From 8e5367f9a829a320721f693210bb6187e9d29144 Mon Sep 17 00:00:00 2001
From: Han Xiao <han.xiao@jina.ai>
Date: Fri, 14 Feb 2025 21:33:00 +0800
Subject: [PATCH] fix: remove references tag in output

---
 src/agent.ts |  3 +--
 src/app.ts   | 28 ++++++++++++++++++++++------
 2 files changed, 23 insertions(+), 8 deletions(-)

diff --git a/src/agent.ts b/src/agent.ts
index dc66d75..b31a05c 100644
--- a/src/agent.ts
+++ b/src/agent.ts
@@ -40,7 +40,7 @@ function getSchema(allowReflect: boolean, allowRead: boolean, allowAnswer: boole
   if (allowAnswer) {
     actions.push("answer");
     properties.answer = z.string()
-      .describe(`Required when action='answer'. Must in ${languageStyle}`).optional();
+      .describe(`Required when action='answer'. Must in ${languageStyle}. Use markdown footnote syntax like [^1], [^2] to refer the corresponding reference item`).optional();
     properties.references = z.array(
       z.object({
         exactQuote: z.string().describe("Exact relevant quote from the document"),
@@ -219,7 +219,6 @@ ${allKeywords.join('\n')}
 <action-answer>
 - If <question> is a simple greeting, chit-chat, or general knowledge, provide the answer directly;
 - Must provide "references" and each must specify "exactQuote" and "url";
-- In the answer, use markdown footnote syntax like [^1], [^2] to refer to the references;
 - Responses must be definitive (no ambiguity, uncertainty, or disclaimers) and in the style of ${languageStyle};
 - Provide final response only when 100% certain;${allowReflect ? '\n- If doubts remain, use <action-reflect> instead' : ''}
 </action-answer>
diff --git a/src/app.ts b/src/app.ts
index 28e6e2b..ee365cb 100644
--- a/src/app.ts
+++ b/src/app.ts
@@ -33,20 +33,36 @@ function buildMdFromAnswer(answer: AnswerAction) {
   }
 
   const references = answer.references.map((ref, i) => {
-    const escapedQuote = ref.exactQuote
-      .replace(/([[\]_*`])/g, '\\$1')
-      .replace(/\n/g, ' ')
+    // Clean up the quote text
+    const cleanQuote = ref.exactQuote
+      // Remove HTML artifacts
+      .replace(/<[^>]+>/g, '')
+      // Remove multiple spaces
+      .replace(/\s+/g, ' ')
+      // Remove special characters and markdown conflicts
+      .replace(/[[\]_*`]/g, '')
+      // Clean up any remaining artifacts
+      .replace(/=+/g, '')
+      .replace(/_[^_]+_/g, '$1')
+      // Remove navigation artifacts
+      .replace(/arrow_drop_down/g, '')
+      .replace(/language/g, '')
+      // Trim whitespace
       .trim();
 
-    return `[^${i + 1}]: [${escapedQuote}](${ref.url})`;
+    // If the quote is too long, truncate it
+    const maxLength = 20;
+    const truncatedQuote = cleanQuote.length > maxLength
+      ? cleanQuote.substring(0, maxLength) + '...'
+      : cleanQuote;
+
+    return `[^${i + 1}]: [${truncatedQuote}](${ref.url})`;
   }).join('\n\n');
 
   return `
 ${answer.answer.replace(/\(REF_(\d+)\)/g, (_, num) => `[^${num}]`)}
 
-
 ${references}
-
 `.trim();
 }