diff options
Diffstat (limited to 'icing/query/query-processor_test.cc')
| -rw-r--r-- | icing/query/query-processor_test.cc | 255 |
1 files changed, 255 insertions, 0 deletions
diff --git a/icing/query/query-processor_test.cc b/icing/query/query-processor_test.cc index 5bbf9ba..fa70de1 100644 --- a/icing/query/query-processor_test.cc +++ b/icing/query/query-processor_test.cc @@ -2646,6 +2646,261 @@ TEST_P(QueryProcessorTest, PropertyFilterTermAndUnrestrictedTerm) { EXPECT_THAT(results.query_terms["foo"], UnorderedElementsAre("animal")); } +TEST_P(QueryProcessorTest, TypePropertyFilter) { + // Create the schema and document store + SchemaProto schema = + SchemaBuilder() + .AddType(SchemaTypeConfigBuilder().SetType("email") + .AddProperty( + PropertyConfigBuilder() + .SetName("foo") + .SetDataTypeString(TERM_MATCH_EXACT, TOKENIZER_PLAIN) + .SetCardinality(CARDINALITY_OPTIONAL)) + .AddProperty( + PropertyConfigBuilder() + .SetName("bar") + .SetDataTypeString(TERM_MATCH_EXACT, TOKENIZER_PLAIN) + .SetCardinality(CARDINALITY_OPTIONAL)) + .AddProperty( + PropertyConfigBuilder() + .SetName("baz") + .SetDataTypeString(TERM_MATCH_EXACT, TOKENIZER_PLAIN) + .SetCardinality(CARDINALITY_OPTIONAL))) + .AddType(SchemaTypeConfigBuilder().SetType("message") + .AddProperty( + PropertyConfigBuilder() + .SetName("foo") + .SetDataTypeString(TERM_MATCH_EXACT, TOKENIZER_PLAIN) + .SetCardinality(CARDINALITY_OPTIONAL)) + .AddProperty( + PropertyConfigBuilder() + .SetName("bar") + .SetDataTypeString(TERM_MATCH_EXACT, TOKENIZER_PLAIN) + .SetCardinality(CARDINALITY_OPTIONAL)) + .AddProperty( + PropertyConfigBuilder() + .SetName("baz") + .SetDataTypeString(TERM_MATCH_EXACT, TOKENIZER_PLAIN) + .SetCardinality(CARDINALITY_OPTIONAL))) + .Build(); + // SectionIds are assigned in ascending order per schema type, + // alphabetically. + int email_bar_section_id = 0; + int email_baz_section_id = 1; + int email_foo_section_id = 2; + int message_bar_section_id = 0; + int message_baz_section_id = 1; + int message_foo_section_id = 2; + ASSERT_THAT(schema_store_->SetSchema( + schema, /*ignore_errors_and_delete_documents=*/false, + /*allow_circular_schema_definitions=*/false), + IsOk()); + + // These documents don't actually match to the tokens in the index. We're + // inserting the documents to get the appropriate number of documents and + // schema types populated. + ICING_ASSERT_OK_AND_ASSIGN(DocumentId email_document_id, + document_store_->Put(DocumentBuilder() + .SetKey("namespace", "1") + .SetSchema("email") + .Build())); + ICING_ASSERT_OK_AND_ASSIGN(DocumentId message_document_id, + document_store_->Put(DocumentBuilder() + .SetKey("namespace", "2") + .SetSchema("message") + .Build())); + + // Poplate the index + TermMatchType::Code term_match_type = TermMatchType::EXACT_ONLY; + + // Email document has content "animal" in all sections + ASSERT_THAT(AddTokenToIndex(email_document_id, email_foo_section_id, + term_match_type, "animal"), + IsOk()); + ASSERT_THAT(AddTokenToIndex(email_document_id, email_bar_section_id, + term_match_type, "animal"), + IsOk()); + ASSERT_THAT(AddTokenToIndex(email_document_id, email_baz_section_id, + term_match_type, "animal"), + IsOk()); + + // Message document has content "animal" in all sections + ASSERT_THAT(AddTokenToIndex(message_document_id, message_foo_section_id, + term_match_type, "animal"), + IsOk()); + ASSERT_THAT(AddTokenToIndex(message_document_id, message_bar_section_id, + term_match_type, "animal"), + IsOk()); + ASSERT_THAT(AddTokenToIndex(message_document_id, message_baz_section_id, + term_match_type, "animal"), + IsOk()); + + SearchSpecProto search_spec; + search_spec.set_query("animal"); + search_spec.set_term_match_type(term_match_type); + search_spec.set_search_type(GetParam()); + + // email has property filters for foo and baz properties + TypePropertyMask *email_mask = search_spec.add_type_property_filters(); + email_mask->set_schema_type("email"); + email_mask->add_paths("foo"); + email_mask->add_paths("baz"); + + // message has property filters for bar and baz properties + TypePropertyMask *message_mask = search_spec.add_type_property_filters(); + message_mask->set_schema_type("message"); + message_mask->add_paths("bar"); + message_mask->add_paths("baz"); + + ICING_ASSERT_OK_AND_ASSIGN( + QueryResults results, + query_processor_->ParseSearch( + search_spec, ScoringSpecProto::RankingStrategy::RELEVANCE_SCORE, + fake_clock_.GetSystemTimeMilliseconds())); + + // Ordered by descending DocumentId, so message comes first since it was + // inserted last + DocHitInfo expected_doc_hit_info1(message_document_id); + expected_doc_hit_info1.UpdateSection(message_bar_section_id); + expected_doc_hit_info1.UpdateSection(message_baz_section_id); + DocHitInfo expected_doc_hit_info2(email_document_id); + expected_doc_hit_info2.UpdateSection(email_foo_section_id); + expected_doc_hit_info2.UpdateSection(email_baz_section_id); + EXPECT_THAT(GetDocHitInfos(results.root_iterator.get()), + ElementsAre(expected_doc_hit_info1, expected_doc_hit_info2)); + EXPECT_THAT(results.query_term_iterators, SizeIs(1)); + + EXPECT_THAT(results.query_terms, SizeIs(1)); + EXPECT_THAT(results.query_terms[""], UnorderedElementsAre("animal")); +} + +TEST_P(QueryProcessorTest, TypePropertyFilterWithSectionRestrict) { + // Create the schema and document store + SchemaProto schema = + SchemaBuilder() + .AddType(SchemaTypeConfigBuilder().SetType("email") + .AddProperty( + PropertyConfigBuilder() + .SetName("foo") + .SetDataTypeString(TERM_MATCH_EXACT, TOKENIZER_PLAIN) + .SetCardinality(CARDINALITY_OPTIONAL)) + .AddProperty( + PropertyConfigBuilder() + .SetName("bar") + .SetDataTypeString(TERM_MATCH_EXACT, TOKENIZER_PLAIN) + .SetCardinality(CARDINALITY_OPTIONAL)) + .AddProperty( + PropertyConfigBuilder() + .SetName("baz") + .SetDataTypeString(TERM_MATCH_EXACT, TOKENIZER_PLAIN) + .SetCardinality(CARDINALITY_OPTIONAL))) + .AddType(SchemaTypeConfigBuilder().SetType("message") + .AddProperty( + PropertyConfigBuilder() + .SetName("foo") + .SetDataTypeString(TERM_MATCH_EXACT, TOKENIZER_PLAIN) + .SetCardinality(CARDINALITY_OPTIONAL)) + .AddProperty( + PropertyConfigBuilder() + .SetName("bar") + .SetDataTypeString(TERM_MATCH_EXACT, TOKENIZER_PLAIN) + .SetCardinality(CARDINALITY_OPTIONAL)) + .AddProperty( + PropertyConfigBuilder() + .SetName("baz") + .SetDataTypeString(TERM_MATCH_EXACT, TOKENIZER_PLAIN) + .SetCardinality(CARDINALITY_OPTIONAL))) + .Build(); + // SectionIds are assigned in ascending order per schema type, + // alphabetically. + int email_bar_section_id = 0; + int email_baz_section_id = 1; + int email_foo_section_id = 2; + int message_bar_section_id = 0; + int