Merge remote-tracking branch 'origin/main' into piotrm/otel_db

.gitignore

@@ -66,3 +66,7 @@ htmlcov/
 # testing: parts of old trulens_eval package for backwards compatibility testing
 _trulens_eval/
 data/
+
+# Build objects
+**/poetry.lock
+!poetry.lock

.pre-commit-config.yaml

@@ -5,7 +5,11 @@ repos:
     hooks:
       - id: trailing-whitespace
       - id: check-added-large-files
-        exclude: src/dashboard/trulens/dashboard/.*/dist/.*
+        exclude: |
+          (?x)^(
+            src/dashboard/trulens/dashboard/.*/dist/.*|
+            poetry.lock
+          )$
       - id: check-yaml
         exclude: meta.yaml
       - id: destroyed-symlinks
@@ -44,42 +48,3 @@ repos:
     hooks:
       - id: poetry-check
         name: trulens-poetry-check
-      - id: poetry-check
-        name: trulens-core-poetry-check
-        args: [-C src/core]
-      - id: poetry-check
-        name: trulens-benchmark-poetry-check
-        args: [-C src/benchmark]
-      - id: poetry-check
-        name: trulens-feedback-poetry-check
-        args: [-C src/feedback]
-      - id: poetry-check
-        name: trulens-apps-langchain-poetry-check
-        args: [-C src/apps/langchain]
-      - id: poetry-check
-        name: trulens-apps-llamaindex-poetry-check
-        args: [-C src/apps/llamaindex]
-      - id: poetry-check
-        name: trulens-apps-nemo-poetry-check
-        args: [-C src/apps/nemo]
-      - id: poetry-check
-        name: trulens-providers-bedrock-poetry-check
-        args: [-C src/providers/bedrock]
-      - id: poetry-check
-        name: trulens-providers-cortex-poetry-check
-        args: [-C src/providers/cortex]
-      - id: poetry-check
-        name: trulens-providers-huggingface-poetry-check
-        args: [-C src/providers/huggingface]
-      - id: poetry-check
-        name: trulens-providers-langchain-poetry-check
-        args: [-C src/providers/langchain]
-      - id: poetry-check
-        name: trulens-providers-litellm-poetry-check
-        args: [-C src/providers/litellm]
-      - id: poetry-check
-        name: trulens-providers-openai-poetry-check
-        args: [-C src/providers/openai]
-      - id: poetry-check
-        name: trulens-connectors-snowflake-poetry-check
-        args: [-C src/connectors/snowflake]

Makefile

@@ -39,8 +39,8 @@ env-tests:
 		jsondiff
 
 env-tests-required:
-	poetry install --only required
-	make env-tests
+	poetry install --only required \
+		&& make env-tests
 
 env-tests-optional: env env-tests
 	poetry run pip install \

examples/expositional/models/snowflake_cortex/cortex_finetuning_experiments.ipynb

@@ -6,7 +6,7 @@
    "source": [
     "# Cortex Finetuning Experiments\n",
     "\n",
-    "This notebook takes you through evaluating a series of "
+    "This notebook takes you through evaluating a series of fine-tuning experiments with Snowflake Cortex, and uses TruLens to evaluate the fine-tuning effectiveness."
    ]
   },
   {

pyproject.toml

@@ -26,6 +26,10 @@ classifiers = [
   "License :: OSI Approved :: MIT License",
 ]
 
+[[tool.poetry.source]]
+name = "pypi-public"
+url = "https://pypi.org/simple/"
+
 [tool.poetry.dependencies]
 python = "^3.8.1,!=3.9.7"
 trulens-core = { version = "^1.0.0", extras = [
@@ -91,10 +95,6 @@ trulens-semconv = { path = "src/semconv", develop = true }
 # Remove after deprecation period.
 trulens_eval = { path = "src/trulens_eval", develop = true }
 
-[[tool.poetry.source]]
-name = "pypi-public"
-url = "https://pypi.org/simple/"
-
 [tool.ruff]
 line-length = 80
 extend-exclude = [

src/connectors/snowflake/trulens/connectors/snowflake/connector.py

@@ -155,6 +155,20 @@ def _validate_snowpark_session_with_connection_parameters(
             self.password_known = True
         return snowpark_session_connection_parameters
 
+    @staticmethod
+    def _validate_snowpark_session_paramstyle(
+        snowpark_session: Session,
+    ) -> None:
+        if snowpark_session.connection._paramstyle == "pyformat":
+            # If this is the case, sql executions with bindings will fail later
+            # on so we fail fast here.
+            raise ValueError(
+                "The Snowpark session must have paramstyle 'qmark'! To ensure"
+                " this, during `snowflake.connector.connect` pass in"
+                " `paramstyle='qmark'` or set"
+                " `snowflake.connector.paramstyle = 'qmark'` beforehand."
+            )
+
     def _init_with_snowpark_session(
         self,
         snowpark_session: Session,
@@ -166,6 +180,7 @@ def _init_with_snowpark_session(
         database_check_revision: bool,
         connection_parameters: Dict[str, str],
     ):
+        self._validate_snowpark_session_paramstyle(snowpark_session)
         database_args = self._set_up_database_args(
             database_args,
             snowpark_session,

