Do not store the hypothetically produced mimetypes always

boefjes/boefjes/api.py

@@ -125,7 +125,7 @@ def boefje_output(
     bytes_client.save_boefje_meta(boefje_meta)
 
     if boefje_output.files:
-        mime_types = _default_mime_types(boefje_meta.boefje).union(plugin.produces)
+        mime_types = _default_mime_types(boefje_meta.boefje)
         for file in boefje_output.files:
             raw = base64.b64decode(file.content)
             # when supported, also save file.name to Bytes

boefjes/boefjes/job_handler.py

@@ -119,13 +119,11 @@ def handle(self, boefje_meta: BoefjeMeta) -> None:
         boefje_meta.runnable_hash = plugin.runnable_hash
         boefje_meta.environment = get_environment_settings(boefje_meta, plugin.boefje_schema)
 
-        mime_types = _default_mime_types(boefje_meta.boefje).union(plugin.produces)
-
         logger.info("Starting boefje %s[%s]", boefje_meta.boefje.id, str(boefje_meta.id))
 
         boefje_meta.started_at = datetime.now(timezone.utc)
 
-        boefje_results: list[tuple[set, bytes | str]]
+        boefje_results: list[tuple[set, bytes | str]] = []
 
         try:
             boefje_results = self.job_runner.run(boefje_meta, boefje_meta.environment)
@@ -152,7 +150,9 @@ def handle(self, boefje_meta: BoefjeMeta) -> None:
                             )
                         else:
                             valid_mimetypes.add(mimetype)
-                    raw_file_id = self.bytes_client.save_raw(boefje_meta.id, output, mime_types.union(valid_mimetypes))
+                    raw_file_id = self.bytes_client.save_raw(
+                        boefje_meta.id, output, _default_mime_types(boefje_meta.boefje).union(valid_mimetypes)
+                    )
                     logger.info(
                         "Saved raw file %s for boefje %s[%s]", raw_file_id, boefje_meta.boefje.id, boefje_meta.id
                     )

bytes/tests/client.py

@@ -9,7 +9,7 @@
 from httpx import HTTPError, HTTPStatusError
 
 from bytes.api.models import BoefjeOutput
-from bytes.models import BoefjeMeta, NormalizerMeta
+from bytes.models import BoefjeMeta, NormalizerMeta, RawDataMeta
 from bytes.repositories.meta_repository import BoefjeMetaFilter, NormalizerMetaFilter, RawDataFilter
 
 BYTES_API_CLIENT_VERSION = "0.2"
@@ -164,6 +164,13 @@ def get_raw(self, raw_id: UUID) -> bytes:
 
         return response.content
 
+    @retry_with_login
+    def get_raw_meta(self, raw_id: UUID) -> RawDataMeta:
+        response = self.client.get(f"/bytes/raw/{raw_id}/meta")
+        self._verify_response(response)
+
+        return RawDataMeta.model_validate(response.json())
+
     @retry_with_login
     def get_raws(self, query_filter: RawDataFilter) -> dict[str, str]:
         params = query_filter.model_dump(exclude_none=True)

bytes/tests/integration/test_bytes_api.py

@@ -315,8 +315,18 @@ def test_raw_mimes(bytes_api_client: BytesAPIClient) -> None:
 
     raw = b"test 123456"
     second_raw = b"second test 200"
-    bytes_api_client.save_raw(boefje_meta.id, raw, mime_types)
-    bytes_api_client.save_raw(boefje_meta.id, second_raw, second_mime_types)
+    first_id = bytes_api_client.save_raw(boefje_meta.id, raw, mime_types)
+    second_id = bytes_api_client.save_raw(boefje_meta.id, second_raw, second_mime_types)
+
+    first_meta = bytes_api_client.get_raw_meta(first_id)
+    second_meta = bytes_api_client.get_raw_meta(second_id)
+
+    assert first_meta.id != second_meta.id
+    assert {x.value for x in first_meta.mime_types} == set(mime_types)
+    assert {x.value for x in second_meta.mime_types} == set(second_mime_types)
+
+    assert bytes_api_client.get_raw(first_id) == raw
+    assert bytes_api_client.get_raw(second_id) == second_raw
 
     retrieved_raws = bytes_api_client.get_raws(
         RawDataFilter(
@@ -386,3 +396,6 @@ def test_save_multiple_raw_files(bytes_api_client: BytesAPIClient) -> None:
 
     assert bytes_api_client.get_raw(ids["first"]) == first_raw
     assert bytes_api_client.get_raw(ids["second"]) == second_raw
+
+    assert bytes_api_client.get_raw_meta(ids["first"]).mime_types == set()
+    assert {x.value for x in bytes_api_client.get_raw_meta(ids["second"]).mime_types} == {"mime", "type"}