Differentiate between top-level items and those that appear in statement blocks

grammar.js

@@ -29,20 +29,17 @@ module.exports = grammar({
   ],
 
   inline: $ => [
-    $._statement
+    $._statement,
+    $._top_level_item,
+    $._compound_statement_item
   ],
 
   conflicts: $ => [
-    [$._declarator, $._type_specifier, $._expression, $.macro_type_specifier],
+    [$._type_specifier, $._declarator],
+    [$._type_specifier, $._declarator, $.macro_type_specifier],
+    [$._type_specifier, $._expression],
     [$._type_specifier, $._expression, $.macro_type_specifier],
-    [$._declarator, $._type_specifier, $.macro_type_specifier],
-    [$._field_declarator, $._type_specifier, $.macro_type_specifier],
-    [$._declarator, $._type_specifier, $._expression],
-    [$._field_declarator, $._type_specifier],
-    [$._declarator, $._type_specifier],
     [$._type_specifier, $.macro_type_specifier],
-    [$._type_specifier, $._expression],
-    [$._declarator, $._expression],
     [$.sized_type_specifier],
   ],
 
@@ -50,17 +47,10 @@ module.exports = grammar({
     translation_unit: $ => repeat($._top_level_item),
 
     _top_level_item: $ => choice(
-      $._preproc_statement,
       $.function_definition,
+      $.linkage_specification,
       $.declaration,
-      $._statement,
       $._empty_declaration,
-      $.linkage_specification
-    ),
-
-    // Preprocesser
-
-    _preproc_statement: $ => choice(
       $.preproc_if,
       $.preproc_ifdef,
       $.preproc_include,
@@ -69,26 +59,34 @@ module.exports = grammar({
       $.preproc_call
     ),
 
+    _compound_statement_item: $ => choice(
+      $._statement,
+      $.declaration,
+      $._empty_declaration,
+      rename($.preproc_if_in_compound_statement, 'preproc_if'),
+      rename($.preproc_ifdef_in_compound_statement, 'preproc_ifdef'),
+      $.preproc_include,
+      $.preproc_def,
+      $.preproc_function_def,
+      $.preproc_call
+    ),
+
+    // Preprocesser
+
     preproc_include: $ => seq(
-      /#[ \t]*include/,
-      choice(
-        $.string_literal,
-        $.system_lib_string
-      )
+      $._pound_include,
+      choice($.string_literal, $.system_lib_string)
     ),
 
     preproc_def: $ => seq(
-      /#[ \t]*define/,
+      $._pound_define,
       $.identifier,
-      optional(seq(
-        /[ \t]+/,
-        $.preproc_arg
-      )),
+      optional(seq(/[ \t]+/, $.preproc_arg)),
       '\n'
     ),
 
     preproc_function_def: $ => seq(
-      /#[ \t]*define/,
+      $._pound_define,
       $.identifier,
       $.preproc_params,
       optional($.preproc_arg),
@@ -104,38 +102,61 @@ module.exports = grammar({
       $.preproc_arg
     ),
 
-    preproc_arg: $ => token(prec(-1, repeat1(choice(/./, '\\\n')))),
-
     preproc_if: $ => seq(
-      /#[ \t]*if/,
+      $._pound_if,
       $.preproc_arg,
       repeat($._top_level_item),
       optional($.preproc_else),
-      /#[ \t]*endif/
+      $._pound_endif
+    ),
+
+    preproc_if_in_compound_statement: $ => seq(
+      $._pound_if,
+      $.preproc_arg,
+      repeat($._compound_statement_item),
+      optional(rename($.preproc_else_in_compound_statement, 'preproc_else')),
+      $._pound_endif
     ),
 
     preproc_ifdef: $ => seq(
-      choice(
-        /#[ \t]*ifdef/,
-        /#[ \t]*ifndef/
-      ),
+      $._pound_ifdef,
       $.identifier,
       repeat($._top_level_item),
       optional($.preproc_else),
-      /#[ \t]*endif/
+      $._pound_endif
+    ),
+
+    preproc_ifdef_in_compound_statement: $ => seq(
+      $._pound_ifdef,
+      $.identifier,
+      repeat($._compound_statement_item),
+      optional(rename($.preproc_else_in_compound_statement, 'preproc_else')),
+      $._pound_endif
     ),
 
     preproc_else: $ => seq(
-      /#[ \t]*else/,
+      $._pound_else,
       repeat($._top_level_item)
     ),
 
