, listening on port 8086

Once you've started the Addrable server demo, you can use, for example curl to resolve Addrables.

$ curl http://127.0.0.1:8086/http%3A%2F%2F127.0.0.1%3A8086%2Fdata%2Ftable1.csv%23where%3Acity%3DBerlin

This Addrable is the same as in the introductory example above - it selects a slice with the two remaining columns person and visits and returns it encoded in JSON as so:

{
   "header": ["person", "visits"],
   "rows": [
             ["Richard", "Michael"], 
             ["20", "4"]
           ]
}

To dig deeper into the server-side demo, you might want to have a look at the Addrable Server demo documentation.

I've tested it with node-v0.2.6 and curl under MacOS X 10.5.8. Note that the server-side implementation, especially the JSON format, is not stable, yet.

The following people influenced the design of Addrable and came up with improvements: Richard Cyganiak for the initial idea of how to render slices as well as for his proposal to extend Addrables to address rows; KevBurnsJr for pointing out the similarity with JSON hyper-schema and for the Addrable sales-pitch 'a CSV query interface implemented in URL fragments'.

Addrable is Public Domain - see UNLICENSE for more details.