feat(nvim): Deduplicate tool results using chunk_id (wip)

Author: Zhe Yu

lua/vectorcode/integrations/codecompanion/common.lua

@@ -1,3 +1,5 @@
+---@module "codecompanion"
+
 local job_runner
 local vc_config = require("vectorcode.config")
 local notify_opts = vc_config.notify_opts
@@ -15,8 +17,10 @@ local default_options = {
   chunk_mode = false,
 }
 
+local TOOL_RESULT_SOURCE = "VectorCodeToolResult"
+
 return {
-  tool_result_source = "VectorCodeToolResult",
+  tool_result_source = TOOL_RESULT_SOURCE,
   ---@param t table|string
   ---@return string
   flatten_table_to_string = function(t)
@@ -88,6 +92,7 @@ return {
     end
     return llm_message
   end,
+
   ---@param use_lsp boolean
   ---@return VectorCode.JobRunner
   initialise_runner = function(use_lsp)
@@ -111,4 +116,53 @@ return {
     end
     return job_runner
   end,
+
+  ---@param results VectorCode.Result[]
+  ---@param chat CodeCompanion.Chat
+  ---@return VectorCode.Result[]
+  filter_results = function(results, chat)
+    local existing_refs = chat.refs
+    if existing_refs == nil then
+      return results
+    end
+    existing_refs = vim
+      .iter(existing_refs)
+      :filter(
+        ---@param ref CodeCompanion.Chat.Ref
+        function(ref)
+          return ref.source == TOOL_RESULT_SOURCE or ref.path or ref.bufnr
+        end
+      )
+      :map(
+        ---@param ref CodeCompanion.Chat.Ref
+        function(ref)
+          if ref.source == TOOL_RESULT_SOURCE then
+            return ref.id
+          elseif ref.path then
+            return ref.path
+          elseif ref.bufnr then
+            return vim.api.nvim_buf_get_name(ref.bufnr)
+          end
+        end
+      )
+      :totable()
+
+    return vim
+      .iter(results)
+      :filter(
+        ---@param res VectorCode.Result
+        function(res)
+          -- return true if res is not in refs
+          if res.chunk then
+            if res.chunk_id == nil then
+              return true
+            end
+            return not vim.tbl_contains(existing_refs, res.chunk_id)
+          else
+            return not vim.tbl_contains(existing_refs, res.path)
+          end
+        end
+      )
+      :totable()
+  end,
 }

lua/vectorcode/integrations/codecompanion/func_calling_tool.lua

@@ -273,6 +273,8 @@ return check_cli_wrap(function(opts)
         )
         local user_message
         if cmd.command == "query" then
+          ---@cast stdout +VectorCode.Result[]
+          stdout = cc_common.filter_results(stdout, agent.chat)
           local max_result = #stdout
           if opts.max_num > 0 then
             max_result = math.min(opts.max_num or 1, max_result)
@@ -297,15 +299,17 @@ return check_cli_wrap(function(opts)
                 cc_common.process_result(file),
                 user_message
               )
-              if not opts.chunk_mode then
+
+              if (not opts.chunk_mode) or file.chunk_id ~= nil then
                 -- skip referencing because there will be multiple chunks with the same path (id).
                 -- TODO: figure out a way to deduplicate.
-                agent.chat.references:add({
+                local ref = {
                   source = cc_common.tool_result_source,
-                  id = file.path,
+                  id = file.chunk_id or file.path,
                   path = file.path,
                   opts = { visible = false },
-                })
+                }
+                agent.chat.references:add(ref)
               end
             end
           end

lua/vectorcode/types.lua

@@ -5,6 +5,7 @@
 ---@field chunk string?
 ---@field start_line integer?
 ---@field end_line integer?
+---@field chunk_id string?
 
 ---Type definitions for the cache of a buffer.
 ---@class VectorCode.Cache