Made sdist/wheel 99.7% smaller (#39)

* Made sdist/wheel 99.7% smaller

* Linter

* Removed `lite` tests

* Removed pq-dim test

* Updated test workflow

* Updated test workflow

Author: dupontcyborg

.github/workflows/test.yml

@@ -82,49 +82,6 @@ jobs:
           # Install optional test dependencies
           pip install langchain langchain-community langchain-core sentence-transformers requests
 
-      - name: Test with Lite version
-        run: |
-          echo "=== Testing with Lite version ==="
-          # Set up environment for Lite
-          export CYBORGDB_API_KEY="${{ secrets.CYBORGDB_FREE_API_KEY }}"
-          export CYBORGDB_DB_TYPE=postgres
-          export CYBORGDB_CONNECTION_STRING="host=localhost port=5432 dbname=cyborgdb user=postgres password=postgres"
-          export CYBORGDB_SERVICE_LOG_LEVEL=DEBUG
-          
-          # Start the service with Lite API key
-          nohup cyborgdb-service > server-lite.log 2>&1 &
-          echo $! > server-lite.pid
-          
-          # Wait for service to be ready
-          for i in {1..30}; do
-            if curl -fs http://localhost:8000/v1/health > /dev/null; then
-              echo "Lite service is up!"
-              break
-            fi
-            sleep 1
-          done
-          
-          if ! curl -fs http://localhost:8000/v1/health > /dev/null; then
-            echo "Lite service failed to start"
-            cat server-lite.log
-            exit 1
-          fi
-          
-          # Run lite-compatible tests
-          # pytest tests/test_client_lite.py tests/test_langchain_integration.py -v --cov=cyborgdb --cov-report=term-missing --cov-append
-          pytest tests/test_client_lite.py -v --cov=cyborgdb --cov-report=term-missing --cov-append
-          
-          # Stop lite server
-          kill $(cat server-lite.pid) || true
-
-          # Remove cyborgdb-lite
-          pip uninstall -y cyborgdb-lite
-          
-          # Clean up PostgreSQL database to avoid conflicts between lite and standard
-          PGPASSWORD=postgres psql -h localhost -U postgres -d cyborgdb -c "DROP SCHEMA public CASCADE; CREATE SCHEMA public;"
-
-          sleep 2
-          
       - name: Test with Standard version
         run: |
           echo "=== Testing with Standard version ==="
@@ -153,8 +110,12 @@ jobs:
             exit 1
           fi
           
-          # Run all tests (full version tests will run, lite tests will also work)
+          # Run all tests
+          if [[ "${{ github.event_name }}" == "pull_request" ]]; then
+            pytest tests/quick_flow_test.py -v --cov=cyborgdb --cov-report=term-missing --cov-append
+          else
             pytest tests/ -v --cov=cyborgdb --cov-report=term-missing --cov-append --ignore=tests/test_api_contract.py
+          fi
           
           # Store the server pid for later cleanup
           echo "Standard server PID: $(cat server-standard.pid)"
@@ -167,9 +128,6 @@ jobs:
 
       - name: Kill Server
         run: |
-          if [ -f server-lite.pid ]; then
-            kill $(cat server-lite.pid) || true
-          fi
           if [ -f server-standard.pid ]; then
             kill $(cat server-standard.pid) || true
           fi

MANIFEST.in

@@ -0,0 +1,15 @@
+include README.md LICENSE pyproject.toml
+recursive-include cyborgdb *.py py.typed
+
+# Exclude everything else
+prune tests
+prune .github
+prune .openapi-generator
+global-exclude .gitignore
+global-exclude CODEOWNERS
+global-exclude *.json
+global-exclude *.sh
+exclude .openapi-generator-ignore
+exclude openapi.json
+exclude update-openapi-client.sh
+exclude CODEOWNERS

pyproject.toml

@@ -8,7 +8,7 @@ dynamic = ["version"]
 description = "Python Client SDK for CyborgDB: The Confidential Vector Database"
 readme = "README.md"
 requires-python = ">= 3.10"
-license = {text = "MIT"}
+license = "MIT"
 authors = [
     {name = "Cyborg Inc."}
 ]
@@ -53,6 +53,25 @@ langchain = [
 [tool.setuptools.packages.find]
 where = ["."]
 include = ["cyborgdb*"]
+exclude = ["tests*", "*.tests*", "*.tests"]
+
+[tool.setuptools]
+# Exclude files from source distribution
+include-package-data = false
+
+[tool.setuptools.package-data]
+cyborgdb = ["py.typed"]
+
+[tool.setuptools.exclude-package-data]
+"*" = [
+    "tests/*",
+    ".github/*",
+    ".openapi-generator/*",
+    "*.json",
+    "*.sh",
+    ".gitignore",
+    "CODEOWNERS",
+]
 
