refactor(reference_picker): simplify and harden patterns

Author: sudo-tee

lua/opencode/ui/formatter.lua

@@ -500,14 +500,14 @@ end
 ---@param metadata FileToolMetadata Metadata for the tool use
 function M._format_file_tool(output, tool_type, input, metadata)
   local file_name = ''
-  if input then
+  if input and input.filePath then
     local cwd = vim.fn.getcwd()
     local absolute = vim.fn.fnamemodify(input.filePath, ':p')
-    
+
     if vim.startswith(absolute, cwd .. '/') then
       file_name = absolute:sub(#cwd + 2)
     else
-      file_name = vim.fn.fnamemodify(input.filePath, ':t')
+      file_name = absolute
     end
   end
 

lua/opencode/ui/reference_picker.lua

@@ -6,12 +6,18 @@ local icons = require('opencode.ui.icons')
 
 local M = {}
 
----Check if a file exists
----@param file_path string
+---Check if a file reference is valid
+---@param path string File path
+---@param context string Surrounding text
 ---@return boolean
-local function file_exists(file_path)
-  local path = file_path
-  return vim.fn.filereadable(path) == 1
+local function is_valid_file_reference(path, context)
+  -- Reject URLs
+  if context:lower():match('https?://') or context:lower():match('www%.') then
+    return false
+  end
+
+  -- Must have extension and exist
+  return path:match('%.[%w]+$') and vim.fn.filereadable(path) == 1
 end
 
 ---@class CodeReference
@@ -79,7 +85,7 @@ local function parse_position_info(line_str, col_or_end_str, end_line_str)
   return line, column, end_line
 end
 
----Generic function to parse file references using a given pattern
+---Parse file references using a pattern
 ---@param text string The text to parse
 ---@param pattern string Lua pattern to match
 ---@param message_id string The message ID for tracking
@@ -94,9 +100,11 @@ local function parse_references_with_pattern(text, pattern, message_id)
       break
     end
 
-    if file_exists(path) then
-      local line, column, end_line = parse_position_info(line_str, col_or_end_str, end_line_str)
+    local context_start = math.max(1, match_start - 30)
+    local context = text:sub(context_start, match_end + 10)
 
+    if is_valid_file_reference(path, context) then
+      local line, column, end_line = parse_position_info(line_str, col_or_end_str, end_line_str)
       local ref = create_code_reference(path, line, column, end_line, message_id, match_start, match_end)
       table.insert(references, ref)
     end
@@ -107,65 +115,34 @@ local function parse_references_with_pattern(text, pattern, message_id)
   return references
 end
 
----Parse backtick-wrapped file references like `path/file.lua:42`
----@param text string
----@param message_id string
----@return CodeReference[]
-local function parse_backtick_refs(text, message_id)
-  local pattern = '`([^`]+%.[%w]+):?(%d*):?(%d*)-?(%d*)`'
-  return parse_references_with_pattern(text, pattern, message_id)
-end
-
----Parse file:// URI references (backward compatibility)
----@param text string
----@param message_id string
----@return CodeReference[]
-local function parse_file_uri_refs(text, message_id)
-  local pattern = 'file://([%w_./%-]+):?(%d*):?(%d*)-?(%d*)'
-  return parse_references_with_pattern(text, pattern, message_id)
-end
-
----Parse plain file paths like path/file.lua:42 or path/file.lua
----@param text string
----@param message_id string
----@return CodeReference[]
-local function parse_plain_path_refs(text, message_id)
-  -- Match file paths with optional extension and line/column info
-  -- Relies on file_exists() to filter out false positives like URLs
-  local pattern = '([%w_%./-]+%.[%w]+):?(%d*):?(%d*)-?(%d*)'
-  return parse_references_with_pattern(text, pattern, message_id)
-end
-
 ---Parse file references from text using multiple pattern strategies
 ---@param text string The text to parse
 ---@param message_id string The message ID for tracking
 ---@return CodeReference[]
 function M.parse_references(text, message_id)
   local all_refs = {}
 
