Follow introduction of ColumnDisplayType enum in data explorer comm

crates/amalthea/src/comm/data_explorer_comm.rs

@@ -74,7 +74,7 @@ pub struct ColumnSchema {
 	pub type_name: String,
 
 	/// Canonical Positron display name of data type
-	pub type_display: ColumnSchemaTypeDisplay,
+	pub type_display: ColumnDisplayType,
 
 	/// Column annotation / description
 	pub description: Option<String>,
@@ -198,26 +198,60 @@ pub struct ColumnProfileResult {
 	pub frequency_table: Option<ColumnFrequencyTable>
 }
 
-/// ColumnSummaryStats in Schemas
+/// Profile result containing summary stats for a column based on the data
+/// type
 #[derive(Clone, Debug, Serialize, Deserialize, PartialEq)]
 pub struct ColumnSummaryStats {
+	/// Canonical Positron display name of data type
+	pub type_display: ColumnDisplayType,
+
+	/// Statistics for a numeric data type
+	pub number_stats: Option<SummaryStatsNumber>,
+
+	/// Statistics for a string-like data type
+	pub string_stats: Option<SummaryStatsString>,
+
+	/// Statistics for a boolean data type
+	pub boolean_stats: Option<SummaryStatsBoolean>
+}
+
+/// SummaryStatsNumber in Schemas
+#[derive(Clone, Debug, Serialize, Deserialize, PartialEq)]
+pub struct SummaryStatsNumber {
 	/// Minimum value as string
 	pub min_value: String,
 
 	/// Maximum value as string
 	pub max_value: String,
 
 	/// Average value as string
-	pub mean_value: Option<String>,
+	pub mean: String,
 
 	/// Sample median (50% value) value as string
-	pub median: Option<String>,
+	pub median: String,
 
-	/// 25th percentile value as string
-	pub q25: Option<String>,
+	/// Sample standard deviation as a string
+	pub stdev: String
+}
+
+/// SummaryStatsBoolean in Schemas
+#[derive(Clone, Debug, Serialize, Deserialize, PartialEq)]
+pub struct SummaryStatsBoolean {
+	/// The number of non-null true values
+	pub true_count: i64,
+
+	/// The number of non-null false values
+	pub false_count: i64
+}
+
+/// SummaryStatsString in Schemas
+#[derive(Clone, Debug, Serialize, Deserialize, PartialEq)]
+pub struct SummaryStatsString {
+	/// The number of empty / length-zero values
+	pub num_empty: i64,
 
-	/// 75th percentile value as string
-	pub q75: Option<String>
+	/// The exact number of distinct values
+	pub num_unique: i64
 }
 
 /// Result from a histogram profile request
@@ -274,9 +308,9 @@ pub struct ColumnSortKey {
 	pub ascending: bool
 }
 
-/// Possible values for TypeDisplay in ColumnSchema
+/// Possible values for ColumnDisplayType
 #[derive(Clone, Debug, Serialize, Deserialize, PartialEq)]
-pub enum ColumnSchemaTypeDisplay {
+pub enum ColumnDisplayType {
 	#[serde(rename = "number")]
 	Number,
 

crates/ark/src/data_explorer/r_data_explorer.rs

@@ -11,7 +11,7 @@ use amalthea::comm::comm_channel::CommMsg;
 use amalthea::comm::data_explorer_comm::ColumnProfileRequestType;
 use amalthea::comm::data_explorer_comm::ColumnProfileResult;
 use amalthea::comm::data_explorer_comm::ColumnSchema;
-use amalthea::comm::data_explorer_comm::ColumnSchemaTypeDisplay;
+use amalthea::comm::data_explorer_comm::ColumnDisplayType;
 use amalthea::comm::data_explorer_comm::ColumnSortKey;
 use amalthea::comm::data_explorer_comm::DataExplorerBackendReply;
 use amalthea::comm::data_explorer_comm::DataExplorerBackendRequest;
@@ -684,61 +684,61 @@ impl RDataExplorer {
 // This returns the type of an _element_ of the column. In R atomic
 // vectors do not have a distinct internal type but we pretend that they
 // do for the purpose of integrating with Positron types.
-fn display_type(x: SEXP) -> ColumnSchemaTypeDisplay {
+fn display_type(x: SEXP) -> ColumnDisplayType {
     if r_is_s4(x) {
-        return ColumnSchemaTypeDisplay::Unknown;
+        return ColumnDisplayType::Unknown;
     }
 
     if r_is_object(x) {
         if r_inherits(x, "logical") {
-            return ColumnSchemaTypeDisplay::Boolean;
+            return ColumnDisplayType::Boolean;
         }
 
         if r_inherits(x, "integer") {
-            return ColumnSchemaTypeDisplay::Number;
+            return ColumnDisplayType::Number;
         }
         if r_inherits(x, "double") {
-            return ColumnSchemaTypeDisplay::Number;
+            return ColumnDisplayType::Number;
         }
         if r_inherits(x, "complex") {
-            return ColumnSchemaTypeDisplay::Number;
+            return ColumnDisplayType::Number;
         }
         if r_inherits(x, "numeric") {
-            return ColumnSchemaTypeDisplay::Number;
+            return ColumnDisplayType::Number;
         }
 
         if r_inherits(x, "character") {
-            return ColumnSchemaTypeDisplay::String;
+            return ColumnDisplayType::String;
         }
         if r_inherits(x, "factor") {
-            return ColumnSchemaTypeDisplay::String;
+            return ColumnDisplayType::String;
         }
 
         if r_inherits(x, "Date") {
-            return ColumnSchemaTypeDisplay::Date;
+            return ColumnDisplayType::Date;
         }
         if r_inherits(x, "POSIXct") {
-            return ColumnSchemaTypeDisplay::Datetime;
+            return ColumnDisplayType::Datetime;
         }
         if r_inherits(x, "POSIXlt") {
-            return ColumnSchemaTypeDisplay::Datetime;
+            return ColumnDisplayType::Datetime;
         }
 
         // TODO: vctrs's list_of
         if r_inherits(x, "list") {
-            return ColumnSchemaTypeDisplay::Unknown;
+            return ColumnDisplayType::Unknown;
         }
 
         // Catch-all, including for data frame
-        return ColumnSchemaTypeDisplay::Unknown;
+        return ColumnDisplayType::Unknown;
     }
 
     match r_typeof(x) {
-        LGLSXP => return ColumnSchemaTypeDisplay::Boolean,
-        INTSXP | REALSXP | CPLXSXP => return ColumnSchemaTypeDisplay::Number,
-        STRSXP => return ColumnSchemaTypeDisplay::String,
-        VECSXP => return ColumnSchemaTypeDisplay::Unknown,
-        _ => return ColumnSchemaTypeDisplay::Unknown,
+        LGLSXP => return ColumnDisplayType::Boolean,
+        INTSXP | REALSXP | CPLXSXP => return ColumnDisplayType::Number,
+        STRSXP => return ColumnDisplayType::String,
+        VECSXP => return ColumnDisplayType::Unknown,
+        _ => return ColumnDisplayType::Unknown,
     }
 }