fairagro-middleware-api-client 9.0.1.dev13__tar.gz → 9.1.1.dev14__tar.gz

{fairagro_middleware_api_client-9.0.1.dev13 → fairagro_middleware_api_client-9.1.1.dev14}/.gitignore

@@ -227,3 +227,6 @@ helmchart/**/client_ext.conf
 .cache_ggshield
 
 docker/Dockerfile.api.bak
+# Docker DinD: ignore runtime state in docker-config, keep minimal config.json tracked
+.devcontainer/docker-config/*
+!.devcontainer/docker-config/config.json

{fairagro_middleware_api_client-9.0.1.dev13 → fairagro_middleware_api_client-9.1.1.dev14}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: fairagro-middleware-api-client
-Version: 9.0.1.dev13
+Version: 9.1.1.dev14
 Summary: The FAIRagro advanced middleware API client
 Requires-Python: >=3.12
 Requires-Dist: httpx>=0.28.1

{fairagro_middleware_api_client-9.0.1.dev13 → fairagro_middleware_api_client-9.1.1.dev14}/src/middleware/api_client/api_client.py

@@ -15,6 +15,7 @@ import httpx
 from pydantic import BaseModel, ValidationError
 
 from middleware.shared.api_models.common.models import HarvestStatus as SharedHarvestStatus
+from middleware.shared.api_models.common.rocrate import RoCratePayload
 from middleware.shared.api_models.v3.models import (
     CreateArcRequest,
     CreateHarvestRequest,
@@ -295,7 +296,7 @@ class ApiClient:
         seen_identifiers: set[str] = set()
 
         async def submit_one(arc_item: dict[str, Any]) -> None:
-            request = SubmitHarvestArcRequest(arc=arc_item)
+            request = SubmitHarvestArcRequest(arc=self._validate_rocrate(arc_item))
             await self._post(f"v3/harvests/{harvest_id}/arcs", request)
 
         async for arc in arcs:
@@ -305,7 +306,7 @@ class ApiClient:
                 if identifier in seen_identifiers:
                     logger.error(
                         "Duplicate ARC identifier '%s' in harvest %s — "
-                        "two ARCs share the same identifier (client-side data error).",
+                        "several ARCs share the same identifier (client-side data error).",
                         identifier,
                         harvest_id,
                     )
@@ -504,23 +505,20 @@ class ApiClient:
     # ------------------------------------------------------------------
 
     @staticmethod
-    def _extract_identifier_from_rocrate(arc_content: dict[str, Any]) -> str | None:
-        """Extract the RO-Crate identifier from a serialized ARC dict.
+    def _validate_rocrate(arc_content: dict[str, Any]) -> RoCratePayload:
+        """Validate a RO-Crate dict before sending it to the API."""
+        try:
+            return RoCratePayload.model_validate(arc_content)
+        except ValidationError as e:
+            raise ApiClientError(f"Invalid RO-Crate JSON: {e}") from e
 
-        Looks for the Root Data Entity (``@id == "./"``), then returns its
-        ``identifier`` field.  Returns ``None`` when the field is absent or
-        the dict does not follow the RO-Crate structure — validation is left
-        to the server.
-        """
-        graph = arc_content.get("@graph")
-        if isinstance(graph, list):
-            for item in graph:
-                if isinstance(item, dict) and item.get("@id") == "./":
-                    identifier = item.get("identifier")
-                    if isinstance(identifier, list):
-                        identifier = identifier[0] if identifier else None
-                    return str(identifier) if identifier else None
-        return None
+    @staticmethod
+    def _extract_identifier_from_rocrate(arc_content: dict[str, Any]) -> str | None:
+        """Return the RO-Crate identifier when the payload is structurally valid."""
+        try:
+            return RoCratePayload.model_validate(arc_content).identifier
+        except ValidationError:
+            return None
 
     @classmethod
     def _serialize_arc(cls, arc: "ARC | dict[str, Any] | str") -> dict[str, Any]:
@@ -575,7 +573,7 @@ class ApiClient:
         """
         logger.info("Creating/updating ARC for RDI: %s", rdi)
         serialized = self._serialize_arc(arc)
-        request = CreateArcRequest(rdi=rdi, arc=serialized)
+        request = CreateArcRequest(rdi=rdi, arc=self._validate_rocrate(serialized))
         data = await self._post("v3/arcs", request)
         return self._parse_arc_response(data)
 
