add workspace config and manual LSP root management

fixup documentation

Co-authored-by: LeoniePhiline <22329650+LeoniePhiline@users.noreply.github.com>

fixup typo

Co-authored-by: LeoniePhiline <22329650+LeoniePhiline@users.noreply.github.com>

book/src/configuration.md

@@ -30,6 +30,9 @@ You can use a custom configuration file by specifying it with the `-c` or
 Additionally, you can reload the configuration file by sending the USR1
 signal to the Helix process on Unix operating systems, such as by using the command `pkill -USR1 hx`.
 
+Finally, you can have a `config.toml` local to a project by putting it under a `.helix` directory in your repository.
+Its settings will be merged with the configuration directory `config.toml` and the built-in configuration.
+
 ## Editor
 
 ### `[editor]` Section
@@ -58,6 +61,7 @@ signal to the Helix process on Unix operating systems, such as by using the comm
 | `bufferline` | Renders a line at the top of the editor displaying open buffers. Can be `always`, `never` or `multiple` (only shown if more than one buffer is in use) | `never` |
 | `color-modes` | Whether to color the mode indicator with different colors depending on the mode itself | `false` |
 | `text-width` | Maximum line length. Used for the `:reflow` command and soft-wrapping if `soft-wrap.wrap_at_text_width` is set | `80` |
+| `workspace-lsp-roots` | Directories relative to the workspace root that are treated as LSP roots. Should only be set in `.helix/config.toml` | `[]` |
 
 ### `[editor.statusline]` Section
 

book/src/generated/typable-cmd.md

@@ -70,6 +70,7 @@
 | `:tree-sitter-subtree`, `:ts-subtree` | Display tree sitter subtree under cursor, primarily for debugging queries. |
 | `:config-reload` | Refresh user config. |
 | `:config-open` | Open the user config.toml file. |
+| `:config-open-workspace` | Open the workspace config.toml file. |
 | `:log-open` | Open the helix log file. |
 | `:insert-output` | Run shell command, inserting output before each selection. |
 | `:append-output` | Run shell command, appending output after each selection. |

book/src/languages.md

@@ -64,6 +64,7 @@ These configuration keys are available:
 | `grammar`             | The tree-sitter grammar to use (defaults to the value of `name`) |
 | `formatter`           | The formatter for the language, it will take precedence over the lsp when defined. The formatter must be able to take the original file as input from stdin and write the formatted file to stdout |
 | `text-width`          |  Maximum line length. Used for the `:reflow` command and soft-wrapping if `soft-wrap.wrap_at_text_width` is set, defaults to `editor.text-width`   |
+| `workspace-lsp-roots`     | Directories relative to the workspace root that are treated as LSP roots. Should only be set in `.helix/config.toml`. Overwrites the setting of the same name in `config.toml` if set. | `` |
 
 ### File-type detection and the `file-types` key
 

helix-core/src/lib.rs

@@ -36,55 +36,12 @@ pub mod unicode {
     pub use unicode_width as width;
 }
 
+pub use helix_loader::find_workspace;
+
 pub fn find_first_non_whitespace_char(line: RopeSlice) -> Option<usize> {
     line.chars().position(|ch| !ch.is_whitespace())
 }
 
-/// Find project root.
-///
-/// Order of detection:
-/// * Top-most folder containing a root marker in current git repository
-/// * Git repository root if no marker detected
-/// * Top-most folder containing a root marker if not git repository detected
-/// * Current working directory as fallback
-pub fn find_root(root: Option<&str>, root_markers: &[String]) -> std::path::PathBuf {
-    let current_dir = std::env::current_dir().expect("unable to determine current directory");
-
-    let root = match root {
-        Some(root) => {
-            let root = std::path::Path::new(root);
-            if root.is_absolute() {
-                root.to_path_buf()
-            } else {
-                current_dir.join(root)
-            }
-        }
-        None => current_dir.clone(),
-    };
-
-    let mut top_marker = None;
-    for ancestor in root.ancestors() {
-        if root_markers
-            .iter()
-            .any(|marker| ancestor.join(marker).exists())
-        {
-            top_marker = Some(ancestor);
-        }
-
-        if ancestor.join(".git").exists() {
-            // Top marker is repo root if not root marker was detected yet
-            if top_marker.is_none() {
-                top_marker = Some(ancestor);
-            }
-            // Don't go higher than repo if we're in one
-            break;
-        }
-    }
-
-    // Return the found top marker or the current_dir as fallback
-    top_marker.map_or(current_dir, |a| a.to_path_buf())
-}
-
 pub use ropey::{self, str_utils, Rope, RopeBuilder, RopeSlice};
 
