Awesome

This is a mini LDP Server in Go.

Linked Data Platform (LDP) is a W3C recommendation that defines rules for how to implement an HTTP API for read-write Linked Data. The official recommendation can be found here. You can also find a more gentle introduction to LDP in my blog.

Warning: This is my sandbox project as I learn both Go and LDP. The code in this repo very likely does not follow Go's best practices and it certainly does not conform to the LDP spec (yet).

Compile and run the server

If Go is installed on your machine:

cd ~/src
git clone git@github.com:hectorcorrea/ldpserver.git
cd ldpserver
go build
./ldpserver

If you are new to Go follow these steps instead:

# Download and install Go from: http://golang.org/doc/install
#
# Go is very picky about the location of the code (e.g. the code must be
# inside an src folder.) Here is a setup that will work with minimal effort
# and configuration on your part. You can skip the first step if you
# already have an ~/src folder.
#
mkdir ~/src
export GOPATH=~/
cd ~/src
git clone git@github.com:hectorcorrea/ldpserver.git
cd ldpserver
go build
./ldpserver

If you don't care about the source code, the fastest way to get started is to download the executable for your platform from the release tab, make it an executable on your box, and run it.

Operations supported

With the server running, you can use cURL to submit requests to it. For example, to fetch the root node

curl localhost:9001

POST to the root (the Slug defaults to "node" + a sequential number)

curl -X POST localhost:9001

Fetch the node created

curl localhost:9001/node1

POST a non-RDF to the root

curl -X POST --header "Content-Type: text/plain" --data "hello world" localhost:9001

curl -X POST --header "Content-Type: image/jpeg" --data-binary "@filename.jpg" localhost:9001

Fetch the non-RDF created

curl localhost:9001/node2

HTTP HEAD operations are supported

curl -I localhost:9001/
curl -I localhost:9001/node1
curl -I localhost:9001/node2

Add an RDF source to add a child node (you can only add to RDF sources)

curl -X POST localhost:9001/node1

See that the child was added

curl localhost:9001/node1

Fetch the child

curl localhost:9001/node1/node3

Create a node with a custom Slug

curl -X POST --header "Slug: demo" localhost:9001

Fetch node created

curl localhost:9001/demo

Create an LDP Direct Container /dc1 that uses /node1 as its membershipResource and someRel as the member relation...

curl -X POST --header "Content-Type: text/turtle" --header "Slug: dc1" -d "<> <http://www.w3.org/ns/ldp#hasMemberRelation> someRel ; <http://www.w3.org/ns/ldp#membershipResource> <http://localhost:9001/node1> ." localhost:9001

...add a node to the direct container

curl -X POST --header "Slug: child1" localhost:9001/dc1

...fetch /node1 and notice that it references /dc1/child1 with the predicate someRel that we defined in the direct container:

curl localhost:9001/dc1

Demo

Take a look at demo.sh file for an example of a shell script that executes some of the operations supported. To run this demo make sure the LDP Server is running in a separate terminal window, for example:

# Run the LDP Server in one terminal window
./ldpserver

# Run the demo script in a separate terminal window
./demo.sh

Storage

Every resource (RDF or non-RDF) is saved in a folder inside the data folder.

Every RDF source is saved on its own folder with single file inside of it. This file is always meta.rdf and it has the triples of the node.

Non-RDF are also saved on their own folder and with a meta.rdf file for their metadata but also a file data.bin with the non-RDF content.

For example, if we have two nodes (blog1 and blog2) and blog1 is an RDF node and blog2 is a non-RDF then the data structure would look as follow:

/data/meta.rdf          (root node)
/data/blog1/meta.rdf    (RDF for blog1)
/data/blog2/meta.rdf    (RDF for blog2)
/data/blog2/data.bin    (binary for blog2)

Overview of the Code

main.go is the launcher program. It's only job is to kick off the web server.
web/web.go is the web server. It's job is to handle HTTP requests and responses. This is the only part of the code that is aware of the web.
server/server.go handles most of the operations like creating new nodes and fetching existing ones.
ldp/node.go handles operations at the individual node level (fetching and saving.)
rdf/ contains utilities to parse and update RDF triples and graphs.

TODO

A lot.

Add validation to make sure the data in the root node matches the URL (host:port) where the server is running.
Support isMemberOfRelation in Direct Containers.
Support Indirect Containers.

LDP Test Suite

The W3C provides a test suite to make sure LDP server implementations meet a minimum criteria. The test suite can be found at http://w3c.github.io/ldp-testsuite/

In order to run the suite against this repo you need to do the following:

Clone the ldp-testsuite repo: git clone https://github.com/w3c/ldp-testsuite
Download maven from http://maven.apache.org/download.cgi
Unzip maven to the ldp-testsuite folder
cd ldp-testsuite
Run bin/mvn package

...and then you can run the following command against the LDP Server to execute an individual test:

java -jar target/ldp-testsuite-0.2.0-SNAPSHOT-shaded.jar --server http://localhost:9001 --test name_of_test_goes_here --basic

...or as follow to run all basic container tests (including support for non-RDF):

java -jar target/ldp-testsuite-0.2.0-SNAPSHOT-shaded.jar --server http://localhost:9001 --basic --non-rdf

As of 1/9/2016 these are the results of all basic container tests (including support for non-RDF):

LDP Test Suite
Total tests run: 112, Failures: 4, Skips: 28

TODO: Document how to test DC and the results 97/5/27