Fix distinction between numeric args and redirect file descriptors

corpus/commands.txt

@@ -23,7 +23,7 @@ cat -n file1.txt file2.txt
   (command (command_name) (argument) (argument) (argument)))
 
 ===============================
-Commands with string arguments
+Commands with quoted arguments
 ===============================
 
 echo "hello $(whoami), this is $(uname)"
@@ -37,6 +37,17 @@ echo 'hi'
     (command_substitution (command (command_name)))))
   (command (command_name) (single_quoted_argument)))
 
+===============================
+Commands with numeric arguments
+===============================
+
+exit 1
+
+---
+
+(program
+  (command (command_name) (argument)))
+
 ===================================
 Commands with environment variables
 ===================================
@@ -99,7 +110,7 @@ File redirects
 ===============================
 
 whoami > /dev/null
-cat a b 2> /dev/null
+cat a b > /dev/null
 2>&1 whoami
 
 ---
@@ -107,14 +118,14 @@ cat a b 2> /dev/null
 (program
   (command
     (command_name)
-    (file_redirect (file_name)))
+    (file_redirect (argument)))
   (command
     (command_name)
     (argument)
     (argument)
-    (file_redirect (file_descriptor) (file_name)))
+    (file_redirect (argument)))
   (command
-    (file_redirect (file_descriptor) (file_descriptor))
+    (file_redirect (file_descriptor) (argument))
     (command_name)))
 
 ===============================

grammar.js

@@ -7,7 +7,8 @@ module.exports = grammar({
     $._simple_heredoc,
     $._heredoc_beginning,
     $._heredoc_middle,
-    $._heredoc_end
+    $._heredoc_end,
+    $.file_descriptor
   ],
 
   extras: $ => [
@@ -171,11 +172,8 @@ module.exports = grammar({
 
     file_redirect: $ => seq(
       optional($.file_descriptor),
-      choice('<', '>', '<&', '>&'),
+      choice('<', '>', '>>', '&>', '&>>', '<&', '>&'),
-      choice(
+      $.value
-        $.file_descriptor,
-        rename($.word, 'file_name')
-      )
     ),
 
     heredoc_redirect: $ => seq(
@@ -196,8 +194,6 @@ module.exports = grammar({
       )
     ),
 
-    file_descriptor: $ => token(prec(1, /\d+/)),
-
     leading_word: $ => /[^"\\\s#=|;:{}()]+/,
 
     word: $ => /[^"#\\\s$<>{}&;()]+/,

src/grammar.json

@@ -646,6 +646,18 @@
               "type": "STRING",
               "value": ">"
             },
+            {
+              "type": "STRING",
+              "value": ">>"
+            },
+            {
+              "type": "STRING",
+              "value": "&>"
+            },
+            {
+              "type": "STRING",
+              "value": "&>>"
+            },
             {
               "type": "STRING",
               "value": "<&"
@@ -657,21 +669,8 @@
           ]
         },
         {
-          "type": "CHOICE",
+          "type": "SYMBOL",
-          "members": [
+          "name": "value"
-            {
-              "type": "SYMBOL",
-              "name": "file_descriptor"
-            },
-            {
-              "type": "RENAME",
-              "content": {
-                "type": "SYMBOL",
-                "name": "word"
-              },
-              "value": "file_name"
-            }
-          ]
         }
       ]
     },
@@ -739,17 +738,6 @@
         }
       ]
     },
-    "file_descriptor": {
-      "type": "TOKEN",
-      "content": {
-        "type": "PREC",
-        "value": 1,
-        "content": {
-          "type": "PATTERN",
-          "value": "\\d+"
-        }
-      }
-    },
     "leading_word": {
       "type": "PATTERN",
       "value": "[^\"\\\\\\s#=|;:{}()]+"
@@ -819,6 +807,10 @@
     {
       "type": "SYMBOL",
       "name": "_heredoc_end"
+    },
+    {
+      "type": "SYMBOL",
+      "name": "file_descriptor"
     }
   ],
   "inline": [

src/scanner.cc

@@ -11,6 +11,7 @@ enum TokenType {
   HEREDOC_BEGINNING,
   HEREDOC_MIDDLE,
   HEREDOC_END,
+  FILE_DESCRIPTOR,
 };
 
 struct Scanner {
@@ -74,23 +75,35 @@ struct Scanner {
   bool scan(TSLexer *lexer, const bool *valid_symbols) {
     if (valid_symbols[HEREDOC_MIDDLE]) {
       return scan_heredoc_content(lexer, HEREDOC_MIDDLE, HEREDOC_END);
-    }
+    } else if (valid_symbols[HEREDOC_BEGINNING]) {
+      heredoc_identifier.clear();
+      while (iswalpha(lexer->lookahead)) {
+        heredoc_identifier += lexer->lookahead;
+        advance(lexer);
+      }
 
-    heredoc_identifier.clear();
+      if (lexer->lookahead != '\n') return false;
-    while (iswalpha(lexer->lookahead)) {
-      heredoc_identifier += lexer->lookahead;
       advance(lexer);
+
+      if (scan_heredoc_end_identifier(lexer)) {
+        lexer->result_symbol = SIMPLE_HEREDOC;
+        return true;
+      }
+
+      return scan_heredoc_content(lexer, HEREDOC_BEGINNING, SIMPLE_HEREDOC);
+    } else if (valid_symbols[FILE_DESCRIPTOR]) {
+      while (lexer->lookahead == ' ' || lexer->lookahead == '\t') skip(lexer);
+      if (iswdigit(lexer->lookahead)) {
+        advance(lexer);
+        while (iswdigit(lexer->lookahead)) advance(lexer);
+        if (lexer->lookahead == '>' || lexer->lookahead == '<') {
+          lexer->result_symbol = FILE_DESCRIPTOR;
+          return true;
+        }
+      }
     }
 
-    if (lexer->lookahead != '\n') return false;
+    return false;
-    advance(lexer);
-
-    if (scan_heredoc_end_identifier(lexer)) {
-      lexer->result_symbol = SIMPLE_HEREDOC;
-      return true;
-    }
-
-    return scan_heredoc_content(lexer, HEREDOC_BEGINNING, SIMPLE_HEREDOC);
   }
 
   wstring heredoc_identifier;