 // pub use tendril::StrTendril as Tendril;

helix-core/src/syntax.rs

@@ -20,7 +20,7 @@ use std::{
     fmt,
     hash::{Hash, Hasher},
     mem::{replace, transmute},
-    path::Path,
+    path::{Path, PathBuf},
     str::FromStr,
     sync::Arc,
 };
@@ -127,6 +127,10 @@ pub struct LanguageConfiguration {
     pub auto_pairs: Option<AutoPairs>,
 
     pub rulers: Option<Vec<u16>>, // if set, override editor's rulers
+
+    /// Hardcoded LSP root directories relative to the workspace root, like `examples` or `tools/fuzz`.
+    /// Falling back to the current working directory if none are configured.
+    pub workspace_lsp_roots: Option<Vec<PathBuf>>,
 }
 
 #[derive(Debug, PartialEq, Eq, Hash)]

helix-loader/src/config.rs

@@ -9,9 +9,8 @@ pub fn default_lang_config() -> toml::Value {
 
 /// User configured languages.toml file, merged with the default config.
 pub fn user_lang_config() -> Result<toml::Value, toml::de::Error> {
-    let config = crate::local_config_dirs()
+    let config = [crate::config_dir(), crate::find_workspace().join(".helix")]
         .into_iter()
-        .chain([crate::config_dir()].into_iter())
         .map(|path| path.join("languages.toml"))
         .filter_map(|file| {
             std::fs::read_to_string(file)
@@ -20,8 +19,7 @@ pub fn user_lang_config() -> Result<toml::Value, toml::de::Error> {
         })
         .collect::<Result<Vec<_>, _>>()?
         .into_iter()
-        .chain([default_lang_config()].into_iter())
-        .fold(toml::Value::Table(toml::value::Table::default()), |a, b| {
+        .fold(default_lang_config(), |a, b| {
             // combines for example
             // b:
             //   [[language]]
@@ -38,7 +36,7 @@ pub fn user_lang_config() -> Result<toml::Value, toml::de::Error> {
             //   language-server = { command = "/usr/bin/taplo" }
             //
             // thus it overrides the third depth-level of b with values of a if they exist, but otherwise merges their values
-            crate::merge_toml_values(b, a, 3)
+            crate::merge_toml_values(a, b, 3)
         });
 
     Ok(config)

helix-loader/src/lib.rs

@@ -42,7 +42,7 @@ fn prioritize_runtime_dirs() -> Vec<PathBuf> {
     let mut rt_dirs = Vec::new();
     if let Ok(dir) = std::env::var("CARGO_MANIFEST_DIR") {
         // this is the directory of the crate being run by cargo, we need the workspace path so we take the parent
-        let path = std::path::PathBuf::from(dir).parent().unwrap().join(RT_DIR);
+        let path = PathBuf::from(dir).parent().unwrap().join(RT_DIR);
         log::debug!("runtime dir: {}", path.to_string_lossy());
         rt_dirs.push(path);
     }
@@ -113,15 +113,6 @@ pub fn config_dir() -> PathBuf {
     path
 }
 
-pub fn local_config_dirs() -> Vec<PathBuf> {
-    let directories = find_local_config_dirs()
-        .into_iter()
-        .map(|path| path.join(".helix"))
-        .collect();
-    log::debug!("Located configuration folders: {:?}", directories);
-    directories
-}
-
 pub fn cache_dir() -> PathBuf {
     // TODO: allow env var override
     let strategy = choose_base_strategy().expect("Unable to find the config directory!");
@@ -137,6 +128,10 @@ pub fn config_file() -> PathBuf {
         .unwrap_or_else(|| config_dir().join("config.toml"))
 }
 
