feat: Initial version of CleverSwarm Python SDK

Extend CleverSwarm codebase with PDF and Machine and Worker IDs

ISSUES CLOSED: #1

tests/test_cleverswarm_client.py

@@ -111,27 +111,6 @@ class TestCleverSwarmClient:
         assert client._timeout == 30
         assert client._triplets_template == "custom_{}_output.jsonl"
 
-    def test_client_file_operations_with_real_files(self):
-        """Test client file operations with real files."""
-        with tempfile.TemporaryDirectory() as temp_dir:
-            # Create test files
-            text_file = Path(temp_dir) / "test.txt"
-            json_file = Path(temp_dir) / "test.json"
-            owl_file = Path(temp_dir) / "test.owl"
-            
-            text_file.write_text("Test content")
-            json_file.write_text('{"test": "data"}')
-            owl_file.write_text('<rdf:RDF>test</rdf:RDF>')
-            
-            # Test that files exist and can be used
-            assert text_file.exists()
-            assert json_file.exists()
-            assert owl_file.exists()
-            
-            # Test that we can create a client and it recognizes the files
-            client = CleverSwarmClient("http://localhost:8000/api/v0/")
-            assert client._base_url == "http://localhost:8000/api/v0/"
-
     def test_file_validation_with_missing_files(self):
         """Test file validation with missing files."""
         # Test with non-existent files - this should raise InvalidFileException before calling update_token
@@ -157,10 +136,27 @@ class TestCleverSwarmClient:
             json_file.write_text('{"test": "data"}')
             jsonl_file.write_text('{"test": "data"}')
             
-            # These should not raise exceptions
+            # Test that files exist and are valid files
             assert text_file.exists()
+            assert text_file.is_file()
             assert json_file.exists()
+            assert json_file.is_file()
             assert jsonl_file.exists()
+            assert jsonl_file.is_file()
+            
+            # Test that the client can validate these files by calling create_benchmark_job
+            # We need to mock the update_token and the HTTP request to avoid actual API calls
+            with patch.object(self.client, 'update_token'):
+                with patch.object(self.client, '_generic_http_request_executor') as mock_executor:
+                    # Mock the executor to return the job_id directly
+                    mock_executor.return_value = "test_job_123"
+                    
+                    # This should not raise InvalidFileException
+                    result = self.client.create_benchmark_job(
+                        text_file, json_file, jsonl_file, True
+                    )
+                    
+                    assert result == "test_job_123"
 
     def test_client_state_management(self):
         """Test client state management."""
@@ -273,6 +269,7 @@ class TestCleverSwarmClient:
         assert FileTypeAPI.AutoDetect == "AutoDetect"
         assert FileTypeAPI.Markdown == "Markdown"
         assert FileTypeAPI.PlainText == "PlainText"
+        assert FileTypeAPI.PDF == "PDF"
 
     def test_response_type_enum_values(self):
         """Test ResponseTypeAPI enum values."""
@@ -850,4 +847,4 @@ class TestCleverSwarmClientHTTP:
             raise ValueError("Unexpected error")
         
         with pytest.raises(UnexpectedConditionException):
-            self.client._generic_http_request_executor(test_func)
+            self.client._generic_http_request_executor(test_func)

tests/test_cswarm_benchmark_client.py

@@ -448,13 +448,15 @@ class TestBenchmarkCLIUtilityMethods:
                 "type": "Benchmark",
                 "status": "Completed",
                 "retries_count": 0,
-                "file_type": "JSONL"
+                "file_type": "JSONL",
+                "machine_id": "machine-001",
+                "worker_id": "worker-001"
             }
         ]
         mock_client.get_jobs_list.return_value = mock_jobs
         mock_client_class.return_value = mock_client
         self.cli._client = mock_client
-
+    
         self.cli.print_server_jobs()
         
         captured = capsys.readouterr()
@@ -473,6 +475,8 @@ class TestBenchmarkCLIUtilityMethods:
                 "status": "Completed",
                 "retries_count": 0,
                 "file_type": "JSONL",
+                "machine_id": "machine-001",
+                "worker_id": "worker-001",
                 "unstructured_sources": ["test.txt"],
                 "ontology_sources": ["test.json"],
                 "ground_truth_sources": ["gt.jsonl"]
@@ -481,7 +485,7 @@ class TestBenchmarkCLIUtilityMethods:
         mock_client.get_jobs_list.return_value = mock_jobs
         mock_client_class.return_value = mock_client
         self.cli._client = mock_client
-
+    
         self.cli.print_server_jobs(detailed=True)
         
         captured = capsys.readouterr()

