Add additional assertions to live tests for new tools

- Now evaluates:
  - File names
  - MIME types
  - File sizes
  - Warnings in response
  - Input ID

Assisted-by: Codex

Author: datalogics-cgreen

tests/live/test_live_convert_to_excel.py

@@ -50,6 +50,8 @@ def test_live_convert_to_excel_success(
         output_file.type
         == "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet"
     )
+    assert output_file.size > 0
+    assert response.warning is None
     assert str(response.input_id) == str(uploaded_pdf_for_excel.id)
     if output_name is not None:
         assert output_file.name.startswith(output_name)
@@ -76,6 +78,8 @@ async def test_live_async_convert_to_excel_success(
         output_file.type
         == "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet"
     )
+    assert output_file.size > 0
+    assert response.warning is None
     assert str(response.input_id) == str(uploaded_pdf_for_excel.id)
 
 

tests/live/test_live_convert_to_markdown.py

@@ -22,6 +22,11 @@ def test_live_convert_to_markdown_success(
 
     assert isinstance(response, PdfRestFileBasedResponse)
     assert response.output_files
+    output_file = response.output_file
+    assert output_file.name.endswith(".md")
+    assert output_file.type == "text/markdown"
+    assert output_file.size > 0
+    assert response.warning is None
     assert response.input_id == uploaded.id
 
 
@@ -40,7 +45,11 @@ async def test_live_async_convert_to_markdown_success(
 
     assert isinstance(response, PdfRestFileBasedResponse)
     assert response.output_files
-    assert response.output_file.name.startswith("async-md")
+    output_file = response.output_file
+    assert output_file.name.startswith("async-md")
+    assert output_file.type == "text/markdown"
+    assert output_file.size > 0
+    assert response.warning is None
     assert response.input_id == uploaded.id
 
 

tests/live/test_live_convert_to_powerpoint.py

@@ -50,6 +50,8 @@ def test_live_convert_to_powerpoint_success(
         output_file.type
         == "application/vnd.openxmlformats-officedocument.presentationml.presentation"
     )
+    assert output_file.size > 0
+    assert response.warning is None
     assert str(response.input_id) == str(uploaded_pdf_for_powerpoint.id)
     if output_name is not None:
         assert output_file.name.startswith(output_name)
@@ -78,6 +80,8 @@ async def test_live_async_convert_to_powerpoint_success(
         output_file.type
         == "application/vnd.openxmlformats-officedocument.presentationml.presentation"
     )
+    assert output_file.size > 0
+    assert response.warning is None
     assert str(response.input_id) == str(uploaded_pdf_for_powerpoint.id)
 
 

tests/live/test_live_convert_xfa_to_acroforms.py

@@ -47,6 +47,8 @@ def test_live_convert_xfa_to_acroforms_success(
     assert response.output_files
     output_file = response.output_file
     assert output_file.type == "application/pdf"
+    assert output_file.size > 0
+    assert response.warning is None
     assert str(response.input_id) == str(uploaded_pdf_for_acroforms.id)
     if output_name is not None:
         assert output_file.name.startswith(output_name)
@@ -90,6 +92,8 @@ async def test_live_async_convert_xfa_to_acroforms_success(
     output_file = response.output_file
     assert output_file.name.startswith("async")
     assert output_file.type == "application/pdf"
+    assert output_file.size > 0
+    assert response.warning is None
     assert str(response.input_id) == str(uploaded_pdf_for_acroforms.id)
 
 

tests/live/test_live_extract_images.py

@@ -21,7 +21,15 @@ def test_live_extract_images_success(
         response = client.extract_images(uploaded)
 
     assert isinstance(response, PdfRestFileBasedResponse)
-    assert response.output_files
+    output_files = response.output_files
+    assert output_files
+    assert all(file.name for file in output_files)
+    assert all(
+        file.type and (file.type.startswith("image/") or file.type == "application/zip")
+        for file in output_files
+    )
+    assert all(file.size > 0 for file in output_files)
+    assert response.warning is None
     assert response.input_id == uploaded.id
 
 
@@ -39,8 +47,16 @@ async def test_live_async_extract_images_success(
         response = await client.extract_images(uploaded, output="async-images")
 
     assert isinstance(response, PdfRestFileBasedResponse)
-    assert response.output_files
-    assert response.output_file.name.startswith("async-images")
+    output_files = response.output_files
+    assert output_files
+    assert output_files[0].name.startswith("async-images")
+    assert all(file.name for file in output_files)
+    assert all(
+        file.type and (file.type.startswith("image/") or file.type == "application/zip")
+        for file in output_files
+    )
+    assert all(file.size > 0 for file in output_files)
+    assert response.warning is None
     assert response.input_id == uploaded.id
 
