Fix grammar for allowing white space to appear everywher

corpus/comment.txt

@@ -41,7 +41,8 @@ Line comment [comment]
 Message with Line comment [comment]
 ===================================
 message(STATUS #Some line comment
-message #Some other line comment
+
+Second #Some other line comment
 )
 
 ---

corpus/condition.txt

@@ -1,7 +1,7 @@
 ====================
 Empty if [condition]
 ====================
-if(cond)
+if ( cond )
 endif()
 
 ---

corpus/foreach.txt

@@ -90,6 +90,7 @@ endforeach()
   (foreach_command
     (foreach)
     (argument (unquoted_argument))
+    (argument (unquoted_argument))
   )
   (endforeach_command (endforeach))
  )
@@ -109,6 +110,7 @@ endforeach()
   (foreach_command
     (foreach)
     (argument (unquoted_argument))
+    (argument (unquoted_argument))
   )
   (endforeach_command (endforeach))
  )

corpus/message.txt

@@ -75,3 +75,19 @@ message(STATUS
   (argument (bracket_argument))
  )
  )
+
+======================================================
+Message with STATUS and an unquoted argument [message]
+======================================================
+
+message(STATUS argument)
+
+---
+
+(source_file
+ (normal_command
+  (identifier)
+  (argument (unquoted_argument))
+  (argument (unquoted_argument))
+ )
+ )

corpus/unquoted_argument.txt

@@ -25,6 +25,7 @@ message(STATUS Hello)
  (normal_command
   (identifier)
   (argument (unquoted_argument))
+  (argument (unquoted_argument))
  )
  )
 

grammar.js

@@ -17,10 +17,10 @@ module.exports = grammar({
   name: "cmake",
 
   externals: ($) => [$.bracket_argument, $.bracket_comment, $.line_comment],
-  extras: ($) => [/[\s\n\r]/, $.bracket_comment, $.line_comment],
+  extras: ($) => [$.bracket_comment, $.line_comment],
 
   rules: {
-    source_file: ($) => repeat($._command_invocation),
+    source_file: ($) => repeat($._untrimmed_command_invocation),
 
     escape_sequence: ($) => choice($._escape_identity, $._escape_encoded, $._escape_semicolon),
     _escape_identity: (_) => /\\[^A-Za-z0-9;]/,
@@ -34,39 +34,45 @@ module.exports = grammar({
     cache_var: ($) => seq("$CACHE", "{", $.variable, "}"),
 
     argument: ($) => choice($.bracket_argument, $.quoted_argument, $.unquoted_argument),
+    _untrimmed_argument: ($) => choice(/\s/, $.argument),
 
     quoted_argument: ($) => seq('"', optional($.quoted_element), '"'),
     quoted_element: ($) => repeat1(choice($.variable_ref, /[^\\"]/, $.escape_sequence)),
 
-    unquoted_argument: ($) => prec.right(repeat1(choice($.variable_ref, /[^\s\n\r()#\"\\]/, $.escape_sequence))),
+    unquoted_argument: ($) => prec.right(repeat1(choice($.variable_ref, /[^\s()#\"\\]/, $.escape_sequence))),
 
-    if_command: ($) => command($.if, repeat(choice($.argument))),
+    if_command: ($) => command($.if, repeat($._untrimmed_argument)),
-    elseif_command: ($) => command($.elseif, repeat(choice($.argument))),
+    elseif_command: ($) => command($.elseif, repeat($._untrimmed_argument)),
-    else_command: ($) => command($.else, optional(choice($.argument))),
+    else_command: ($) => command($.else, optional(seq(/\s*/, $.argument, /\s*/))),
-    endif_command: ($) => command($.endif, optional(choice($.argument))),
+    endif_command: ($) => command($.endif, optional(seq(/\s*/, $.argument, /\s*/))),
     if_condition: ($) =>
-      seq($.if_command, repeat(choice($._command_invocation, $.elseif_command, $.else_command)), $.endif_command),
+      seq(
+        $.if_command,
+        repeat(choice($._untrimmed_command_invocation, $.elseif_command, $.else_command)),
+        $.endif_command
+      ),
 
