def test_distinct(self):
self._create(z=0)
self._create(z=0)
self._create(z=1)
self._create(z=1)
self._create(z=1)
distinct = query.distinct(query.match(self.z_index_ref))
self.assertEqual(self._set_to_list(distinct), [0, 1])
def wordPartsGenerator(word):
return q.let(
{
"indexes":
q.map_(
# Reduce this array if you want less ngrams per word.
# Setting it to [ 0 ] would only create the word itself, Setting it to [0, 1] would result in the word itself
# and all ngrams that are one character shorter, etc..
lambda index: q.subtract(q.length(word), index),
maxNgrams),
"indexesFiltered":
q.filter_(
# left min parts length 3
lambda l: q.gte(l, 3),
q.var('indexes')),
"ngramsArray":
q.distinct(
q.union(
q.map_(lambda l: q.ngram(q.lowercase(word), l, l),
q.var('indexesFiltered'))))
},
q.var('ngramsArray'))
def test_distinct(self):
self.assertJson(
query.distinct(SetRef({"match": query.index("widget")})),
'{"distinct":{"@set":{"match":{"index":"widget"}}}}')
def translate_select(sql_query: sql.SQLQuery) -> QueryExpression:
"""Translate a SELECT SQL query into an equivalent FQL query.
Params:
-------
sql_query: An SQLQuery instance.
Returns:
--------
An FQL query expression based on the SQL query.
"""
document_pages = _define_document_pages(sql_query)
selected_table = next(table for table in sql_query.tables
if table.has_columns)
get_field_value = lambda function_value, raw_value: q.if_(
q.equals(function_value, common.NULL),
q.if_(q.equals(raw_value, common.NULL), None, raw_value),
q.select([common.DATA, 0], function_value),
)
calculate_function_value = lambda document_set, function_name: q.if_(
q.is_null(function_name),
common.NULL,
q.if_(
q.equals(function_name, sql.Function.COUNT.value),
q.count(document_set),
common.NULL,
),
)
# With aggregation functions, standard behaviour is to include the first value
# if any column selections are part of the query, at least until we add support
# for GROUP BY
get_first_document = lambda documents: q.if_(q.is_empty(documents), [{}],
q.take(1, documents))
translate_document_fields = lambda maybe_documents: q.let(
{
# We map over selected_fields to build document object
# to maintain the order of fields as queried. Otherwise,
# SQLAlchemy gets confused and assigns values to the incorrect keys.
"selected_column_info":
[[col.table_name, col.name, col.function_name]
for col in sql_query.columns],
"has_functions":
any(col.function_name for col in sql_query.columns),
"maybe_document_set":
q.if_(
q.var("has_functions"),
get_first_document(maybe_documents),
maybe_documents,
),
"field_alias_map":
sql_query.alias_map,
},
q.map_(
q.lambda_(
"maybe_document",
q.let(
{
"document":
q.if_(
q.is_ref(q.var("maybe_document")),
{
# We use the selected table name here instead of deriving
# the collection name from the document ref in order to
# save a 'get' call from inside of a map, which could get
# expensive.
selected_table.name:
q.merge(
q.select(
common.DATA,
q.get(q.var("maybe_document")),
),
{"ref": q.var("maybe_document")},
),
},
q.var("maybe_document"),
),
},
q.to_object(
q.map_(
q.lambda_(
[
"collection_name", "field_name",
"function_name"
],
q.let(
{
"function_value":
calculate_function_value(
maybe_documents,
q.var("function_name")),
"raw_value":
q.select(
[
q.var("collection_name"),
q.var("field_name"),
],
q.var("document"),
default=common.NULL,
),
},
[
q.select(
[
q.var("collection_name"),
q.var("field_name"),
],
q.var("field_alias_map"),
),
get_field_value(
q.var("function_value"),
q.var("raw_value")),
],
),
),
q.var("selected_column_info"),
)),
),
),
q.var("maybe_document_set"),
),
)
return q.let(
{
"maybe_documents":
document_pages,
"translated_documents":
translate_document_fields(q.var("maybe_documents")),
"result":
q.distinct(q.var("translated_documents"))
if sql_query.distinct else q.var("translated_documents"),
},
# Paginated sets hold an array of results in a 'data' field, so we try to flatten it
# in case we're dealing with pages instead of an array of results which doesn't
# have such nesting
{common.DATA: q.select(common.DATA, q.var("result"), q.var("result"))},
)
