Awesome Open Source
Awesome Open Source

A multi-tool shell script for doing Semantic Web jobs on the command line.

Build Status



manually is a single bash shell script so installation is trivial ... :-) Just copy or link it to you path, e.g. with

$ sudo ln -s /path/to/ /usr/local/bin/rdf

debian / ubuntu

You can download a debian package from the release section and install it as root with the following commands:

$ sudo dpkg -i /path/to/your/rdf.sh_X.Y_all.deb
$ sudo apt-get -f install

The dpkg run will probably fail due to missing dependencies but the apt-get run will install all dependencies as well as rdf.

Currently, zsh is a hard dependency since the zsh completion "needs" it.

brew based

You can install by using the provided recipe:

brew install

This will install the latest stable version. In case you want to install the latest develop version, use this command:

brew install --HEAD

docker based

You can install by using the provided docker image:

docker pull seebi/

After that, you can e.g. run this command:

docker run -i -t --rm seebi/ desc foaf:Person


Required tools currently are:

  • roqet (from rasqal-utils)
  • rapper (from raptor-utils or raptor2-utils)
  • curl

Suggested tools are:

  • zsh (without the autocompletion, it is not the same)


These files are available in the repository:

  • - this file
  • _rdf - zsh autocompletion file
  • - version change log
  • doap.ttl - doap description of
  • rdf.1 - man page
  • - the script
  • Screenshot.png - a screeny of in action
  • example.rc - an example config file which can be copied

These files are used by

  • $HOME/.cache/ - history of all processed resources
  • $HOME/.cache/ - a cache of all fetched namespaces
  • $HOME/.config/ - locally defined prefix / namespaces
  • $HOME/.config/ - config file follows the XDG Base Directory Specification in order to allow different cache and config directories.

usage / features

overview currently provides these subcommands:

  • color: get a html color for a resource URI
  • count: count distinct triples
  • delete: deletes an existing linked data resource via LDP
  • desc: outputs description of the given resource in a given format (default: turtle)
  • diff: diff of triples from two RDF files
  • edit: edit the content of an existing linked data resource via LDP (GET + PUT)
  • get: fetches an URL as RDF to stdout (tries accept header)
  • get-ntriples: curls rdf and transforms to ntriples
  • gsp-delete: delete a graph via SPARQL 1.1 Graph Store HTTP Protocol
  • gsp-get: get a graph via SPARQL 1.1 Graph Store HTTP Protocol
  • gsp-put: delete and re-create a graph via SPARQL 1.1 Graph Store HTTP Protocol
  • head: curls only the http header but accepts only rdf
  • headn: curls only the http header
  • help: outputs the manpage of rdf
  • list: list resources which start with the given URI
  • ns: curls the namespace from
  • nscollect: collects prefix declarations of a list of ttl/n3 files
  • nsdist: distributes prefix declarations from one file to a list of other ttl/n3 files
  • put: replaces an existing linked data resource via LDP
  • split: split an RDF file into pieces of max X triple and output the file names
  • turtleize: outputs an RDF file in turtle, using as much as possible prefix declarations

namespace lookup (ns) allows you to quickly lookup namespaces from as well as locally defined prefixes:

$ rdf ns foaf

These namespace lookups are cached (typically $HOME/.cache/ in order to avoid unneeded network traffic. As a result of this subcommand, all other rdf command can get qnames as parameters (e.g. foaf:Person or skos:Concept).

To define you own lookup table, just add a line


to $HOME/.config/ will use it as a priority lookup table which overwrites cache and lookup. can also output syntax templates (uncached):

$ rdf ns skos sparql
PREFIX skos: <>

  ?s ?p ?o .

$ rdf ns dct n3    
@prefix dct: <>.

resource description (desc)

Describe a resource by querying for statements where the resource is the subject. This is extremly useful to fastly check schema details.

$ rdf desc foaf:Person
@prefix rdf: <> .
@prefix rdfs: <> .
@prefix owl: <> .
@prefix foaf: <> .
@prefix geo: <> .
@prefix contact: <> .

    a rdfs:Class, owl:Class ;
    rdfs:comment "A person." ;
    rdfs:isDefinedBy <> ;
    rdfs:label "Person" ;
    rdfs:subClassOf contact:Person, geo:SpatialThing, foaf:Agent ;
    owl:disjointWith foaf:Organization, foaf:Project ;
    <> "stable" .

In addition to the textual representation, you can calculate a color for visual resource representation with the color command:

$ rdf color

Refer to the cold webpage for more information :-)

SPARQL graph store protocol client

The SPARQL 1.1 Graph Store HTTP Protocol describes the use of HTTP operations for the purpose of managing a collection of RDF graphs. supports the following commands in order to manipulate graphs:

Syntax: rdf gsp-get <graph URI | Prefix:LocalPart> <store URL | Prefix:LocalPart (optional)>
(get a graph via SPARQL 1.1 Graph Store HTTP Protocol)
Syntax: rdf gsp-put <graph URI | Prefix:LocalPart> <path/to/your/file.rdf> <store URL | Prefix:LocalPart (optional)>
(delete and re-create a graph via SPARQL 1.1 Graph Store HTTP Protocol)
Syntax: rdf gsp-delete <graph URI | Prefix:LocalPart> <store URL | Prefix:LocalPart (optional)>
(delete a graph via SPARQL 1.1 Graph Store HTTP Protocol)

If the store URL is not given, the Direct Graph Identification is assumed, which means the store URL is taken as the graph URL. If the store URL is given, Indirect Graph Identification is used.

linked data platform client

The Linked Data Platform describe a read-write Linked Data architecture, based on HTTP access to web resources that describe their state using the RDF data model. supports DELETE, PUT and edit (GET, followed by an edit command, followed by a PUT request) of Linked Data Platform Resources (LDPRs).

