def _query2xquery(self, query):
"""take a "itools" query and return a "xapian" query
"""
query_class = type(query)
fields = self._fields
metadata = self._metadata
# All Query
if query_class is AllQuery:
return Query('')
# PhraseQuery, the field must be indexed
if query_class is PhraseQuery:
name = query.name
if type(name) is not str:
raise TypeError, "unexpected '%s'" % type(name)
# If there is a problem => an empty result
if name not in metadata:
warn_not_indexed(name)
return Query()
info = metadata[name]
try:
prefix = info['prefix']
except KeyError:
raise ValueError, 'the field "%s" must be indexed' % name
field_cls = _get_field_cls(name, fields, info)
return _make_PhraseQuery(field_cls, query.value, prefix)
# RangeQuery, the field must be stored
if query_class is RangeQuery:
name = query.name
if type(name) is not str:
raise TypeError, "unexpected '%s'" % type(name)
# If there is a problem => an empty result
if name not in metadata:
warn_not_indexed(name)
return Query()
info = metadata[name]
value = info.get('value')
if value is None:
raise AttributeError, MSG_NOT_STORED.format(name=name)
field_cls = _get_field_cls(name, fields, info)
if field_cls.multiple:
error = 'range-query not supported on multiple fields'
raise ValueError, error
left = query.left
if left is not None:
left = _encode_simple_value(field_cls, left)
right = query.right
if right is not None:
right = _encode_simple_value(field_cls, right)
# Case 1: no limits, return everything
if left is None and right is None:
return Query('')
# Case 2: left limit only
if right is None:
return Query(OP_VALUE_GE, value, left)
# Case 3: right limit only
if left is None:
return Query(OP_VALUE_LE, value, right)
# Case 4: left and right
return Query(OP_VALUE_RANGE, value, left, right)
# StartQuery, the field must be stored
if query_class is StartQuery:
name = query.name
if type(name) is not str:
raise TypeError, "unexpected '%s'" % type(name)
# If there is a problem => an empty result
if name not in metadata:
warn_not_indexed(name)
return Query()
info = metadata[name]
value_nb = info.get('value')
if value_nb is None:
raise AttributeError, MSG_NOT_STORED.format(name=name)
field_cls = _get_field_cls(name, fields, info)
value = query.value
value = _encode(field_cls, value)
if value:
# good = {x / x >= value}
good = Query(OP_VALUE_GE, value_nb, value)
# Construct the variable end_value:
# end_value = the word "after" value: toto => totp
# Delete the '\xff' at the end of value
end_value = value
while end_value and ord(end_value[-1]) == 255:
end_value = end_value[:-1]
# Normal case: end_value is not empty
if end_value:
# The world after
end_value = end_value[:-1] + chr(ord(end_value[-1]) + 1)
# bad = {x / x >= end_value}
bad = Query(OP_VALUE_GE, value_nb, end_value)
# Return {x / x in good but x not in bad}
return Query(OP_AND_NOT, good, bad)
# If end_value is empty
else:
# Return {x / x in good}
return good
else:
# If value == '', we return everything
return Query('')
# TextQuery, the field must be indexed
if query_class is TextQuery:
name = query.name
if type(name) is not str:
raise TypeError, "unexpected %s for 'name'" % type(name)
# If there is a problem => an empty result
if name not in metadata:
warn_not_indexed(name)
return Query()
info = metadata[name]
field_cls = _get_field_cls(name, fields, info)
try:
prefix = info['prefix']
except KeyError:
raise ValueError, 'the field "%s" must be indexed' % name
# Remove accents from the value
value = query.value
if type(value) is not unicode:
raise TypeError, "unexpected %s for 'value'" % type(value)
value = value.translate(TRANSLATE_MAP)
qp = QueryParser()
qp.set_database(self._db)
return qp.parse_query(_encode(field_cls, value), TQ_FLAGS, prefix)
i2x = self._query2xquery
# Multiple query with single atom
if isinstance(query, _MultipleQuery) and len(query.atoms) == 1:
return i2x(query.atoms[0])
# And
if query_class is _AndQuery:
return Query(OP_AND, [ i2x(q) for q in query.atoms ])
# Or
if query_class is _OrQuery:
return Query(OP_OR, [ i2x(q) for q in query.atoms ])
# Not
if query_class is NotQuery:
return Query(OP_AND_NOT, Query(''), i2x(query.query))
def _query2xquery(self, query):
"""take a "itools" query and return a "xapian" query
"""
query_class = type(query)
fields = self._fields
metadata = self._metadata
# All Query
if query_class is AllQuery:
return Query('')
