Refa: migrate document preview/download to RESTful API (#14633)

### What problem does this PR solve?

migrate document preview/download to RESTful API

### Type of change
- [x] Refactoring

test/testcases/test_web_api/test_common.py

@@ -26,7 +26,6 @@ from utils.file_utils import create_txt_file
 HEADERS = {"Content-Type": "application/json"}
 
 DATASETS_URL = f"/api/{VERSION}/datasets"
-DOCUMENT_APP_URL = f"/{VERSION}/document"
 CHUNK_APP_URL = f"/{VERSION}/chunk"
 CHUNK_API_URL = f"/api/{VERSION}/datasets/{{dataset_id}}/documents/{{document_id}}/chunks"
 # SESSION_WITH_CHAT_ASSISTANT_API_URL = "/api/v1/chats/{chat_id}/sessions"
@@ -404,10 +403,33 @@ def document_infos(auth, dataset_id, params=None, payload=None, *, headers=HEADE
 
 
 def document_metadata_summary(auth, payload=None, *, headers=HEADERS, data=None):
-    res = requests.post(url=f"{HOST_ADDRESS}{DOCUMENT_APP_URL}/metadata/summary", headers=headers, auth=auth, json=payload, data=data)
+    dataset_id = (payload or {}).get("kb_id")
+    doc_ids = (payload or {}).get("doc_ids")
+    if not dataset_id:
+        return {"code": 101, "message": "KB ID is required"}
+    params = {}
+    if doc_ids:
+        params["doc_ids"] = ",".join(doc_ids)
+    res = requests.get(url=f"{HOST_ADDRESS}{DATASETS_URL}/{dataset_id}/metadata/summary", headers=headers, auth=auth, params=params, data=data)
     return res.json()
 
 
+def document_get(auth, document_id, *, headers=HEADERS, data=None):
+    res = requests.get(url=f"{HOST_ADDRESS}/api/{VERSION}/documents/{document_id}/preview", headers=headers, auth=auth, data=data)
+    return res
+
+
+def document_download(auth, attachment_id, *, ext="markdown", headers=HEADERS, data=None):
+    res = requests.get(
+        url=f"{HOST_ADDRESS}/api/{VERSION}/documents/{attachment_id}/download",
+        headers=headers,
+        auth=auth,
+        params={"ext": ext},
+        data=data,
+    )
+    return res
+
+
 def document_metadata_update(auth, dataset_id, payload=None, *, headers=HEADERS, data=None):
     """New unified API for updating document metadata.
 

test/testcases/test_web_api/test_document_app/conftest.py

@@ -126,11 +126,31 @@ def document_app_module(monkeypatch):
     monkeypatch.setitem(sys.modules, "xgboost", ModuleType("xgboost"))
 
     stub_apps = ModuleType("api.apps")
+    stub_apps.__path__ = [str(repo_root / "api" / "apps")]
     stub_apps.current_user = SimpleNamespace(id="user-1")
     stub_apps.login_required = lambda func: func
     monkeypatch.setitem(sys.modules, "api.apps", stub_apps)
 
-    module_path = repo_root / "api" / "apps" / "document_app.py"
+    stub_apps_services = ModuleType("api.apps.services")
+    stub_apps_services.__path__ = [str(repo_root / "api" / "apps" / "services")]
+    monkeypatch.setitem(sys.modules, "api.apps.services", stub_apps_services)
+
+    document_api_service_mod = ModuleType("api.apps.services.document_api_service")
+    document_api_service_mod.validate_document_update_fields = lambda *_args, **_kwargs: (None, None)
+    document_api_service_mod.map_doc_keys = lambda doc: doc.to_dict() if hasattr(doc, "to_dict") else doc
+
+    def _map_doc_keys_with_run_status(doc, run_status="0"):
+        payload = doc if isinstance(doc, dict) else doc.to_dict()
+        return {**payload, "run": run_status}
+
+    document_api_service_mod.map_doc_keys_with_run_status = _map_doc_keys_with_run_status
+    document_api_service_mod.update_document_name_only = lambda *_args, **_kwargs: None
+    document_api_service_mod.update_chunk_method = lambda *_args, **_kwargs: None
+    document_api_service_mod.update_document_status_only = lambda *_args, **_kwargs: None
+    document_api_service_mod.reset_document_for_reparse = lambda *_args, **_kwargs: None
+    monkeypatch.setitem(sys.modules, "api.apps.services.document_api_service", document_api_service_mod)
+
+    module_path = repo_root / "api" / "apps" / "restful_apis" / "document_api.py"
     spec = importlib.util.spec_from_file_location("test_document_app_unit", module_path)
     module = importlib.util.module_from_spec(spec)
     module.manager = _DummyManager()

test/testcases/test_web_api/test_document_app/test_document_metadata.py

@@ -394,7 +394,7 @@ class TestDocumentMetadataUnit:
             "apply_safe_file_response_headers",
             lambda response, content_type, extension: response.headers.update({"content_type": content_type, "extension": extension}),
         )
-        res = _run(module.download_attachment("att1"))
+        res = _run(module.download_attachment(attachment_id="att1"))
         assert isinstance(res, _DummyResponse)
         assert res.data == b"attachment"
         assert res.headers["content_type"] == "application/abc"
@@ -405,7 +405,7 @@ class TestDocumentMetadataUnit:
 
         monkeypatch.setattr(module, "thread_pool_exec", raise_error)
         monkeypatch.setattr(module, "server_error_response", lambda e: {"code": 500, "message": str(e)})
-        res = _run(module.download_attachment("att1"))
+        res = _run(module.download_attachment(attachment_id="att1"))
         assert res["code"] == 500
         assert "download boom" in res["message"]