Provide a Python wrapper

[chiarcos]

• Start CoNLLRDFManager with a given configuration with subprocess.Popen, e.g.,

    self.parse2graph_call=str(\
    	"bash "+os.path.join(self.path, "run.sh")+" "+\
    	"CoNLLRDFManager -c "+os.path.join(self.path,"..","parse2graph.json"))
    self.parse2graph=subprocess.Popen(self.parse2graph_call.split(),
    		 stdin=subprocess.PIPE,
             stdout=subprocess.PIPE,
             stderr=sys.stderr, 
             universal_newlines=True,
             bufsize=0)
  

• Provide a parse(self, string) method that

  • writes into self.parse2graph.stdin,
  • terminates with a stop symbol that will be preserved by CoNLL-RDF (e.g. \n#_END_\n\n),
  • reads from self.parse2graph.stdout until the stop symbol is encountered
  • return the result

This is implemented already, but it currently fails because of #56.

[chiarcos]

This is operational now. Basically, this only requires a single file. Any preferences where to put this in the overall repository structure? Otherwise, I would put this in examples, but it won't be found there.

[leogott]

src/main/python/conll_rdf.py seems like a reasonable choice?

[chiarcos]

src/main/python/extract.py (no formatter or updater included yet). Tested only for input from stdin without SPARQL execution.
TODO (low prio):

• test for SPARQL exec
• test for processing lists of files
• test for processing single file
• test for processing multi-line strings (= full content of one file)
• test for processing bulk data (tbc whether we may need to restart the extractor process)
• tested on Ubuntu only => test on Fedora-type Linux and MacOS

Output is slightly reformatted (dedup prefixes, one line per word), but not restructured (i.e., no line ordering as provided by CoNLLRDFFormatter).