Correctly parse multiple whitespace characters (#1469)

martijnversluis · Dec 1, 2024 · fad247b · fad247b
1 parent 5c1a631
commit fad247b
Show file tree

Hide file tree

Showing 4 changed files with 21 additions and 3 deletions.
diff --git a/script/debug_parser.ts b/script/debug_parser.ts
@@ -14,10 +14,16 @@ const grammarFile = `${parserFolder}/grammar.pegjs`;
 const helpersFile = `${parserFolder}/helpers.ts`;
 const chordGrammarFile = './src/parser/chord/base_grammar.pegjs';
 const chordSuffixGrammarFile = './src/parser/chord/suffix_grammar.pegjs';
+const whitespaceGrammarFile = './src/parser/whitespace_grammar.pegjs';
+const sectionsGrammarFile = './src/parser/chord_pro/sections_grammar.pegjs';
+const chordDefinitionGrammarFile = './src/parser/chord_definition/grammar.pegjs';
 
 const parserGrammar = fs.readFileSync(grammarFile, 'utf8');
 const chordGrammar = skipChordGrammar ? '' : fs.readFileSync(chordGrammarFile);
 const chordSuffixGrammar = fs.readFileSync(chordSuffixGrammarFile);
+const whitespaceGrammar = fs.readFileSync(whitespaceGrammarFile);
+const sectionsGrammar = fs.readFileSync(sectionsGrammarFile);
+const chordDefinitionGrammar = fs.readFileSync(chordDefinitionGrammarFile);
 
 const result = esbuild.buildSync({
   bundle: true,
@@ -33,6 +39,9 @@ const parserSource = [
   parserGrammar,
   chordGrammar,
   chordSuffixGrammar,
+  chordDefinitionGrammar,
+  sectionsGrammar,
+  whitespaceGrammar,
 ].join('\n\n');
 
 async function run() {

diff --git a/src/parser/whitespace.grammar → src/parser/whitespace_grammar.pegjs b/src/parser/whitespace.grammar → src/parser/whitespace_grammar.pegjs
@@ -8,7 +8,7 @@ WhitespaceCharacter
   = [ \t\n\r]
 
 Space "space"
-  = [ \t]+
+  = $([ \t]+)
 
 NewLine
   = CarriageReturn / LineFeed / CarriageReturnLineFeed

diff --git a/test/parser/chord_pro_parser.test.ts b/test/parser/chord_pro_parser.test.ts
@@ -76,6 +76,15 @@ describe('ChordProParser', () => {
     expect(song.lines[0].items[0]).toBeTag('comment', 'Some {comment}');
   });
 
+  it('correctly parses multiple whitespace characters', () => {
+    const chordSheet = '[C]Let it be         ';
+    const song = new ChordProParser().parse(chordSheet);
+    const { items } = song.lines[0];
+
+    expect(items[0]).toBeChordLyricsPair('C', 'Let ');
+    expect(items[1]).toBeChordLyricsPair('', 'it be         ');
+  });
+
   it('parses directive with empty value', () => {
     const song = new ChordProParser().parse('{c: }');
     expect(song.lines[0].items[0]).toBeTag('comment', '');

diff --git a/unibuild.ts b/unibuild.ts
@@ -70,7 +70,7 @@ unibuild((u: Builder) => {
       'src/parser/chord_pro/grammar.pegjs',
       'src/parser/chord_definition/grammar.pegjs',
       sectionsGrammar,
-      'src/parser/whitespace.grammar',
+      'src/parser/whitespace_grammar.pegjs',
     ],
     outfile: 'src/parser/chord_pro/peg_parser.ts',
     build: ({ release }: BuildOptions, ...grammars: string[]) => {
@@ -82,7 +82,7 @@ unibuild((u: Builder) => {
   const chordDefinitionParser = u.asset('chordDefinitionParser', {
     input: [
       'src/parser/chord_definition/grammar.pegjs',
-      'src/parser/whitespace.grammar',
+      'src/parser/whitespace_grammar.pegjs',
     ],
     outfile: 'src/parser/chord_definition/peg_parser.ts',
     build: ({ release }: BuildOptions, ...grammars: string[]) => (