qtype 0.1.13__py3-none-any.whl → 0.1.14__py3-none-any.whl

qtype/interpreter/conversions.py

@@ -194,16 +194,15 @@ def to_llm(
 
         from qtype.semantic.model import AWSAuthProvider
 
+        creds_kwargs = {}
         if model.auth:
             # Type hint for mypy - we know it's AWSAuthProvider for aws-bedrock
             assert isinstance(model.auth, AWSAuthProvider)
-            with aws(model.auth, secret_manager) as session:
-                session = session._session
-        else:
-            session = None
+            with aws(model.auth, secret_manager) as creds:
+                creds_kwargs = creds.as_kwargs()
 
         brv: BaseLLM = BedrockConverse(
-            botocore_session=session,
+            **creds_kwargs,
             model=model.model_id if model.model_id else model.id,
             system_prompt=system_prompt,
             **(model.inference_params if model.inference_params else {}),
@@ -314,14 +313,14 @@ def to_embedding_model(
             BedrockEmbedding,
         )
 
-        session = None
+        creds_kwargs = {}
         if model.auth is not None:
             assert isinstance(model.auth, AWSAuthProvider)
-            with aws(model.auth, secret_manager) as session:
-                session = session._session
+            with aws(model.auth, secret_manager) as creds:
+                creds_kwargs = creds.as_kwargs()
 
         bedrock_embedding: BaseEmbedding = BedrockEmbedding(
-            botocore_session=session,
+            **creds_kwargs,
             model_name=model.model_id if model.model_id else model.id,
             max_retries=100,
         )
@@ -382,10 +381,12 @@ def to_opensearch_client(
         elif hasattr(index.auth, "type") and index.auth.type == "aws":
             # Use AWS authentication with boto3 session
             # Get AWS credentials from auth provider using context manager
-            with auth(index.auth, secret_manager) as auth_session:
-                # Type checker doesn't know this is a boto3.Session
-                # but runtime validation ensures it for AWS auth
-                credentials = auth_session.get_credentials()  # type: ignore
+            with auth(index.auth, secret_manager) as creds:
+                # Create a boto3 session from credentials to get boto3.Credentials
+                import boto3
+
+                session = boto3.Session(**creds.as_kwargs())
+                credentials = session.get_credentials()
                 if credentials is None:
                     raise InterpreterError(
                         f"Failed to obtain AWS credentials for DocumentIndex '{index.id}'"
@@ -394,7 +395,7 @@ def to_opensearch_client(
                 # Use opensearch-py's async AWS auth
                 aws_auth = AWSV4SignerAsyncAuth(
                     credentials,
-                    auth_session.region_name or "us-east-1",  # type: ignore
+                    creds.region_name or "us-east-1",
                     "aoss",  # service name for OpenSearch Serverless
                 )
 

qtype/interpreter/converters.py

@@ -126,7 +126,7 @@ def read_dataframe_from_file(
     with fsspec.open(file_path, "rb") as file_handle:
         # Read based on MIME type
         if mime_type == "text/csv" or mime_type == "text/plain":
-            df = pd.read_csv(file_handle)  # type: ignore[arg-type]
+            df = pd.read_csv(file_handle, keep_default_na=False)  # type: ignore[arg-type]
         elif mime_type in ["application/json", "application/jsonlines"]:
             # Check if it's JSONL by extension
             if Path(file_path).suffix.lower() == ".jsonl":

qtype/interpreter/executors/bedrock_reranker_executor.py

@@ -2,7 +2,6 @@
 
 from __future__ import annotations
 
-import asyncio
 import logging
 from typing import AsyncIterator
 
@@ -56,10 +55,10 @@ class BedrockRerankerExecutor(StepExecutor):
                 )
                 return
 
-            # Get session for region info
+            # Get region from auth or default session
             if self.step.auth is not None:
-                with aws(self.step.auth, self.context.secret_manager) as s:
-                    region_name = s.region_name
+                with aws(self.step.auth, self.context.secret_manager) as creds:
+                    region_name = creds.region_name
             else:
                 import boto3
 
