#!/usr/bin/env python

"""
Populates a solr index with LCSH concepts from the triple store.
"""

from rdflib import ConjunctiveGraph
from rdflib import RDF
from lcsh.namespaces import SKOS
from lcsh.config import Config
from solr import SolrConnection

config = Config()

# connect to solr
solr = SolrConnection(config.solr)

# clean out the index
solr.delete_query('uri:[*TO *]') 
solr.optimize()

# open triplestore
graph = ConjunctiveGraph('Sleepycat')
graph.open(config.store)

try:
    count = 0

    # find each concept
    for s in graph.subjects(predicate=RDF.type, object=SKOS['Concept']):
    
        # gather properties for each concept
        pref_label = None
        alt_labels = []
        for p, o in graph.predicate_objects(s):
            if p == SKOS['prefLabel']:
                pref_label = o
            elif p == SKOS['altLabel']:
                alt_labels.append(o)

        # add to index
        solr.add(uri=s, pref_label=pref_label, alt_label=alt_labels)

        # diagnostics
        count += 1
        print "%i %s" % (count, s)
        if count % 10000 == 0:
            solr.commit()

finally:
    graph.close()