feat(tree-sitter): support file to language association through globs

* new field (file-globs) in the config allowing to match a language by globs (alternative to file-types)
* added tree-sitter ssh-client-config for the ~/.ssh/config file
* added the .git/config file to tree-sitter git-config

book/src/generated/lang-support.md

@@ -77,6 +77,7 @@
 | scala | ✓ |  | ✓ | `metals` |
 | solidity | ✓ |  |  | `solc` |
 | sql | ✓ |  |  |  |
+| sshclientconfig | ✓ |  |  |  |
 | svelte | ✓ |  | ✓ | `svelteserver` |
 | swift | ✓ |  |  | `sourcekit-lsp` |
 | tablegen | ✓ | ✓ | ✓ |  |

helix-core/Cargo.toml

@@ -43,5 +43,7 @@ chrono = { version = "0.4", default-features = false, features = ["alloc", "std"
 etcetera = "0.3"
 textwrap = "0.15.0"
 
+globset = "0.4"
+
 [dev-dependencies]
 quickcheck = { version = "1", default-features = false }

helix-core/src/syntax.rs

@@ -8,6 +8,7 @@ use crate::{
 };
 
 use arc_swap::{ArcSwap, Guard};
+use globset::{Glob, GlobSet, GlobSetBuilder};
 use slotmap::{DefaultKey as LayerId, HopSlotMap};
 
 use std::{
@@ -25,6 +26,16 @@ use serde::{Deserialize, Serialize};
 
 use helix_loader::grammar::{get_language, load_runtime_file};
 
+fn deserialize_glob_set<'de, D: serde::Deserializer<'de>>(
+    deserializer: D,
+) -> Result<GlobSet, D::Error> {
+    let globs = Vec::<String>::deserialize(deserializer)?;
+    let mut builder = GlobSetBuilder::new();
+    for glob in globs {
+        builder.add(Glob::new(&glob).map_err(serde::de::Error::custom)?);
+    }
+    builder.build().map_err(serde::de::Error::custom)
+}
 fn deserialize_regex<'de, D>(deserializer: D) -> Result<Option<Regex>, D::Error>
 where
     D: serde::Deserializer<'de>,
@@ -61,11 +72,14 @@ pub struct Configuration {
 pub struct LanguageConfiguration {
     #[serde(rename = "name")]
     pub language_id: String, // c-sharp, rust
-    pub scope: String,           // source.rust
+    pub scope: String, // source.rust
+    #[serde(default)]
     pub file_types: Vec<String>, // filename ends_with? <Gemfile, rb, etc>
+    #[serde(default, skip_serializing, deserialize_with = "deserialize_glob_set")]
+    pub file_globs: GlobSet,
     #[serde(default)]
     pub shebangs: Vec<String>, // interpreter(s) associated with language
-    pub roots: Vec<String>,      // these indicate project roots <.git, Cargo.toml>
+    pub roots: Vec<String>, // these indicate project roots <.git, Cargo.toml>
     pub comment_token: Option<String>,
     pub max_line_length: Option<usize>,
 
@@ -436,6 +450,7 @@ pub struct Loader {
     // highlight_names ?
     language_configs: Vec<Arc<LanguageConfiguration>>,
     language_config_ids_by_file_type: HashMap<String, usize>, // Vec<usize>
+    language_config_ids_by_globset: Vec<(GlobSet, usize)>,
     language_config_ids_by_shebang: HashMap<String, usize>,
 
     scopes: ArcSwap<Vec<String>>,
@@ -447,6 +462,7 @@ impl Loader {
             language_configs: Vec::new(),
             language_config_ids_by_file_type: HashMap::new(),
             language_config_ids_by_shebang: HashMap::new(),
+            language_config_ids_by_globset: Vec::new(),
             scopes: ArcSwap::from_pointee(Vec::new()),
         };
 
@@ -465,6 +481,9 @@ impl Loader {
                     .language_config_ids_by_shebang
                     .insert(shebang.clone(), language_id);
             }
+            loader
+                .language_config_ids_by_globset
+                .push((config.file_globs.clone(), language_id));
 
             loader.language_configs.push(Arc::new(config));
         }
@@ -478,16 +497,27 @@ impl Loader {
         let configuration_id = path
             .file_name()
             .and_then(|n| n.to_str())
-            .and_then(|file_name| self.language_config_ids_by_file_type.get(file_name))
+            .and_then(|file_name| {
+                self.language_config_ids_by_file_type
+                    .get(file_name)
+                    .copied()
+            })
             .or_else(|| {
                 path.extension()
                     .and_then(|extension| extension.to_str())
-                    .and_then(|extension| self.language_config_ids_by_file_type.get(extension))
-            });
-
-        configuration_id.and_then(|&id| self.language_configs.get(id).cloned())
+                    .and_then(|extension| {
+                        self.language_config_ids_by_file_type
+                            .get(extension)
+                            .copied()
+                    })
+            })
+            .or_else(|| {
+                self.language_config_ids_by_globset
+                    .iter()
+                    .find_map(|(gs, id)| if gs.is_match(path) { Some(*id) } else { None })
+            })?;
 
-        // TODO: content_regex handling conflict resolution
+        self.language_configs.get(configuration_id).cloned()
     }
 
     pub fn language_config_for_shebang(&self, source: &Rope) -> Option<Arc<LanguageConfiguration>> {

languages.toml

@@ -946,8 +946,8 @@ source = { git = "https://github.com/tree-sitter/tree-sitter-regex", rev = "e1cf
 name = "git-config"
 scope = "source.gitconfig"
 roots = []
-# TODO: allow specifying file-types as a regex so we can read directory names (e.g. `.git/config`)
 file-types = [".gitmodules", ".gitconfig"]
+file-globs = ["*.git/config"]
 injection-regex = "git-config"
 comment-token = "#"
 indent = { tab-width = 4, unit = "\t" }
@@ -1342,3 +1342,14 @@ indent = { tab-width = 4, unit = "  " }
 [[grammar]]
 name = "odin"
 source = { git = "https://github.com/MineBill/tree-sitter-odin", rev = "da885f4a387f169b9b69fe0968259ee257a8f69a" }
+
+[[language]]
+name = "sshclientconfig"
+scope = "source.sshclientconfig"
+file-globs = ["*.ssh/config"] 
+roots = []
+
+[[grammar]]
+name = "sshclientconfig"
+source = { git = "https://github.com/metio/tree-sitter-ssh-client-config", rev = "769d7a01a2e5493b4bb5a51096c6bf4be130b024" }
+