+pub fn workspace_config_file() -> PathBuf {
+    find_workspace().join(".helix").join("config.toml")
+}
+
 pub fn lang_config_file() -> PathBuf {
     config_dir().join("languages.toml")
 }
@@ -145,22 +140,6 @@ pub fn log_file() -> PathBuf {
     cache_dir().join("helix.log")
 }
 
-pub fn find_local_config_dirs() -> Vec<PathBuf> {
-    let current_dir = std::env::current_dir().expect("unable to determine current directory");
-    let mut directories = Vec::new();
-
-    for ancestor in current_dir.ancestors() {
-        if ancestor.join(".git").exists() {
-            directories.push(ancestor.to_path_buf());
-            // Don't go higher than repo if we're in one
-            break;
-        } else if ancestor.join(".helix").is_dir() {
-            directories.push(ancestor.to_path_buf());
-        }
-    }
-    directories
-}
-
 /// Merge two TOML documents, merging values from `right` onto `left`
 ///
 /// When an array exists in both `left` and `right`, `right`'s array is
@@ -302,3 +281,16 @@ mod merge_toml_tests {
         )
     }
 }
+
+/// Finds the current workspace folder.
+/// Used as a ceiling dir for root resolve, for the filepicker and other related
+pub fn find_workspace() -> PathBuf {
+    let current_dir = std::env::current_dir().expect("unable to determine current directory");
+    for ancestor in current_dir.ancestors() {
+        if ancestor.join(".git").exists() || ancestor.join(".helix").exists() {
+            return ancestor.to_owned();
+        }
+    }
+
+    current_dir
+}

helix-lsp/src/client.rs

@@ -1,22 +1,22 @@
 use crate::{
-    jsonrpc,
+    find_root, jsonrpc,
     transport::{Payload, Transport},
     Call, Error, OffsetEncoding, Result,
 };
 
-use helix_core::{find_root, ChangeSet, Rope};
+use helix_core::{ChangeSet, Rope};
 use helix_loader::{self, VERSION_AND_GIT_HASH};
 use lsp::PositionEncodingKind;
 use lsp_types as lsp;
 use serde::Deserialize;
 use serde_json::Value;
-use std::collections::HashMap;
 use std::future::Future;
 use std::process::Stdio;
 use std::sync::{
     atomic::{AtomicU64, Ordering},
     Arc,
 };
+use std::{collections::HashMap, path::PathBuf};
 use tokio::{
     io::{BufReader, BufWriter},
     process::{Child, Command},
@@ -49,6 +49,7 @@ impl Client {
         config: Option<Value>,
         server_environment: HashMap<String, String>,
         root_markers: &[String],
+        manual_roots: &[PathBuf],
         id: usize,
         req_timeout: u64,
         doc_path: Option<&std::path::PathBuf>,
@@ -77,8 +78,11 @@ impl Client {
             Transport::start(reader, writer, stderr, id);
 
         let root_path = find_root(
-            doc_path.and_then(|x| x.parent().and_then(|x| x.to_str())),
+            doc_path
+                .and_then(|x| x.parent().and_then(|x| x.to_str()))
+                .unwrap_or("."),
             root_markers,
+            manual_roots,
         );
 
         let root_uri = lsp::Url::from_file_path(root_path.clone()).ok();

helix-lsp/src/lib.rs

@@ -10,11 +10,15 @@ pub use lsp::{Position, Url};
 pub use lsp_types as lsp;
 
 use futures_util::stream::select_all::SelectAll;
-use helix_core::syntax::{LanguageConfiguration, LanguageServerConfiguration};
+use helix_core::{
+    find_workspace,
+    syntax::{LanguageConfiguration, LanguageServerConfiguration},
+};
 use tokio::sync::mpsc::UnboundedReceiver;
 
