def search(self,
q,
models=None,
order_by=RELEVANCE,
limit=None,
offset=None):
"""Perform a search."""
original_query = q
q = query.convert(original_query, LuceneQueryConverter)
if models:
models_queries = []
for m in models:
if hasattr(m, "_meta"):
models_queries.append('%s:"%s"' % (MODEL_FIELD, m._meta))
else:
models_queries.append('%s:"%s"' % (MODEL_FIELD, m))
q += ' AND (%s)' % (' '.join(models_queries))
searcher = PyLucene.IndexSearcher(settings.SEARCH_INDEX_PATH)
analyzer = PorterStemmerAnalyzer()
compiled_query = PyLucene.QueryParser(CONTENTS_FIELD,
analyzer).parse(q)
if order_by is RELEVANCE:
sort = PyLucene.Sort.RELEVANCE
else:
reversed = order_by.startswith('-')
sort_field = PyLucene.SortField(order_by.lstrip('-'), reversed)
sort = PyLucene.Sort(sort_field)
hits = searcher.search(compiled_query, sort)
return self._get_search_results(original_query, hits, limit, offset)
def search(self, phrase, keywords=None, sortAscending=True):
if not phrase and not keywords:
return []
# XXX Colons in phrase will screw stuff up. Can they be quoted or
# escaped somehow? Probably by using a different QueryParser.
if keywords:
fieldPhrase = u' '.join(u':'.join((k, v))
for (k, v) in keywords.iteritems())
if phrase:
phrase = phrase + u' ' + fieldPhrase
else:
phrase = fieldPhrase
phrase = phrase.translate({
ord(u'@'): u' ',
ord(u'-'): u' ',
ord(u'.'): u' '
})
qp = PyLucene.QueryParser('text', self.analyzer)
qp.setDefaultOperator(qp.Operator.AND)
query = qp.parseQuery(phrase)
sort = PyLucene.Sort(PyLucene.SortField('sortKey', not sortAscending))
try:
hits = self.searcher.search(query, sort)
except PyLucene.JavaError, err:
if 'no terms in field sortKey' in str(err):
hits = []
else:
raise