-    preproc_directive: $ => /#[ \t]*\a\w*/,
+    preproc_else_in_compound_statement: $ => seq(
+      $._pound_else,
+      repeat($._compound_statement_item)
+    ),
+
+    _pound_include: $ => preprocessor('include'),
+    _pound_define: $ => preprocessor('define'),
+    _pound_if: $ => preprocessor('if'),
+    _pound_ifdef: $ => preprocessor('ifn?def'),
+    _pound_endif: $ => preprocessor('endif'),
+    _pound_else: $ => preprocessor('else'),
+    preproc_directive: $ => preprocessor('\\a\\w*'),
+    preproc_arg: $ => token(prec(-1, repeat1(choice(/./, '\\\n')))),
 
     // Main Grammar
 
     function_definition: $ => seq(
-      optional($._declaration_specifiers),
+      $._declaration_specifiers,
       $._declarator,
       $.compound_statement
     ),
@@ -244,7 +265,7 @@ module.exports = grammar({
 
     compound_statement: $ => seq(
       '{',
-      repeat($._top_level_item),
+      repeat($._compound_statement_item),
       '}'
     ),
 
@@ -329,7 +350,7 @@ module.exports = grammar({
     ),
 
     field_declaration: $ => seq(
-      optional($._declaration_specifiers),
+      $._declaration_specifiers,
       commaSep($._field_declarator),
       optional(seq(':', $._expression)),
       ';'
@@ -700,6 +721,10 @@ module.exports = grammar({
 
 module.exports.PREC = PREC
 
+function preprocessor (pattern) {
+  return new RegExp('#[ \t]*' + pattern)
+}
+
 function commaSep (rule) {
   return optional(commaSep1(rule))
 }