 [tool.setuptools_scm]
 version_scheme = "guess-next-dev"

tests/comprehensive_test.py

@@ -205,29 +205,27 @@ def test_ivfpq_index_creation_and_operations(self):
         query_vector = self.test_vectors[0]
         results = self.index.query(query_vectors=[query_vector], top_k=5)
 
-        
-
         self.assertGreater(len(results[0]), 0)
         self.assertTrue("id" in results[0])
 
-    def test_ivfpq_parameter_validation(self):
-        """Test IVFPQ parameter validation"""
-        # Test invalid pq_dim = 0
-        invalid_config = cyborgdb.IndexIVFPQ(
-            dimension=self.dimension, pq_dim=0, pq_bits=8
-        )
+    # def test_ivfpq_parameter_validation(self):
+    #     """Test IVFPQ parameter validation"""
+    #     # Test invalid pq_dim = 0
+    #     invalid_config = cyborgdb.IndexIVFPQ(
+    #         dimension=self.dimension, pq_dim=0, pq_bits=8
+    #     )
 
-        with self.assertRaises(Exception) as context:
-            invalid_index = self.client.create_index(
-                generate_unique_name(),
-                self.client.generate_key(),
-                invalid_config,
-                metric="euclidean",
-            )
-            invalid_index.delete_index()
+    #     with self.assertRaises(Exception) as context:
+    #         invalid_index = self.client.create_index(
+    #             generate_unique_name(),
+    #             self.client.generate_key(),
+    #             invalid_config,
+    #             metric="euclidean",
+    #         )
+    #         invalid_index.delete_index()
 
-        # Verify the error is about pq_dim
-        self.assertIn("pq_dim", str(context.exception).lower())
+    #     # Verify the error is about pq_dim
+    #     self.assertIn("pq_dim", str(context.exception).lower())
 
 
 class TestErrorHandling(unittest.TestCase):
@@ -394,7 +392,7 @@ def test_mismatched_parameter_lengths(self):
 
         # Test with empty items list
         with self.assertRaises(Exception):
-            result = self.index.upsert([])
+            self.index.upsert([])
 
     def test_content_preservation_through_operations(self):
         """Test that content is preserved through various operations"""
@@ -468,7 +466,7 @@ async def concurrent_upsert():
 
 
 class TestBackendCompatibility(unittest.TestCase):
-    """Test backend compatibility (Lite vs Full)"""
+    """Test backend compatibility"""
 
     def setUp(self):
         self.client = create_client()
@@ -490,12 +488,6 @@ def tearDown(self):
         except Exception:
             pass
 
-    def test_lite_backend_compatibility(self):
-        """Test operations with lite backend"""
-        client = create_client()
-        health = client.get_health()
-        self.assertIsInstance(health, (dict, bool, str, type(None)))
-
     def test_feature_availability_differences(self):
         """Test feature availability between backend variants"""
         client = create_client()

tests/test_api_contract.py

@@ -708,9 +708,7 @@ def test_15_encrypted_index_query(self):
 
         self.assertIsInstance(results, list)
         # Single query returns flat list of results
-        self.assertGreater(
-            len(results), 0, "Query should return at least one result"
-        )
+        self.assertGreater(len(results), 0, "Query should return at least one result")
 
         # Check exact structure with default include - id is always included, plus distance and metadata by default
         for result in results:
@@ -823,7 +821,9 @@ def test_16_encrypted_index_query_patterns(self):
         self.assertIsInstance(results, list)
         self.assertGreater(len(results), 0, "Should have results for single query")
         if len(results) > 0:
-            self.assertIsInstance(results[0], dict, "Single query should return flat list of dicts")
+            self.assertIsInstance(
+                results[0], dict, "Single query should return flat list of dicts"
+            )
             self.assertIn("id", results[0])
             self.assertIn("distance", results[0])
 

tests/test_client.py

@@ -33,13 +33,6 @@ def setUp(self):
         self.index = self.client.create_index(
             self.index_name, self.test_key, self.index_config, metric="euclidean"
         )
-        # except Exception as e:
-        #     # If IndexIVF fails (likely using lite backend), skip these tests
-        #     if "cyborgdb_lite" in str(e) or "IndexIVF" in str(e):
-        #         self.skipTest(
-        #             "Server is using lite backend, skipping full version tests"
-        #         )
-        #     raise
 
     def tearDown(self):
         """Clean up after tests."""

tests/test_client_lite.py

@@ -159,9 +159,7 @@ def test_get_demo_api_key_connection_error(self, mock_post):
         # Mock connection error
         import requests
 
-        mock_post.side_effect = requests.exceptions.ConnectionError(
-            "Failed to connect"
-        )
+        mock_post.side_effect = requests.exceptions.ConnectionError("Failed to connect")
 
         # Call the function and expect ValueError
         with self.assertRaises(ValueError) as context: