Refactoring and more group-by flexibility.

crates/nu-cli/src/utils/data/group.rs

@@ -1,61 +1,28 @@
 use indexmap::IndexMap;
 use nu_errors::ShellError;
 use nu_protocol::{TaggedDictBuilder, UntaggedValue, Value};
-use nu_source::{Tag, Tagged, TaggedItem};
-use nu_value_ext::{as_string, get_data_by_key};
+use nu_source::Tag;
+use nu_value_ext::as_string;
 
 #[allow(clippy::type_complexity)]
 pub fn group(
-    column_name: Option<Tagged<String>>,
-    values: &[Value],
-    grouper: Option<Box<dyn Fn(&Value) -> Result<String, ShellError> + Send>>,
+    values: &Value,
+    grouper: &Option<Box<dyn Fn(&Value) -> Result<String, ShellError> + Send>>,
     tag: impl Into<Tag>,
 ) -> Result<Value, ShellError> {
     let tag = tag.into();
 
     let mut groups: IndexMap<String, Vec<Value>> = IndexMap::new();
 
-    for value in values {
-        let group_key = if let Some(ref column_name) = column_name {
-            get_data_by_key(&value, column_name.borrow_spanned())
+    for value in values.table_entries() {
+        let group_key = if let Some(ref grouper) = grouper {
+            grouper(&value)
         } else {
-            Some(value.clone())
+            as_string(&value)
         };
 
-        if let Some(group_key) = group_key {
-            let group_key = if let Some(ref grouper) = grouper {
-                grouper(&group_key)
-            } else {
-                as_string(&group_key)
-            };
-            let group = groups.entry(group_key?).or_insert(vec![]);
-            group.push((*value).clone());
-        } else {
-            let column_name = column_name.unwrap_or_else(|| String::from("").tagged(&tag));
-
-            let possibilities = value.data_descriptors();
-
-            let mut possible_matches: Vec<_> = possibilities
-                .iter()
-                .map(|x| (natural::distance::levenshtein_distance(x, &column_name), x))
-                .collect();
-
-            possible_matches.sort();
-
-            if !possible_matches.is_empty() {
-                return Err(ShellError::labeled_error(
-                    "Unknown column",
-                    format!("did you mean '{}'?", possible_matches[0].1),
-                    column_name.tag(),
-                ));
-            } else {
-                return Err(ShellError::labeled_error(
-                    "Unknown column",
-                    "row does not contain this column",
-                    column_name.tag(),
-                ));
-            }
-        }
+        let group = groups.entry(group_key?).or_insert(vec![]);
+        group.push((*value).clone());
     }
 
     let mut out = TaggedDictBuilder::new(&tag);

crates/nu-cli/src/utils/data/mod.rs

@@ -1,3 +1,5 @@
 pub mod group;
+pub mod split;
 
 pub use crate::utils::data::group::group;
+pub use crate::utils::data::split::split;

crates/nu-cli/src/utils/data_processing.rs

@@ -12,7 +12,7 @@ use num_traits::Zero;
 const ERR_EMPTY_DATA: &str = "Cannot perform aggregate math operation on empty data";
 
 pub fn columns_sorted(
-    _group_by_name: Option<String>,
+    _group_by_name: Option<Tagged<String>>,
     value: &Value,
     tag: impl Into<Tag>,
 ) -> Vec<Tagged<String>> {
@@ -61,7 +61,7 @@ pub fn columns_sorted(
 }
 
 pub fn t_sort(
-    group_by_name: Option<String>,
+    group_by_name: Option<Tagged<String>>,
     split_by_name: Option<String>,
     value: &Value,
     tag: impl Into<Tag>,
@@ -454,12 +454,13 @@ mod tests {
     }
 
     fn nu_releases_grouped_by_date() -> Result<Value, ShellError> {
-        let key = String::from("date").tagged_unknown();
-        group(&key, nu_releases_commiters(), Tag::unknown())
+        let key = Some(String::from("date").tagged_unknown());
+        let sample = table(&nu_releases_committers());
+        group(&key, &sample, Tag::unknown())
     }
 
     fn nu_releases_sorted_by_date() -> Result<Value, ShellError> {
-        let key = String::from("date");
+        let key = String::from("date").tagged(Tag::unknown());
 
         t_sort(
             Some(key),
@@ -481,7 +482,7 @@ mod tests {
         )
     }
 
-    fn nu_releases_commiters() -> Vec<Value> {
+    fn nu_releases_committers() -> Vec<Value> {
         vec![
             row(
                 indexmap! {"name".into() => string("AR"), "country".into() => string("EC"), "date".into() => string("August 23-2019")},
@@ -515,7 +516,7 @@ mod tests {
 
     #[test]
     fn show_columns_sorted_given_a_column_to_sort_by() -> Result<(), ShellError> {
-        let by_column = String::from("date");
+        let by_column = String::from("date").tagged(Tag::unknown());
 
         assert_eq!(
             columns_sorted(
@@ -535,7 +536,7 @@ mod tests {
 
     #[test]
     fn sorts_the_tables() -> Result<(), ShellError> {
-        let group_by = String::from("date");
+        let group_by = String::from("date").tagged(Tag::unknown());
 
         assert_eq!(
             t_sort(