 use std::{
     collections::{hash_map::Entry, HashMap},
+    path::PathBuf,
     sync::{
         atomic::{AtomicUsize, Ordering},
         Arc,
@@ -641,6 +645,7 @@ impl Registry {
         &mut self,
         language_config: &LanguageConfiguration,
         doc_path: Option<&std::path::PathBuf>,
+        root_dirs: &[PathBuf],
     ) -> Result<Option<Arc<Client>>> {
         let config = match &language_config.language_server {
             Some(config) => config,
@@ -656,7 +661,7 @@ impl Registry {
                 let id = self.counter.fetch_add(1, Ordering::Relaxed);
 
                 let NewClientResult(client, incoming) =
-                    start_client(id, language_config, config, doc_path)?;
+                    start_client(id, language_config, config, doc_path, root_dirs)?;
                 self.incoming.push(UnboundedReceiverStream::new(incoming));
 
                 let (_, old_client) = entry.insert((id, client.clone()));
@@ -684,6 +689,7 @@ impl Registry {
         &mut self,
         language_config: &LanguageConfiguration,
         doc_path: Option<&std::path::PathBuf>,
+        root_dirs: &[PathBuf],
     ) -> Result<Option<Arc<Client>>> {
         let config = match &language_config.language_server {
             Some(config) => config,
@@ -697,7 +703,7 @@ impl Registry {
                 let id = self.counter.fetch_add(1, Ordering::Relaxed);
 
                 let NewClientResult(client, incoming) =
-                    start_client(id, language_config, config, doc_path)?;
+                    start_client(id, language_config, config, doc_path, root_dirs)?;
                 self.incoming.push(UnboundedReceiverStream::new(incoming));
 
                 entry.insert((id, client.clone()));
@@ -798,13 +804,15 @@ fn start_client(
     config: &LanguageConfiguration,
     ls_config: &LanguageServerConfiguration,
     doc_path: Option<&std::path::PathBuf>,
+    root_dirs: &[PathBuf],
 ) -> Result<NewClientResult> {
     let (client, incoming, initialize_notify) = Client::start(
         &ls_config.command,
         &ls_config.args,
         config.config.clone(),
         ls_config.environment.clone(),
         &config.roots,
+        config.workspace_lsp_roots.as_deref().unwrap_or(root_dirs),
         id,
         ls_config.timeout,
         doc_path,
@@ -842,6 +850,48 @@ fn start_client(
     Ok(NewClientResult(client, incoming))
 }
 
+/// Find an LSP root of a file using the following mechansim:
+/// * start at `file` (either an absolute path or relative to CWD)
+/// * find the top most directory containing a root_marker
+/// * inside the current workspace
+///     * stop the search at the first root_dir that contains `file` or the workspace (obtained from `helix_core::find_workspace`)
+///     * root_dirs only apply inside the workspace. For files outside of the workspace they are ignored
+/// * outside the current workspace: keep searching to the top of the file hiearchy
+pub fn find_root(file: &str, root_markers: &[String], root_dirs: &[PathBuf]) -> PathBuf {
+    let file = std::path::Path::new(file);
+    let workspace = find_workspace();
+    let file = if file.is_absolute() {
+        file.to_path_buf()
+    } else {
+        let current_dir = std::env::current_dir().expect("unable to determine current directory");
+        current_dir.join(file)
+    };
+
+    let inside_workspace = file.strip_prefix(&workspace).is_ok();
+
+    let mut top_marker = None;
+    for ancestor in file.ancestors() {
+        if root_markers
+            .iter()
+            .any(|marker| ancestor.join(marker).exists())
+        {
+            top_marker = Some(ancestor);
+        }
+
+        if inside_workspace
+            && (ancestor == workspace
+                || root_dirs
+                    .iter()
+                    .any(|root_dir| root_dir == ancestor.strip_prefix(&workspace).unwrap()))
+        {
+            return top_marker.unwrap_or(ancestor).to_owned();
+        }
+    }
+
+    // If no root was found use the workspace as a fallback
+    workspace
+}
+
 #[cfg(test)]
 mod tests {
     use super::{lsp, util::*, OffsetEncoding};