feat: update notebooks (#892)

👋 Thanks for submitting a Pull Request to EvaDB!

🙌 We want to make contributing to EvaDB as easy and transparent as
possible. Here are a few tips to get you started:

- 🔍 Search existing EvaDB
[PRs](https://github.com/georgia-tech-db/eva/pulls) to see if a similar
PR already exists.
- 🔗 Link this PR to a EvaDB
[issue](https://github.com/georgia-tech-db/eva/issues) to help us
understand what bug fix or feature is being implemented.
- 📈 Provide before and after profiling results to help us quantify the
improvement your PR provides (if applicable).

👉 Please see our ✅ [Contributing
Guide](https://evadb.readthedocs.io/en/stable/source/contribute/index.html)
for more details.

---------

Co-authored-by: Gaurav <gaurav21776@gmail.com>

.circleci/config.yml

@@ -7,64 +7,114 @@ orbs:
 workflows:
   main:
     jobs:
+        ################################
+        #### PIP
+        ################################
+        ################################
         - Pip-PyPI
         - Pip-Local
-        - Docker-Local
-        - AWS
+        ################################
         #### UNIT TESTS
+        ################################
+        ################################
         - Linux:
             name: "Test | v3.8 | Linux"
             v: "3.8"
-            mode: "COV"
+            mode: "TEST"
+            ray: "DISABLED"
         - Linux:
             name: "Test | v3.9 | Linux"
             v: "3.9"
-            mode: "COV"
+            mode: "TEST"
+            ray: "DISABLED"
         - Linux:
             name: "Test | v3.10 | Linux"
             v: "3.10"
-            mode: "COV"
+            mode: "TEST"
+            ray: "DISABLED"
+        - Linux:
+            name: "Test | Ray | v3.8 | Linux"
+            v: "3.8"
+            mode: "TEST"
+            ray: "ENABLED"
+        - Linux:
+            name: "Test | Ray | v3.9 | Linux"
+            v: "3.9"
+            mode: "TEST"
+            ray: "ENABLED"
+        - Linux:
+            name: "Test | Ray | v3.10 | Linux"
+            v: "3.10"
+            mode: "TEST"
+            ray: "ENABLED"            
         # Ray does not work on 3.11
         # https://github.com/ray-project/ray/issues/33232
         # - Linux:
         #     name: "Test | v3.11 | Linux"
         #     v: "3.11"
         #     mode: "TEST"
+        #     ray: "DISABLED"
+        ################################
         ### NOTEBOOKS
+        ################################
+        ################################
         - Linux:
             name: "Notebook | v3.8 | Linux"
             v: "3.8"
             mode: "NOTEBOOK"
+            ray: "DISABLED"
         - Linux:
             name: "Notebook | v3.9 | Linux"
             v: "3.9"
             mode: "NOTEBOOK"
+            ray: "DISABLED"
         - Linux:
             name: "Notebook | v3.10 | Linux"
             v: "3.10"
             mode: "NOTEBOOK"
+            ray: "DISABLED"
+        - Linux:
+            name: "Notebook | Ray | v3.8 | Linux"
+            v: "3.8"
+            mode: "NOTEBOOK"
+            ray: "ENABLED"
+        - Linux:
+            name: "Notebook | Ray | v3.9 | Linux"
+            v: "3.9"
+            mode: "NOTEBOOK"
+            ray: "ENABLED"            
+        - Linux:
+            name: "Notebook |  Ray | v3.10 | Linux"
+            v: "3.10"
+            mode: "NOTEBOOK"
+            ray: "ENABLED"            
         # - Linux:
         #     name: "Notebook | v3.11 | Linux"
         #     v: "3.11"
         #     mode: "NOTEBOOK"
+        ################################
         ## LINTER
+        ################################
+        ################################
         - Linux:
             name: "Linter | Linux"
             v: "3.10"
             mode: "LINTER"
-        ### RAY
-        - Linux:
-            name: "Test | Ray | v3.10 | Linux"
-            v: "3.10"
-            mode: "RAY"
-        - Windows:
-            name: "Windows | v3.10"
-        - MacOS:
-           name: "MacOS | v3.10"
-        # missing Torchvision
-        #- Linux:
-        #    name: "Linux -  v3.11"  
-        #    v: "3.11"
+            ray: "DISABLED"
+        ################################
+        # OTHER OPERATING SYSTEMS
+        ################################
+        ################################
+        #- Windows:
+        #    name: "Windows | v3.10"
+        # - MacOS:
+        #    name: "MacOS | v3.10"
+        ################################
+        # DOCKER AND CLOUD
+        ################################
+        ################################
+        - Docker-Local
+        - AWS
 
