Start work on heredocs

binding.gyp

@@ -8,7 +8,8 @@
       ],
       "sources": [
         "src/parser.c",
-        "src/binding.cc"
+        "src/binding.cc",
+        "src/scanner.cc",
       ],
       "cflags_c": [
         "-std=c99",

corpus/commands.txt

@@ -101,6 +101,29 @@ cat a b 2> /dev/null
     (file_redirect (file_descriptor) (file_descriptor))
     (command_name))))
 
+===============================
+Heredoc redirects
+===============================
+
+node <<JS
+console.log("hi")
+JS
+
+bash -c <<JS
+echo hi
+JS
+
+---
+
+(program
+  (command (simple_command
+    (command_name)
+    (heredoc_redirect (heredoc))))
+  (command (simple_command
+    (command_name)
+    (argument)
+    (heredoc_redirect (heredoc)))))
+
 ===============================
 Variable expansions
 ===============================

grammar.js

@@ -3,6 +3,10 @@ module.exports = grammar({
 
   inline: $ => [$.control_operator],
 
+  externals: $ => [
+    $.heredoc
+  ],
+
   rules: {
     program: $ => repeat($.command),
 
@@ -29,9 +33,10 @@ module.exports = grammar({
           $.operator_expansion
         ))
       )),
-      repeat(
+      repeat(choice(
-        $.file_redirect
+        $.file_redirect,
-      )
+        $.heredoc_redirect
+      ))
     ),
 
     pipeline: $ => prec.left(seq(
@@ -83,6 +88,11 @@ module.exports = grammar({
       )
     ),
 
+    heredoc_redirect: $ => seq(
+      choice('<<', '<<-'),
+      $.heredoc
+    ),
+
     file_descriptor: $ => token(prec(1, /\d+/)),
 
     leading_word: $ => /[^\s=|;:{}]+/,

src/grammar.json

@@ -105,8 +105,17 @@
         {
           "type": "REPEAT",
           "content": {
+            "type": "CHOICE",
+            "members": [
+              {
                 "type": "SYMBOL",
                 "name": "file_redirect"
+              },
+              {
+                "type": "SYMBOL",
+                "name": "heredoc_redirect"
+              }
+            ]
           }
         }
       ]
@@ -349,6 +358,28 @@
         }
       ]
     },
+    "heredoc_redirect": {
+      "type": "SEQ",
+      "members": [
+        {
+          "type": "CHOICE",
+          "members": [
+            {
+              "type": "STRING",
+              "value": "<<"
+            },
+            {
+              "type": "STRING",
+              "value": "<<-"
+            }
+          ]
+        },
+        {
+          "type": "SYMBOL",
+          "name": "heredoc"
+        }
+      ]
+    },
     "file_descriptor": {
       "type": "TOKEN",
       "content": {
@@ -389,7 +420,12 @@
     }
   ],
   "conflicts": [],
-  "externals": [],
+  "externals": [
+    {
+      "type": "SYMBOL",
+      "name": "heredoc"
+    }
+  ],
   "inline": [
     "control_operator"
   ]

src/scanner.cc

@@ -0,0 +1,90 @@
+#include <tree_sitter/parser.h>
+#include <string>
+#include <cwctype>
+
+namespace {
+
+using std::wstring;
+
+enum TokenType {
+  HEREDOC
+};
+
+struct Scanner {
+  void skip(TSLexer *lexer) {
+    lexer->advance(lexer, true);
+  }
+
+  void advance(TSLexer *lexer) {
+    lexer->advance(lexer, false);
+  }
+
+  void reset() {}
+
+  bool serialize(TSExternalTokenState state) { return true; }
+
+  void deserialize(TSExternalTokenState state) {}
+
+  bool scan(TSLexer *lexer, const bool *valid_symbols) {
+    wstring heredoc_content;
+
+    while (iswalpha(lexer->lookahead)) {
+      heredoc_content += lexer->lookahead;
+      advance(lexer);
+    }
+
+    if (lexer->lookahead != '\n') return false;
+
+    wstring leading_word;
+
+    for (;;) {
+      advance(lexer);
+
+      while (iswalpha(lexer->lookahead)) {
+        leading_word += lexer->lookahead;
+        advance(lexer);
+      }
+
+      if (leading_word == heredoc_content || lexer->lookahead == '\0') break;
+      if (lexer->lookahead == '\n') leading_word.clear();
+    }
+
+    return true;
+  }
+};
+
+}
+
+extern "C" {
+
+void *tree_sitter_bash_external_scanner_create() {
+  return new Scanner();
+}
+
+bool tree_sitter_bash_external_scanner_scan(void *payload, TSLexer *lexer,
+                                            const bool *valid_symbols) {
+  Scanner *scanner = static_cast<Scanner *>(payload);
+  return scanner->scan(lexer, valid_symbols);
+}
+
+void tree_sitter_bash_external_scanner_reset(void *payload) {
+  Scanner *scanner = static_cast<Scanner *>(payload);
+  scanner->reset();
+}
+
+bool tree_sitter_bash_external_scanner_serialize(void *payload, TSExternalTokenState state) {
+  Scanner *scanner = static_cast<Scanner *>(payload);
+  return scanner->serialize(state);
+}
+
+void tree_sitter_bash_external_scanner_deserialize(void *payload, TSExternalTokenState state) {
+  Scanner *scanner = static_cast<Scanner *>(payload);
+  scanner->deserialize(state);
+}
+
+void tree_sitter_bash_external_scanner_destroy(void *payload) {
+  Scanner *scanner = static_cast<Scanner *>(payload);
+  delete scanner;
+}
+
+}