src/core/trulens/experimental/otel_tracing/core/trace/__init__.py

@@ -1,7 +1,9 @@
 """
-Modules in this folder depend on each other than makes it impossible to import
-them linearly. Because of this, they need to be imported here and the delayed
-typing information needs to be filled in afterwards with model_rebuild`.
+Modules in this folder depend on each other which makes it impossible to import
+them without circular import errors. Because of this, some imports need to be
+put into `if TYPE_CHECKING` blocks and classes that depend on those imports need
+to be "rebuilt" with `model_rebuild`. This only applies to `pydantic.BaseModel`
+classes.
 """
 
 from . import context as core_context
@@ -10,12 +12,12 @@
 from . import span as core_span
 from . import trace as core_trace
 
-core_trace.Tracer.model_rebuild()
-core_span.Span.model_rebuild()
-core_otel.Span.model_rebuild()
 core_context.SpanContext.model_rebuild()
 core_context.TraceState.model_rebuild()
-core_sem.TypedSpan.model_rebuild()
-core_trace.TracerProvider.model_rebuild()
+core_otel.Span.model_rebuild()
 core_otel.Tracer.model_rebuild()
 core_otel.TracerProvider.model_rebuild()
+core_span.Span.model_rebuild()
+core_sem.TypedSpan.model_rebuild()
+core_trace.Tracer.model_rebuild()
+core_trace.TracerProvider.model_rebuild()

tests/e2e/test_snowflake_connection.py

@@ -5,6 +5,9 @@
 from unittest import main
 import uuid
 
+import snowflake.connector
+from snowflake.snowpark import Session
+from trulens.connectors.snowflake import SnowflakeConnector
 from trulens.dashboard import run_dashboard
 from trulens.dashboard import stop_dashboard
 
@@ -68,6 +71,37 @@ def test_run_leaderboard_without_password(self):
             except Exception:
                 pass
 
+    @optional_test
+    def test_paramstyle_pyformat(self):
+        default_paramstyle = snowflake.connector.paramstyle
+        try:
+            # pyformat paramstyle should fail fast.
+            snowflake.connector.paramstyle = "pyformat"
+            schema_name = self.create_and_use_schema(
+                "test_paramstyle_pyformat", append_uuid=True
+            )
+            snowflake_connection = snowflake.connector.connect(
+                **self._snowflake_connection_parameters, schema=schema_name
+            )
+            snowpark_session = Session.builder.configs({
+                "connection": snowflake_connection
+            }).create()
+            with self.assertRaisesRegex(
+                ValueError, "The Snowpark session must have paramstyle 'qmark'!"
+            ):
+                SnowflakeConnector(snowpark_session=snowpark_session)
+            # qmark paramstyle should be fine.
+            snowflake.connector.paramstyle = "qmark"
+            snowflake_connection = snowflake.connector.connect(
+                **self._snowflake_connection_parameters, schema=schema_name
+            )
+            snowpark_session = Session.builder.configs({
+                "connection": snowflake_connection
+            }).create()
+            SnowflakeConnector(snowpark_session=snowpark_session)
+        finally:
+            snowflake.connector.paramstyle = default_paramstyle
+
 
 if __name__ == "__main__":
     main()

tests/e2e/test_snowflake_notebooks.py

@@ -1,7 +1,6 @@
 import tempfile
 from typing import Sequence
 from unittest import main
-import uuid
 
 from trulens.connectors.snowflake.utils.server_side_evaluation_artifacts import (
     _TRULENS_PACKAGES,
@@ -18,8 +17,7 @@
 
 class TestSnowflakeNotebooks(SnowflakeTestCase):
     def test_simple(self) -> None:
-        schema_name = f"test_simple_{str(uuid.uuid4()).replace('-', '_')}"
-        self.create_and_use_schema(schema_name)
+        self.create_and_use_schema("test_simple", append_uuid=True)
         self._upload_and_run_notebook("simple", _TRULENS_PACKAGES)
 
     def test_staged_packages(self) -> None:

tests/util/snowflake_test_case.py

@@ -117,10 +117,19 @@ def run_query(
     ) -> List[Row]:
         return self._snowpark_session.sql(q, bindings).collect()
 
-    def create_and_use_schema(self, schema_name: str) -> None:
+    def create_and_use_schema(
+        self, schema_name: str, append_uuid: bool = False
+    ) -> str:
+        schema_name = schema_name.upper()
+        if append_uuid:
+            schema_name = (
+                f"{schema_name}__{str(uuid.uuid4()).replace('-', '_')}"
+            )
+        self._schema = schema_name
         self.run_query("CREATE SCHEMA IDENTIFIER(?)", [schema_name])
         self._snowflake_schemas_to_delete.add(schema_name)
         self._snowpark_session.use_schema(schema_name)
+        return schema_name
 
 
 if __name__ == "__main__":