 jobs:
   Linux:
@@ -75,6 +125,9 @@ jobs:
       mode:
         type: string
         default: "ALL"
+      ray:
+        type: string
+        default: "DISABLED"
     resource_class: large
     docker:
       # https://circleci.com/docs/circleci-images#language-image-variants
@@ -88,23 +141,25 @@ jobs:
             - v1-model_cache-{{ checksum "setup.py" }}
 
       - run:
-          name: Install EVA package from GitHub repo with all dependencies
+          name: Install EvaDB package from GitHub repo with all dependencies
           command: |
             "python<< parameters.v >>" -m venv test_evadb
             pip install --upgrade pip
             source test_evadb/bin/activate
             pip install ".[dev]"
+            pip uninstall -y ray
 
-      # Enable Ray
+      # Enable Ray (update evadb.yml file and install Ray package)
       - when:
           condition:
-            equal: [ RAY, << parameters.mode >> ]
+            equal: [ ENABLED, << parameters.ray >> ]
           steps:
             - run:
                 name: Enable Ray setting in the config.yml file
                 command: |
                   source test_evadb/bin/activate
                   python -c "import yaml;f = open('evadb/evadb.yml', 'r+');config_obj = yaml.load(f, Loader=yaml.FullLoader);config_obj['experimental']['ray'] = True;f.seek(0);f.write(yaml.dump(config_obj));f.truncate();"
+                  pip install ".[dev]"
 
       - run:
           name: Test and upload coverage report to coveralls

evadb/catalog/catalog_manager.py

@@ -91,6 +91,14 @@ def reset(self):
         """
         self._clear_catalog_contents()
 
+    def close(self):
+        """
+        This method closes all the connections
+        """
+        if self.sql_config is not None:
+            sqlalchemy_engine = self.sql_config.engine
+            sqlalchemy_engine.dispose()
+
     def _bootstrap_catalog(self):
         """Bootstraps catalog.
         This method runs all tasks required for using catalog. Currently,

evadb/catalog/sql_config.py

@@ -61,7 +61,7 @@ def __init__(self, uri):
 
         self.worker_uri = str(uri)
         # set echo=True to log SQL
-        self.engine = create_engine(self.worker_uri)
+        self.engine = create_engine(self.worker_uri, connect_args={"timeout": 1000})
 
         if self.engine.url.get_backend_name() == "sqlite":
             # enforce foreign key constraint and wal logging for sqlite
@@ -80,3 +80,4 @@ def _enable_sqlite_pragma(dbapi_con, con_record):
             event.listen(self.engine, "connect", _enable_sqlite_pragma)
         # statements
         self.session = scoped_session(sessionmaker(bind=self.engine))
+        self.session.close()

evadb/evadb.yml

@@ -25,7 +25,7 @@ server:
   socket_timeout: 60
 
 experimental:
-  ray: False
+  ray: False 
 
 third_party:
   OPENAI_KEY: ""

evadb/interfaces/relational/db.py

@@ -41,6 +41,7 @@
     parse_table_clause,
 )
 from evadb.udfs.udf_bootstrap_queries import init_builtin_udfs
+from evadb.utils.generic_utils import is_ray_enabled_and_installed
 from evadb.utils.logging_manager import logger
 
 
@@ -508,6 +509,25 @@ def rename(self, table_name, new_table_name, **kwargs) -> EvaDBQuery:
         stmt = parse_rename(table_name, new_table_name, **kwargs)
         return EvaDBQuery(self._evadb, stmt)
 
+    def close(self):
+        """
+        Closes the connection.
+
+        Args: None
+
+        Returns:  None
+
+        Examples:
+            >>> cursor.close()
+        """
+        self._evadb.catalog().close()
+
+        ray_enabled = self._evadb.config.get_value("experimental", "ray")
+        if is_ray_enabled_and_installed(ray_enabled):
+            import ray
+
+            ray.shutdown()
+
 
 def connect(
     evadb_dir: str = EvaDB_DATABASE_DIR, sql_backend: str = None

evadb/optimizer/rules/rules_manager.py

@@ -62,7 +62,7 @@
     XformLateralJoinToLinearFlow,
 )
 from evadb.optimizer.rules.rules_base import Rule
-from evadb.utils.generic_utils import is_ray_available
+from evadb.utils.generic_utils import is_ray_enabled_and_installed
 
 
 class RulesManager:
@@ -116,8 +116,13 @@ def __init__(self, config: ConfigurationManager):
             LogicalVectorIndexScanToPhysical(),
         ]
 