-  local backtick_refs = parse_backtick_refs(text, message_id)
-  local file_uri_refs = parse_file_uri_refs(text, message_id)
-  local plain_refs = parse_plain_path_refs(text, message_id)
+  local patterns = {
+    '`([^`\n]+%.%w+):?(%d*):?(%d*)-?(%d*)`', -- Backticks: `file.ext:line`
+    'file://([%S]+%.%w+):?(%d*):?(%d*)-?(%d*)', -- file:// URIs
+    '([%w_./%-]+/[%w_./%-]*%.%w+):?(%d*):?(%d*)-?(%d*)', -- Paths with /
+    '([%w_%-]+%.%w+):?(%d*):?(%d*)-?(%d*)', -- Top-level files
+  }
 
-  vim.list_extend(all_refs, backtick_refs)
-  vim.list_extend(all_refs, file_uri_refs)
-  vim.list_extend(all_refs, plain_refs)
+  for _, pattern in ipairs(patterns) do
+    local refs = parse_references_with_pattern(text, pattern, message_id)
+    vim.list_extend(all_refs, refs)
+  end
 
+  -- Sort by position and deduplicate
   table.sort(all_refs, function(a, b)
     return a.match_start < b.match_start
   end)
 
   local deduplicated = {}
   for _, ref in ipairs(all_refs) do
-    local overlaps = false
-    for _, existing in ipairs(deduplicated) do
-      if ref.match_start <= existing.match_end and existing.match_start <= ref.match_end then
-        overlaps = true
-        break
-      end
-    end
-
-    if not overlaps then
+    local last = deduplicated[#deduplicated]
+    if not last or ref.match_start > last.match_end then
       table.insert(deduplicated, ref)
     end
   end
@@ -229,7 +206,7 @@ function M._parse_message_references(msg)
 
     if part.type == 'tool' then
       local file_path = vim.tbl_get(part, 'state', 'input', 'filePath')
-      if file_path and file_exists(file_path) then
+      if file_path and vim.fn.filereadable(file_path) == 1 then
         local relative_path = vim.fn.fnamemodify(file_path, ':~:.')
         local ref = create_code_reference(relative_path, nil, nil, nil, message_id, 0, 0)
         table.insert(refs, ref)

