override nested arrays when merging TOML (#2145)

We merge the elements of arrays for the top-level array. For
`languages.toml`, this is the array of languages. For any nested
arrays, we simply take the `right` array as-is instead of using
the union of `left` and `right`.

closes #1000

helix-loader/src/config.rs

@@ -19,7 +19,7 @@ pub fn user_lang_config() -> Result<toml::Value, toml::de::Error> {
         .into_iter()
         .chain([default_lang_config()].into_iter())
         .fold(toml::Value::Table(toml::value::Table::default()), |a, b| {
-            crate::merge_toml_values(b, a)
+            crate::merge_toml_values(b, a, false)
         });
 
     Ok(config)

helix-loader/src/lib.rs

@@ -98,8 +98,24 @@ pub fn find_root_impl(root: Option<&str>, root_markers: &[String]) -> Vec<std::p
     directories
 }
 
-// right overrides left
+/// Merge two TOML documents, merging values from `right` onto `left`
-pub fn merge_toml_values(left: toml::Value, right: toml::Value) -> toml::Value {
+///
+/// When an array exists in both `left` and `right`, `right`'s array is
+/// used. When a table exists in both `left` and `right`, the merged table
+/// consists of all keys in `left`'s table unioned with all keys in `right`
+/// with the values of `right` being merged recursively onto values of
+/// `left`.
+///
+/// `merge_toplevel_arrays` controls whether a top-level array in the TOML
+/// document is merged instead of overridden. This is useful for TOML
+/// documents that use a top-level array of values like the `languages.toml`,
+/// where one usually wants to override or add to the array instead of
+/// replacing it altogether.
+pub fn merge_toml_values(
+    left: toml::Value,
+    right: toml::Value,
+    merge_toplevel_arrays: bool,
+) -> toml::Value {
     use toml::Value;
 
     fn get_name(v: &Value) -> Option<&str> {
@@ -108,24 +124,35 @@ pub fn merge_toml_values(left: toml::Value, right: toml::Value) -> toml::Value {
 
     match (left, right) {
         (Value::Array(mut left_items), Value::Array(right_items)) => {
-            left_items.reserve(right_items.len());
+            // The top-level arrays should be merged but nested arrays should
-            for rvalue in right_items {
+            // act as overrides. For the `languages.toml` config, this means
-                let lvalue = get_name(&rvalue)
+            // that you can specify a sub-set of languages in an overriding
-                    .and_then(|rname| left_items.iter().position(|v| get_name(v) == Some(rname)))
+            // `languages.toml` but that nested arrays like Language Server
-                    .map(|lpos| left_items.remove(lpos));
+            // arguments are replaced instead of merged.
-                let mvalue = match lvalue {
+            if merge_toplevel_arrays {
-                    Some(lvalue) => merge_toml_values(lvalue, rvalue),
+                left_items.reserve(right_items.len());
-                    None => rvalue,
+                for rvalue in right_items {
-                };
+                    let lvalue = get_name(&rvalue)
-                left_items.push(mvalue);
+                        .and_then(|rname| {
+                            left_items.iter().position(|v| get_name(v) == Some(rname))
+                        })
+                        .map(|lpos| left_items.remove(lpos));
+                    let mvalue = match lvalue {
+                        Some(lvalue) => merge_toml_values(lvalue, rvalue, false),
+                        None => rvalue,
+                    };
+                    left_items.push(mvalue);
+                }
+                Value::Array(left_items)
+            } else {
+                Value::Array(right_items)
             }
-            Value::Array(left_items)
         }
         (Value::Table(mut left_map), Value::Table(right_map)) => {
             for (rname, rvalue) in right_map {
                 match left_map.remove(&rname) {
                     Some(lvalue) => {
-                        let merged_value = merge_toml_values(lvalue, rvalue);
+                        let merged_value = merge_toml_values(lvalue, rvalue, merge_toplevel_arrays);
                         left_map.insert(rname, merged_value);
                     }
                     None => {
@@ -143,23 +170,22 @@ pub fn merge_toml_values(left: toml::Value, right: toml::Value) -> toml::Value {
 #[cfg(test)]
 mod merge_toml_tests {
     use super::merge_toml_values;
+    use toml::Value;
 
     #[test]
-    fn language_tomls() {
+    fn language_toml_map_merges() {
-        use toml::Value;
+        const USER: &str = r#"
-
-        const USER: &str = "
         [[language]]
-        name = \"nix\"
+        name = "nix"
-        test = \"bbb\"
+        test = "bbb"
-        indent = { tab-width = 4, unit = \"    \", test = \"aaa\" }
+        indent = { tab-width = 4, unit = "    ", test = "aaa" }
-        ";
+        "#;
 
         let base: Value = toml::from_slice(include_bytes!("../../languages.toml"))
             .expect("Couldn't parse built-in languages config");
         let user: Value = toml::from_str(USER).unwrap();
 
-        let merged = merge_toml_values(base, user);
+        let merged = merge_toml_values(base, user, true);
         let languages = merged.get("language").unwrap().as_array().unwrap();
         let nix = languages
             .iter()
@@ -179,4 +205,33 @@ mod merge_toml_tests {
         // We didn't change comment-token so it should be same
         assert_eq!(nix.get("comment-token").unwrap().as_str().unwrap(), "#");
     }
+
+    #[test]
+    fn language_toml_nested_array_merges() {
+        const USER: &str = r#"
+        [[language]]
+        name = "typescript"
+        language-server = { command = "deno", args = ["lsp"] }
+        "#;
+
+        let base: Value = toml::from_slice(include_bytes!("../../languages.toml"))
+            .expect("Couldn't parse built-in languages config");
+        let user: Value = toml::from_str(USER).unwrap();
+
+        let merged = merge_toml_values(base, user, true);
+        let languages = merged.get("language").unwrap().as_array().unwrap();
+        let ts = languages
+            .iter()
+            .find(|v| v.get("name").unwrap().as_str().unwrap() == "typescript")
+            .unwrap();
+        assert_eq!(
+            ts.get("language-server")
+                .unwrap()
+                .get("args")
+                .unwrap()
+                .as_array()
+                .unwrap(),
+            &vec![Value::String("lsp".into())]
+        )
+    }
 }