Fetch variable metadata from Spanner

internal/server/spanner/client.go

@@ -38,6 +38,7 @@ type SpannerClient interface {
 	SearchNodes(ctx context.Context, query string, types []string) ([]*SearchNode, error)
 	ResolveByID(ctx context.Context, nodes []string, in, out string) (map[string][]string, error)
 	Sparql(ctx context.Context, nodes []types.Node, queries []*types.Query, opts *types.QueryOptions) ([][]string, error)
+	GetVariableMetadata(ctx context.Context, ids []string) (map[string][]*VariableMetadata, error)
 	Id() string
 	Start()
 	Close()

internal/server/spanner/golden/query/get_variable_metadata.json

@@ -0,0 +1,87 @@
+{
+  "Count_Household_FamilyHousehold": [
+    {
+      "VariableMeasured": "Count_Household_FamilyHousehold",
+      "ImportName": "CensusACS5YearSurvey",
+      "FacetId": "1457828370",
+      "ObservationPeriod": "",
+      "MeasurementMethod": "CensusACS5yrSurvey",
+      "Unit": "",
+      "ScalingFactor": "",
+      "IsDcAggregate": false,
+      "TotalObservations": 2547106,
+      "ObservedPlaces": 220438,
+      "MinDate": "2011",
+      "MaxDate": "2023",
+      "PlaceTypes": [
+        "AdministrativeArea1",
+        "AdministrativeArea2",
+        "AdministrativeArea4",
+        "AdministrativeArea5",
+        "Borough",
+        "CensusCoreBasedStatisticalArea",
+        "CensusCountyDivision",
+        "CensusDivision",
+        "CensusTract",
+        "CensusZipCodeTabulationArea",
+        "City",
+        "CongressionalDistrict",
+        "Country",
+        "County",
+        "ElementarySchoolDistrict",
+        "HighSchoolDistrict",
+        "Neighborhood",
+        "Place",
+        "ProvisionalNode",
+        "SchoolDistrict",
+        "State",
+        "StateComponent",
+        "Town",
+        "Village"
+      ]
+    }
+  ],
+  "Count_Household_HasComputer": [
+    {
+      "VariableMeasured": "Count_Household_HasComputer",
+      "ImportName": "CensusACS5YearSurvey",
+      "FacetId": "1457828370",
+      "ObservationPeriod": "",
+      "MeasurementMethod": "CensusACS5yrSurvey",
+      "Unit": "",
+      "ScalingFactor": "",
+      "IsDcAggregate": false,
+      "TotalObservations": 2805865,
+      "ObservedPlaces": 511517,
+      "MinDate": "2017",
+      "MaxDate": "2023",
+      "PlaceTypes": [
+        "AdministrativeArea1",
+        "AdministrativeArea2",
+        "AdministrativeArea4",
+        "AdministrativeArea5",
+        "Borough",
+        "CensusBlockGroup",
+        "CensusCoreBasedStatisticalArea",
+        "CensusCountyDivision",
+        "CensusDivision",
+        "CensusTract",
+        "CensusZipCodeTabulationArea",
+        "City",
+        "CongressionalDistrict",
+        "Country",
+        "County",
+        "ElementarySchoolDistrict",
+        "HighSchoolDistrict",
+        "Neighborhood",
+        "Place",
+        "ProvisionalNode",
+        "SchoolDistrict",
+        "State",
+        "StateComponent",
+        "Town",
+        "Village"
+      ]
+    }
+  ]
+}

internal/server/spanner/golden/query_builder/get_variable_metadata.sql

@@ -0,0 +1,18 @@
+		SELECT
+			variable_measured,
+			import_name,
+			facet_id,
+			observation_period,
+			measurement_method,
+			unit,
+			scaling_factor,
+			is_dc_aggregate,
+			total_observations,
+			observed_places,
+			min_date,
+			max_date,
+			place_types
+		FROM
+			VariableMetadata
+		WHERE
+			variable_measured IN ('Count_Household_FamilyHousehold','Count_Household_HasComputer')

internal/server/spanner/golden/query_builder_test.go

@@ -125,6 +125,18 @@ func TestSparqlQuery(t *testing.T) {
 	}
 }
 
+func TestGetVariableMetadataQuery(t *testing.T) {
+	t.Parallel()
+
+	for _, c := range variableMetadataTestCases {
+		goldenFile := c.golden + ".sql"
+
+		runQueryBuilderGoldenTest(t, goldenFile, func(ctx context.Context) (interface{}, error) {
+			return spanner.GetVariableMetadataQuery(c.ids), nil
+		})
+	}
+}
+
 // runQueryBuilderGoldenTest is a helper function that performs the golden file validation.
 func runQueryBuilderGoldenTest(t *testing.T, goldenFile string, fn goldenTestFunc) {
 	t.Helper()

internal/server/spanner/golden/query_cases_test.go

@@ -410,3 +410,13 @@ var sparqlTestCases = []struct {
 		golden: "sparql_dcid_california",
 	},
 }
+
+var variableMetadataTestCases = []struct {
+	ids    []string
+	golden string
+}{
+	{
+		ids:    []string{"Count_Household_FamilyHousehold", "Count_Household_HasComputer"},
+		golden: "get_variable_metadata",
+	},
+}

internal/server/spanner/golden/query_test.go

@@ -190,6 +190,23 @@ func TestSparql(t *testing.T) {
 	}
 }
 
+func TestGetVariableMetadata(t *testing.T) {
+	client := test.NewSpannerClient()
+	if client == nil {
+		return
+	}
+
+	t.Parallel()
+
+	for _, c := range variableMetadataTestCases {
+		goldenFile := c.golden + ".json"
+
+		runQueryGoldenTest(t, goldenFile, func(ctx context.Context) (interface{}, error) {
+			return client.GetVariableMetadata(ctx, c.ids)
+		})
+	}
+}
+
 // runQueryGoldenTest is a helper function that performs the golden file validation.
 func runQueryGoldenTest(t *testing.T, goldenFile string, fn goldenTestFunc) {
 	t.Helper()

internal/server/spanner/model.go

@@ -108,6 +108,23 @@ type ResolutionCandidate struct {
 	Candidate string `spanner:"candidate"`
 }
 
+// VariableMetadata struct represents a single row in the VariableMetadata table.
+type VariableMetadata struct {
+	VariableMeasured  string   `spanner:"variable_measured"`
+	ImportName        string   `spanner:"import_name"`
+	FacetId           string   `spanner:"facet_id"`
+	ObservationPeriod string   `spanner:"observation_period"`
+	MeasurementMethod string   `spanner:"measurement_method"`
+	Unit              string   `spanner:"unit"`
+	ScalingFactor     string   `spanner:"scaling_factor"`
+	IsDcAggregate     bool     `spanner:"is_dc_aggregate"`
+	TotalObservations int64    `spanner:"total_observations"`
+	ObservedPlaces    int64    `spanner:"observed_places"`
+	MinDate           string   `spanner:"min_date"`
+	MaxDate           string   `spanner:"max_date"`
+	PlaceTypes        []string `spanner:"place_types"`
+}
+
 // SpannerConfig struct to hold the YAML configuration to a spanner database.
 type SpannerConfig struct {
 	Project  string `yaml:"project"`

internal/server/spanner/query.go

@@ -214,6 +214,31 @@ func (sc *spannerDatabaseClient) Sparql(ctx context.Context, nodes []types.Node,
 	return sc.queryDynamic(ctx, *query)
 }
 
+func (sc *spannerDatabaseClient) GetVariableMetadata(ctx context.Context, ids []string) (map[string][]*VariableMetadata, error) {
+	var idToMetadata = make(map[string][]*VariableMetadata)
+	if len(ids) == 0 {
+		return idToMetadata, nil
+	}
+
+	err := sc.queryStructs(
+		ctx,
+		*GetVariableMetadataQuery(ids),
+		func() interface{} {
+			return &VariableMetadata{}
+		},
+		func(rowStruct interface{}) {
+			varMeta := rowStruct.(*VariableMetadata)
+			id := varMeta.VariableMeasured
+			idToMetadata[id] = append(idToMetadata[id], varMeta)
+		},
+	)
+	if err != nil {
+		return nil, err
+	}
+
+	return idToMetadata, nil
+}
+
 // fetchAndUpdateTimestamp queries Spanner and updates the timestamp.
 func (sc *spannerDatabaseClient) fetchAndUpdateTimestamp(ctx context.Context) error {
 	iter := sc.client.Single().Query(ctx, *GetCompletionTimestampQuery())

internal/server/spanner/query_builder.go

@@ -340,6 +340,15 @@ func SparqlQuery(nodes []types.Node, queries []*types.Query, opts *types.QueryOp
 	}, nil
 }
 
+func GetVariableMetadataQuery(ids []string) *spanner.Statement {
+	getIds, params := getIdStatement(ids)
+
+	return &spanner.Statement{
+		SQL:    fmt.Sprintf(statements.getVariableMetadata, getIds),
+		Params: params,
+	}
+}
+
 // generateSafeAliasMap generates a map of safe aliases for SPARQL queries.
 func generateSafeAliasMap(queries []*types.Query) map[string]string {
 	safeAliasMap := make(map[string]string)

internal/server/spanner/statements.go

@@ -85,6 +85,8 @@ var statements = struct {
 	nodeFilter string
 	// Generic triple pattern.
 	triple string
+	// Get variable metadata.
+	getVariableMetadata string
 }{
 	getCompletionTimestamp: `		SELECT
 		CompletionTimestamp
@@ -292,4 +294,22 @@ var statements = struct {
 		WHERE
 			%[1]s.subject_id IN UNNEST(@%[1]s)`,
 	triple: `(%[1]s:Node%[2]s)-[:Edge {predicate: @predicate%[3]d}]->(%[4]s:Node%[5]s)`,
+	getVariableMetadata: `		SELECT
+			variable_measured,
+			import_name,
+			facet_id,
+			observation_period,
+			measurement_method,
+			unit,
+			scaling_factor,
+			is_dc_aggregate,
+			total_observations,
+			observed_places,
+			min_date,
+			max_date,
+			place_types
+		FROM
+			VariableMetadata
+		WHERE
+			variable_measured %s`,
 }