Skip to content

~95% on SimpleQA (e.g. Qwen3.6-27B on a 3090). Supports all local and cloud LLMs (llama.cpp, Ollama, Google, ...). 10+ search engines - arXiv, PubMed, your private documents. Everything Local & Encrypted.

chore(security): SSRF housekeeping cleanups (#3931)

LearningCircuitfbc1a442dafe
diff --git a/src/local_deep_research/web_search_engines/engines/search_engine_searxng.py b/src/local_deep_research/web_search_engines/engines/search_engine_searxng.py
index 593a361c65..d3de802368 100644
--- a/src/local_deep_research/web_search_engines/engines/search_engine_searxng.py
+++ b/src/local_deep_research/web_search_engines/engines/search_engine_searxng.py
@@ -352,7 +352,12 @@ class SearXNGSearchEngine(BaseSearchEngine):
url = ""
if url_element and url_element.has_attr("href"):
- url = str(url_element["href"])
+ # .strip() is defensive against HTML scrapes
+ # that include surrounding whitespace; the
+ # SSRF validator strips internally too, but
+ # cleaning here keeps URLs clean for logs
+ # and downstream consumers.
+ url = str(url_element["href"]).strip()
elif url_element:
url = url_element.get_text(strip=True)
@@ -457,7 +462,9 @@ class SearXNGSearchEngine(BaseSearchEngine):
previews = []
for i, result in enumerate(results):
title = result.get("title", "")
- url = result.get("url", "")
+ # .strip() is defensive against API responses with surrounding
+ # whitespace; the SSRF validator strips internally too.
+ url = (result.get("url") or "").strip()
content = result.get("content", "")
preview = {
@@ -540,7 +547,7 @@ class SearXNGSearchEngine(BaseSearchEngine):
formatted_results.append(
{
"title": result.get("title", ""),
- "link": result.get("url", ""),
+ "link": (result.get("url") or "").strip(),
"snippet": result.get("content", ""),
}
)
diff --git a/src/local_deep_research/web_search_engines/engines/search_engine_tavily.py b/src/local_deep_research/web_search_engines/engines/search_engine_tavily.py
index c722d01919..f9308c212d 100644
--- a/src/local_deep_research/web_search_engines/engines/search_engine_tavily.py
+++ b/src/local_deep_research/web_search_engines/engines/search_engine_tavily.py
@@ -153,14 +153,18 @@ class TavilySearchEngine(BaseSearchEngine):
# Format results as previews
previews = []
for i, result in enumerate(results):
+ # .strip() is defensive against API responses with
+ # surrounding whitespace; the SSRF validator strips
+ # internally too.
+ url = (result.get("url") or "").strip()
preview = {
- "id": result.get("url", str(i)), # Use URL as ID
+ "id": url or str(i), # Use URL as ID
"title": result.get("title", ""),
- "link": result.get("url", ""),
+ "link": url,
"snippet": result.get(
"content", ""
), # Tavily calls it "content"
- "displayed_link": result.get("url", ""),
+ "displayed_link": url,
"position": i,
}