src/grammar.json

@@ -13,11 +13,11 @@
       "members": [
         {
           "type": "SYMBOL",
-          "name": "_preproc_statement"
+          "name": "function_definition"
         },
         {
           "type": "SYMBOL",
-          "name": "function_definition"
+          "name": "linkage_specification"
         },
         {
           "type": "SYMBOL",
@@ -25,28 +25,64 @@
         },
         {
           "type": "SYMBOL",
-          "name": "_statement"
+          "name": "_empty_declaration"
         },
         {
           "type": "SYMBOL",
-          "name": "_empty_declaration"
+          "name": "preproc_if"
         },
         {
           "type": "SYMBOL",
-          "name": "linkage_specification"
+          "name": "preproc_ifdef"
+        },
+        {
+          "type": "SYMBOL",
+          "name": "preproc_include"
+        },
+        {
+          "type": "SYMBOL",
+          "name": "preproc_def"
+        },
+        {
+          "type": "SYMBOL",
+          "name": "preproc_function_def"
+        },
+        {
+          "type": "SYMBOL",
+          "name": "preproc_call"
         }
       ]
     },
-    "_preproc_statement": {
+    "_compound_statement_item": {
       "type": "CHOICE",
       "members": [
         {
           "type": "SYMBOL",
-          "name": "preproc_if"
+          "name": "_statement"
         },
         {
           "type": "SYMBOL",
-          "name": "preproc_ifdef"
+          "name": "declaration"
+        },
+        {
+          "type": "SYMBOL",
+          "name": "_empty_declaration"
+        },
+        {
+          "type": "RENAME",
+          "content": {
+            "type": "SYMBOL",
+            "name": "preproc_if_in_compound_statement"
+          },
+          "value": "preproc_if"
+        },
+        {
+          "type": "RENAME",
+          "content": {
+            "type": "SYMBOL",
+            "name": "preproc_ifdef_in_compound_statement"
+          },
+          "value": "preproc_ifdef"
         },
         {
           "type": "SYMBOL",
@@ -70,8 +106,8 @@
       "type": "SEQ",
       "members": [
         {
-          "type": "PATTERN",
-          "value": "#[ \\t]*include"
+          "type": "SYMBOL",
+          "name": "_pound_include"
         },
         {
           "type": "CHOICE",
@@ -92,8 +128,8 @@
       "type": "SEQ",
       "members": [
         {
-          "type": "PATTERN",
-          "value": "#[ \\t]*define"
+          "type": "SYMBOL",
+          "name": "_pound_define"
         },
         {
           "type": "SYMBOL",
@@ -130,8 +166,8 @@
       "type": "SEQ",
       "members": [
         {
-          "type": "PATTERN",
-          "value": "#[ \\t]*define"
+          "type": "SYMBOL",
+          "name": "_pound_define"
         },
         {
           "type": "SYMBOL",
@@ -236,35 +272,12 @@
         }
       ]
     },
-    "preproc_arg": {
-      "type": "TOKEN",
-      "content": {
-        "type": "PREC",
-        "value": -1,
-        "content": {
-          "type": "REPEAT1",
-          "content": {
-            "type": "CHOICE",
-            "members": [
-              {
-                "type": "PATTERN",
-                "value": "."
-              },
-              {
-                "type": "STRING",
-                "value": "\\\n"
-              }
-            ]
-          }
-        }
-      }
-    },
     "preproc_if": {
       "type": "SEQ",
       "members": [
         {
-          "type": "PATTERN",
-          "value": "#[ \\t]*if"
+          "type": "SYMBOL",
+          "name": "_pound_if"
         },
         {
           "type": "SYMBOL",
@@ -290,27 +303,58 @@
           ]
         },
         {
-          "type": "PATTERN",
-          "value": "#[ \\t]*endif"
+          "type": "SYMBOL",
+          "name": "_pound_endif"
         }
       ]
     },
-    "preproc_ifdef": {
+    "preproc_if_in_compound_statement": {
       "type": "SEQ",
       "members": [
+        {
+          "type": "SYMBOL",
+          "name": "_pound_if"
+        },
+        {
+          "type": "SYMBOL",
+          "name": "preproc_arg"
+        },
+        {
+          "type": "REPEAT",
+          "content": {
+            "type": "SYMBOL",
+            "name": "_compound_statement_item"
+          }
+        },
         {
           "type": "CHOICE",
           "members": [
             {
-              "type": "PATTERN",
-              "value": "#[ \\t]*ifdef"
+              "type": "RENAME",
+              "content": {
+                "type": "SYMBOL",
+                "name": "preproc_else_in_compound_statement"
+              },
+              "value": "preproc_else"
             },
             {
-              "type": "PATTERN",
-              "value": "#[ \\t]*ifndef"
+              "type": "BLANK"
             }
           ]
         },
+        {
+          "type": "SYMBOL",
+          "name": "_pound_endif"
+        }
+      ]
+    },
+    "preproc_ifdef": {
+      "type": "SEQ",
+      "members": [
+        {
+          "type": "SYMBOL",
+          "name": "_pound_ifdef"
+        },
         {
           "type": "SYMBOL",
           "name": "identifier"
@@ -335,8 +379,48 @@
           ]
         },
