Merge pull request #119 from misdocumeno/warn-missing-loc-option

Warning about capture characters

Author: DmitrySoshnikov

src/bin/syntax.js

@@ -554,7 +554,10 @@ function getGrammar(grammarFile, mode) {
     return null;
   }
 
-  const grammarData = Grammar.dataFromGrammarFile(grammarFile, 'bnf');
+  const grammarData = Grammar.dataFromGrammarFile(grammarFile, {
+    grammarType: 'bnf',
+    useLocation: options.loc,
+  });
 
   // If explicit lexical grammar file was passed, use it.
   const lexGrammarData = getLexGrammarData(options);
@@ -584,7 +587,7 @@ function getLexGrammarData(options) {
 
   // If explicit lexical grammar file was passed, use it.
   if (options.lex) {
-    data = Grammar.dataFromGrammarFile(options.lex, 'lex');
+    data = Grammar.dataFromGrammarFile(options.lex, { grammarType: 'lex' });
   }
 
   if (options['ignore-whitespaces'] && !data) {

src/grammar/grammar.js

@@ -168,19 +168,41 @@ export default class Grammar {
    * for the specific options.
    */
   static fromGrammarFile(grammarFile, options = {}, grammarType = 'bnf') {
-    const grammarData = Grammar.dataFromGrammarFile(grammarFile, grammarType);
+    const grammarData = Grammar.dataFromGrammarFile(grammarFile, { grammarType });
     return Grammar.fromData(grammarData, options);
   }
 
   /**
    * Reads grammar file data. Supports reading `bnf`,
    * and `lex` grammars based on mode.
    */
-  static dataFromGrammarFile(grammarFile, grammarType = 'bnf') {
-    return Grammar.dataFromString(
-      fs.readFileSync(grammarFile, 'utf-8'),
-      grammarType
-    );
+  static dataFromGrammarFile(grammarFile, { grammarType = 'bnf', useLocation = false }) {
+    const grammar = fs.readFileSync(grammarFile, 'utf8');
+
+    // check if the bnf grammar contains location capture characters
+    if (grammarType === 'bnf' && !useLocation) {
+      const bnf = grammar
+        // lexer rules
+        .replace(/%lex[\n\s\S]*?\/lex/g, '')
+        // comments
+        .replace(/\/\*[\n\s\S]*?\*\//g, '')
+        .replace(/\/\/.*?\n/g, '')
+        //strings
+        .replace(/'(\\.|[^'\\])*'/g, '')
+        .replace(/"(\\.|[^"\\])*"/g, '')
+        .replace(/`(\\.|[^`\\])*`/g, '')
+        // module include
+        .replace(/%{[\n\s\S]*?%}/g, '');
+
+      if (/@\w+/.test(bnf)) {
+        console.info(colors.red(
+          'The grammar file contains location capture characters (@), which require the ' +
+          '"--loc" option, but it has not been provided. The generated parser will throw an error.'
+        ));
+      }
+    }
+
+    return Grammar.dataFromString(grammar, grammarType);
   }
 
   /**