wasm: update to latest liquid grammars, fix crash due to default start rule

Author: pdubroy

packages/wasm/src/Compiler.js

@@ -847,7 +847,12 @@ export class Compiler {
       Object.hasOwn(grammar.rules, name)
     );
     const rules = ownRuleNames.map(name => [name, lookUpRule(name)]);
-    rules.push(['spaces', lookUpRule('spaces')]); // Ensure 'spaces' is always present.
+
+    // Ensure the certain rules are always included. (The default start rule
+    // might be inherited from the supergrammar, so not there yet.)
+    for (const name of ['spaces', grammar.defaultStartRule]) {
+      rules.push([name, lookUpRule(name)]);
+    }
 
     const liftedTerminals = new IndexedSet();
 
@@ -1045,7 +1050,11 @@ export class Compiler {
 
           // Inline these. TODO: Handle this elsewhere.
           // We need this to avoid having >256 rules in the Liquid grammar.
-          if (['liquidRawTagImpl', 'liquidTagRule'].includes(ruleName)) {
+          if (
+            ['liquidRawTagImpl', 'liquidTagRule', 'anyExceptStar', 'anyExceptPlus'].includes(
+              ruleName
+            )
+          ) {
             return specialize(ir.substituteParams(ruleInfo.body, children));
           }
 

packages/wasm/test/data/liquid-html.ohm

@@ -237,7 +237,7 @@ Liquid <: Helpers {
   // because we'd otherwise positively match the following string
   // instead of falling back to the other rule:
   // {{ 'string' | some_filter }}
-  liquidVariable<delim> = liquidExpression<delim> liquidFilter<delim>* space* &delim
+  liquidVariable<delim> = liquidComplexExpression<delim> liquidFilter<delim>* space* &delim
 
   liquidExpression<delim> =
     | liquidString<delim>
@@ -246,6 +246,18 @@ Liquid <: Helpers {
     | liquidRange<delim>
     | liquidVariableLookup<delim>
 
+  liquidComplexExpression<delim> =
+    | liquidBooleanExpression<delim>
+    | liquidExpression<delim>
+
+  liquidBooleanExpression<delim> = booleanExpressionCondition<delim> listOf<booleanExpressionSubsequentCondition<delim>, conditionSeparator>
+
+  // This might over-capture things since the conditions below can contain any `liquidExpression`s.
+  // We will need to clean this up in CST. If we restrict the conditions to only contain comparisons and
+  // variable lookups, we can't support "truthy" expressions like `{{ some_var and 'this' }}`
+  booleanExpressionSubsequentCondition<delim> = space* logicalOperator space* (comparison<delim> | liquidExpression<delim>) space*
+  booleanExpressionCondition<delim> = comparison<delim> | liquidExpression<delim>
+
   liquidString<delim> = liquidSingleQuotedString<delim> | liquidDoubleQuotedString<delim>
   liquidSingleQuotedString<delim> = "'" anyExceptStar<("'"| delim)> "'"
   liquidDoubleQuotedString<delim> = "\"" anyExceptStar<("\""| delim)> "\""

packages/wasm/test/test-liquid-html.js

@@ -10,16 +10,21 @@ import {matchWithInput, unparse, toWasmGrammar} from './_helpers.js';
 const scriptRel = relPath => new URL(relPath, import.meta.url);
 const grammarSource = fs.readFileSync(scriptRel('data/liquid-html.ohm'), 'utf8');
 
-const liquid = ohm.grammars(grammarSource);
+const grammars = ohm.grammars(grammarSource);
+
+test('basic compilation', async t => {
+  Object.values(grammars).forEach(async g => await toWasmGrammar(g));
+  t.pass();
+});
 
 test('basic matching (small)', async t => {
   const input = `---
     layout: default
     ---
     {% assign year = page.started | date: '%Y' %}`;
-  t.is(liquid.LiquidHTML.match(input).succeeded(), true);
+  t.is(grammars.LiquidHTML.match(input).succeeded(), true);
 
-  const g = await toWasmGrammar(liquid.LiquidHTML);
+  const g = await toWasmGrammar(grammars.LiquidHTML);
   t.is(matchWithInput(g, input), 1);
   t.is(unparse(g), input);
 });
@@ -31,25 +36,25 @@ test('swatch.liquid', async t => {
     class="{% if x == 'x' %}x{% endif %}"
   {% endif %}
 >`;
-  const g = await toWasmGrammar(liquid.LiquidHTML);
+  const g = await toWasmGrammar(grammars.LiquidHTML);
   t.is(matchWithInput(g, input), 1);
 });
 
 test('html comment', async t => {
   const input = `{% if x %}
     <!-- x -->
   {% endif %}`;
-  const g = await toWasmGrammar(liquid.LiquidHTML);
+  const g = await toWasmGrammar(grammars.LiquidHTML);
   t.is(matchWithInput(g, input), 1);
 });
 
 test('book-review.liquid', async t => {
   const input = fs.readFileSync(scriptRel('data/book-review.liquid'), 'utf8');
   let start = performance.now();
-  t.is(liquid.LiquidHTML.match(input).succeeded(), true); // Trigger fillInputBuffer
+  t.is(grammars.LiquidHTML.match(input).succeeded(), true); // Trigger fillInputBuffer
   t.log(`Ohm.js: ${(performance.now() - start).toFixed(2)}ms`);
 
-  const g = await toWasmGrammar(liquid.LiquidHTML);
+  const g = await toWasmGrammar(grammars.LiquidHTML);
   start = performance.now();
   t.is(matchWithInput(g, input), 1);
   t.log(`Wasm: ${(performance.now() - start).toFixed(2)}ms`);