tests/test_cswarm_benchmark_client_extended.py

@@ -60,6 +60,8 @@ class TestBenchmarkCLIExtended:
                 "status": "Completed",
                 "retries_count": 0,
                 "file_type": "JSONL",
+                "machine_id": "machine-001",
+                "worker_id": "worker-001",
                 "unstructured_sources": ["test.txt"],
                 "ontology_sources": ["test.json"],
                 "ontology_spec_sources": ["test.owl"],
@@ -67,7 +69,7 @@ class TestBenchmarkCLIExtended:
                 "ground_truth_sources": ["test_gt.jsonl"]
             }
         ]
-        
+
         with patch.object(self.cli, 'list_server_jobs') as mock_list:
             mock_list.return_value = mock_jobs
             self.cli.print_server_jobs(detailed=True)
@@ -82,13 +84,15 @@ class TestBenchmarkCLIExtended:
                 "status": "Completed",
                 "retries_count": 0,
                 "file_type": "JSONL",
+                "machine_id": "machine-001",
+                "worker_id": "worker-001",
                 "unstructured_sources": ["test.txt"],
                 "ontology_sources": ["test.json"],
                 "ontology_spec_sources": ["test.owl"],
                 "ground_truth_sources": ["test_gt.jsonl"]
             }
         ]
-        
+
         with patch.object(self.cli, 'list_server_jobs') as mock_list:
             mock_list.return_value = mock_jobs
             self.cli.print_server_jobs(detailed=True)

tests/test_cswarm_text_to_kg_client.py

@@ -534,11 +534,13 @@ class TestTextoToKGCLIJobPrinting:
                 "type": "UnstructuredWithOntology",
                 "status": "Completed",
                 "retries_count": 0,
-                "file_type": "JSONL"
+                "file_type": "JSONL",
+                "machine_id": "machine-001",
+                "worker_id": "worker-001"
             }
         ]
         mock_list_jobs.return_value = mock_jobs
-        
+    
         self.cli.print_server_jobs()
         
         captured = capsys.readouterr()
@@ -556,13 +558,15 @@ class TestTextoToKGCLIJobPrinting:
                 "status": "Completed",
                 "retries_count": 0,
                 "file_type": "JSONL",
+                "machine_id": "machine-001",
+                "worker_id": "worker-001",
                 "unstructured_sources": ["test.txt"],
                 "ontology_sources": ["test.json"],
                 "ontology_spec_sources": ["test.owl"]
             }
         ]
         mock_list_jobs.return_value = mock_jobs
-        
+    
         self.cli.print_server_jobs(is_detailed=True)
         
         captured = capsys.readouterr()
@@ -582,6 +586,8 @@ class TestTextoToKGCLIJobPrinting:
                 "status": "Completed",
                 "retries_count": 0,
                 "file_type": "JSONL",
+                "machine_id": "machine-001",
+                "worker_id": "worker-001",
                 "unstructured_sources": ["test.txt"],
                 "ontology_sources": ["test.json"],
                 "ontology_spec_sources": ["test.owl"],
@@ -589,7 +595,7 @@ class TestTextoToKGCLIJobPrinting:
             }
         ]
         mock_list_jobs.return_value = mock_jobs
-        
+    
         self.cli.print_server_jobs(is_detailed=True)
         
         captured = capsys.readouterr()

tests/test_cswarm_text_to_kg_client_extended.py

@@ -182,13 +182,15 @@ class TestTextoToKGCLIExtended:
                     "status": "Completed",
                     "retries_count": 0,
                     "file_type": "JSONL",
+                    "machine_id": "machine-001",
+                    "worker_id": "worker-001",
                     "unstructured_sources": ["test.txt"],
                     "ontology_sources": ["test.json"],
                     "ontology_spec_sources": ["test.owl"]
                 }
             ]
             mock_client.get_jobs_list.return_value = mock_jobs
-            
+    
             with patch('builtins.print') as mock_print:
                 self.cli.print_server_jobs()
                 assert mock_print.call_count > 0
@@ -204,6 +206,8 @@ class TestTextoToKGCLIExtended:
                     "status": "Completed",
                     "retries_count": 0,
                     "file_type": "JSONL",
+                    "machine_id": "machine-001",
+                    "worker_id": "worker-001",
                     "unstructured_sources": ["test.txt"],
                     "ontology_sources": ["test.json"],
                     "ontology_spec_sources": ["test.owl"],
@@ -211,7 +215,7 @@ class TestTextoToKGCLIExtended:
                 }
             ]
             mock_client.get_jobs_list.return_value = mock_jobs
-            
+    
             with patch('builtins.print') as mock_print:
                 self.cli.print_server_jobs(is_detailed=True)
                 assert mock_print.call_count > 0
@@ -227,13 +231,15 @@ class TestTextoToKGCLIExtended:
                     "status": "Completed",
                     "retries_count": 0,
                     "file_type": "JSONL",
+                    "machine_id": "machine-001",
+                    "worker_id": "worker-001",
                     "unstructured_sources": ["test.txt"],
                     "ontology_sources": ["test.json"],
                     "ontology_spec_sources": ["test.owl"]
                 }
             ]
             mock_client.get_jobs_list.return_value = mock_jobs
-            
+    
             with patch('builtins.print') as mock_print:
                 self.cli.print_server_jobs(is_detailed=True)
                 assert mock_print.call_count > 0