tests/data/diagnostics.expected.json

@@ -1 +1 @@
-{"timestamp":1769093505,"actions":[{"display_line":18,"text":"[R]evert file","key":"R","args":["1f593f7ed419c95d3995f8ef4b98d4e571c3a492"],"range":{"to":18,"from":18},"type":"diff_revert_selected_file"},{"display_line":18,"text":"Revert [A]ll","key":"A","args":["1f593f7ed419c95d3995f8ef4b98d4e571c3a492"],"range":{"to":18,"from":18},"type":"diff_revert_all"},{"display_line":18,"text":"[D]iff","key":"D","args":["1f593f7ed419c95d3995f8ef4b98d4e571c3a492"],"range":{"to":18,"from":18},"type":"diff_open"}],"lines":["----","","","can you add \"great\" before \"string\" in @diff-test.txt?","","[diff-test.txt](diff-test.txt)","","----","","","**  edit** `diff-test.txt`","","`````txt"," this is a string"," this is a great string","","`````","","**󰻛  Created Snapshot** `1f593f7e`","","----","","",""],"extmarks":[[1,1,0,{"ns_id":3,"virt_text":[["▌󰭻 ","OpencodeMessageRoleUser"],[" "],["USER","OpencodeMessageRoleUser"],["","OpencodeHint"],[" (2025-10-12 06:42:56)","OpencodeHint"],[" [msg_9d7287269001C5gRusYfX7A1w1]","OpencodeHint"]],"virt_text_hide":false,"virt_text_win_col":-3,"priority":10,"virt_text_repeat_linebreak":false,"virt_text_pos":"win_col","right_gravity":true}],[2,2,0,{"ns_id":3,"virt_text":[["▌","OpencodeMessageRoleUser"]],"virt_text_hide":false,"virt_text_win_col":-3,"priority":4096,"virt_text_repeat_linebreak":true,"virt_text_pos":"win_col","right_gravity":true}],[3,3,0,{"ns_id":3,"virt_text":[["▌","OpencodeMessageRoleUser"]],"virt_text_hide":false,"virt_text_win_col":-3,"priority":4096,"virt_text_repeat_linebreak":true,"virt_text_pos":"win_col","right_gravity":true}],[4,3,39,{"end_col":53,"end_right_gravity":false,"right_gravity":true,"hl_eol":false,"priority":1000,"hl_group":"OpencodeMention","ns_id":3,"end_row":3}],[5,4,0,{"ns_id":3,"virt_text":[["▌","OpencodeMessageRoleUser"]],"virt_text_hide":false,"virt_text_win_col":-3,"priority":4096,"virt_text_repeat_linebreak":true,"virt_text_pos":"win_col","right_gravity":true}],[6,5,0,{"ns_id":3,"virt_text":[["▌","OpencodeMessageRoleUser"]],"virt_text_hide":false,"virt_text_win_col":-3,"priority":4096,"virt_text_repeat_linebreak":true,"virt_text_pos":"win_col","right_gravity":true}],[7,8,0,{"ns_id":3,"virt_text":[[" ","OpencodeMessageRoleAssistant"],[" "],["BUILD","OpencodeMessageRoleAssistant"],[" claude-sonnet-4.5","OpencodeHint"],[" (2025-10-12 06:42:56)","OpencodeHint"],[" [msg_9d7287287001HVwpPaH7WkRVdN]","OpencodeHint"]],"virt_text_hide":false,"virt_text_win_col":-3,"priority":10,"virt_text_repeat_linebreak":false,"virt_text_pos":"win_col","right_gravity":true}],[8,10,0,{"ns_id":3,"virt_text":[["▌","OpencodeToolBorder"]],"virt_text_hide":false,"virt_text_win_col":-1,"priority":4096,"virt_text_repeat_linebreak":true,"virt_text_pos":"win_col","right_gravity":true}],[9,11,0,{"ns_id":3,"virt_text":[["▌","OpencodeToolBorder"]],"virt_text_hide":false,"virt_text_win_col":-1,"priority":4096,"virt_text_repeat_linebreak":true,"virt_text_pos":"win_col","right_gravity":true}],[10,12,0,{"ns_id":3,"virt_text":[["▌","OpencodeToolBorder"]],"virt_text_hide":false,"virt_text_win_col":-1,"priority":4096,"virt_text_repeat_linebreak":true,"virt_text_pos":"win_col","right_gravity":true}],[11,13,0,{"end_col":0,"virt_text":[["-","OpencodeDiffDelete"]],"virt_text_pos":"overlay","ns_id":