Add support for structured headers for SDOs

src/Clause.ts

@@ -88,7 +88,6 @@ export default class Clause extends Builder {
     // if we find such a DL, treat this as a structured header
 
     const type = this.node.getAttribute('type');
-    this.node.removeAttribute('type'); // TODO maybe leave it in; this is just to minimize the diff
 
     const { name, formattedHeader, formattedParams } = parseStructuredHeaderH1(this.spec, header);
     if (type === 'numeric method' && name != null && !name.includes('::')) {
@@ -101,7 +100,11 @@ export default class Clause extends Builder {
         nodeRelativeColumn: 1,
       });
     }
-    if (formattedHeader != null) {
+    if (type === 'sdo' && (formattedHeader ?? header.innerHTML).includes('(')) {
+      // SDOs are rendered without parameter lists in the header, for the moment
+      const currentHeader = formattedHeader ?? header.innerHTML;
+      header.innerHTML = currentHeader.substring(0, currentHeader.indexOf('(')).trim();
+    } else if (formattedHeader != null) {
       header.innerHTML = formattedHeader;
     }
 
@@ -132,6 +135,8 @@ export default class Clause extends Builder {
       type != null &&
       [
         'abstract operation',
+        'sdo',
+        'syntax-directed operation',
         'host-defined abstract operation',
         'implementation-defined abstract operation',
         'numeric method',

src/Spec.ts

@@ -1235,7 +1235,9 @@ ${await utils.readFile(path.join(__dirname, '../js/multipage.js'))}
       }
     }
 
-    const sdos = this.doc.querySelectorAll('emu-clause[type=sdo]');
+    const sdos = this.doc.querySelectorAll(
+      'emu-clause[type=sdo],emu-clause[type="syntax-directed operation"]'
+    );
     outer: for (const sdo of sdos) {
       let header: Element | undefined;
       for (const child of sdo.children) {

src/header-parser.ts

@@ -13,7 +13,9 @@ export function parseStructuredHeaderH1(
     headerText = headerText.substring(prefix[0].length);
   }
 
-  const parsed = headerText.match(/^(?<beforeParams>\s*(?<name>[^(\s]+)\s*\()(?<params>.*)\)\s*$/s);
+  const parsed = headerText.match(
+    /^(?<beforeParams>\s*(?<name>[^(\s]+)\s*)(?:\((?<params>.*)\)\s*)?$/s
+  );
   if (parsed == null) {
     spec.warn({
       type: 'contents',
@@ -28,7 +30,7 @@ export function parseStructuredHeaderH1(
 
   type Param = { name: string; type: string | null };
   const name = parsed.groups!.name;
-  let paramText = parsed.groups!.params;
+  let paramText = parsed.groups!.params ?? '';
   const params: Array<Param> = [];
   const optionalParams: Array<Param> = [];
   let formattedHeader = null;
@@ -49,6 +51,7 @@ export function parseStructuredHeaderH1(
         return (
           (prefix?.[0].length ?? 0) +
           parsed!.groups!.beforeParams.length +
+          1 + // `beforeParams` does not include the leading `(`
           (offset - line.length) + // we've already updated offset to include line.length at this point
           index + // to account for the `\n`s eaten by the .split
           line.match(/^\s*/)![0].length
@@ -265,7 +268,8 @@ export function formatPreamble(
 ): Array<Element> {
   const para = spec.doc.createElement('p');
   const paras = [para];
-  switch (type?.toLowerCase()) {
+  type = (type ?? '').toLowerCase();
+  switch (type) {
     case 'numeric method':
     case 'abstract operation': {
       // TODO tests (for each type of parametered thing) which have HTML in the parameter type
@@ -280,6 +284,11 @@ export function formatPreamble(
       para.innerHTML += `The implementation-defined abstract operation ${name} takes ${formattedParams}.`;
       break;
     }
+    case 'sdo':
+    case 'syntax-directed operation': {
+      para.innerHTML += `The syntax-directed operation ${name} takes ${formattedParams}.`;
+      break;
+    }
     case 'internal method':
     case 'concrete method': {
       if (_for == null) {
@@ -326,23 +335,30 @@ export function formatPreamble(
       para.append(' ', ...description.childNodes);
     }
   }
+  const isSdo = type === 'sdo' || type === 'syntax-directed operation';
+  const lastSentence = isSdo
+    ? 'It is defined piecewise over the following productions:'
+    : 'It performs the following steps when called:';
   let next = dl.nextElementSibling;
   while (next != null && next.tagName === 'EMU-NOTE') {
     next = next.nextElementSibling;
   }
-  if (next?.tagName == 'EMU-ALG' && !next.hasAttribute('replaces-step')) {
+  if (
+    (isSdo && next?.tagName === 'EMU-GRAMMAR') ||
+    (!isSdo && next?.tagName === 'EMU-ALG' && !next.hasAttribute('replaces-step'))
+  ) {
     if (paras.length > 1 || next !== dl.nextElementSibling) {
       const whitespace = next.previousSibling;
       const after = spec.doc.createElement('p');
-      after.append('It performs the following steps when called:');
+      after.append(lastSentence);
       next.parentElement!.insertBefore(after, next);
 