         {
-          "type": "PATTERN",
-          "value": "#[ \\t]*endif"
+          "type": "SYMBOL",
+          "name": "_pound_endif"
+        }
+      ]
+    },
+    "preproc_ifdef_in_compound_statement": {
+      "type": "SEQ",
+      "members": [
+        {
+          "type": "SYMBOL",
+          "name": "_pound_ifdef"
+        },
+        {
+          "type": "SYMBOL",
+          "name": "identifier"
+        },
+        {
+          "type": "REPEAT",
+          "content": {
+            "type": "SYMBOL",
+            "name": "_compound_statement_item"
+          }
+        },
+        {
+          "type": "CHOICE",
+          "members": [
+            {
+              "type": "RENAME",
+              "content": {
+                "type": "SYMBOL",
+                "name": "preproc_else_in_compound_statement"
+              },
+              "value": "preproc_else"
+            },
+            {
+              "type": "BLANK"
+            }
+          ]
+        },
+        {
+          "type": "SYMBOL",
+          "name": "_pound_endif"
         }
       ]
     },
@@ -344,8 +428,8 @@
       "type": "SEQ",
       "members": [
         {
-          "type": "PATTERN",
-          "value": "#[ \\t]*else"
+          "type": "SYMBOL",
+          "name": "_pound_else"
         },
         {
           "type": "REPEAT",
@@ -356,24 +440,79 @@
         }
       ]
     },
+    "preproc_else_in_compound_statement": {
+      "type": "SEQ",
+      "members": [
+        {
+          "type": "SYMBOL",
+          "name": "_pound_else"
+        },
+        {
+          "type": "REPEAT",
+          "content": {
+            "type": "SYMBOL",
+            "name": "_compound_statement_item"
+          }
+        }
+      ]
+    },
+    "_pound_include": {
+      "type": "PATTERN",
+      "value": "#[ \t]*include"
+    },
+    "_pound_define": {
+      "type": "PATTERN",
+      "value": "#[ \t]*define"
+    },
+    "_pound_if": {
+      "type": "PATTERN",
+      "value": "#[ \t]*if"
+    },
+    "_pound_ifdef": {
+      "type": "PATTERN",
+      "value": "#[ \t]*ifn?def"
+    },
+    "_pound_endif": {
+      "type": "PATTERN",
+      "value": "#[ \t]*endif"
+    },
+    "_pound_else": {
+      "type": "PATTERN",
+      "value": "#[ \t]*else"
+    },
     "preproc_directive": {
       "type": "PATTERN",
-      "value": "#[ \\t]*\\a\\w*"
+      "value": "#[ \t]*\\a\\w*"
+    },
+    "preproc_arg": {
+      "type": "TOKEN",
+      "content": {
+        "type": "PREC",
+        "value": -1,
+        "content": {
+          "type": "REPEAT1",
+          "content": {
+            "type": "CHOICE",
+            "members": [
+              {
+                "type": "PATTERN",
+                "value": "."
+              },
+              {
+                "type": "STRING",
+                "value": "\\\n"
+              }
+            ]
+          }
+        }
+      }
     },
     "function_definition": {
       "type": "SEQ",
       "members": [
         {
-          "type": "CHOICE",
-          "members": [
-            {
-              "type": "SYMBOL",
-              "name": "_declaration_specifiers"
-            },
-            {
-              "type": "BLANK"
-            }
-          ]
+          "type": "SYMBOL",
+          "name": "_declaration_specifiers"
         },
         {
           "type": "SYMBOL",
@@ -972,7 +1111,7 @@
           "type": "REPEAT",
           "content": {
             "type": "SYMBOL",
-            "name": "_top_level_item"
+            "name": "_compound_statement_item"
           }
         },
         {
@@ -1312,16 +1451,8 @@
       "type": "SEQ",
       "members": [
         {
-          "type": "CHOICE",
-          "members": [
-            {
-              "type": "SYMBOL",
-              "name": "_declaration_specifiers"
-            },
-            {
-              "type": "BLANK"
-            }
-          ]
+          "type": "SYMBOL",
+          "name": "_declaration_specifiers"
         },
         {
           "type": "CHOICE",
@@ -3429,50 +3560,26 @@
   ],
   "conflicts": [
     [
-      "_declarator",
       "_type_specifier",
-      "_expression",
-      "macro_type_specifier"
+      "_declarator"
     ],
     [
       "_type_specifier",
-      "_expression",
-      "macro_type_specifier"
-    ],
-    [
       "_declarator",
-      "_type_specifier",
       "macro_type_specifier"
     ],
     [
-      "_field_declarator",
-      "_type_specifier",
-      "macro_type_specifier"
-    ],
-    [
-      "_declarator",
       "_type_specifier",
       "_expression"
     ],
-    [
-      "_field_declarator",
-      "_type_specifier"
-    ],
-    [
-      "_declarator",
-      "_type_specifier"
-    ],
     [
       "_type_specifier",
+      "_expression",
       "macro_type_specifier"
     ],
     [
       "_type_specifier",
-      "_expression"
-    ],
-    [
-      "_declarator",
-      "_expression"
+      "macro_type_specifier"
     ],
     [
       "sized_type_specifier"
@@ -3480,7 +3587,9 @@
   ],
   "externals": [],
   "inline": [
-    "_statement"
+    "_statement",
+    "_top_level_item",
+    "_compound_statement_item"
   ],
   "PREC": {
     "ASSIGNMENT": -1,