# PhraseQuery, the field must be indexed
if query_class is PhraseQuery:
name = query.name
if type(name) is not str:
raise TypeError, "unexpected '%s'" % type(name)
# If there is a problem => an empty result
if name not in metadata:
warn_not_indexed(name)
return Query()
info = metadata[name]
try:
prefix = info['prefix']
except KeyError:
raise ValueError, 'the field "%s" must be indexed' % name
field_cls = _get_field_cls(name, fields, info)
return _make_PhraseQuery(field_cls, query.value, prefix)
# RangeQuery, the field must be stored
if query_class is RangeQuery:
name = query.name
if type(name) is not str:
raise TypeError, "unexpected '%s'" % type(name)
# If there is a problem => an empty result
if name not in metadata:
warn_not_indexed(name)
return Query()
info = metadata[name]
value = info.get('value')
if value is None:
raise AttributeError, MSG_NOT_STORED.format(name=name)
field_cls = _get_field_cls(name, fields, info)
if field_cls.multiple:
error = 'range-query not supported on multiple fields'
raise ValueError, error
left = query.left
if left is not None:
left = _encode_simple_value(field_cls, left)
right = query.right
if right is not None:
right = _encode_simple_value(field_cls, right)
# Case 1: no limits, return everything
if left is None and right is None:
return Query('')
# Case 2: left limit only
if right is None:
return Query(OP_VALUE_GE, value, left)
# Case 3: right limit only
if left is None:
return Query(OP_VALUE_LE, value, right)
# Case 4: left and right
return Query(OP_VALUE_RANGE, value, left, right)
# StartQuery, the field must be stored
if query_class is StartQuery:
name = query.name
if type(name) is not str:
raise TypeError, "unexpected '%s'" % type(name)
# If there is a problem => an empty result
if name not in metadata:
warn_not_indexed(name)
return Query()
info = metadata[name]
value_nb = info.get('value')
if value_nb is None:
raise AttributeError, MSG_NOT_STORED.format(name=name)
field_cls = _get_field_cls(name, fields, info)
value = query.value
value = _encode(field_cls, value)
if value:
# good = {x / x >= value}
good = Query(OP_VALUE_GE, value_nb, value)
# Construct the variable end_value:
# end_value = the word "after" value: toto => totp
# Delete the '\xff' at the end of value
end_value = value
while end_value and ord(end_value[-1]) == 255:
end_value = end_value[:-1]
# Normal case: end_value is not empty
if end_value:
# The world after
end_value = end_value[:-1] + chr(ord(end_value[-1]) + 1)
# bad = {x / x >= end_value}
bad = Query(OP_VALUE_GE, value_nb, end_value)
# Return {x / x in good but x not in bad}
return Query(OP_AND_NOT, good, bad)
# If end_value is empty
else:
# Return {x / x in good}
return good
else:
# If value == '', we return everything
return Query('')
# TextQuery, the field must be indexed
if query_class is TextQuery:
name = query.name
if type(name) is not str:
raise TypeError, "unexpected %s for 'name'" % type(name)
# If there is a problem => an empty result
if name not in metadata:
warn_not_indexed(name)
return Query()
info = metadata[name]
field_cls = _get_field_cls(name, fields, info)
try:
prefix = info['prefix']
except KeyError:
raise ValueError, 'the field "%s" must be indexed' % name
# Remove accents from the value
value = query.value
if type(value) is not unicode:
raise TypeError, "unexpected %s for 'value'" % type(value)
value = value.translate(TRANSLATE_MAP)
qp = QueryParser()
qp.set_database(self._db)
return qp.parse_query(_encode(field_cls, value), TQ_FLAGS, prefix)
i2x = self._query2xquery
# Multiple query with single atom
if isinstance(query, _MultipleQuery) and len(query.atoms) == 1:
return i2x(query.atoms[0])
# And
if query_class is _AndQuery:
return Query(OP_AND, [ i2x(q) for q in query.atoms ])
# Or
if query_class is _OrQuery:
return Query(OP_OR, [ i2x(q) for q in query.atoms ])
# Not
if query_class is NotQuery:
return Query(OP_AND_NOT, Query(''), i2x(query.query))
def _query_parser(self):
query_parser = QueryParser()
query_parser.set_database(self.database)
query_parser.set_default_op(Query.OP_AND)
query_parser.set_stemmer(Stem("none"))
query_parser.add_boolean_prefix('model', 'M:')
query_parser.add_boolean_prefix('modelid', 'I:')
if self.prefixes is not None:
for name, prefix in self.prefixes.items():
if prefix == 'M:' or prefix == 'I:':
continue
query_parser.add_boolean_prefix(name, prefix)
query_parser.set_stemming_strategy(self.stem)
return query_parser