       // fix up the whitespace in the generated HTML
       if (whitespace?.nodeType === 3 /* TEXT_NODE */ && /^\s+$/.test(whitespace.nodeValue!)) {
         next.parentElement!.insertBefore(whitespace.cloneNode(), next);
       }
     } else {
-      para.append(' It performs the following steps when called:');
+      para.append(' ' + lastSentence);
     }
   }
   return paras;

test/baselines/generated-reference/structured-headers.html

@@ -1,7 +1,7 @@
 <!doctype html>
 <head><meta charset="utf-8"></head><body><div id="spec-container">
 
-<emu-clause id="sec-exampleao" aoid="ExampleAO">
+<emu-clause id="sec-exampleao" type="abstract operation" aoid="ExampleAO">
   <h1><span class="secnum">1</span> ExampleAO ( param [ , param2 ] )</h1>
   <p>The abstract operation ExampleAO takes argument param (an <emu-xref href="#integer"><a href="https://tc39.es/ecma262/#integer">integer</a></emu-xref>) and optional argument param2 (a String). It is an example.</p>
   <emu-note><span class="note">Note</span><div class="note-contents">
@@ -10,4 +10,41 @@ <h1><span class="secnum">1</span> ExampleAO ( param [ , param2 ] )</h1>
   <p>It performs the following steps when called:</p>
   <emu-alg><ol><li>Algorithm steps go here.</li></ol></emu-alg>
 </emu-clause>
+
+<emu-clause id="sec-example-grammar">
+  <h1><span class="secnum">2</span> This and That</h1>
+  <emu-grammar type="definition"><emu-production name="PrimaryExpression" id="prod-PrimaryExpression">
+    <emu-nt><a href="#prod-PrimaryExpression">PrimaryExpression</a></emu-nt> <emu-geq>:</emu-geq> <emu-rhs a="jo4mwtvh" id="prod-Joc9-5i9"><emu-t>this</emu-t></emu-rhs>
+    <emu-rhs a="_ditw-4g" id="prod-QC7Z2SC4"><emu-t>that</emu-t></emu-rhs>
+</emu-production>
+</emu-grammar>
+  <emu-clause id="sec-isthis" type="sdo" aoid="IsThis">
+    <h1><span class="secnum">2.1</span> IsThis</h1>
+    <p>The <emu-xref href="#sec-algorithm-conventions-syntax-directed-operations"><a href="https://tc39.es/ecma262/#sec-algorithm-conventions-syntax-directed-operations">syntax-directed operation</a></emu-xref> IsThis takes no arguments. It is an example. It is defined piecewise over the following productions:</p>
+    <emu-grammar><emu-production name="PrimaryExpression" collapsed="">
+    <emu-nt><a href="#prod-PrimaryExpression">PrimaryExpression</a></emu-nt> <emu-geq>:</emu-geq> <emu-rhs a="jo4mwtvh" id="prod-oDKnYE2s"><emu-t>this</emu-t></emu-rhs>
+</emu-production>
+</emu-grammar>
+    <emu-alg><ol><li>Return <emu-val>true</emu-val>.</li></ol></emu-alg>
+    <emu-grammar><emu-production name="PrimaryExpression" collapsed="">
