add ability to rename column with --rename-col arg on oxen df

Author: gschoeni

src/cli/src/cmd/df.rs

@@ -219,6 +219,12 @@ impl RunCmd for DFCmd {
                 .help("Add a row and cast to the values data types to match the current schema. If used with --add-col, row is added first, then column. Format 'comma,separated,vals'")
                 .action(clap::ArgAction::Set),
         )
+        .arg(
+            Arg::new("rename-col")
+                .long("rename-col")
+                .help("Rename a column in the data frame. Format: 'old_name:new_name'")
+                .action(clap::ArgAction::Set),
+        )
         .arg(
             Arg::new("delete-row")
                 .long("delete-row")
@@ -286,6 +292,7 @@ impl DFCmd {
         liboxen::opts::DFOpts {
             add_col: args.get_one::<String>("add-col").map(String::from),
             add_row: args.get_one::<String>("add-row").map(String::from),
+            rename_col: args.get_one::<String>("rename-col").map(String::from),
             at: args
                 .get_one::<String>("at")
                 .map(|x| x.parse::<usize>().expect("at must be valid int")),

src/lib/src/core/df/tabular.rs

@@ -498,6 +498,21 @@ pub fn transform_lazy(mut df: LazyFrame, opts: DFOpts) -> Result<LazyFrame, Oxen
         }
     }
 
+    if let Some(names) = &opts.rename_col {
+        if names.contains(":") {
+            let parts = names.split(":").collect::<Vec<&str>>();
+            let old_name = parts[0];
+            let new_name = parts[1];
+            let mut mut_df = df
+                .collect()
+                .map_err(|e| OxenError::basic_str(format!("{e:?}")))?;
+            rename_col(&mut mut_df, old_name, new_name)?;
+            df = mut_df.lazy();
+        } else {
+            log::error!("Invalid rename_col format: {}", names);
+        }
+    }
+
     // These ops should be the last ops since they depends on order
     if let Some(indices) = opts.take_indices() {
         match take(df.clone(), indices) {
@@ -604,6 +619,19 @@ fn slice(df: LazyFrame, opts: &DFOpts) -> LazyFrame {
     }
 }
 
+fn rename_col(
+    df: &mut DataFrame,
+    old_name: impl AsRef<str>,
+    new_name: impl AsRef<str>,
+) -> Result<(), OxenError> {
+    let old_name = old_name.as_ref();
+    let new_name = new_name.as_ref();
+    log::debug!("Renaming column {:?} to {:?}", old_name, new_name);
+    df.rename(old_name, PlSmallStr::from_str(new_name))
+        .map_err(|e| OxenError::basic_str(format!("{e:?}")))?;
+    Ok(())
+}
+
 pub fn df_add_row_num(df: DataFrame) -> Result<DataFrame, OxenError> {
     df.with_row_index(PlSmallStr::from_str(constants::ROW_NUM_COL_NAME), Some(0))
         .map_err(|e| OxenError::basic_str(format!("{e:?}")))

src/lib/src/opts/df_opts.rs

@@ -31,6 +31,7 @@ pub struct IndexedItem {
 pub struct DFOpts {
     pub add_col: Option<String>,
     pub add_row: Option<String>,
+    pub rename_col: Option<String>,
     pub at: Option<usize>,
     pub columns: Option<String>,
     pub delete_row: Option<String>,
@@ -80,6 +81,7 @@ impl DFOpts {
         DFOpts {
             add_col: None,
             add_row: None,
+            rename_col: None,
             at: None,
             columns: None,
             delete_row: None,
@@ -159,6 +161,7 @@ impl DFOpts {
     pub fn has_transform(&self) -> bool {
         self.add_col.is_some()
             || self.add_row.is_some()
+            || self.rename_col.is_some()
             || self.item.is_some()
             || self.columns.is_some()
             || self.filter.is_some()

src/lib/src/repositories/workspaces.rs

@@ -259,10 +259,7 @@ pub fn list(repo: &LocalRepository) -> Result<Vec<Workspace>, OxenError> {
     let workspaces_hashes = util::fs::list_dirs_in_dir(&workspaces_dir)
         .map_err(|e| OxenError::basic_str(format!("Error listing workspace directories: {}", e)))?;
 
-    log::debug!(
-        "workspace::list got {} workspaces",
-        workspaces_hashes.len()
-    );
+    log::debug!("workspace::list got {} workspaces", workspaces_hashes.len());
 
     let mut workspaces = Vec::new();
     for workspace_hash in workspaces_hashes {