tests/test_detailed_eval_extended.py

@@ -43,7 +43,10 @@ class TestDetailedEvalExtendedSimple:
         encoder = CustomEncoder()
         test_set = {1, 2, 3, "test"}
         result = encoder.default(test_set)
-        assert result == [1, 2, 3, "test"]
+        # Convert result back to set for order-independent comparison
+        assert set(result) == test_set
+        # Also verify it's a list
+        assert isinstance(result, list)
 
     def test_custom_encoder_with_non_set(self):
         """Test CustomEncoder with non-set objects."""

tests/test_file_type_enum.py

@@ -30,6 +30,7 @@ class TestFileTypeAPI:
         assert FileTypeAPI.AutoDetect == "AutoDetect"
         assert FileTypeAPI.Markdown == "Markdown"
         assert FileTypeAPI.PlainText == "PlainText"
+        assert FileTypeAPI.PDF == "PDF"
 
     def test_enum_membership(self):
         """Test that FileTypeAPI members can be checked for membership."""
@@ -37,6 +38,7 @@ class TestFileTypeAPI:
         assert FileTypeAPI.AutoDetect in FileTypeAPI
         assert FileTypeAPI.Markdown in FileTypeAPI
         assert FileTypeAPI.PlainText in FileTypeAPI
+        assert FileTypeAPI.PDF in FileTypeAPI
         
         # Test string value membership (Python 3.12+ only)
         import sys
@@ -44,13 +46,14 @@ class TestFileTypeAPI:
             assert "AutoDetect" in FileTypeAPI
             assert "Markdown" in FileTypeAPI
             assert "PlainText" in FileTypeAPI
+            assert "PDF" in FileTypeAPI
             assert "InvalidType" not in FileTypeAPI
 
     def test_enum_iteration(self):
         """Test that FileTypeAPI can be iterated over."""
         values = list(FileTypeAPI)
-        expected_values = ["AutoDetect", "Markdown", "PlainText"]
-        assert len(values) == 3
+        expected_values = ["AutoDetect", "Markdown", "PlainText", "PDF"]
+        assert len(values) == 4
         for value in expected_values:
             assert value in values
 
@@ -59,6 +62,7 @@ class TestFileTypeAPI:
         assert FileTypeAPI.AutoDetect == "AutoDetect"
         assert FileTypeAPI.Markdown == "Markdown"
         assert FileTypeAPI.PlainText == "PlainText"
+        assert FileTypeAPI.PDF == "PDF"
         assert FileTypeAPI.AutoDetect != FileTypeAPI.Markdown
 
     def test_enum_string_representation(self):
@@ -66,6 +70,7 @@ class TestFileTypeAPI:
         assert str(FileTypeAPI.AutoDetect) == "AutoDetect"
         assert str(FileTypeAPI.Markdown) == "Markdown"
         assert str(FileTypeAPI.PlainText) == "PlainText"
+        assert str(FileTypeAPI.PDF) == "PDF"
 
     def test_enum_repr(self):
         """Test repr representation of FileTypeAPI values."""
@@ -73,18 +78,14 @@ class TestFileTypeAPI:
         assert "FileTypeAPI.AutoDetect" in repr(FileTypeAPI.AutoDetect)
         assert "FileTypeAPI.Markdown" in repr(FileTypeAPI.Markdown)
         assert "FileTypeAPI.PlainText" in repr(FileTypeAPI.PlainText)
+        assert "FileTypeAPI.PDF" in repr(FileTypeAPI.PDF)
 
     def test_enum_inheritance(self):
         """Test that FileTypeAPI inherits from StrEnum."""
-        from enum import StrEnum
+        # Import StrEnum from the same module where FileTypeAPI is defined
+        from cleverswarm_python_client.libs.file_type_enum import StrEnum
         assert issubclass(FileTypeAPI, StrEnum)
 