3,"right_gravity":true,"hl_eol":true,"end_right_gravity":false,"virt_text_hide":false,"virt_text_repeat_linebreak":false,"end_row":14,"hl_group":"OpencodeDiffDelete","priority":5000}],[12,13,0,{"ns_id":3,"virt_text":[["▌","OpencodeToolBorder"]],"virt_text_hide":false,"virt_text_win_col":-1,"priority":4096,"virt_text_repeat_linebreak":true,"virt_text_pos":"win_col","right_gravity":true}],[13,14,0,{"end_col":0,"virt_text":[["+","OpencodeDiffAdd"]],"virt_text_pos":"overlay","ns_id":3,"right_gravity":true,"hl_eol":true,"end_right_gravity":false,"virt_text_hide":false,"virt_text_repeat_linebreak":false,"end_row":15,"hl_group":"OpencodeDiffAdd","priority":5000}],[14,14,0,{"ns_id":3,"virt_text":[["▌","OpencodeToolBorder"]],"virt_text_hide":false,"virt_text_win_col":-1,"priority":4096,"virt_text_repeat_linebreak":true,"virt_text_pos":"win_col","right_gravity":true}],[15,15,0,{"ns_id":3,"virt_text":[["▌","OpencodeToolBorder"]],"virt_text_hide":false,"virt_text_win_col":-1,"priority":4096,"virt_text_repeat_linebreak":true,"virt_text_pos":"win_col","right_gravity":true}],[16,16,0,{"ns_id":3,"virt_text":[["▌","OpencodeToolBorder"]],"virt_text_hide":false,"virt_text_win_col":-1,"priority":4096,"virt_text_repeat_linebreak":true,"virt_text_pos":"win_col","right_gravity":true}],[17,21,0,{"ns_id":3,"virt_text":[[" ","OpencodeMessageRoleAssistant"],[" "],["BUILD","OpencodeMessageRoleAssistant"],[" claude-sonnet-4.5","OpencodeHint"],[" (2025-10-12 06:43:03)","OpencodeHint"],[" [msg_9d7288f2f001hW6NqqhtBc72UU]","OpencodeHint"]],"virt_text_hide":false,"virt_text_win_col":-3,"priority":10,"virt_text_repeat_linebreak":false,"virt_text_pos":"win_col","right_gravity":true}]]}
+{"extmarks":[[1,1,0,{"ns_id":3,"priority":10,"virt_text":[["▌󰭻 ","OpencodeMessageRoleUser"],[" "],["USER","OpencodeMessageRoleUser"],["","OpencodeHint"],[" (2025-10-12 06:42:56)","OpencodeHint"],[" [msg_9d7287269001C5gRusYfX7A1w1]","OpencodeHint"]],"virt_text_pos":"win_col","virt_text_win_col":-3,"virt_text_hide":false,"right_gravity":true,"virt_text_repeat_linebreak":false}],[2,2,0,{"ns_id":3,"priority":4096,"virt_text":[["▌","OpencodeMessageRoleUser"]],"virt_text_pos":"win_col","virt_text_win_col":-3,"virt_text_hide":false,"right_gravity":true,"virt_text_repeat_linebreak":true}],[3,3,0,{"ns_id":3,"priority":4096,"virt_text":[["▌","OpencodeMessageRoleUser"]],"virt_text_pos":"win_col","virt_text_win_col":-3,"virt_text_hide":false,"right_gravity":true,"virt_text_repeat_linebreak":true}],[4,3,39,{"ns_id":3,"hl_group":"OpencodeMention","end_row":3,"end_col":53,"end_right_gravity":false,"priority":1000,"right_gravity":true,"hl_eol":false}],[5,4,0,{"ns_id":3,"priority":4096,"virt_text":[["▌","OpencodeMessageRoleUser"]],"virt_text_pos":"win_col","virt_text_win_col":-3,"virt_text_hide":false,"right_gravity":true,"virt_text_repeat_linebreak":true}],[6,5,0,{"ns_id":3,"priority":4096,"virt_text":[["▌","OpencodeMessageRoleUser"]],"virt_text_pos":"win_col","virt_text_win_col":-3,"virt_text_hide":false,"right_gravity":true,"virt_text_repeat_linebreak":true}],[7,8,0,{"ns_id":3,"priority":10,"virt_text":[[" ","OpencodeMessageRoleAssistant"],[" "],["BUILD","OpencodeMessageRoleAssistant"],[" claude-sonnet-4.5","OpencodeHint"],[" (2025-10-12 06:42:56)","OpencodeHint"],[" [msg_9d7287287001HVwpPaH7WkRVdN]","OpencodeHint"]],"virt_text_pos":"win_col","virt_text_win_col":-3,"virt_text_hide":false,"right_gravity":true,"virt_text_repeat_linebreak":false}],[8,10,0,{"ns_id":3,"priority":4096,"virt_text":[["▌","OpencodeToolBorder"]],"virt_text_pos":"win_col","virt_text_win_col":-1,"virt_text_hide":false,"right_gravity":true,"virt_text_repeat_linebreak":true}],[9,11,0,{"ns_id":3,"priority":4096,"virt_text":[["▌","OpencodeToolBorder"]],"virt_text_pos":"win_col","virt_text_win_col":-1,"virt_text_hide":false,"right_gravity":true,"virt_text_repeat_linebreak":true}],[10,12,0,{"ns_id":3,"priority":4096,"virt_text":[["▌","OpencodeToolBorder"]],"virt_text_pos":"win_col","virt_text_win_col":-1,"virt_text_hide":false,"right_gravity":true,"virt_text_repeat_linebreak":true}],[11,13,0,{"ns_id":3,"end_col":0,"virt_text_pos":"overlay","virt_text":[["-","OpencodeDiffDelete"]],"priority":5000,"hl_group":"OpencodeDiffDelete","end_row":14,"hl_eol":true,"end_right_gravity":false,"virt_text_hide":false,"right_gravity":true,"virt_text_repeat_linebreak":false}],[12,13,0,{"ns_id":3,"priority":4096,"virt_text":[["▌","OpencodeToolBorder"]],"virt_text_pos":"win_col","virt_text_win_col":-1,"virt_text_hide":false,"right_gravity":true,"virt_text_repeat_linebreak":true}],[13,14,0,{"ns_id":3,"end_col":0,"virt_text_pos":"overlay","virt_text":[["+","OpencodeDiffAdd"]],"priority":5000,"hl_group":"OpencodeDiffAdd","end_row":15,"hl_eol":true,"end_right_gravity":false,"virt_text_hide":false,"right_gravity":true,"virt_text_repeat_linebreak":false}],[14,14,0,{"ns_id":3,"priority":4096,"virt_text":[["▌","OpencodeToolBorder"]],"virt_text_pos":"win_col","virt_text_win_col":-1,"virt_text_hide":false,"right_gravity":true,"virt_text_repeat_linebreak":true}],[15,15,0,{"ns_id":3,"priority":4096,"virt_text":[["▌","OpencodeToolBorder"]],"virt_text_pos":"win_col","virt_text_win_col":-1,"virt_text_hide":false,"right_gravity":true,"virt_text_repeat_linebreak":true}],[16,16,0,{"ns_id":3,"priority":4096,"virt_text":[["▌","OpencodeToolBorder"]],"virt_text_pos":"win_col","virt_text_win_col":-1,"virt_text_hide":false,"right_gravity":true,"virt_text_repeat_linebreak":true}],[17,21,0,{"ns_id":3,"priority":10,"virt_text":[[" ","OpencodeMessageRoleAssistant"],[" "],["BUILD","OpencodeMessageRoleAssistant"],[" claude-sonnet-4.5","OpencodeHint"],[" (2025-10-12 06:43:03)","OpencodeHint"],[" [msg_9d7288f2f001hW6NqqhtBc72UU]","OpencodeHint"]],"virt_text_pos":"win_col","virt_text_win_col":-3,"virt_text_hide":false,"right_gravity":true,"virt_text_repeat_linebreak":false}]],"timestamp":1769172144,"actions":[{"key":"R","args":["1f593f7ed419c95d3995f8ef4b98d4e571c3a492"],"display_line":18,"range":{"from":18,"to":18},"type":"diff_revert_selected_file","text":"[R]evert file"},{"key":"A","args":["1f593f7ed419c95d3995f8ef4b98d4e571c3a492"],"display_line":18,"range":{"from":18,"to":18},"type":"diff_revert_all","text":"Revert [A]ll"},{"key":"D","args":["1f593f7ed419c95d3995f8ef4b98d4e571c3a492"],"display_line":18,"range":{"from":18,"to":18},"type":"diff_open","text":"[D]iff"}],"lines":["----","","","can you add \"great\" before \"string\" in @diff-test.txt?","","[diff-test.txt](diff-test.txt)","","----","","","**  edit** `/Users/cam/tmp/a/diff-test.txt`","","`````txt"," this is a string"," this is a great string","","`````","","**󰻛  Created Snapshot** `1f593f7e`","","----","","",""]}