qtype 0.1.2__tar.gz → 0.1.4__tar.gz

{qtype-0.1.2/qtype.egg-info → qtype-0.1.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: qtype
-Version: 0.1.2
+Version: 0.1.4
 Summary: DSL for Generative AI Prototyping
 Author-email: Lou Kratz <lou.kratz+qtype@bazaarvoice.com>
 License-Expression: Apache-2.0

{qtype-0.1.2 → qtype-0.1.4}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "qtype"
-version = "0.1.2"
+version = "0.1.4"
 description = "DSL for Generative AI Prototyping"
 authors = [{ name="Lou Kratz", email="lou.kratz+qtype@bazaarvoice.com" }]
 readme = "README.md"

{qtype-0.1.2 → qtype-0.1.4}/qtype/dsl/model.py

@@ -765,8 +765,9 @@ class AWSAuthProvider(AuthorizationProvider):
         has_keys = self.access_key_id and self.secret_access_key
         has_profile = self.profile_name
         has_role = self.role_arn
+        has_region = self.region
 
-        if not (has_keys or has_profile or has_role):
+        if not (has_keys or has_profile or has_role or has_region):
             raise ValueError(
                 "AWSAuthProvider must specify at least one authentication method: "
                 "access keys, profile name, or role ARN."

{qtype-0.1.2 → qtype-0.1.4}/qtype/dsl/parser.py

@@ -153,12 +153,12 @@ def _format_validation_errors(
         error_msg = "Validation failed (see details above)"
     else:
         error_msg = "Validation failed:\n"
-        for error in relevant_errors[:5]:  # Show max 5 errors
+        for error in relevant_errors[:30]:  # Show max 5 errors
             loc_path = _simplify_field_path(error["loc"])
             error_msg += f"  {loc_path}: {error['msg']}\n"
 
-        if len(relevant_errors) > 5:
-            error_msg += f"  ... and {len(relevant_errors) - 5} more errors\n"
+        if len(relevant_errors) > 30:
+            error_msg += f"  ... and {len(relevant_errors) - 30} more errors\n"
 
     if source_name:
         error_msg = f"In {source_name}:\n{error_msg}"

{qtype-0.1.2 → qtype-0.1.4}/qtype/interpreter/base/base_step_executor.py

@@ -241,7 +241,7 @@ class StepExecutor(ABC):
 
                 # Track message counts for telemetry
                 message_count = 0
-                error_count = len(failed_messages)
+                error_count = 0
 
                 # Stream results and track progress
                 async with all_results.stream() as streamer:

{qtype-0.1.2 → qtype-0.1.4}/qtype/interpreter/conversions.py

@@ -34,6 +34,7 @@ from qtype.interpreter.base.secrets import SecretManagerBase
 from qtype.interpreter.types import InterpreterError
 from qtype.semantic.model import (
     APIKeyAuthProvider,
+    AWSAuthProvider,
     DocumentIndex,
     DocumentSplitter,
     Model,
@@ -302,7 +303,9 @@ def to_vector_store(
 
 
 @cached_resource
-def to_embedding_model(model: Model) -> BaseEmbedding:
+def to_embedding_model(
+    model: Model, secret_manager: SecretManagerBase
+) -> BaseEmbedding:
     """Convert a qtype Model to a LlamaIndex embedding model."""
 
     if model.provider == "aws-bedrock":
@@ -310,7 +313,14 @@ def to_embedding_model(model: Model) -> BaseEmbedding:
             BedrockEmbedding,
         )
 
+        session = None
+        if model.auth is not None:
+            assert isinstance(model.auth, AWSAuthProvider)
+            with aws(model.auth, secret_manager) as session:
+                session = session._session
+
         bedrock_embedding: BaseEmbedding = BedrockEmbedding(
+            botocore_session=session,
             model_name=model.model_id if model.model_id else model.id,
             max_retries=100,
         )
@@ -320,8 +330,20 @@ def to_embedding_model(model: Model) -> BaseEmbedding:
             OpenAIEmbedding,
         )
 
+        api_key = None
+        if model.auth:
+            with auth(model.auth, secret_manager) as provider:
+                if not isinstance(provider, APIKeyAuthProvider):
+                    raise InterpreterError(
+                        f"OpenAI provider requires APIKeyAuthProvider, "
+                        f"got {type(provider).__name__}"
+                    )
+                # api_key is guaranteed to be str after auth() resolves it
+                api_key = provider.api_key  # type: ignore[assignment]
+
         openai_embedding: BaseEmbedding = OpenAIEmbedding(
-            model_name=model.model_id if model.model_id else model.id
+            api_key=api_key,
+            model_name=model.model_id if model.model_id else model.id,
         )
         return openai_embedding
     else:
@@ -566,7 +588,7 @@ def to_llama_vector_store_and_retriever(
     vector_store = to_vector_store(index, secret_manager)
 
     # Get the embedding model
-    embedding_model = to_embedding_model(index.embedding_model)
+    embedding_model = to_embedding_model(index.embedding_model, secret_manager)
 
     # Create a VectorStoreIndex with the vector store and embedding model
     vector_index = VectorStoreIndex.from_vector_store(

{qtype-0.1.2 → qtype-0.1.4}/qtype/interpreter/executors/document_embedder_executor.py

