def test_geo():
# Geo location search is not supported yet.
# Links:
# - https://cloud.google.com/appengine/docs/standard/python/search/query_strings#Python_Queries_on_geopoint_fields (GAE)
# - https://lucene.apache.org/solr/guide/7_6/spatial-search.html (Solr).
prepare_solr_query('distance(location, geopoint(35.2, 40.5)) < 100',
FIELDS, GROUPED_FIELDS)
def test_bool_operator_priorities():
solr_query_options = prepare_solr_query('NOT cat AND dogs OR horses',
FIELDS, GROUPED_FIELDS)
assert solr_query_options.query_string == \
'(NOT "cat" AND ("dogs" OR "horses"))'
solr_query_options = prepare_solr_query('NOT cat OR dogs AND horses',
FIELDS, GROUPED_FIELDS)
assert solr_query_options.query_string == \
'((NOT "cat" OR "dogs") AND "horses")'
def test_field_match_any_value():
# Value like number OR value like date
solr_query_options = prepare_solr_query(
'description:(-123.6 OR 1999-08-15)', FIELDS, GROUPED_FIELDS)
assert solr_query_options.query_string == (
'((description_txt_en:"-123.6" OR description_txt_fr:"-123.6" OR'
' description_number:"-123.6") OR '
'(description_txt_en:"1999-08-15" OR description_txt_fr:"1999-08-15" OR'
' description_date:"1999-08-15"))')
assert solr_query_options.query_fields == []
def test_single_word():
solr_query_options = prepare_solr_query('foo', FIELDS, GROUPED_FIELDS)
assert solr_query_options.query_string == '"foo"'
assert set(solr_query_options.query_fields) == {
'tag_atom',
'name_atom',
'description_txt_en',
'description_txt_fr',
}
assert solr_query_options.def_type == 'edismax'
def test_word_and_word():
solr_query_options = prepare_solr_query('hello AND world', FIELDS,
GROUPED_FIELDS)
assert solr_query_options.query_string == '("hello" AND "world")'
assert set(solr_query_options.query_fields) == {
'tag_atom',
'name_atom',
'description_txt_en',
'description_txt_fr',
}
def test_single_number():
solr_query_options = prepare_solr_query('-123.4', FIELDS, GROUPED_FIELDS)
assert solr_query_options.query_string == '"-123.4"'
assert set(solr_query_options.query_fields) == {
'tag_atom',
'name_atom',
'description_txt_en',
'description_txt_fr',
'description_number',
'price_number',
}
def test_quoted():
solr_query_options = prepare_solr_query(
'"tag:hello AND whateve OR (it does NOT matter)"', FIELDS,
GROUPED_FIELDS)
assert solr_query_options.query_string == \
'"tag:hello AND whateve OR (it does NOT matter)"'
assert set(solr_query_options.query_fields) == {
'tag_atom',
'name_atom',
'description_txt_en',
'description_txt_fr',
}
def test_word_and_word_or_number():
solr_query_options = prepare_solr_query('hello AND world OR 321.6', FIELDS,
GROUPED_FIELDS)
assert solr_query_options.query_string == '("hello" AND ("world" OR "321.6"))'
assert set(solr_query_options.query_fields) == {
'tag_atom',
'name_atom',
'description_txt_en',
'description_txt_fr',
'description_number',
'price_number',
}
def test_field_compare():
solr_query_options = prepare_solr_query(
'description>100.2 description<=200 price<999.9 price>=100 '
'created_at>=1990-06-13', FIELDS, GROUPED_FIELDS)
assert solr_query_options.query_string == (
'(description_number:{"100.2" TO *] AND description_number:[* TO "200"]'
' AND price_number:[* TO "999.9"} AND price_number:["100" TO *]'
' AND created_at_date:["1990-06-13" TO *])'
# Searching date on description_number would lead to error,
# so it should be skipped.
)
assert solr_query_options.query_fields == []
def test_two_words():
solr_query_options = prepare_solr_query('hello world', FIELDS,
GROUPED_FIELDS)
# It's currently expected to fail as we convert it to '"hello" AND "world"'
# Links:
# - https://cloud.google.com/appengine/docs/standard/python/search/query_strings#Python_Multi-value_queries
assert solr_query_options.query_string == '"hello world"'
assert set(solr_query_options.query_fields) == {
'tag_atom',
'name_atom',
'description_txt_en',
'description_txt_fr',
}
def test_field_with_many_types():
# Value like number
solr_query_options = prepare_solr_query('description:-123.6', FIELDS,
GROUPED_FIELDS)
assert solr_query_options.query_string == (
'(description_txt_en:"-123.6" OR description_txt_fr:"-123.6" OR'
' description_number:"-123.6")'
# Searching number on description_date would lead to error,
# so it should be skipped.
)
assert solr_query_options.query_fields == []
# Value like date
solr_query_options = prepare_solr_query('description:1999-08-15', FIELDS,
GROUPED_FIELDS)
assert solr_query_options.query_string == (
'(description_txt_en:"1999-08-15" OR description_txt_fr:"1999-08-15" OR'
' description_date:"1999-08-15")'