-    foreach_command: ($) => command($.foreach, repeat(choice($.argument))),
+    foreach_command: ($) => command($.foreach, repeat($._untrimmed_argument)),
     endforeach_command: ($) => command($.endforeach, optional($.argument)),
-    foreach_loop: ($) => seq($.foreach_command, repeat($._command_invocation), $.endforeach_command),
+    foreach_loop: ($) => seq($.foreach_command, repeat($._untrimmed_command_invocation), $.endforeach_command),
 
-    while_command: ($) => command($.while, repeat(choice($.argument))),
+    while_command: ($) => command($.while, repeat($._untrimmed_argument)),
-    endwhile_command: ($) => command($.endwhile, optional(choice($.argument))),
+    endwhile_command: ($) => command($.endwhile, optional(seq(/\s*/, $.argument, /\s*/))),
-    while_loop: ($) => seq($.while_command, repeat($._command_invocation), $.endwhile_command),
+    while_loop: ($) => seq($.while_command, repeat($._untrimmed_command_invocation), $.endwhile_command),
 
-    function_command: ($) => command($.function, repeat($.argument)),
+    function_command: ($) => command($.function, repeat($._untrimmed_argument)),
-    endfunction_command: ($) => command($.endfunction, repeat($.argument)),
+    endfunction_command: ($) => command($.endfunction, repeat($._untrimmed_argument)),
-    function_def: ($) => seq($.function_command, repeat($._command_invocation), $.endfunction_command),
+    function_def: ($) => seq($.function_command, repeat($._untrimmed_command_invocation), $.endfunction_command),
 
-    macro_command: ($) => command($.macro, repeat($.argument)),
+    macro_command: ($) => command($.macro, repeat($._untrimmed_argument)),
-    endmacro_command: ($) => command($.endmacro, repeat($.argument)),
+    endmacro_command: ($) => command($.endmacro, repeat($._untrimmed_argument)),
-    macro_def: ($) => seq($.macro_command, repeat($._command_invocation), $.endmacro_command),
+    macro_def: ($) => seq($.macro_command, repeat($._untrimmed_command_invocation), $.endmacro_command),
 
-    normal_command: ($) => command($.identifier, repeat($.argument)),
+    normal_command: ($) => command($.identifier, repeat($._untrimmed_argument)),
 
     _command_invocation: ($) =>
       choice($.normal_command, $.if_condition, $.foreach_loop, $.while_loop, $.function_def, $.macro_def),
+    _untrimmed_command_invocation: ($) => choice(/\s/, $._command_invocation),
 
     ...commandNames(...commands),
     identifier: (_) => /[A-Za-z_][A-Za-z0-9_]*/,
@@ -87,5 +93,5 @@ function commandNames(...names) {
 }
 
 function command(name_rule, arg_rule) {
-  return seq(name_rule, "(", arg_rule, ")");
+  return seq(name_rule, repeat(/[\t ]/), "(", arg_rule, ")");
 }