@@ -710,7 +708,7 @@ class ApiClient:
             :class:`ArcResult` with the result of the operation.
         """
         serialized = self._serialize_arc(arc)
-        request = SubmitHarvestArcRequest(arc=serialized)
+        request = SubmitHarvestArcRequest(arc=self._validate_rocrate(serialized))
         data = await self._post(f"v3/harvests/{harvest_id}/arcs", request)
         return self._parse_arc_response(data)
 

{fairagro_middleware_api_client-9.0.1.dev13 → fairagro_middleware_api_client-9.1.1.dev14}/tests/unit/test_client.py

@@ -2,6 +2,7 @@
 
 import asyncio
 import http
+import json
 import ssl
 from collections.abc import AsyncGenerator
 from pathlib import Path
@@ -40,6 +41,15 @@ _ARC_RESPONSE = {
     "events": [],
 }
 
+
+def _rocrate_dict(identifier: str = "mock-arc") -> dict[str, Any]:
+    """Minimal valid RO-Crate payload for client tests."""
+    return {
+        "@context": "https://w3id.org/ro/crate/1.1/context",
+        "@graph": [{"@id": "./", "identifier": identifier}],
+    }
+
+
 _HARVEST_RESPONSE: dict[str, str | None | dict] = {
     "client_id": "test-client",
     "message": "Harvest created",
@@ -229,7 +239,7 @@ async def test_create_or_update_arc_with_dict(client_config: Config) -> None:
         return_value=httpx.Response(http.HTTPStatus.OK, json=_ARC_RESPONSE)
     )
     async with ApiClient(client_config) as client:
-        response = await client.create_or_update_arc(rdi="test-rdi", arc={"id": "mock-arc"})
+        response = await client.create_or_update_arc(rdi="test-rdi", arc=_rocrate_dict())
     assert isinstance(response, ArcResult)
 
 
@@ -241,7 +251,7 @@ async def test_create_or_update_arc_with_json_string(client_config: Config) -> N
         return_value=httpx.Response(http.HTTPStatus.OK, json=_ARC_RESPONSE)
     )
     async with ApiClient(client_config) as client:
-        response = await client.create_or_update_arc(rdi="test-rdi", arc='{"id": "mock-arc"}')
+        response = await client.create_or_update_arc(rdi="test-rdi", arc=json.dumps(_rocrate_dict()))
     assert route.called
     assert isinstance(response, ArcResult)
     assert response.arc_id == "arc-123"
@@ -252,7 +262,7 @@ async def test_create_or_update_arc_with_invalid_json_string(client_config: Conf
     """Test create_or_update_arc with an invalid JSON string."""
     async with ApiClient(client_config) as client:
         with pytest.raises(ApiClientError, match="Invalid JSON string provided for ARC"):
-            await client.create_or_update_arc(rdi="test-rdi", arc='{"id": "mock-arc"')
+            await client.create_or_update_arc(rdi="test-rdi", arc='{"@context":')
 
 
 @pytest.mark.asyncio
@@ -290,6 +300,14 @@ async def test_create_or_update_arc_invalid_response(client_config: Config) -> N
     )
     async with ApiClient(client_config) as client:
         with pytest.raises(ApiClientError, match="Invalid ARC response"):
+            await client.create_or_update_arc(rdi="test-rdi", arc=_rocrate_dict("mock"))
+
+
+@pytest.mark.asyncio
+async def test_create_or_update_arc_invalid_rocrate(client_config: Config) -> None:
+    """Test create_or_update_arc rejects structurally invalid RO-Crate JSON."""
+    async with ApiClient(client_config) as client:
+        with pytest.raises(ApiClientError, match="Invalid RO-Crate JSON"):
             await client.create_or_update_arc(rdi="test-rdi", arc={"id": "mock"})
 
