feat(fuzzy): add custom regex option

lua/blink/cmp/config/fuzzy.lua

@@ -4,6 +4,7 @@
 --- @field use_frecency boolean Tracks the most recently/frequently used items and boosts the score of the item. Note, this does not apply when using the Lua implementation.
 --- @field use_proximity boolean Boosts the score of items matching nearby words. Note, this does not apply when using the Lua implementation.
 --- @field use_unsafe_no_lock boolean UNSAFE!! When enabled, disables the lock and fsync when writing to the frecency database. This should only be used on unsupported platforms (i.e. alpine termux). Note, this does not apply when using the Lua implementation.
+--- @field custom_regex? string Override the default regex with a custom pattern for word matching.
 --- @field sorts blink.cmp.Sort[] Controls which sorts to use and in which order.
 --- @field prebuilt_binaries blink.cmp.PrebuiltBinariesConfig
 
@@ -37,6 +38,7 @@ local fuzzy = {
     use_frecency = true,
     use_proximity = true,
     use_unsafe_no_lock = false,
+    custom_regex = nil,
     sorts = { 'score', 'sort_text' },
     prebuilt_binaries = {
       download = true,
@@ -65,6 +67,7 @@ function fuzzy.validate(config)
     use_frecency = { config.use_frecency, 'boolean' },
     use_proximity = { config.use_proximity, 'boolean' },
     use_unsafe_no_lock = { config.use_unsafe_no_lock, 'boolean' },
+    custom_regex = { config.custom_regex, { 'string', 'nil' } },
     sorts = {
       config.sorts,
       function(sorts)

lua/blink/cmp/fuzzy/init.lua

@@ -20,7 +20,11 @@ end
 function fuzzy.init_db()
   if fuzzy.has_init_db then return end
 
-  fuzzy.implementation.init_db(vim.fn.stdpath('data') .. '/blink/cmp/fuzzy.db', config.use_unsafe_no_lock)
+  fuzzy.implementation.init_db(
+    vim.fn.stdpath('data') .. '/blink/cmp/fuzzy.db',
+    config.use_unsafe_no_lock,
+    config.fuzzy.custom_regex
+  )
 
   vim.api.nvim_create_autocmd('VimLeavePre', {
     callback = fuzzy.implementation.destroy_db,

lua/blink/cmp/fuzzy/lua/init.lua

@@ -1,22 +1,26 @@
+local config_fuzzy = require('blink.cmp.config').fuzzy
 local match = require('blink.cmp.fuzzy.lua.match')
 local match_indices = require('blink.cmp.fuzzy.lua.match_indices')
 local get_keyword_range = require('blink.cmp.fuzzy.lua.keyword').get_keyword_range
 local guess_keyword_range = require('blink.cmp.fuzzy.lua.keyword').guess_keyword_range
 
+local words_regex = vim.regex(
+  [[\%(-\?\d\+\%(\.\d\+\)\?\|\h\%(\w\|á\|Á\|é\|É\|í\|Í\|ó\|Ó\|ú\|Ú\)*\%(-\%(\w\|á\|Á\|é\|É\|í\|Í\|ó\|Ó\|ú\|Ú\)*\)*\)]]
+)
 --- @type blink.cmp.FuzzyImplementation
 --- @diagnostic disable-next-line: missing-fields
 local fuzzy = {
   --- @type table<string, blink.cmp.CompletionItem[]>
   provider_items = {},
 }
 
-function fuzzy.init_db() end
+function fuzzy.init_db()
+  if config_fuzzy.custom_regex then words_regex = vim.regex(config_fuzzy.custom_regex) end
+end
+
 function fuzzy.destroy_db() end
-function fuzzy.access() end
 
-local words_regex = vim.regex(
-  [[\%(-\?\d\+\%(\.\d\+\)\?\|\h\%(\w\|á\|Á\|é\|É\|í\|Í\|ó\|Ó\|ú\|Ú\)*\%(-\%(\w\|á\|Á\|é\|É\|í\|Í\|ó\|Ó\|ú\|Ú\)*\)*\)]]
-)
+function fuzzy.access() end
 
 --- Takes ~0.25ms for 1200 characters split over 40 lines
 function fuzzy.get_words(text)

lua/blink/cmp/fuzzy/rust/lib.rs

@@ -17,13 +17,24 @@ mod keyword;
 mod lsp_item;
 mod sort;
 
-static REGEX: LazyLock<Regex> =
-    LazyLock::new(|| Regex::new(r"[\p{L}_][\p{L}0-9_\\-]{2,}").unwrap());
+static REGEX: LazyLock<RwLock<Regex>> =
+    LazyLock::new(|| RwLock::new(Regex::new(r"[\p{L}0-9_\\-]{2,}").unwrap()));
 static FRECENCY: LazyLock<RwLock<Option<FrecencyTracker>>> = LazyLock::new(|| RwLock::new(None));
 static HAYSTACKS_BY_PROVIDER: LazyLock<RwLock<HashMap<String, Vec<LspItem>>>> =
     LazyLock::new(|| RwLock::new(HashMap::new()));
 
-pub fn init_db(_: &Lua, (db_path, use_unsafe_no_lock): (String, bool)) -> LuaResult<bool> {
+fn update_regex(new_pattern: &str) {
+    let mut regex = REGEX.write().unwrap();
+    *regex = Regex::new(new_pattern).unwrap();
+}
+
+pub fn init_db(
+    _: &Lua,
+    (db_path, use_unsafe_no_lock, custom_regex): (String, bool, Option<String>),
+) -> LuaResult<bool> {
+    if let Some(regex) = custom_regex {
+        update_regex(&regex);
+    }
     let mut frecency = FRECENCY.write().map_err(|_| Error::AcquireFrecencyLock)?;
     if frecency.is_some() {
         return Ok(false);
@@ -211,6 +222,8 @@ pub fn guess_edit_range(
 
 pub fn get_words(_: &Lua, text: mlua::String) -> LuaResult<Vec<String>> {
     Ok(REGEX
+        .read()
+        .unwrap()
         .find_iter(&text.to_string_lossy())
         .map(|m| m.as_str().to_string())
         .filter(|s| s.len() < 512)