From d5efd2c13f1646ec40fe711c057c8d7401f70a2f Mon Sep 17 00:00:00 2001
From: Tim Li <tianleli@berkeley.edu>
Date: Mon, 30 Dec 2024 00:50:39 -0800
Subject: [PATCH] fix first page only parsing issue

---
 fastchat/serve/gradio_block_arena_vision.py | 11 +++++------
 1 file changed, 5 insertions(+), 6 deletions(-)

diff --git a/fastchat/serve/gradio_block_arena_vision.py b/fastchat/serve/gradio_block_arena_vision.py
index 09139a2ec..c6c6c0393 100644
--- a/fastchat/serve/gradio_block_arena_vision.py
+++ b/fastchat/serve/gradio_block_arena_vision.py
@@ -215,8 +215,7 @@ def wrap_pdfchat_query(query, document):
     document_context = f"""
     The following is the content of a document:
     {document}
-    Based on this document, answer the following question:
-    {query}
+    Based on this document, answer the user query.
     """
 
     reformatted_query_context = (
@@ -236,13 +235,13 @@ def parse_pdf(file_path):
     assert (
         "LLAMA_CLOUD_API_KEY" in os.environ
     ), "Make sure to specify LlamaParse API key."
-    document = LlamaParse(
+    documents = LlamaParse(
         result_type="markdown",
         verbose=True,
     ).load_data(file_path)
-    
-    assert len(document) > 0
-    output = document[0].text
+
+    assert len(documents) > 0
+    output = "\n".join([doc.text for doc in documents])
 
     return output