-    def test_enum_equality_with_string(self):
-        """Test that FileTypeAPI values are equal to their string representations."""
-        assert FileTypeAPI.AutoDetect == "AutoDetect"
-        assert FileTypeAPI.Markdown == "Markdown"
-        assert FileTypeAPI.PlainText == "PlainText"
-
     def test_enum_case_sensitivity(self):
         """Test that FileTypeAPI values are case sensitive."""
         assert FileTypeAPI.AutoDetect != "autodetect"
@@ -100,6 +101,7 @@ class TestFileType:
         assert FileType.JSONL == "JSONL"
         assert FileType.Markdown == "Markdown"
         assert FileType.PlainText == "PlainText"
+        assert FileType.PDF == "PDF"
         assert FileType.Unknown == "Unknown"
 
     def test_enum_membership(self):
@@ -108,6 +110,7 @@ class TestFileType:
         assert FileType.JSONL in FileType
         assert FileType.Markdown in FileType
         assert FileType.PlainText in FileType
+        assert FileType.PDF in FileType
         assert FileType.Unknown in FileType
         
         # Test string value membership (Python 3.12+ only)
@@ -116,14 +119,15 @@ class TestFileType:
             assert "JSONL" in FileType
             assert "Markdown" in FileType
             assert "PlainText" in FileType
+            assert "PDF" in FileType
             assert "Unknown" in FileType
             assert "InvalidType" not in FileType
 
     def test_enum_iteration(self):
         """Test that FileType can be iterated over."""
         values = list(FileType)
-        expected_values = ["JSONL", "Markdown", "PlainText", "Unknown"]
-        assert len(values) == 4
+        expected_values = ["JSONL", "Markdown", "PlainText", "PDF", "Unknown"]
+        assert len(values) == 5
         for value in expected_values:
             assert value in values
 
@@ -132,6 +136,7 @@ class TestFileType:
         assert FileType.JSONL == "JSONL"
         assert FileType.Markdown == "Markdown"
         assert FileType.PlainText == "PlainText"
+        assert FileType.PDF == "PDF"
         assert FileType.Unknown == "Unknown"
         assert FileType.JSONL != FileType.Markdown
 
@@ -140,6 +145,7 @@ class TestFileType:
         assert str(FileType.JSONL) == "JSONL"
         assert str(FileType.Markdown) == "Markdown"
         assert str(FileType.PlainText) == "PlainText"
+        assert str(FileType.PDF) == "PDF"
         assert str(FileType.Unknown) == "Unknown"
 
     def test_enum_repr(self):
@@ -148,20 +154,15 @@ class TestFileType:
         assert "FileType.JSONL" in repr(FileType.JSONL)
         assert "FileType.Markdown" in repr(FileType.Markdown)
         assert "FileType.PlainText" in repr(FileType.PlainText)
+        assert "FileType.PDF" in repr(FileType.PDF)
         assert "FileType.Unknown" in repr(FileType.Unknown)
 
     def test_enum_inheritance(self):
         """Test that FileType inherits from StrEnum."""
-        from enum import StrEnum
+        # Import StrEnum from the same module where FileType is defined
+        from cleverswarm_python_client.libs.file_type_enum import StrEnum
         assert issubclass(FileType, StrEnum)
 
-    def test_enum_equality_with_string(self):
-        """Test that FileType values are equal to their string representations."""
-        assert FileType.JSONL == "JSONL"
-        assert FileType.Markdown == "Markdown"
-        assert FileType.PlainText == "PlainText"
-        assert FileType.Unknown == "Unknown"
-
     def test_enum_case_sensitivity(self):
         """Test that FileType values are case sensitive."""
         assert FileType.JSONL != "jsonl"

tests/test_job_enums.py

@@ -76,7 +76,8 @@ class TestJobType:
 
     def test_enum_inheritance(self):
         """Test that JobType inherits from StrEnum."""
-        from enum import StrEnum
+        # Import StrEnum from the same module where JobType is defined
+        from cleverswarm_python_client.libs.job_enums import StrEnum
         assert issubclass(JobType, StrEnum)
 
     def test_enum_equality_with_string(self):
@@ -165,7 +166,8 @@ class TestJobStatus:
 
     def test_enum_inheritance(self):
         """Test that JobStatus inherits from StrEnum."""
-        from enum import StrEnum
+        # Import StrEnum from the same module where JobStatus is defined
+        from cleverswarm_python_client.libs.job_enums import StrEnum
         assert issubclass(JobStatus, StrEnum)
 
     def test_enum_equality_with_string(self):

tests/test_response_type_enum.py

@@ -70,7 +70,8 @@ class TestResponseTypeAPI:
 
     def test_enum_inheritance(self):
         """Test that ResponseTypeAPI inherits from StrEnum."""
-        from enum import StrEnum
+        # Import StrEnum from the same module where ResponseTypeAPI is defined
+        from cleverswarm_python_client.libs.response_type_enum import StrEnum
         assert issubclass(ResponseTypeAPI, StrEnum)
 
     def test_enum_equality_with_string(self):