This package consists of a set of command-line tools that do interesting things with RDF and SPARQL.
The functionality is provided by RDFLib, and while that provides a set of commands those provided here are somewhat more extensive and also based upon a common command framework that can be extended easily for more cases.
The tooling uses a common starting command, rdf
, that then executes sub-commands. As expected, the command has a help function and lists the supported sub-commands as positional arguments. These sub-commands also have their own help.
$ rdf -h
usage: rdf [-h] [-v] {validate,convert,select,query} ...
RDF tool
positional arguments:
{validate,convert,select,shell,query}
subargs
optional arguments:
-h, --help show this help message and exit
-v, --verbose
The currently supported sub-commands are as follows.
convert
- convert files between different RDF representations (NTriples, Notation3, XML, ...).query
- execute SPARQL queries over RDF files.select
- simple projections from RDF files.shell
- run an interactive shell session.validate
- validate an RDF file.
An example, running a SPARQL query over a downloaded file is shown below.
$ rdf query -i ~/social.n3 -r n3 -q "SELECT DISTINCT ?person ?topic WHERE { ?person <http://example.org/social/relationship/1.0/likes> ?topic. }"
person topic
============================================== =============================================
http://amazon.com/cprm/customers/1.0/Alice http://amazon.com/cprm/entities/1.0/Diving
http://amazon.com/cprm/customers/1.0/Bob http://amazon.com/cprm/entities/1.0/Diving
http://amazon.com/cprm/customers/1.0/Alice http://amazon.com/cprm/entities/1.0/Shoes
3 rows returned in 1.629622 seconds.
The -v
parameter to either rdf
or one of the sub-commands controls the standard Python logging level. It can be stated multiple times to increase the logging; -v
for warnings, -vv
for informational, -vvv
for debug.
For a more interactive exploration of RDF data you can run rdf shell
which gives you access to a lot of the same functions in the separate tools. The shell has a single common graph into which you can load data from external files (and stores in the future), and run SPARQL queries. The shell also has a default initialization file, so commonly used prefixes, common data, etc. can be loaded before you start your session.
$ rdf shell
RDF Shell, v0.1.0.
reading commands from file /Users/simonjo/.rdfshrc
Graph updated with 40 statements.
>
As you might expect, the shell supports a help
function and command completion as well as a persistent history.
The default location for this is ~/.rdfshrc
, all commands are read as if you typed them into the shell.
The default location for this is ~/.rdfsh_hist
, it will be read at startup and updated on closing the shell.
New commands are added as modules in the rdftools/scripts
folder and have the following structure.
import rdftools
def main():
(LOG, cmd) = rdftools.startup('Tool description.', add_args=None)
...
The add_args
parameter is used to add additional command-line arguments to the common argparse
structure. The function, if required, takes in a parser object and returns it. The common command line arguments include verbosity, help, and reading files.
def add_args(parser):
return parser
The results from startup
are a standard logger and an (ArgumentParser
) Namespace
object. The tool can then use the functions read
, read_into
, read_all
, write
, and query
to perform common operations on RDF files.
Extending the shell is also pretty simple, you add a function of the following form, it always takes a context object first, and the doc string will be used by default as the displayed help for your command. Arguments may be parsed for more structure, and print()
is used extensively for user feedback. Note that you must always return the context, whether you updated it or not. The add_command
function will install it into the shell, enabling help and command completion.
def echo(context, args):
""" echo text
Echo back the following text."""
print(args)
return context
add_command(echo)