feat(treesitter): improve query error message

5 files changed, 122 insertions, 13 deletions

diff --git a/runtime/doc/news.txt b/runtime/doc/news.txt
index a233b66d1f..7607d218c3 100644
--- a/runtime/doc/news.txt
+++ b/runtime/doc/news.txt
@@ -135,6 +135,7 @@ The following new APIs and features were added.
     or the parent LanguageTree's language, respectively.
   • Added `vim.treesitter.preview_query()`, for live editing of treesitter
     queries.
+  • Improved error messages for query parsing.
 
 • |vim.ui.open()| opens URIs using the system default handler (macOS `open`,
   Windows `explorer`, Linux `xdg-open`, etc.)
diff --git a/src/nvim/lua/treesitter.c b/src/nvim/lua/treesitter.c
index 49e920f748..48057b0c65 100644
--- a/src/nvim/lua/treesitter.c
+++ b/src/nvim/lua/treesitter.c
@@ -1543,8 +1543,9 @@ int tslua_parse_query(lua_State *L)
   TSQuery *query = ts_query_new(lang, src, (uint32_t)len, &error_offset, &error_type);
 
   if (!query) {
-    return luaL_error(L, "query: %s at position %d for language %s",
-                      query_err_string(error_type), (int)error_offset, lang_name);
+    char err_msg[IOSIZE];
+    query_err_string(src, (int)error_offset, error_type, err_msg, sizeof(err_msg));
+    return luaL_error(L, "%s", err_msg);
   }
 
   TSQuery **ud = lua_newuserdata(L, sizeof(TSQuery *));  // [udata]
@@ -1554,24 +1555,85 @@ int tslua_parse_query(lua_State *L)
   return 1;
 }
 
-static const char *query_err_string(TSQueryError err)
+static const char *query_err_to_string(TSQueryError error_type)
 {
-  switch (err) {
+  switch (error_type) {
   case TSQueryErrorSyntax:
-    return "invalid syntax";
+    return "Invalid syntax:\n";
   case TSQueryErrorNodeType:
-    return "invalid node type";
+    return "Invalid node type ";
   case TSQueryErrorField:
-    return "invalid field";
+    return "Invalid field name ";
   case TSQueryErrorCapture:
-    return "invalid capture";
+    return "Invalid capture name ";
   case TSQueryErrorStructure:
-    return "invalid structure";
+    return "Impossible pattern:\n";
   default:
     return "error";
   }
 }
 
+static void query_err_string(const char *src, int error_offset, TSQueryError error_type, char *err,
+                             size_t errlen)
+{
+  int line_start = 0;
+  int row = 0;
+  const char *error_line = NULL;
+  int error_line_len = 0;
+
+  const char *end_str;
+  const char *src_tmp = src;
+  while ((end_str = strchr(src_tmp, '\n')) != NULL) {
+    int line_length = (int)(end_str - src_tmp) + 1;
+    int line_end = line_start + line_length;
+    if (line_end > error_offset) {
+      error_line = src_tmp;
+      error_line_len = line_length;
+      break;
+    }
+    line_start = line_end;
+    row++;
+    src_tmp += line_length;
+  }
+
+  // Additional check for the last line
+  if (line_start <= error_offset) {
+    error_line = src_tmp;
+    error_line_len = (int)strlen(src_tmp);
+  }
+
+  int column = error_offset - line_start;
+
+  const char *type_msg = query_err_to_string(error_type);
+  snprintf(err, errlen, "Query error at %d:%d. %s", row + 1, column + 1, type_msg);
+  size_t offset = strlen(err);
+  errlen = errlen - offset;
+  err = err + offset;
+
+  // Error types that report names
+  if (error_type == TSQueryErrorNodeType
+      || error_type == TSQueryErrorField
+      || error_type == TSQueryErrorCapture) {
+    const char *suffix = src + error_offset;
+    int suffix_len = 0;
+    char c = suffix[suffix_len];
+    while (isalnum(c) || c == '_' || c == '-' || c == '.') {
+      c = suffix[++suffix_len];
+    }
+    snprintf(err, errlen, "\"%.*s\":\n", suffix_len, suffix);
+    offset = strlen(err);
+    errlen = errlen - offset;
+    err = err + offset;
+  }
+
+  if (!error_line) {
+    snprintf(err, errlen, "Unexpected EOF\n");
+    return;
+  }
+
+  snprintf(err, errlen, "%.*s\n%*s^\n", error_line_len, error_line, column, "");
+}
+
 static TSQuery *query_check(lua_State *L, int index)
 {
   TSQuery **ud = luaL_checkudata(L, index, TS_META_QUERY);
diff --git a/test/functional/lua/overrides_spec.lua b/test/functional/lua/overrides_spec.lua
index 1777dd078d..c08f3d06a9 100644
--- a/test/functional/lua/overrides_spec.lua
+++ b/test/functional/lua/overrides_spec.lua
@@ -54,7 +54,7 @@ describe('print', function()
     -- TODO(bfredl): these look weird, print() should not use "E5114:" style errors..
     eq('Vim(lua):E5108: Error executing lua E5114: Error while converting print argument #2: [NULL]',
        pcall_err(command, 'lua print("foo", v_nilerr, "bar")'))
-    eq('Vim(lua):E5108: Error executing lua E5114: Error while converting print argument #2: Xtest-functional-lua-overrides-luafile:0: abc',
+    eq('Vim(lua):E5108: Error executing lua E5114: Error while converting print argument #2: Xtest-functional-lua-overrides-luafile:2: abc',
        pcall_err(command, 'lua print("foo", v_abcerr, "bar")'))
     eq('Vim(lua):E5108: Error executing lua E5114: Error while converting print argument #2: <Unknown error: lua_tolstring returned NULL for tostring result>',
        pcall_err(command, 'lua print("foo", v_tblout, "bar")'))
@@ -84,9 +84,9 @@ describe('print', function()
     end
     ]])
     eq('', exec_capture('luafile ' .. fname))
-    eq('Vim(lua):E5108: Error executing lua Xtest-functional-lua-overrides-luafile:0: my mistake',
+    eq('Vim(lua):E5108: Error executing lua Xtest-functional-lua-overrides-luafile:1: my mistake',
       pcall_err(command, 'lua string_error()'))
-    eq('Vim(lua):E5108: Error executing lua Xtest-functional-lua-overrides-luafile:0: 1234',
+    eq('Vim(lua):E5108: Error executing lua Xtest-functional-lua-overrides-luafile:2: 1234',
       pcall_err(command, 'lua number_error()'))
     eq('Vim(lua):E5108: Error executing lua [NULL]',
       pcall_err(command, 'lua nil_error()'))
diff --git a/test/functional/treesitter/parser_spec.lua b/test/functional/treesitter/parser_spec.lua
index 56af0c8738..37dde37a64 100644
--- a/test/functional/treesitter/parser_spec.lua
+++ b/test/functional/treesitter/parser_spec.lua
@@ -1055,4 +1055,44 @@ int x = INT_MAX;
     ]])
   end)
 
