Fix chat stream recovery and PDF library indexing (#468)

tests/test_personal_docs_pdf_index.py

@@ -0,0 +1,24 @@
+from pathlib import Path
+
+from src import personal_docs
+
+
+def test_personal_index_includes_pdf_uploads(tmp_path, monkeypatch):
+    pdf_path = tmp_path / "notes.pdf"
+    pdf_path.write_bytes(b"%PDF-1.4 fake test pdf")
+
+    monkeypatch.setattr(
+        personal_docs,
+        "extract_pdf_text",
+        lambda path: "readable pdf text" if Path(path) == pdf_path else "",
+    )
+
+    files = personal_docs.load_personal_index(str(tmp_path))
+
+    assert [item["name"] for item in files] == ["notes.pdf"]
+    assert files[0]["path"] == str(pdf_path)
+    assert files[0]["chunks"] == ["readable pdf text"]
+
+
+def test_personal_index_default_extensions_advertise_pdf_support():
+    assert ".pdf" in personal_docs.config.DEFAULT_EXTENSIONS