feat: implement hybrid sort

lua/blink/cmp/config/fuzzy.lua

@@ -2,7 +2,7 @@
 --- @field use_typo_resistance boolean When enabled, allows for a number of typos relative to the length of the query. Disabling this matches the behavior of fzf
 --- @field use_frecency boolean Tracks the most recently/frequently used items and boosts the score of the item
 --- @field use_proximity boolean Boosts the score of items matching nearby words
---- @field sorts ("label" | "sort_text" | "kind" | "score" | blink.cmp.SortFunction)[] Controls which sorts to use and in which order, these three are currently the only allowed options
+--- @field sort blink.cmp.FuzzySortConfig
 --- @field prebuilt_binaries blink.cmp.PrebuiltBinariesConfig
 
 --- @class (exact) blink.cmp.PrebuiltBinariesConfig
@@ -11,16 +11,24 @@
 --- @field force_system_triple? string When downloading a prebuilt binary, force the downloader to use this system triple. If this is unset then the downloader will attempt to infer the system triple from `jit.os` and `jit.arch`. Check the latest release for all available system triples. WARN: Beware that `main` may be incompatible with the version you select
 --- @field extra_curl_args? string[] Extra arguments that will be passed to curl like { 'curl', ..extra_curl_args, ..built_in_args }
 
---- @alias blink.cmp.SortFunction fun(a: blink.cmp.CompletionItem, b: blink.cmp.CompletionItem): boolean | nil
+--- @alias blink.cmp.SortFunction fun(a: blink.cmp.CompletionItem, b: blink.cmp.CompletionItem): boolean | nil Fallbacks to the next sort function when returning nil
+--- @alias blink.cmp.SortFunctions ("label" | "sort_text" | "kind" | "score" | blink.cmp.SortFunction)[] Controls which sorts to use and in which order, falling back when the sort function returns nil
+
+--- @class blink.cmp.FuzzySortConfig
+--- @field strong_match blink.cmp.SortFunctions Controls which sorts to use and in which order, for strong matches (based on fuzzy match score)
+--- @field weak_match blink.cmp.SortFunctions Controls which sorts to use and in which order, for weak matches (based on fuzzy match score)
 
 local validate = require('blink.cmp.config.utils').validate
 local fuzzy = {
   --- @type blink.cmp.FuzzyConfig
   default = {
     use_typo_resistance = true,
-    use_frecency = true,
-    use_proximity = true,
-    sorts = { 'score', 'sort_text' },
+    use_frecency = false,
+    use_proximity = false,
+    sort = {
+      strong_match = { 'score', 'sort_text' },
+      weak_match = { 'sort_text', 'score' },
+    },
     prebuilt_binaries = {
       download = true,
       force_version = nil,
@@ -35,18 +43,7 @@ function fuzzy.validate(config)
     use_typo_resistance = { config.use_typo_resistance, 'boolean' },
     use_frecency = { config.use_frecency, 'boolean' },
     use_proximity = { config.use_proximity, 'boolean' },
-    sorts = {
-      config.sorts,
-      function(sorts)
-        for _, sort in ipairs(sorts) do
-          if not vim.tbl_contains({ 'label', 'sort_text', 'kind', 'score' }, sort) and type(sort) ~= 'function' then
-            return false
-          end
-        end
-        return true
-      end,
-      'one of: "label", "sort_text", "kind", "score" or a function',
-    },
+    sort = { config.sort, 'table' },
     prebuilt_binaries = { config.prebuilt_binaries, 'table' },
   }, config)
   validate('fuzzy.prebuilt_binaries', {
@@ -55,6 +52,28 @@ function fuzzy.validate(config)
     force_system_triple = { config.prebuilt_binaries.force_system_triple, { 'string', 'nil' } },
     extra_curl_args = { config.prebuilt_binaries.extra_curl_args, { 'table' } },
   }, config.prebuilt_binaries)
+
+  --- @param sorts blink.cmp.SortFunctions
+  local function validate_sort(sorts)
+    for _, sort in ipairs(sorts) do
+      if not vim.tbl_contains({ 'label', 'sort_text', 'kind', 'score' }, sort) and type(sort) ~= 'function' then
+        return false
+      end
+    end
+    return true
+  end
+  validate('fuzzy.sort', {
+    strong_match = {
+      config.sort.strong_match,
+      validate_sort,
+      'one of: "label", "sort_text", "kind", "score" or a function',
+    },
+    weak_match = {
+      config.sort.weak_match,
+      validate_sort,
+      'one of: "label", "sort_text", "kind", "score" or a function',
+    },
+  }, config.sort)
 end
 
 return fuzzy

lua/blink/cmp/fuzzy/init.lua

@@ -65,7 +65,6 @@ function fuzzy.fuzzy(needle, haystacks_by_provider)
       use_typo_resistance = config.fuzzy.use_typo_resistance,
       use_frecency = config.fuzzy.use_frecency and #needle > 0,
       use_proximity = config.fuzzy.use_proximity and #needle > 0,
-      sorts = config.fuzzy.sorts,
       nearby_words = nearby_words,
     })
 
@@ -76,7 +75,12 @@ function fuzzy.fuzzy(needle, haystacks_by_provider)
     end
   end
 
-  return require('blink.cmp.fuzzy.sort').sort(filtered_items, config.fuzzy.sorts)
+  return require('blink.cmp.fuzzy.sort').sort(
+    filtered_items,
+    6 * needle:len(),
+    config.fuzzy.sort.strong_match,
+    config.fuzzy.sort.weak_match
+  )
 end
 
 return fuzzy

lua/blink/cmp/fuzzy/sort.lua

@@ -1,9 +1,40 @@
 local sort = {}
 
+--- Similar to Zed, we split the list into two buckets, sort them separately and combine.
+--- By default, the strong matches will be sorted by score and then sort_text, while the weak
+--- matches will be sorted by sort_text and then score.
+--- https://github.com/zed-industries/zed/blob/f64fcedab/crates/editor/src/code_context_menus.rs#L553-L566
 --- @param list blink.cmp.CompletionItem[]
---- @param funcs ("label" | "sort_text" | "kind" | "score" | blink.cmp.SortFunction)[]
+--- @param score_threshold number
+--- @param strong_match_funcs blink.cmp.SortFunctions
+--- @param weak_match_funcs blink.cmp.SortFunctions
 --- @return blink.cmp.CompletionItem[]
-function sort.sort(list, funcs)
+function sort.sort(list, score_threshold, strong_match_funcs, weak_match_funcs)
+  local strong_matches, weak_matches = sort.partition_by_score(list, score_threshold)
+
+  sort.list(strong_matches, strong_match_funcs)
+  sort.list(weak_matches, weak_match_funcs)
+
+  return vim.list_extend(strong_matches, weak_matches)
+end
+
+function sort.partition_by_score(list, score_threshold)
+  local above = {}
+  local below = {}
+  for _, item in ipairs(list) do
+    if item.score >= score_threshold then
+      table.insert(above, item)
+    else
+      table.insert(below, item)
+    end
+  end
+  return above, below
+end
+
+--- @param list blink.cmp.CompletionItem[]
+--- @param funcs blink.cmp.SortFunctions
+--- @return blink.cmp.CompletionItem[]
+function sort.list(list, funcs)
   local sorting_funcs = vim.tbl_map(
     function(name_or_func) return type(name_or_func) == 'string' and sort[name_or_func] or name_or_func end,
     funcs