message_baz_section_id = 1; + int message_foo_section_id = 2; + ASSERT_THAT(schema_store_->SetSchema( + schema, /*ignore_errors_and_delete_documents=*/false, + /*allow_circular_schema_definitions=*/false), + IsOk()); + + // These documents don't actually match to the tokens in the index. We're + // inserting the documents to get the appropriate number of documents and + // schema types populated. + ICING_ASSERT_OK_AND_ASSIGN(DocumentId email_document_id, + document_store_->Put(DocumentBuilder() + .SetKey("namespace", "1") + .SetSchema("email") + .Build())); + ICING_ASSERT_OK_AND_ASSIGN(DocumentId message_document_id, + document_store_->Put(DocumentBuilder() + .SetKey("namespace", "2") + .SetSchema("message") + .Build())); + + // Poplate the index + TermMatchType::Code term_match_type = TermMatchType::EXACT_ONLY; + + // Email document has content "animal" in all sections + ASSERT_THAT(AddTokenToIndex(email_document_id, email_foo_section_id, + term_match_type, "animal"), + IsOk()); + ASSERT_THAT(AddTokenToIndex(email_document_id, email_bar_section_id, + term_match_type, "animal"), + IsOk()); + ASSERT_THAT(AddTokenToIndex(email_document_id, email_baz_section_id, + term_match_type, "animal"), + IsOk()); + + // Message document has content "animal" in all sections + ASSERT_THAT(AddTokenToIndex(message_document_id, message_foo_section_id, + term_match_type, "animal"), + IsOk()); + ASSERT_THAT(AddTokenToIndex(message_document_id, message_bar_section_id, + term_match_type, "animal"), + IsOk()); + ASSERT_THAT(AddTokenToIndex(message_document_id, message_baz_section_id, + term_match_type, "animal"), + IsOk()); + + SearchSpecProto search_spec; + // Create a section filter '<section name>:<query term>' + search_spec.set_query("foo:animal"); + search_spec.set_term_match_type(term_match_type); + search_spec.set_search_type(GetParam()); + + // email has property filters for foo and baz properties + TypePropertyMask *email_mask = search_spec.add_type_property_filters(); + email_mask->set_schema_type("email"); + email_mask->add_paths("foo"); + email_mask->add_paths("baz"); + + // message has property filters for bar and baz properties + TypePropertyMask *message_mask = search_spec.add_type_property_filters(); + message_mask->set_schema_type("message"); + message_mask->add_paths("bar"); + message_mask->add_paths("baz"); + + ICING_ASSERT_OK_AND_ASSIGN( + QueryResults results, + query_processor_->ParseSearch( + search_spec, ScoringSpecProto::RankingStrategy::RELEVANCE_SCORE, + fake_clock_.GetSystemTimeMilliseconds())); + + // Only hits in sections allowed by both the property filters and section + // restricts should be returned. Message document should not be returned since + // section foo specified in the section restrict is not allowed by the + // property filters. + DocHitInfo expected_doc_hit_info(email_document_id); + expected_doc_hit_info.UpdateSection(email_foo_section_id); + EXPECT_THAT(GetDocHitInfos(results.root_iterator.get()), + ElementsAre(expected_doc_hit_info)); + EXPECT_THAT(results.query_term_iterators, SizeIs(1)); + + EXPECT_THAT(results.query_terms, SizeIs(1)); + EXPECT_THAT(results.query_terms["foo"], UnorderedElementsAre("animal")); +} + TEST_P(QueryProcessorTest, DocumentBeforeTtlNotFilteredOut) { // Create the schema and document store SchemaProto schema = SchemaBuilder() |