# Searching date on description_number would lead to error,
# so it should be skipped.
)
assert solr_query_options.query_fields == []
def test_single_date():
solr_query_options = prepare_solr_query('2019-01-23', FIELDS,
GROUPED_FIELDS)
assert solr_query_options.query_string == '"2019-01-23"'
assert set(solr_query_options.query_fields) == {
'tag_atom',
'name_atom',
'description_txt_en',
'description_txt_fr',
'description_date',
'created_at_date',
'modified_at_date',
}
def test_date_and_word_or_number():
solr_query_options = prepare_solr_query('2019-02-02 AND world OR 321.6',
FIELDS, GROUPED_FIELDS)
assert solr_query_options.query_string == \
'("2019-02-02" AND ("world" OR "321.6"))'
assert set(solr_query_options.query_fields) == {
'tag_atom',
'name_atom',
'description_txt_en',
'description_txt_fr',
'description_number',
'description_date',
'created_at_date',
'modified_at_date',
'price_number',
}
def test_complex_composition():
solr_query_options = prepare_solr_query(
'name:(-123.6 OR 1999-08-15 AND (foo AND NOT bar dogs (hello))) '
'AND tag:phone "some () quoted text NOT 123" '
'OR (one two OR three) OR price>321.6', FIELDS, GROUPED_FIELDS)
assert solr_query_options.query_string == (
'(((name_atom:"-123.6" OR name_atom:"1999-08-15") AND '
'(name_atom:"foo" AND NOT name_atom:"bar" AND name_atom:"dogs" AND '
'name_atom:"hello")) AND '
'tag_atom:"phone" AND ("some () quoted text NOT 123" OR '
'("one" AND ("two" OR "three")) OR '
'price_number:{"321.6" TO *]))')
assert set(solr_query_options.query_fields) == {
'tag_atom',
'name_atom',
'description_txt_en',
'description_txt_fr',
}
async def query(self, app_id, namespace, index_name, query,
projection_fields, sort_expressions, limit, offset, cursor,
keys_only, auto_discover_facet_count, facet_requests,
facet_refinements, facet_auto_detect_limit):
""" Retrieves documents which matches query from Solr collection
and converts it to unified documents.
Args:
app_id: a str representing Application ID.
namespace: a str representing GAE namespace or None.
index_name: a str representing name of Search API index.
query: a str containing Search API query.
projection_fields: a list of field names to return in results.
sort_expressions: a list of sort expressions, e.g.: ("field1", "asc").
limit: an int specifying maximum number of results to return.
offset: an int specifying number of first document to skip.
cursor: a str representing query cursor.
keys_only: a bool indicating if only document IDs should be returned.
auto_discover_facet_count: An int - number of top facets to discover.
facet_requests: A list of FacetRequest.
facet_refinements: A list of FacetRefinement.
facet_auto_detect_limit: An int - number of top terms to return.
Returns (asynchronously):
An instance of models.SearchResult.
"""
index_schema = await self._get_schema_info(app_id, namespace,
index_name)
# Convert Search API query to Solr query with a list of fields to search.
query_options = query_converter.prepare_solr_query(
query, index_schema.fields, index_schema.grouped_fields)
# Process GAE projection fields
solr_projection_fields = self._convert_projection(
keys_only, projection_fields, index_schema)
# Process GAE sort expressions
solr_sort_fields = self._convert_sort_expressions(
sort_expressions, index_schema)
# Process GAE facet-related parameters
refinement_filter = None
if facet_refinements:
# Determine if we need to filter by refinement.
refinement_filter = facet_converter.generate_refinement_filter(
index_schema.grouped_facet_indexes, facet_refinements)
facet_items, stats_items = await self._convert_facet_args(
auto_discover_facet_count, facet_auto_detect_limit, facet_requests,
index_schema, query_options, refinement_filter)
stats_fields = [stats_line for solr_field, stats_line in stats_items]
# DO ACTUAL QUERY:
solr_result = await self.solr.query_documents(
collection=index_schema.collection,
query=query_options.query_string,
offset=offset,
limit=limit,
cursor=cursor,
fields=solr_projection_fields,
sort=solr_sort_fields,
def_type=query_options.def_type,
query_fields=query_options.query_fields,
facet_dict=dict(facet_items) if facet_items else None,
stats_fields=stats_fields or None,
filter_=refinement_filter)
# Convert Solr results to unified models
docs = [
_from_solr_document(solr_doc) for solr_doc in solr_result.documents
]
# Read stats results
stats_results = []
for solr_field, stats_line in stats_items:
stats_info = solr_result.stats_results[solr_field.solr_name]
stats_results.append((solr_field.gae_name, stats_info))
# Convert facet results from Solr facets and stats
facet_results = facet_converter.convert_facet_results(
solr_result.facet_results, stats_results)
result = SearchResult(num_found=solr_result.num_found,
scored_documents=docs,
cursor=cursor,
facet_results=facet_results)
return result
def test_stem_text():
solr_query_options = prepare_solr_query('~hello', FIELDS, GROUPED_FIELDS)
assert solr_query_options.query_string == '"hello"~'
def test_field_quoted_value():
solr_query_options = prepare_solr_query('tag:"hello"', FIELDS,
GROUPED_FIELDS)
assert solr_query_options.query_string == 'tag_atom:"hello"'
assert solr_query_options.query_fields == []