+    <emu-nt><a href="#prod-PrimaryExpression">PrimaryExpression</a></emu-nt> <emu-geq>:</emu-geq> <emu-rhs a="_ditw-4g" id="prod-_G-zA61Z"><emu-t>that</emu-t></emu-rhs>
+</emu-production>
+</emu-grammar>
+    <emu-alg><ol><li>Return <emu-val>false</emu-val>.</li></ol></emu-alg>
+  </emu-clause>
+  <emu-clause id="sec-isthat" type="sdo" aoid="IsThat">
+    <h1><span class="secnum">2.2</span> IsThat</h1>
+    <p>The <emu-xref href="#sec-algorithm-conventions-syntax-directed-operations"><a href="https://tc39.es/ecma262/#sec-algorithm-conventions-syntax-directed-operations">syntax-directed operation</a></emu-xref> IsThat takes argument <var>ignored</var> (an example). It is defined piecewise over the following productions:</p>
+    <emu-grammar><emu-production name="PrimaryExpression" collapsed="">
+    <emu-nt><a href="#prod-PrimaryExpression">PrimaryExpression</a></emu-nt> <emu-geq>:</emu-geq> <emu-rhs a="jo4mwtvh" id="prod-QDgm4qgx"><emu-t>this</emu-t></emu-rhs>
+</emu-production>
+</emu-grammar>
+    <emu-alg><ol><li>Return <emu-val>false</emu-val>.</li></ol></emu-alg>
+    <emu-grammar><emu-production name="PrimaryExpression" collapsed="">
+    <emu-nt><a href="#prod-PrimaryExpression">PrimaryExpression</a></emu-nt> <emu-geq>:</emu-geq> <emu-rhs a="_ditw-4g" id="prod-6qiWXvFG"><emu-t>that</emu-t></emu-rhs>
+</emu-production>
+</emu-grammar>
+    <emu-alg><ol><li>Return <emu-val>true</emu-val>.</li></ol></emu-alg>
+  </emu-clause>
+</emu-clause>
 </div></body>

test/baselines/sources/structured-headers.html

@@ -22,3 +22,44 @@ <h1>
     1. Algorithm steps go here.
   </emu-alg>
 </emu-clause>
+
+<emu-clause id="sec-example-grammar">
+  <h1>This and That</h1>
+  <emu-grammar type="definition">
+    PrimaryExpression :
+      `this`
+      `that`
+  </emu-grammar>
+  <emu-clause id="sec-isthis" type="sdo">
+    <h1>IsThis</h1>
+    <dl class='header'>
+      <dt>description</dt>
+      <dd>It is an example.</dd>
+    </dl>
+    <emu-grammar>PrimaryExpression : `this`</emu-grammar>
+    <emu-alg>
+      1. Return *true*.
+    </emu-alg>
+    <emu-grammar>PrimaryExpression : `that`</emu-grammar>
+    <emu-alg>
+      1. Return *false*.
+    </emu-alg>
+  </emu-clause>
+  <emu-clause id="sec-isthat" type="sdo">
+    <h1>
+      IsThat (
+        _ignored_: an example,
+      )
+    </h1>
+    <dl class='header'>
+    </dl>
+    <emu-grammar>PrimaryExpression : `this`</emu-grammar>
+    <emu-alg>
+      1. Return *false*.
+    </emu-alg>
+    <emu-grammar>PrimaryExpression : `that`</emu-grammar>
+    <emu-alg>
+      1. Return *true*.
+    </emu-alg>
+  </emu-clause>
+</emu-clause>