@@ -120,31 +119,21 @@ class BedrockRerankerExecutor(StepExecutor):
                 },
             }
 
-            def _call_bedrock_rerank():
-                """Create client and call rerank in executor thread."""
-                if self.step.auth is not None:
-                    with aws(self.step.auth, self.context.secret_manager) as s:
-                        client = s.client("bedrock-agent-runtime")
-                        return client.rerank(
-                            queries=queries,
-                            sources=documents,
-                            rerankingConfiguration=reranking_configuration,
-                        )
-                else:
-                    import boto3
-
-                    session = boto3.Session()
-                    client = session.client("bedrock-agent-runtime")
-                    return client.rerank(
-                        queries=queries,
-                        sources=documents,
-                        rerankingConfiguration=reranking_configuration,
-                    )
+            # Create async bedrock client and call rerank
+            import aioboto3
 
-            loop = asyncio.get_running_loop()
-            response = await loop.run_in_executor(
-                self.context.thread_pool, _call_bedrock_rerank
-            )
+            creds_kwargs = {}
+            if self.step.auth is not None:
+                with aws(self.step.auth, self.context.secret_manager) as creds:
+                    creds_kwargs = creds.as_kwargs()
+
+            session = aioboto3.Session(**creds_kwargs)
+            async with session.client("bedrock-agent-runtime") as client:
+                response = await client.rerank(
+                    queries=queries,
+                    sources=documents,
+                    rerankingConfiguration=reranking_configuration,
+                )
 
             results = []
             for d in response["results"]:

qtype/interpreter/executors/document_embedder_executor.py

@@ -1,4 +1,3 @@
-import asyncio
 import logging
 from typing import AsyncIterator
 
@@ -60,17 +59,7 @@ class DocumentEmbedderExecutor(StepExecutor):
         Returns:
             The embedding vector as a list of floats.
         """
-
-        # TODO: switch back to async once aws auth supports it.
-        # https://github.com/bazaarvoice/qtype/issues/108
-        def _call():
-            return self.embedding_model.get_text_embedding(text=text)
-
-        loop = asyncio.get_running_loop()
-        response = await loop.run_in_executor(self.context.thread_pool, _call)
-
-        return response
-        # return await self.embedding_model.aget_text_embedding(text=text)
+        return await self.embedding_model.aget_text_embedding(text=text)
 
     async def process_message(
         self,

qtype/interpreter/executors/invoke_embedding_executor.py

@@ -1,4 +1,3 @@
-import asyncio
 from typing import AsyncIterator
 
 from openinference.semconv.trace import OpenInferenceSpanKindValues
@@ -54,41 +53,32 @@ class InvokeEmbeddingExecutor(StepExecutor):
             # Get the input value
             input_value = message.get_variable(input_id)
 
-            def _call(input_value=input_value):
-                # Generate embedding based on input type
-                if input_type == PrimitiveTypeEnum.text:
-                    if not isinstance(input_value, str):
-                        input_value = str(input_value)
-                    vector = self.embedding_model.get_text_embedding(
-                        text=input_value
-                    )
-                    content = input_value
-                elif input_type == PrimitiveTypeEnum.image:
-                    # For image embeddings
-                    vector = self.embedding_model.get_image_embedding(
-                        image_path=input_value
-                    )
-                    content = input_value
-                else:
-                    raise ValueError(
-                        (
-                            f"Unsupported input type for embedding: "
-                            f"{input_type}. Must be 'text' or 'image'."
-                        )
+            # Generate embedding based on input type
+            if input_type == PrimitiveTypeEnum.text:
+                if not isinstance(input_value, str):
+                    input_value = str(input_value)
+                vector = await self.embedding_model.aget_text_embedding(
+                    text=input_value
+                )
+                content = input_value
+            elif input_type == PrimitiveTypeEnum.image:
+                # For image embeddings
+                vector = await self.embedding_model.aget_image_embedding(
+                    image_path=input_value
+                )
+                content = input_value
+            else:
+                raise ValueError(
+                    (
+                        f"Unsupported input type for embedding: "
+                        f"{input_type}. Must be 'text' or 'image'."
                     )
-
-                # Create the Embedding object
-                embedding = Embedding(
-                    vector=vector,
-                    content=content,
                 )
-                return embedding
 