@@ -41,7 +41,7 @@ class DocumentEmbedderExecutor(StepExecutor):
         self.step: DocumentEmbedder = step
         # Initialize the embedding model once for the executor
         self.embedding_model: BaseEmbedding = to_embedding_model(
-            self.step.model
+            self.step.model, context.secret_manager
         )
 
     # TODO: properly abstract this into a mixin

{qtype-0.1.2 → qtype-0.1.4}/qtype/interpreter/executors/document_splitter_executor.py

@@ -71,7 +71,7 @@ class DocumentSplitterExecutor(StepExecutor):
             # Convert to LlamaIndex Document for splitting
             llama_doc = LlamaDocument(
                 text=content_text,
-                metadata=document.metadata or {},
+                metadata={},  # ommit metadata here as it's added back later and we don't want the chunk size checks
                 doc_id=document.file_id,
             )
 

{qtype-0.1.2 → qtype-0.1.4}/qtype/interpreter/executors/invoke_embedding_executor.py

@@ -30,7 +30,9 @@ class InvokeEmbeddingExecutor(StepExecutor):
             )
         self.step: InvokeEmbedding = step
         # Initialize the embedding model once for the executor
-        self.embedding_model = to_embedding_model(self.step.model)
+        self.embedding_model = to_embedding_model(
+            self.step.model, context.secret_manager
+        )
 
     async def process_message(
         self,

{qtype-0.1.2 → qtype-0.1.4}/qtype/interpreter/flow.py

@@ -11,9 +11,12 @@ from openinference.semconv.trace import (
 from opentelemetry import context as otel_context
 from opentelemetry import trace
 from opentelemetry.trace import Status, StatusCode
+from rich.console import Console
+from transformers import ProgressCallback
 
 from qtype.interpreter.base import factory
 from qtype.interpreter.base.executor_context import ExecutorContext
+from qtype.interpreter.logging_progress import LoggingProgressCallback
 from qtype.interpreter.rich_progress import RichProgressCallback
 from qtype.interpreter.types import FlowMessage
 from qtype.semantic.model import Flow
@@ -42,9 +45,17 @@ async def run_flow(
     """
     from qtype.interpreter.base.secrets import NoOpSecretManager
 
+    # Wire up progress callback if requested
+    progress_callback: ProgressCallback | None = None
+    if show_progress:
+        console = Console()
+        if console.is_terminal:
+            progress_callback = RichProgressCallback()
+        else:
+            progress_callback = LoggingProgressCallback(log_every_seconds=120)
+
     # Extract or create ExecutorContext
     exec_context = kwargs.pop("context", None)
-    progress_callback = RichProgressCallback() if show_progress else None
     if exec_context is None:
         exec_context = ExecutorContext(
             secret_manager=NoOpSecretManager(),

qtype-0.1.4/qtype/interpreter/logging_progress.py

@@ -0,0 +1,61 @@
+import logging
+import time
+
+from qtype.interpreter.types import ProgressCallback
+
+
+class LoggingProgressCallback(ProgressCallback):
+    def __init__(self, log_every_seconds: float = 120.0) -> None:
+        super().__init__()
+        self.log_every_seconds = log_every_seconds
+        self._last_log: dict[str, float] = {}
+        self._totals: dict[str, int | None] = {}
+
+    def __call__(
+        self,
+        step_id: str,
+        items_processed: int,
+        items_in_error: int,
+        items_succeeded: int,
+        total_items: int | None,
+        cache_hits: int | None,
+        cache_misses: int | None,
+    ) -> None:
+        logger = logging.getLogger(__name__)
+        now = time.monotonic()
+        last = self._last_log.get(step_id, 0.0)
+
+        self._totals[step_id] = total_items
+
+        if now - last < self.log_every_seconds:
+            return
+
+        self._last_log[step_id] = now
+        total_str = (
+            f"{items_processed}/{total_items}"
+            if total_items is not None
+            else f"{items_processed}"
+        )
+        if cache_hits is not None or cache_misses is not None:
+            logger.info(
+                "Step %s: processed=%s, succeeded=%s, errors=%s, "
+                "cache_hits=%s, cache_misses=%s",
+                step_id,
+                total_str,
+                items_succeeded,
+                items_in_error,
+                cache_hits if cache_hits is not None else "-",
+                cache_misses if cache_misses is not None else "-",
+            )
+        else:
+            logger.info(
+                "Step %s: processed=%s, succeeded=%s, errors=%s",
+                step_id,
+                total_str,
+                items_succeeded,
+                items_in_error,
+            )
+
+    def close(self) -> None:
+        # optional: final summary logging
+        pass

{qtype-0.1.2 → qtype-0.1.4/qtype.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: qtype
-Version: 0.1.2
+Version: 0.1.4
 Summary: DSL for Generative AI Prototyping
 Author-email: Lou Kratz <lou.kratz+qtype@bazaarvoice.com>
 License-Expression: Apache-2.0

{qtype-0.1.2 → qtype-0.1.4}/qtype.egg-info/SOURCES.txt

@@ -43,6 +43,7 @@ qtype/interpreter/conversions.py
 qtype/interpreter/converters.py
 qtype/interpreter/endpoints.py
 qtype/interpreter/flow.py
+qtype/interpreter/logging_progress.py
 qtype/interpreter/metadata_api.py
 qtype/interpreter/resource_cache.py
 qtype/interpreter/rich_progress.py