 
@@ -301,7 +319,7 @@ async def test_create_or_update_arc_sends_correct_headers(client_config: Config)
         return_value=httpx.Response(http.HTTPStatus.OK, json=_ARC_RESPONSE)
     )
     async with ApiClient(client_config) as client:
-        await client.create_or_update_arc(rdi="test", arc={"id": "mock-arc"})
+        await client.create_or_update_arc(rdi="test", arc=_rocrate_dict())
 
     assert route.called
     req = route.calls.last.request
@@ -499,7 +517,7 @@ async def test_submit_arc_in_harvest(client_config: Config) -> None:
         return_value=httpx.Response(http.HTTPStatus.OK, json=_ARC_RESPONSE)
     )
     async with ApiClient(client_config) as client:
-        response = await client.submit_arc_in_harvest("harvest-456", arc={"id": "mock-arc"})
+        response = await client.submit_arc_in_harvest("harvest-456", arc=_rocrate_dict())
     assert isinstance(response, ArcResult)
     assert response.arc_id == "arc-123"
 
@@ -513,6 +531,14 @@ async def test_submit_arc_in_harvest_invalid_response(client_config: Config) ->
     )
     async with ApiClient(client_config) as client:
         with pytest.raises(ApiClientError, match="Invalid ARC response"):
+            await client.submit_arc_in_harvest("harvest-456", arc=_rocrate_dict("mock"))
+
+
+@pytest.mark.asyncio
+async def test_submit_arc_in_harvest_invalid_rocrate(client_config: Config) -> None:
+    """Test submit_arc_in_harvest rejects structurally invalid RO-Crate JSON."""
+    async with ApiClient(client_config) as client:
+        with pytest.raises(ApiClientError, match="Invalid RO-Crate JSON"):
             await client.submit_arc_in_harvest("harvest-456", arc={"id": "mock"})
 
 
@@ -524,7 +550,7 @@ async def test_submit_arc_in_harvest_with_json_string(client_config: Config) ->
         return_value=httpx.Response(http.HTTPStatus.OK, json=_ARC_RESPONSE)
     )
     async with ApiClient(client_config) as client:
-        response = await client.submit_arc_in_harvest("harvest-456", arc='{"id": "mock-arc"}')
+        response = await client.submit_arc_in_harvest("harvest-456", arc=json.dumps(_rocrate_dict()))
     assert route.called
     assert isinstance(response, ArcResult)
     assert response.arc_id == "arc-123"
@@ -535,7 +561,7 @@ async def test_submit_arc_in_harvest_with_invalid_json_string(client_config: Con
     """Test submit_arc_in_harvest with an invalid JSON string."""
     async with ApiClient(client_config) as client:
         with pytest.raises(ApiClientError, match="Invalid JSON string provided for ARC"):
-            await client.submit_arc_in_harvest("harvest-456", arc='{"id": "mock-arc"')
+            await client.submit_arc_in_harvest("harvest-456", arc='{"@context":')
 
 
 # ---------------------------------------------------------------------------
@@ -564,7 +590,7 @@ async def test_harvest_arcs_success(client_config: Config) -> None:
         return_value=httpx.Response(http.HTTPStatus.OK, json=completed_response)
     )
 
-    arcs = _arc_gen({"id": "arc-1"}, {"id": "arc-2"}, {"id": "arc-3"})
+    arcs = _arc_gen(_rocrate_dict("arc-1"), _rocrate_dict("arc-2"), _rocrate_dict("arc-3"))
     async with ApiClient(client_config) as client:
         result = await client.harvest_arcs("test-rdi", arcs, expected_datasets=3)
 
@@ -589,7 +615,7 @@ async def test_harvest_arcs_success_with_parallelism(client_config: Config) -> N
         return_value=httpx.Response(http.HTTPStatus.OK, json=completed_response)
     )
 
-    arcs = _arc_gen({"id": "arc-1"}, {"id": "arc-2"}, {"id": "arc-3"})
+    arcs = _arc_gen(_rocrate_dict("arc-1"), _rocrate_dict("arc-2"), _rocrate_dict("arc-3"))
     async with ApiClient(client_config) as client:
         result = await client.harvest_arcs("test-rdi", arcs)
 
@@ -614,7 +640,7 @@ async def test_harvest_arcs_uses_config_default_concurrency(client_config: Confi
         return_value=httpx.Response(http.HTTPStatus.OK, json=completed_response)
     )
 
-    arcs = _arc_gen({"id": "arc-1"}, {"id": "arc-2"}, {"id": "arc-3"})
+    arcs = _arc_gen(_rocrate_dict("arc-1"), _rocrate_dict("arc-2"), _rocrate_dict("arc-3"))
     async with ApiClient(client_config) as client:
         result = await client.harvest_arcs("test-rdi", arcs)
 
