feat: Handle example comments for default start inside the grammar

Author: pdubroy

packages/ohm-js/extras/extractExamples.js

@@ -9,13 +9,7 @@ export const grammarsSource = String.raw`
   // Example:
   //+ "//+ \"x\"\nG {\n//- \"\"\nstart = \"x\"}"
   OhmWithExamples <: Ohm {
-    // The default start rule for Ohm is 'Grammars', which is syntactic rule.
-    // When the start rule is a syntactic rule, there's no way to get access to
-    // leading space (including comments). So, for this grammar to be useful,
-    // you have to explicit use this rule as the start rule.
-    grammarsWithExamples = (exampleComments applySyntactic<Grammar>)*
-
-    Grammar := ident SuperGrammar? "{" (#exampleComments Rule)* "}"
+    Grammar := ident SuperGrammar? "{" (#exampleComments Rule)* #exampleComments "}"
 
     exampleComments = (spacesNoExampleComment exampleComment)*
 
@@ -79,19 +73,10 @@ const semantics = grammars.OhmWithExamples.createSemantics().addOperation('hasEx
 });
 
 semantics.addOperation('examples', {
-  grammarsWithExamples(exampleCommentsIter, grammarIter) {
-    const result = [];
-    for (const [i, child] of Object.entries(grammarIter.children)) {
-      if (exampleCommentsIter.hasExamples()) {
-        const defaultExamples = exampleCommentsIter.child(i).examples();
-        const grammar = child.grammarName();
-        result.push(...defaultExamples.map(ex => ({...ex, grammar, rule: ''})));
-      }
-      result.push(...child.examples());
-    }
-    return result;
+  Grammars(grammarIter) {
+    return grammarIter.children.flatMap(c => c.examples());
   },
-  Grammar(name, _, _open, exampleCommentsIter, ruleIter, _close) {
+  Grammar(name, _, _open, exampleCommentsIter, ruleIter, trailingCommentsIter, _close) {
     const result = [];
     const grammar = this.grammarName();
     for (let i = 0; i < ruleIter.numChildren; i++) {
@@ -103,6 +88,10 @@ semantics.addOperation('examples', {
 
       result.push(...augmentedExamples);
     }
+    if (trailingCommentsIter.hasExamples()) {
+      const defaultExamples = trailingCommentsIter.examples();
+      result.push(...defaultExamples.map(ex => ({...ex, grammar, rule: ''})));
+    }
     return result;
   },
   exampleComments(_, commentIter) {
@@ -128,7 +117,7 @@ semantics.addOperation('examples', {
 });
 
 semantics.addOperation('grammarName', {
-  Grammar(name, _, _open, exampleCommentsIter, ruleIter, _close) {
+  Grammar(name, _, _open, exampleCommentsIter, ruleIter, trailingCommentsIter, _close) {
     return name.sourceString;
   },
 });
@@ -152,7 +141,7 @@ semantics.addOperation('ruleName', {
  * @return {[Example]}
  */
 export function extractExamples(grammarsDef) {
-  const matchResult = grammars.OhmWithExamples.match(grammarsDef, 'grammarsWithExamples');
+  const matchResult = grammars.OhmWithExamples.match(grammarsDef);
   if (matchResult.failed()) {
     throw new Error(matchResult.message);
   }

packages/ohm-js/test/extras/test-extractExamples.js

@@ -35,35 +35,32 @@ test('simple positive examples', t => {
 
 test('examples for default start rule', t => {
   let examples = extractExamples(`
-    //+ "hey"
     G {
       //+ ""
       start = ""
+      //+ "hey"
     }
   `);
   t.deepEqual(examples, [
-    {grammar: 'G', rule: '', example: 'hey', shouldMatch: true},
     {grammar: 'G', rule: 'start', example: '', shouldMatch: true},
+    {grammar: 'G', rule: '', example: 'hey', shouldMatch: true},
   ]);
 
   examples = extractExamples(`
-    //+ "hey"
+    //+ ""
     G {
-      //+ ""
       start = ""
+      //+ "hey"
     }
   `);
-  t.deepEqual(examples, [
-    {grammar: 'G', rule: '', example: 'hey', shouldMatch: true},
-    {grammar: 'G', rule: 'start', example: '', shouldMatch: true},
-  ]);
+  t.deepEqual(examples, [{grammar: 'G', rule: '', example: 'hey', shouldMatch: true}]);
 });
 
-test('top-level whitespace', t => {
+test('whitespace in trailing example comments', t => {
   const expected = [{grammar: 'G', rule: '', example: '', shouldMatch: true}];
-  t.deepEqual(extractExamples('  //+ ""\n  G{}'), expected);
-  t.deepEqual(extractExamples('  //+ "" \nG{}'), expected);
-  t.deepEqual(extractExamples('\n\n//+ ""\n\nG{}'), expected);
+  t.deepEqual(extractExamples('G{  //+ ""\n  }'), expected);
+  t.deepEqual(extractExamples('G{  //+ "" \n}'), expected);
+  t.deepEqual(extractExamples('G{\n\n//+ ""\n\n}'), expected);
 });
 
 function getExamples(input) {