get_next_uri.py

"""
Mint random number based IRIs.
"""

BU = Namespace('http://vivo.brown.edu/individual/')


def get_next_uri(self, prefix='n', number=1, used_uris=[], max=99999):
        """
        Mint new URIs in the VIVO pattern (e.g. n1234).
        SPARQL interpretation of getNextURI here:
        https://github.com/vivo-project/Vitro/blob/maint-rel-1.7/webapp/src/edu/cornell/mannlib/vitro/webapp/utils/jena/JenaIngestUtils.java#L250
        """
        count = 0
        out = []
        minted_uris = 0
        while True:
            next_uri = URIRef(BU + prefix + str(random.randint(1, max)))
            #Make sure we haven't used this URI already.
            if next_uri in used_uris:
                continue
            if next_uri in out:
                continue
            count += 1
            q = """
            SELECT ?o ?s2
            WHERE {{
                OPTIONAL{{<{0}> ?p ?o}}.
                OPTIONAL{{?s2 ?p2 <{0}>}}.
            }}
            """.format(next_uri)
            results = [row for row in self.query(q)]
            #Make sure result set is empty.
            if results == [(None, None)]:
                minted_uris += 1
                out.append(next_uri)
                #Don't create more uris than was requested.
                if minted_uris >= number:
                    break
            else:
                #Raise an exception if we try 50 URIs and none are 'new'.
                #This probably means something is wrong.
                vlog.debug("{0} is an assigned URI.  Trying again.".format(next_uri))
                if count == 50:
                    raise VIVODataError("""
                        50 random uris were tried and none returned
                        a unique uri.  Verify or increase max random uri.""")
        if number == 1:
            return out[0]
        else:
            return out