@@ -664,7 +690,8 @@ async def test_harvest_arcs_continues_on_item_error(client_config: Config) -> No
     )
 
     async with ApiClient(client_config) as client:
-        result = await client.harvest_arcs("test-rdi", _arc_gen({"id": "arc-1"}, {"id": "arc-2"}, {"id": "arc-3"}))
+        arcs = _arc_gen(_rocrate_dict("arc-1"), _rocrate_dict("arc-2"), _rocrate_dict("arc-3"))
+        result = await client.harvest_arcs("test-rdi", arcs)
 
     assert isinstance(result, HarvestResult)
     assert route_submit.call_count == EXPECTED_ARC_UPLOADS
@@ -672,7 +699,7 @@ async def test_harvest_arcs_continues_on_item_error(client_config: Config) -> No
     assert not cancel_route.called
     assert len(result.errors) == 1
     assert result.errors[0].error_type == HarvestErrorType.SUBMISSION_FAILED
-    assert result.errors[0].arc_id is None
+    assert result.errors[0].arc_id == "arc-1"
     assert "HTTP error 400" in result.errors[0].message
 
 
@@ -698,7 +725,7 @@ async def test_harvest_arcs_cancels_on_catastrophic_error(client_config: Config)
 
     async with ApiClient(client_config) as client:
         with pytest.raises(ApiClientError, match="HTTP error 409"):
-            await client.harvest_arcs("test-rdi", _arc_gen({"id": "arc-1"}))
+            await client.harvest_arcs("test-rdi", _arc_gen(_rocrate_dict("arc-1")))
 
     assert fail_route.called
 
@@ -760,8 +787,8 @@ async def test_harvest_arcs_with_json_string(client_config: Config) -> None:
     )
 
     arcs = _arc_gen(
-        '{"id": "arc-1-string"}',
-        {"id": "arc-2-dict"},
+        json.dumps(_rocrate_dict("arc-1-string")),
+        _rocrate_dict("arc-2-dict"),
         ARC.from_arc_investigation(ArcInvestigation.create(identifier="test", title="Test")),
     )
     async with ApiClient(client_config) as client:
@@ -807,7 +834,7 @@ async def test_harvest_arcs_cancel_failure_does_not_mask_original_error(client_c
 
     async with ApiClient(client_config) as client:
         with pytest.raises(ApiClientError, match="HTTP error 409"):
-            await client.harvest_arcs("test-rdi", _arc_gen({"id": "arc-1"}))
+            await client.harvest_arcs("test-rdi", _arc_gen(_rocrate_dict("arc-1")))
 
 
 @pytest.mark.asyncio
@@ -855,7 +882,7 @@ async def test_harvest_arcs_all_exceptions_retrieved_when_multiple_tasks_catastr
         with pytest.raises(ApiClientError, match="HTTP error 409"):
             await client.harvest_arcs(
                 "test-rdi",
-                _arc_gen({"id": "arc-1"}, {"id": "arc-2"}, {"id": "arc-3"}),
+                _arc_gen(_rocrate_dict("arc-1"), _rocrate_dict("arc-2"), _rocrate_dict("arc-3")),
             )
 
     assert fail_route.called
@@ -892,7 +919,7 @@ async def test_harvest_arcs_500_is_submission_failed(client_config: Config) -> N
     async with ApiClient(client_config) as client:
         result = await client.harvest_arcs(
             "test-rdi",
-            _arc_gen({"id": "arc-1"}, {"id": "arc-2"}, {"id": "arc-3"}),
+            _arc_gen(_rocrate_dict("arc-1"), _rocrate_dict("arc-2"), _rocrate_dict("arc-3")),
         )
 
     assert complete_route.called
@@ -929,7 +956,7 @@ async def test_harvest_arcs_502_is_submission_failed(client_config: Config) -> N
     )
 
     async with ApiClient(client_config) as client:
-        result = await client.harvest_arcs("test-rdi", _arc_gen({"id": "arc-1"}, {"id": "arc-2"}))
+        result = await client.harvest_arcs("test-rdi", _arc_gen(_rocrate_dict("arc-1"), _rocrate_dict("arc-2")))
 
     assert complete_route.called
     assert not fail_route.called