content reference parser - parse references on new lines correctly

elastic · Feb 6, 2025 · 015ac28 · 015ac28
1 parent 55ef292
commit 015ac28
Show file tree

Hide file tree

Showing 2 changed files with 27 additions and 6 deletions.
diff --git a/...solution/public/assistant/get_comments/content_reference/content_reference_parser.test.ts b/...solution/public/assistant/get_comments/content_reference/content_reference_parser.test.ts
@@ -17,7 +17,8 @@ describe('ContentReferenceParser', () => {
 A furry friend, from head to toes.{reference(ccaSI)}
 Loyal companion, always near,{reference(ccaSI)}
 Chasing squirrels, full of cheer.{reference(ccaSI)}
-A paw to hold, a gentle nudge,{reference(ccaSI)}
+A paw to hold, a gentle nudge,
+{reference(ccaSI)}
 A furry alarm, a playful judge.{reference(ccaSI)}
 From golden retrievers to tiny Chihuahuas,{reference(ccaSI)}
 Their love's a gift, that always conquers.{reference(ccaSI)}
@@ -29,6 +30,24 @@ Their love's a beacon, shining bright.{reference(ccaSI)}`) as Parent;
         (child) => (child.type as string) === 'contentReference'
       )
     ).toHaveLength(10);
+    expect(file.children[0].children).toEqual(
+      expect.arrayContaining([
+        expect.objectContaining({ type: 'text', value: '\nA paw to hold, a gentle nudge,\n' }),
+      ])
+    );
+  });
+
+  it('extracts reference after linebreak', async () => {
+    const file = unified().use([[markdown, {}], ContentReferenceParser]).parse(`First line
+{reference(FTQJp)}
+`) as Parent;
+
+    expect(file.children[0].children).toEqual(
+      expect.arrayContaining([
+        expect.objectContaining({ type: 'text', value: 'First line\n' }),
+        expect.objectContaining({ type: 'contentReference' }),
+      ])
+    );
   });
 
   it('eats empty content reference', async () => {

diff --git a/...rity_solution/public/assistant/get_comments/content_reference/content_reference_parser.ts b/...rity_solution/public/assistant/get_comments/content_reference/content_reference_parser.ts
@@ -17,9 +17,9 @@ export interface ContentReferenceNode extends Node {
   contentReferenceBlock: ContentReferenceBlock;
 }
 
-/**
- * Parses `{reference(contentReferenceId)}` or ` {reference(contentReferenceId)}` (notice space prefix) into ContentReferenceNode
- */
+/** Matches `{reference` and ` {reference(` */
+const REFERENCE_START_PATTERN = '\\u0020?\\{reference';
+
 export const ContentReferenceParser: Plugin = function ContentReferenceParser() {
   const Parser = this.Parser;
   const tokenizers = Parser.prototype.inlineTokenizers;
@@ -33,7 +33,7 @@ export const ContentReferenceParser: Plugin = function ContentReferenceParser()
     value,
     silent
   ) {
-    const [match] = value.match(/^\s?{reference/) || [];
+    const [match] = value.match(new RegExp(`^${REFERENCE_START_PATTERN}`)) || [];
 
     if (!match) return false;
 
@@ -115,7 +115,9 @@ export const ContentReferenceParser: Plugin = function ContentReferenceParser()
   tokenizeCustomCitation.notInLink = true;
 
   tokenizeCustomCitation.locator = (value, fromIndex) => {
-    return 1 + (value.substring(fromIndex).match(/\s?{reference/)?.index ?? -2);
+    return (
+      1 + (value.substring(fromIndex).match(new RegExp(`${REFERENCE_START_PATTERN}`))?.index ?? -2)
+    );
   };
 
   tokenizers.contentReference = tokenizeCustomCitation;