Syntax: rdf put <URI | Prefix:LocalPart> <path/to/your/file.rdf>
(replaces an existing linked data resource via LDP)
Syntax: rdf delete <URI | Prefix:LocalPart>
(deletes an existing linked data resource via LDP)
Syntax: rdf edit <URI | Prefix:LocalPart>
(edit the content of an existing linked data resource via LDP (GET + PUT))

The edit command uses the EDITOR variable to start the editor of your choice with a prepared turtle file. You can change the content of that file (add or remove triple) and you can use any prefix you've already declared via config or which is cached. Used prefix declarations are added automatically afterwards and the file is the PUTted to the server.

WebID requests

In order to request ressources with your WebID client certificate, you need to setup the rc file (see configuration section). Curl allows for using client certs with the -E parameter, which needs a pem file with your private key AND the certificate.

To use your proper created WebID pem file, just add this to your rc file:

RDFSH_CURLOPTIONS_ADDITONS="-E $HOME/path/to/your/webid.pem"

syntax highlighting supports the highlighted output of turtle with pygmentize and a proper turtle lexer. If everything is available (pygmentize -l turtle does not throw an error), then it will look like this.

If you do not want syntax highlighting for some reason, you can disable it by setting the shell environment variable RDFSH_HIGHLIGHTING_SUPPRESS to true e.g with


before you start

resource listings (list)

To get a quick overview of an unknown RDF schema, provides the list command which outputs a distinct list of subject resources of the fetched URI:

$ rdf list geo:

You can also provide a starting sequence to constrain the output

$ rdf list skos:C

Note: Here the $GREP_OPTIONS environment applies to the list. In my case, I have a --ignore-case in it, so e.g. skos:changeNote is listed as well.

This feature only works with schema documents which are available by fetching the namespace URI (optionally with linked data headers to be redirected to an RDF document).

resource inspection (get, count, head and headn)

Fetch a resource via linked data and print it to stdout:

$ rdf get >me.rdf

Count all statements of a resource:

$ rdf count

Inspect the header of a resource. Use head for header request with content negotiation suitable for linked data and headn for a normal header request as sent by browsers.

$ rdf head
HTTP/1.1 302 Found

prefix distribution for data projects (nscollect and nsdist)

Often I need to create a lot of n3/ttl files as a data project which consists of schema and instance resources. These projects are split over several files for a better handling and share a set if used namespaces.

When introducing a new namespace to such projects, I need to add the @prefix line to each of the ttl files of this project. has two subcommands which handle this procedure:

  • rdf nscollect collects all prefixes from existing n3/ttl files in the current directory and collect them in the file prefixes.n3
  • rdf nsdist *.n3 firstly removes all @prefix lines from the target files and then add prefixes.n3 on top of them.

Materialize / skolemize bnodes (skolemize)

Blank nodes can be painful, so this command materializes all blank nodes as full IRIs. The first parameter is the RDF file while the second parameter is an optional namespace for the UUID minted IRIs (default is urn:uuid:).

re-format RDF files in turtle (turtleize)

Working with RDF files often requires to convert and reformat different files. With rdf turtleize, its easy to get RDF files in turtle plus they are nicely formatted because all needed prefix declarations are added.

turtleize uses rapper and tries to detect all namespaces which are cached in your prefix.cache file, as well as which a defined in the prefix.local file.

To turtleize your current buffer in vim for example, you can do a :%! rdf turtleize %.

autocompletion and resource history can be used with a zsh command-line completion function. This boosts the usability of this tool to a new level! The completion features support for the base commands as well as for auto-completion of resources. These resources are taken from the resource history. The resource history is written to $HOME/.cache/

When loaded, the completion function could be used in this way:

rdf de<tab> tramp<tab>

This could result in the following commandline:

rdf desc


  • The substring matching feature of the zsh completion system should be turned on.
    • e.g. with zstyle ':completion:*' matcher-list 'r:|[._-]=* r:|=*' 'l:|=* r:|=*'
  • This assumes that at least one resource exists in the history which matches .*tramp.*

configuration imports $HOME/.config/ at the beginning of each execution so this is the place to setup personal configuration options such as

  • WebID support
  • syntax highlighting suppression
  • setup of preferred accept headers
  • setup of alternate ntriples fetch program such as any23's rover (see this feature request for background infos)

Please have a look at the example rc file.

Get A Weekly Email With Trending Projects For These Topics
No Spam. Unsubscribe easily at any time.
shell (10,536
syntax-highlighting (173
rdf (88
semantic-web (42
autocompletion (39
namespace (24