src/grammar.json

@@ -5,7 +5,7 @@
       "type": "REPEAT",
       "content": {
         "type": "SYMBOL",
-        "name": "_command_invocation"
+        "name": "_untrimmed_command_invocation"
       }
     },
     "escape_sequence": {
@@ -163,6 +163,19 @@
         }
       ]
     },
+    "_untrimmed_argument": {
+      "type": "CHOICE",
+      "members": [
+        {
+          "type": "PATTERN",
+          "value": "\\s"
+        },
+        {
+          "type": "SYMBOL",
+          "name": "argument"
+        }
+      ]
+    },
     "quoted_argument": {
       "type": "SEQ",
       "members": [
@@ -222,7 +235,7 @@
             },
             {
               "type": "PATTERN",
-              "value": "[^\\s\\n\\r()#\\\"\\\\]"
+              "value": "[^\\s()#\\\"\\\\]"
             },
             {
               "type": "SYMBOL",
@@ -239,6 +252,13 @@
           "type": "SYMBOL",
           "name": "if"
         },
+        {
+          "type": "REPEAT",
+          "content": {
+            "type": "PATTERN",
+            "value": "[\\t ]"
+          }
+        },
         {
           "type": "STRING",
           "value": "("
@@ -246,13 +266,8 @@
         {
           "type": "REPEAT",
           "content": {
-            "type": "CHOICE",
-            "members": [
-              {
             "type": "SYMBOL",
-                "name": "argument"
+            "name": "_untrimmed_argument"
-              }
-            ]
           }
         },
         {
@@ -268,6 +283,13 @@
           "type": "SYMBOL",
           "name": "elseif"
         },
+        {
+          "type": "REPEAT",
+          "content": {
+            "type": "PATTERN",
+            "value": "[\\t ]"
+          }
+        },
         {
           "type": "STRING",
           "value": "("
@@ -275,13 +297,8 @@
         {
           "type": "REPEAT",
           "content": {
-            "type": "CHOICE",
-            "members": [
-              {
             "type": "SYMBOL",
-                "name": "argument"
+            "name": "_untrimmed_argument"
-              }
-            ]
           }
         },
         {
@@ -297,6 +314,13 @@
           "type": "SYMBOL",
           "name": "else"
         },
+        {
+          "type": "REPEAT",
+          "content": {
+            "type": "PATTERN",
+            "value": "[\\t ]"
+          }
+        },
         {
           "type": "STRING",
           "value": "("
@@ -305,11 +329,19 @@
           "type": "CHOICE",
           "members": [
             {
-              "type": "CHOICE",
+              "type": "SEQ",
               "members": [
+                {
+                  "type": "PATTERN",
+                  "value": "\\s*"
+                },
                 {
                   "type": "SYMBOL",
                   "name": "argument"
+                },
+                {
+                  "type": "PATTERN",
+                  "value": "\\s*"
                 }
               ]
             },
@@ -331,6 +363,13 @@
           "type": "SYMBOL",
           "name": "endif"
         },
+        {
+          "type": "REPEAT",
+          "content": {
+            "type": "PATTERN",
+            "value": "[\\t ]"
+          }
+        },
         {
           "type": "STRING",
           "value": "("
@@ -339,11 +378,19 @@
           "type": "CHOICE",
           "members": [
             {
-              "type": "CHOICE",
+              "type": "SEQ",
               "members": [
+                {
+                  "type": "PATTERN",
+                  "value": "\\s*"
+                },
                 {
                   "type": "SYMBOL",
                   "name": "argument"
+                },
+                {
+                  "type": "PATTERN",
+                  "value": "\\s*"
                 }
               ]
             },
@@ -372,7 +419,7 @@
             "members": [
               {
                 "type": "SYMBOL",
-                "name": "_command_invocation"
+                "name": "_untrimmed_command_invocation"
               },
               {
                 "type": "SYMBOL",
@@ -398,6 +445,13 @@
           "type": "SYMBOL",
           "name": "foreach"
         },
+        {
+          "type": "REPEAT",
+          "content": {
+            "type": "PATTERN",
+            "value": "[\\t ]"
+          }
+        },
         {
           "type": "STRING",
           "value": "("
@@ -405,13 +459,8 @@
         {
           "type": "REPEAT",
           "content": {
-            "type": "CHOICE",
-            "members": [
-              {
             "type": "SYMBOL",
-                "name": "argument"
+            "name": "_untrimmed_argument"
-              }
-            ]
           }
         },
         {
@@ -427,6 +476,13 @@
           "type": "SYMBOL",
           "name": "endforeach"
         },
+        {
+          "type": "REPEAT",
+          "content": {
+            "type": "PATTERN",
+            "value": "[\\t ]"
+          }
+        },
         {
           "type": "STRING",
           "value": "("
@@ -460,7 +516,7 @@
           "type": "REPEAT",
           "content": {
             "type": "SYMBOL",
-            "name": "_command_invocation"
+            "name": "_untrimmed_command_invocation"
           }
         },
         {
@@ -476,6 +532,13 @@
           "type": "SYMBOL",
           "name": "while"
         },
+        {
+          "type": "REPEAT",
+          "content": {
+            "type": "PATTERN",
+            "value": "[\\t ]"
+          }
+        },
         {
           "type": "STRING",
           "value": "("
@@ -483,13 +546,8 @@
         {
           "type": "REPEAT",
           "content": {
-            "type": "CHOICE",
-            "members": [
-              {
             "type": "SYMBOL",
-                "name": "argument"
+            "name": "_untrimmed_argument"
-              }
-            ]
           }
         },
         {
@@ -505,6 +563,13 @@
           "type": "SYMBOL",
           "name": "endwhile"
         },
+        {
+          "type": "REPEAT",
+          "content": {
+            "type": "PATTERN",
+            "value": "[\\t ]"
+          }
+        },
         {
           "type": "STRING",
           "value": "("
@@ -513,11 +578,19 @@
           "type": "CHOICE",
           "members": [
             {
-              "type": "CHOICE",
+              "type": "SEQ",
               "members": [
+                {
+                  "type": "PATTERN",
+                  "value": "\\s*"
+                },
                 {
                   "type": "SYMBOL",
                   "name": "argument"
+                },
+                {
+                  "type": "PATTERN",
+                  "value": "\\s*"
                 }
               ]
             },
@@ -543,7 +616,7 @@
           "type": "REPEAT",
           "content": {
             "type": "SYMBOL",
-            "name": "_command_invocation"
+            "name": "_untrimmed_command_invocation"
           }
         },
         {
@@ -559,6 +632,13 @@
           "type": "SYMBOL",
           "name": "function"
         },
+        {
+          "type": "REPEAT",
+          "content": {
+            "type": "PATTERN",
+            "value": "[\\t ]"
+          }
+        },
         {
           "type": "STRING",
           "value": "("
@@ -567,7 +647,7 @@
           "type": "REPEAT",
           "content": {
             "type": "SYMBOL",
-            "name": "argument"
+            "name": "_untrimmed_argument"
           }
         },
         {
@@ -583,6 +663,13 @@
           "type": "SYMBOL",
           "name": "endfunction"
         },
+        {
+          "type": "REPEAT",
+          "content": {
+            "type": "PATTERN",
+            "value": "[\\t ]"
+          }
+        },
         {
           "type": "STRING",
           "value": "("
@@ -591,7 +678,7 @@
           "type": "REPEAT",
           "content": {
             "type": "SYMBOL",
-            "name": "argument"
+            "name": "_untrimmed_argument"
           }
         },
         {
@@ -611,7 +698,7 @@
           "type": "REPEAT",
           "content": {
             "type": "SYMBOL",
-            "name": "_command_invocation"
+            "name": "_untrimmed_command_invocation"
           }
         },
         {
@@ -627,6 +714,13 @@
           "type": "SYMBOL",
           "name": "macro"
         },
+        {
+          "type": "REPEAT",
+          "content": {
+            "type": "PATTERN",
+            "value": "[\\t ]"
+          }
+        },
         {
           "type": "STRING",
           "value": "("
@@ -635,7 +729,7 @@
           "type": "REPEAT",
           "content": {
             "type": "SYMBOL",
-            "name": "argument"
+            "name": "_untrimmed_argument"
           }
         },
         {
@@ -651,6 +745,13 @@
           "type": "SYMBOL",
           "name": "endmacro"
         },
+        {
+          "type": "REPEAT",
+          "content": {
+            "type": "PATTERN",
+            "value": "[\\t ]"
+          }
+        },
         {
           "type": "STRING",
           "value": "("
@@ -659,7 +760,7 @@
           "type": "REPEAT",
           "content": {
             "type": "SYMBOL",
-            "name": "argument"
+            "name": "_untrimmed_argument"
           }
         },
         {
@@ -679,7 +780,7 @@
           "type": "REPEAT",
           "content": {
             "type": "SYMBOL",
-            "name": "_command_invocation"
+            "name": "_untrimmed_command_invocation"
           }
         },
         {
@@ -695,6 +796,13 @@
           "type": "SYMBOL",
           "name": "identifier"
         },
+        {
+          "type": "REPEAT",
+          "content": {
+            "type": "PATTERN",
+            "value": "[\\t ]"
+          }
+        },
         {
           "type": "STRING",
           "value": "("
@@ -703,7 +811,7 @@
           "type": "REPEAT",
           "content": {
             "type": "SYMBOL",
-            "name": "argument"
+            "name": "_untrimmed_argument"
           }
         },
         {
@@ -741,6 +849,19 @@
         }
       ]
     },
+    "_untrimmed_command_invocation": {
+      "type": "CHOICE",
+      "members": [
+        {
+          "type": "PATTERN",
+          "value": "\\s"
+        },
+        {
+          "type": "SYMBOL",
+          "name": "_command_invocation"
+        }
+      ]
+    },
     "if": {
       "type": "PATTERN",
       "value": "[iI][fF]"
@@ -799,10 +920,6 @@
     }
   },
   "extras": [
-    {
-      "type": "PATTERN",
-      "value": "[\\s\\n\\r]"
-    },
     {
       "type": "SYMBOL",
       "name": "bracket_comment"