fix(context): guard SearchAPIRetriever against None raw_content

SearchAPIRetriever sliced page.get('raw_content', '')[:MAX], but the
default only applies when the key is absent. The scraper explicitly sets
raw_content=None for pages that failed to scrape, and slicing None raises
TypeError, crashing context retrieval. Coerce a missing/None raw_content
to '' before truncating (matching the defensive str() coercion already
used in compression.py).

(cherry picked from commit 2ffe6df27b)

gpt_researcher/context/retriever.py

@@ -23,7 +23,10 @@ class SearchAPIRetriever(BaseRetriever):
 
         docs = [
             Document(
-                page_content=page.get("raw_content", "")[:_MAX_CONTENT_CHARS],
+                # ``raw_content`` may be explicitly None (the scraper sets it to
+                # None for pages that failed to scrape), and slicing None raises
+                # TypeError. Coerce to a string before truncating.
+                page_content=(page.get("raw_content") or "")[:_MAX_CONTENT_CHARS],
                 metadata={
                     "title": page.get("title", ""),
                     "source": page.get("url", ""),

tests/test_searchapi_retriever_none.py

@@ -0,0 +1,32 @@
+"""Regression: SearchAPIRetriever must tolerate raw_content=None.
+
+The scraper sets ``raw_content`` to ``None`` for pages that failed to scrape.
+Slicing ``None`` raises ``TypeError``, so the retriever must coerce a missing
+or None ``raw_content`` to an empty string.
+"""
+
+from gpt_researcher.context.retriever import SearchAPIRetriever
+
+
+def test_none_raw_content_yields_empty_page_content():
+    retriever = SearchAPIRetriever(
+        pages=[{"raw_content": None, "title": "t", "url": "https://u.example"}]
+    )
+    docs = retriever.invoke("q")
+    assert len(docs) == 1
+    assert docs[0].page_content == ""
+    assert docs[0].metadata["source"] == "https://u.example"
+
+
+def test_present_raw_content_is_preserved():
+    retriever = SearchAPIRetriever(
+        pages=[{"raw_content": "hello world", "title": "t", "url": "u"}]
+    )
+    docs = retriever.invoke("q")
+    assert docs[0].page_content == "hello world"
+
+
+def test_missing_raw_content_key_yields_empty():
+    retriever = SearchAPIRetriever(pages=[{"title": "t", "url": "u"}])
+    docs = retriever.invoke("q")
+    assert docs[0].page_content == ""