fix: publication url detection

hieuddo · hieuddo · commit d2d7e098b34c · 2025-12-17T01:39:30.000+08:00
diff --git a/scripts/crawl-publications.ts b/scripts/crawl-publications.ts
@@ -309,7 +309,7 @@ function parsePublicationFromLi(liContent: string): Publication | null {
     const isPaperHost = PAPER_HOST_DOMAINS.some(domain => url.includes(domain));
     const isTitle = isPaperHost &&
       text.length > MIN_TITLE_LINK_TEXT_LENGTH &&
-      !text.toLowerCase().includes('code');
+      (!text.toLowerCase().includes('code') || text.length > 30);
 
     links.push({ text, url, isTitle });
   }

Original file line number	Diff line number	Diff line change
`@@ -309,7 +309,7 @@ function parsePublicationFromLi(liContent: string): Publication \| null {`
`309`	`309`	`const isPaperHost = PAPER_HOST_DOMAINS.some(domain => url.includes(domain));`
`310`	`310`	`const isTitle = isPaperHost &&`
`311`	`311`	`text.length > MIN_TITLE_LINK_TEXT_LENGTH &&`
`312`		`- !text.toLowerCase().includes('code');`
	`312`	`+ (!text.toLowerCase().includes('code') \|\| text.length > 30);`
`313`	`313`
`314`	`314`	`links.push({ text, url, isTitle });`
`315`	`315`	`}`