add pdf cases for in memory integration tests

deepset-ai · May 17, 2024 · 4e1acf0 · 4e1acf0
1 parent 47325d7
commit 4e1acf0
Show file tree

Hide file tree

Showing 4 changed files with 11 additions and 2 deletions.
diff --git a/deepset_cloud_sdk/_service/files_service.py b/deepset_cloud_sdk/_service/files_service.py
@@ -42,7 +42,7 @@
 )
 from deepset_cloud_sdk._s3.upload import S3, S3UploadResult, S3UploadSummary
 from deepset_cloud_sdk._utils.constants import SUPPORTED_TYPE_SUFFIXES
-from deepset_cloud_sdk.models import DeepsetCloudFile, DeepsetCloudFileBase
+from deepset_cloud_sdk.models import DeepsetCloudFileBase
 
 logger = structlog.get_logger(__name__)
 

diff --git a/tests/data/upload_folder/example.pdf b/tests/data/upload_folder/example.pdf
diff --git a/tests/integration/service/test_integration_files_service.py b/tests/integration/service/test_integration_files_service.py
@@ -12,9 +12,9 @@
 from deepset_cloud_sdk._service.files_service import (
     META_SUFFIX,
     SUPPORTED_TYPE_SUFFIXES,
-    DeepsetCloudFile,
     FilesService,
 )
+from deepset_cloud_sdk.models import DeepsetCloudFile, DeepsetCloudFileBytes
 
 
 @pytest.mark.asyncio
@@ -204,13 +204,17 @@ async def test_async_upload_multiple_file_types(
         assert file00_metadata == {"file_name_duplicate_check": "file00.txt", "source": "multiple file types"}
 
     async def test_upload_in_memory(self, integration_config: CommonConfig, workspace_name: str) -> None:
+        with open(Path("./tests/test_data/multiple_file_types/file08.pdf"), "rb") as f:
+            pdf_contents = f.read()
+
         async with FilesService.factory(integration_config) as file_service:
             files = [
                 DeepsetCloudFile("file1", "file1.txt", {"which": 1}),
                 DeepsetCloudFile("file2", "file2.txt", {"which": 2}),
                 DeepsetCloudFile("file3", "file3.txt", {"which": 3}),
                 DeepsetCloudFile("file4", "file4.txt", {"which": 4}),
                 DeepsetCloudFile("file5", "file5.txt", {"which": 5}),
+                DeepsetCloudFileBytes(file_bytes=pdf_contents, name="file6.pdf", meta={"which": 6}),
             ]
             result = await file_service.upload_in_memory(
                 workspace_name=workspace_name,
@@ -228,13 +232,18 @@ async def test_upload_in_memory_less_than_session_threshold(
         self, integration_config: CommonConfig, workspace_name: str, monkeypatch: MonkeyPatch
     ) -> None:
         monkeypatch.setattr("deepset_cloud_sdk._service.files_service.DIRECT_UPLOAD_THRESHOLD", -1)
+
+        with open(Path("./tests/test_data/multiple_file_types/file08.pdf"), "rb") as f:
+            pdf_contents = f.read()
+
         async with FilesService.factory(integration_config) as file_service:
             files = [
                 DeepsetCloudFile("file1", "file1.txt", {"which": 1}),
                 DeepsetCloudFile("file2", "file2.txt", {"which": 2}),
                 DeepsetCloudFile("file3", "file3.txt", {"which": 3}),
                 DeepsetCloudFile("file4", "file4.txt", {"which": 4}),
                 DeepsetCloudFile("file5", "file5.txt", {"which": 5}),
+                DeepsetCloudFileBytes(file_bytes=pdf_contents, name="file6.pdf", meta={"which": 6}),
             ]
             result = await file_service.upload_in_memory(
                 workspace_name=workspace_name,

diff --git a/tests/test_data/multiple_file_types/file08.pdf b/tests/test_data/multiple_file_types/file08.pdf