 

tests/live/test_live_extract_pdf_text_to_file.py

@@ -28,6 +28,11 @@ def test_live_extract_pdf_text_to_file_success(
 
     assert isinstance(response, PdfRestFileBasedResponse)
     assert response.output_files
+    output_file = response.output_file
+    assert output_file.name.endswith(".txt")
+    assert output_file.type == "text/plain"
+    assert output_file.size > 0
+    assert response.warning is None
     assert response.input_id == uploaded.id
 
 
@@ -53,7 +58,11 @@ async def test_live_async_extract_pdf_text_to_file_success(
 
     assert isinstance(response, PdfRestFileBasedResponse)
     assert response.output_files
-    assert response.output_file.name.startswith("async-text")
+    output_file = response.output_file
+    assert output_file.name.startswith("async-text")
+    assert output_file.type == "text/plain"
+    assert output_file.size > 0
+    assert response.warning is None
     assert response.input_id == uploaded.id
 
 

tests/live/test_live_flatten_annotations.py

@@ -47,6 +47,8 @@ def test_live_flatten_annotations_success(
     assert response.output_files
     output_file = response.output_file
     assert output_file.type == "application/pdf"
+    assert output_file.size > 0
+    assert response.warning is None
     assert str(response.input_id) == str(uploaded_pdf_for_annotations.id)
     if output_name is not None:
         assert output_file.name.startswith(output_name)
@@ -72,6 +74,8 @@ async def test_live_async_flatten_annotations_success(
     output_file = response.output_file
     assert output_file.name.startswith("async")
     assert output_file.type == "application/pdf"
+    assert output_file.size > 0
+    assert response.warning is None
     assert str(response.input_id) == str(uploaded_pdf_for_annotations.id)
 
 

tests/live/test_live_flatten_transparencies.py

@@ -50,6 +50,8 @@ def test_live_flatten_transparencies_success(
     assert response.output_files
     output_file = response.output_file
     assert output_file.type == "application/pdf"
+    assert output_file.size > 0
+    assert response.warning is None
     assert str(response.input_id) == str(uploaded_pdf_for_transparencies.id)
     if output_name is not None:
         assert output_file.name.startswith(output_name)
@@ -75,6 +77,8 @@ async def test_live_async_flatten_transparencies_success(
     output_file = response.output_file
     assert output_file.name.startswith("async")
     assert output_file.type == "application/pdf"
+    assert output_file.size > 0
+    assert response.warning is None
     assert str(response.input_id) == str(uploaded_pdf_for_transparencies.id)
 
 

tests/live/test_live_linearize_pdf.py

@@ -47,6 +47,8 @@ def test_live_linearize_pdf(
     assert response.output_files
     output_file = response.output_file
     assert output_file.type == "application/pdf"
+    assert output_file.size > 0
+    assert response.warning is None
     assert str(response.input_id) == str(uploaded_pdf_for_linearize.id)
     if output_name is not None:
         assert output_file.name.startswith(output_name)
@@ -73,6 +75,8 @@ async def test_live_async_linearize_pdf(
     output_file = response.output_file
     assert output_file.name.startswith("async-linearized")
     assert output_file.type == "application/pdf"
+    assert output_file.size > 0
+    assert response.warning is None
     assert str(response.input_id) == str(uploaded_pdf_for_linearize.id)
 
 

tests/live/test_live_ocr_pdf.py

@@ -22,7 +22,11 @@ def test_live_ocr_pdf_success(
 
     assert isinstance(response, PdfRestFileBasedResponse)
     assert response.output_files
-    assert response.output_file.id
+    output_file = response.output_file
+    assert output_file.name.endswith(".pdf")
+    assert output_file.type == "application/pdf"
+    assert output_file.size > 0
+    assert response.warning is None
     assert response.input_id == uploaded.id
 
 
@@ -42,6 +46,9 @@ async def test_live_async_ocr_pdf_success(
     assert isinstance(response, PdfRestFileBasedResponse)
     assert response.output_files
     assert response.output_file.name.startswith("async-ocr")
+    assert response.output_file.type == "application/pdf"
+    assert response.output_file.size > 0
+    assert response.warning is None
     assert response.input_id == uploaded.id