+        # These rules are enabled only if
+        # (1) ray is installed and (2) ray is enabled
+        # Ray must be installed using pip
+        # It must also be enabled in "evadb.yml"
+        # NOTE: By default, it is not enabled
         ray_enabled = config.get_value("experimental", "ray")
-        if ray_enabled and is_ray_available():
+        if is_ray_enabled_and_installed(ray_enabled):
             self._implementation_rules.extend(
                 [
                     LogicalExchangeToPhysical(),

evadb/udfs/chatgpt.py

@@ -111,8 +111,13 @@ def completion_with_backoff(**kwargs):
                 ],
             }
 
-            response = completion_with_backoff(**params)
-            results.append(response.choices[0].message.content)
+            try:
+                response = completion_with_backoff(**params)
+                results.append(response.choices[0].message.content)
+            # https://help.openai.com/en/articles/6897213-openai-library-error-types-guidance
+            # ignore API rate limit error etc.
+            except Exception as e:
+                results.append(f"{e}")
 
         df = pd.DataFrame({"response": results})
 

evadb/udfs/ocr_extractor.py

@@ -36,6 +36,8 @@ def setup(self):
         try_to_import_torch()
         try_to_import_torchvision()
         try_to_import_transformers()
+        # https://developers.google.com/protocol-buffers/docs/news/2022-05-06#python-updates
+
         from transformers import DonutProcessor, VisionEncoderDecoderModel
 
         self.processor = DonutProcessor.from_pretrained(

evadb/udfs/udf_bootstrap_queries.py

@@ -218,10 +218,17 @@ def init_builtin_udfs(db: EvaDBDatabase, mode: str = "debug") -> None:
     # "RuntimeError: random_device could not be read"
     # The suspicion is that importing torch prior to decord resolves this issue
     try:
-        import torch  # noqa
+        import torch  # noqa: F401
     except ImportError:
         pass
 
+    # Enable environment variables
+    # Relevant for ocr and other transformer-based models
+    import os
+
+    os.environ["PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION"] = "python"
+    os.environ["TOKENIZERS_PARALLELISM"] = "false"
+
     # list of UDF queries to load
     queries = [
         mnistcnn_udf_query,
@@ -236,10 +243,10 @@ def init_builtin_udfs(db: EvaDBDatabase, mode: str = "debug") -> None:
         # ocr_udf_query,
         # Mvit_udf_query,
         Sift_udf_query,
-        Yolo_udf_query,
+        yolo8n_query,
     ]
 
-    # if mode is 'debug', add debug UDFs and a smaller Yolo model
+    # if mode is 'debug', add debug UDFs
     if mode == "debug":
         queries.extend(
             [
@@ -248,8 +255,6 @@ def init_builtin_udfs(db: EvaDBDatabase, mode: str = "debug") -> None:
                 DummyFeatureExtractor_udf_query,
             ]
         )
-        queries.remove(Yolo_udf_query)
-        queries.append(yolo8n_query)
 
     # execute each query in the list of UDF queries
     # ignore exceptions during the bootstrapping phase due to missing packages

evadb/utils/generic_utils.py

@@ -239,11 +239,16 @@ def try_to_import_ray():
 def is_ray_available() -> bool:
     try:
         try_to_import_ray()
-        return False
+        return True
     except ValueError:  # noqa: E722
         return False
 
 
+def is_ray_enabled_and_installed(ray_enabled: bool) -> bool:
+    ray_installed = is_ray_available()
+    return ray_enabled and ray_installed
+
+
 ##############################
 ## VISION
 ##############################

script/formatting/formatter.py

@@ -233,6 +233,7 @@ def format_file(file_path, add_header, strip_header, format_code):
 
 # check the notebooks
 def check_notebook_format(notebook_file):
+    print(notebook_file)
     notebook_file_name = os.path.basename(notebook_file)
 
     # Ignore this notebook