+  it('fails to load queries', function()
+    local function test(exp, cquery)
+      eq(exp, pcall_err(exec_lua, "vim.treesitter.query.parse('c', ...)", cquery))
+    end
+
+    -- Invalid node type
+    test(
+      '.../query.lua:0: Query error at 1:2. Invalid node type "dentifier":\n'..
+      '(dentifier) @variable\n'..
+      ' ^',
+      '(dentifier) @variable')
+
+    -- Impossible pattern
+    test(
+      '.../query.lua:0: Query error at 1:13. Impossible pattern:\n'..
+      '(identifier (identifier) @variable)\n'..
+      '            ^',
+      '(identifier (identifier) @variable)')
+
+    -- Invalid syntax
+    test(
+      '.../query.lua:0: Query error at 1:13. Invalid syntax:\n'..
+      '(identifier @variable\n'..
+      '            ^',
+      '(identifier @variable')
+
+    -- Invalid field name
+    test(
+      '.../query.lua:0: Query error at 1:15. Invalid field name "invalid_field":\n'..
+      '((identifier) invalid_field: (identifier))\n'..
+      '              ^',
+      '((identifier) invalid_field: (identifier))')
+
+    -- Invalid capture name
+    test(
+      '.../query.lua:0: Query error at 1:30. Invalid capture name "ok.capture":\n'..
+      '((identifier) @id (#eq? @id @ok.capture))\n'..
+      '                             ^',
+      '((identifier) @id (#eq? @id @ok.capture))')
+  end)
 end)
diff --git a/test/helpers.lua b/test/helpers.lua
index 51114611ab..02192e4924 100644
--- a/test/helpers.lua
+++ b/test/helpers.lua
@@ -188,10 +188,16 @@ function module.pcall(fn, ...)
   local errmsg = tostring(rv):gsub('([%s<])vim[/\\]([^%s:/\\]+):%d+', '%1\xffvim\xff%2:0')
                              :gsub('[^%s<]-[/\\]([^%s:/\\]+):%d+', '.../%1:0')
                              :gsub('\xffvim\xff', 'vim/')
+
   -- Scrub numbers in paths/stacktraces:
   --    shared.lua:0: in function 'gsplit'
   --    shared.lua:0: in function <shared.lua:0>'
-  errmsg = errmsg:gsub('([^%s]):%d+', '%1:0')
+  errmsg = errmsg:gsub('([^%s].lua):%d+', '%1:0')
+  --    [string "<nvim>"]:0:
+  --    [string ":lua"]:0:
+  --    [string ":luado"]:0:
+  errmsg = errmsg:gsub('(%[string "[^"]+"%]):%d+', '%1:0')
+
   -- Scrub tab chars:
   errmsg = errmsg:gsub('\t', '    ')
   -- In Lua 5.1, we sometimes get a "(tail call): ?" on the last line.