feat(avatar): optimize embedded H5 management flow

digital-avatar-app/backend/tests/test_knowledge_storage.py

@@ -2,9 +2,17 @@ from pathlib import Path
 from types import SimpleNamespace
 from unittest.mock import patch
 
+from fastapi.testclient import TestClient
+
+from database import SessionLocal
+from main import app
+from models import KnowledgeChunk, KnowledgeDoc
 from routers.knowledge import _doc_payload
 
 
+client = TestClient(app)
+
+
 def test_doc_payload_reports_whether_the_persisted_file_exists(tmp_path: Path):
     avatar_id = "avatar-1"
     stored_name = "knowledge.md"
@@ -21,3 +29,66 @@ def test_doc_payload_reports_whether_the_persisted_file_exists(tmp_path: Path):
         assert _doc_payload(doc)["filePresent"] is False
         stored_file.write_text("knowledge", encoding="utf-8")
         assert _doc_payload(doc)["filePresent"] is True
+
+
+def test_upload_marks_vectorization_failure_instead_of_staying_processing(
+    tmp_path: Path,
+    authorization_context,
+):
+    context = authorization_context
+    with (
+        patch("routers.knowledge.UPLOAD_DIR", str(tmp_path)),
+        patch("routers.knowledge.embeddings.embed", side_effect=RuntimeError("provider unavailable")),
+    ):
+        response = client.post(
+            f"/api/avatar/{context['avatar'].id}/knowledge/docs",
+            headers=context["owner_headers"],
+            files={"file": ("knowledge.md", b"# Knowledge\n\nTest content", "text/markdown")},
+        )
+
+    payload = response.json()["data"]
+    assert payload["status"] == "failed"
+    assert payload["vectorized"] is False
+    assert payload["chunkCount"] == 0
+
+    db = SessionLocal()
+    try:
+        stored = db.query(KnowledgeDoc).filter(KnowledgeDoc.id == payload["id"]).one()
+        assert stored.status == "failed"
+        assert db.query(KnowledgeChunk).filter(KnowledgeChunk.doc_id == stored.id).count() == 0
+        db.delete(stored)
+        db.commit()
+    finally:
+        db.close()
+
+
+def test_markdown_upload_commits_ready_document_and_chunks_together(
+    tmp_path: Path,
+    authorization_context,
+):
+    context = authorization_context
+    with (
+        patch("routers.knowledge.UPLOAD_DIR", str(tmp_path)),
+        patch("routers.knowledge.embeddings.embed", return_value=[[1.0, 0.0]]),
+    ):
+        response = client.post(
+            f"/api/avatar/{context['avatar'].id}/knowledge/docs",
+            headers=context["owner_headers"],
+            files={"file": ("knowledge.md", b"# Knowledge\n\nTest content", "text/markdown")},
+        )
+
+    payload = response.json()["data"]
+    assert payload["status"] == "ready"
+    assert payload["vectorized"] is True
+    assert payload["chunkCount"] == 1
+
+    db = SessionLocal()
+    try:
+        stored = db.query(KnowledgeDoc).filter(KnowledgeDoc.id == payload["id"]).one()
+        assert stored.status == "ready"
+        assert db.query(KnowledgeChunk).filter(KnowledgeChunk.doc_id == stored.id).count() == 1
+        db.query(KnowledgeChunk).filter(KnowledgeChunk.doc_id == stored.id).delete()
+        db.delete(stored)
+        db.commit()
+    finally:
+        db.close()