refactor: wire up column_values for row_group

read_buffer/src/column.rs

@@ -683,10 +683,21 @@ impl Column {
         }
     }
 
-    /// Determines if the column contains other values than those provided in
-    /// `values`.
-    pub fn has_other_values(&self, values: &BTreeSet<String>) -> bool {
-        todo!()
+    /// Determines if the column contains any string values that are not present
+    /// in the provided `values` argument.
+    pub fn has_other_non_null_string_values(&self, values: &BTreeSet<String>) -> bool {
+        match self {
+            Column::String(_, data) => data.has_other_non_null_values(values),
+            Column::Float(_, _) => unimplemented!("operation not supported on `Float` column"),
+            Column::Integer(_, _) => unimplemented!("operation not supported on `Integer` column"),
+            Column::Unsigned(_, _) => {
+                unimplemented!("operation not supported on `Unsigned` column")
+            }
+            Column::Bool(_, _) => unimplemented!("operation not supported on `Bool` column"),
+            Column::ByteArray(_, _) => {
+                unimplemented!("operation not supported on `ByteArray` column")
+            }
+        }
     }
 }
 

read_buffer/src/column/string.rs

@@ -91,6 +91,15 @@ impl StringEncoding {
         }
     }
 
+    /// Determines if the column contains any values other than those provided.
+    /// Short-circuits execution as soon as it finds a value not in `values`.
+    pub fn has_other_non_null_values(&self, values: &BTreeSet<String>) -> bool {
+        match &self {
+            Self::RLEDictionary(c) => c.has_other_non_null_values(values),
+            Self::Dictionary(c) => c.has_other_non_null_values(values),
+        }
+    }
+
     /// Returns the logical value found at the provided row id.
     pub fn value(&self, row_id: u32) -> Value<'_> {
         match &self {

read_buffer/src/row_group.rs

@@ -1019,7 +1019,7 @@ impl RowGroup {
                 match dst.get(*name) {
                     // process the column if we haven't got all the distinct
                     // values.
-                    Some(values) => column.has_other_values(values),
+                    Some(values) => column.has_other_non_null_string_values(values),
                     // no existing values for this column - we will need to
                     // process it.
                     None => true,
@@ -1039,9 +1039,11 @@ impl RowGroup {
             };
 
             let results = dst.entry(name.clone()).or_default();
-            for value in column.distinct_values(row_itr).iter() {
-                if value.is_some() && !results.contains(value.unwrap()) {
-                    results.insert(value.unwrap().to_owned());
+            for value in column.distinct_values(row_itr).into_iter() {
+                if let Some(v) = value {
+                    if !results.contains(v) {
+                        results.insert(v.to_owned());
+                    }
                 }
             }
         }
@@ -3057,6 +3059,21 @@ west,host-d,11,9
             to_map(vec![("env", &["stag"]), ("region", &["north", "south"])])
         );
 
+        let mut dst = BTreeMap::new();
+        dst.insert(
+            "env".to_owned(),
+            vec!["stag".to_owned()].into_iter().collect::<BTreeSet<_>>(),
+        );
+        let result = rg.column_values(
+            &Predicate::new(vec![BinaryExpr::from(("time", ">", 1_i64))]),
+            &["env", "region"],
+            dst,
+        );
+        assert_eq!(
+            result,
+            to_map(vec![("env", &["stag"]), ("region", &["north", "south"])])
+        );
+
         let result = rg.column_values(
             &Predicate::new(vec![BinaryExpr::from(("time", ">", 4_i64))]),
             &["env", "region"],