From 8c9a182505315854301394a88535099f652f904e Mon Sep 17 00:00:00 2001
From: SyedaAnshrahGillani <syedaanshrah16@gmail.com>
Date: Fri, 25 Jul 2025 15:31:54 +0500
Subject: [PATCH] feat: Make webpage summarization timeout configurable

This commit introduces a configurable timeout for the webpage summarization process.
Previously, the `summarize_webpage` function in `utils.py` had a hardcoded 60-second timeout.
This change allows users to specify a custom timeout via the `summarization_timeout`
field in the `Configuration` class, providing more flexibility for varying content lengths
and network conditions.

- Added `summarization_timeout` field to `Configuration` in `configuration.py`.
- Modified `summarize_webpage` in `utils.py` to accept and use the new timeout.
- Updated `tavily_search` to pass the configurable timeout to `summarize_webpage`.
---
 src/open_deep_research/configuration.py | 10 ++++++++++
 src/open_deep_research/utils.py         |  5 +++--
 2 files changed, 13 insertions(+), 2 deletions(-)

diff --git a/src/open_deep_research/configuration.py b/src/open_deep_research/configuration.py
index b78b991d9..86823fcc2 100644
--- a/src/open_deep_research/configuration.py
+++ b/src/open_deep_research/configuration.py
@@ -128,6 +128,16 @@ class Configuration(BaseModel):
             }
         }
     )
+    summarization_timeout: float = Field(
+        default=60.0,
+        metadata={
+            "x_oap_ui_config": {
+                "type": "number",
+                "default": 60.0,
+                "description": "Timeout in seconds for webpage summarization"
+            }
+        }
+    )
     research_model: str = Field(
         default="openai:gpt-4.1",
         metadata={
diff --git a/src/open_deep_research/utils.py b/src/open_deep_research/utils.py
index d79edd2b4..c1295c4cb 100644
--- a/src/open_deep_research/utils.py
+++ b/src/open_deep_research/utils.py
@@ -74,6 +74,7 @@ async def noop():
         noop() if not result.get("raw_content") else summarize_webpage(
             summarization_model, 
             result['raw_content'][:max_char_to_include],
+            timeout=configurable.summarization_timeout
         )
         for result in unique_results.values()
     ]
@@ -108,11 +109,11 @@ async def tavily_search_async(search_queries, max_results: int = 5, topic: Liter
     search_docs = await asyncio.gather(*search_tasks)
     return search_docs
 
-async def summarize_webpage(model: BaseChatModel, webpage_content: str) -> str:
+async def summarize_webpage(model: BaseChatModel, webpage_content: str, timeout: float = 60.0) -> str:
     try:
         summary = await asyncio.wait_for(
             model.ainvoke([HumanMessage(content=summarize_webpage_prompt.format(webpage_content=webpage_content, date=get_today_str()))]),
-            timeout=60.0
+            timeout=timeout
         )
         return f"""<summary>\n{summary.summary}\n</summary>\n\n<key_excerpts>\n{summary.key_excerpts}\n</key_excerpts>"""
     except (asyncio.TimeoutError, Exception) as e: