Read triples dumped from Jena SDB via SQL and convert to n-triples

read_dumped_triples.py

"""

Used chardet library to find encoding - it reported:
{'confidence': 0.7158930206047512, 'encoding': 'ISO-8859-2'}

Use line writing to print ntriples as they are read to avoid
memory issues with storing all statements in an in memory 
RDFLib Graph.

"""


import csv
import sys

from rdflib import Graph
from rdflib import Literal, URIRef

def clean_uri(raw):
    return raw.lstrip('<').rstrip('>')

def print_triple(s, p, o):
    g = Graph()
    g.add((s, p, o))
    print g.serialize(format='nt').strip('\n')

def unicode_csv_reader(utf8_data, dialect=csv.excel, **kwargs):
    """
    http://stackoverflow.com/a/904085/758157
    """
    csv_reader = csv.reader(utf8_data, dialect=dialect, **kwargs)
    for row in csv_reader:
        #Decode and then encode as utf8
        #decode('iso-8859-1').encode('utf8')
        #http://stackoverflow.com/a/6539919/758157
        yield [cell.decode('iso-8859-1').encode('utf8') for cell in row]

with open(sys.argv[1]) as infile:
    reader = unicode_csv_reader(infile, delimiter="\t")
    # skip header
    reader.next()
    for row in reader:
        s, p, o = row
        subj = URIRef(clean_uri(s))
        pred = URIRef(clean_uri(p))
        try:
            obj = Literal(o)
        except Exception:
            raise
        print_triple(subj, pred, obj)