-            # TODO: switch back to async once aws auth supports it.
-            # https://github.com/bazaarvoice/qtype/issues/108
-            loop = asyncio.get_running_loop()
-            embedding = await loop.run_in_executor(
-                self.context.thread_pool, _call
+            # Create the Embedding object
+            embedding = Embedding(
+                vector=vector,
+                content=content,
             )
 
             # Yield the result

qtype/interpreter/executors/llm_inference_executor.py

@@ -1,3 +1,4 @@
+import logging
 from typing import AsyncIterator
 
 from llama_cloud import MessageRole as LlamaMessageRole
@@ -85,6 +86,7 @@ class LLMInferenceExecutor(StepExecutor):
 
         except Exception as e:
             # Emit error event to stream so frontend can display it
+            logging.error(f"LLM Inference search failed: {e}", exc_info=True)
             await self.stream_emitter.error(str(e))
             yield message.copy_with_error(self.step.id, e)
 

qtype/interpreter/executors/sql_source_executor.py

@@ -42,8 +42,12 @@ class SQLSourceExecutor(StepExecutor):
         connect_args = {}
         if self.step.auth:
             with auth(self.step.auth, self._secret_manager) as creds:
-                if isinstance(creds, boto3.Session):
-                    connect_args["session"] = creds
+                # For AWS auth, create a boto3 session from credentials
+                from qtype.interpreter.auth.aws import AWSCredentials
+
+                if isinstance(creds, AWSCredentials):
+                    session = boto3.Session(**creds.as_kwargs())
+                    connect_args["session"] = session
         engine = create_engine(connection_string, connect_args=connect_args)
 
         output_columns = {output.id for output in self.step.outputs}

qtype/interpreter/flow.py

@@ -17,7 +17,7 @@ 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, ProgressCallback
+from qtype.interpreter.types import FlowMessage, ProgressCallback, Session
 from qtype.semantic.model import Flow
 
 logger = logging.getLogger(__name__)
@@ -106,6 +106,16 @@ async def run_flow(
             initial = [initial]
 
         if isinstance(initial, list):
+            # Handle empty list by creating a single empty message
+            if len(initial) == 0:
+                session_id = kwargs.get("session_id", "default")
+                initial = [
+                    FlowMessage(
+                        session=Session(session_id=session_id),
+                        variables={},
+                    )
+                ]
+
             span.set_attribute("flow.input_count", len(initial))
 
             # convert to async iterator

qtype/mcp/server.py

@@ -1,7 +1,6 @@
 from __future__ import annotations
 
 import json
-import re
 import tempfile
 from functools import lru_cache
 from importlib.resources import files
@@ -12,133 +11,15 @@ import tantivy
 from mcp.server.fastmcp import FastMCP
 from pydantic import BaseModel
 
+from qtype.base.resources import get_docs_resource, get_examples_resource
 from qtype.commands.convert import convert_to_yaml
 
 # Initialize FastMCP server
 mcp = FastMCP("qtype", host="0.0.0.0")
 
-# Regex for pymdownx snippets: --8<-- "path/to/file"
-SNIPPET_REGEX = re.compile(r'--8<--\s+"([^"]+)"')
-
-
-# ============================================================================
-# Resource Abstraction Layer
-# ============================================================================
-
-
-class ResourceDirectory:
-    """Abstraction for accessing resource directories (docs, examples, etc.)."""
-
-    def __init__(
-        self, name: str, file_extension: str, resolve_snippets: bool = False
-    ):
-        """Initialize a resource directory.
-
-        Args:
-            name: Directory name (e.g., "docs", "examples")
-            file_extension: File extension to search for (e.g., ".md", ".yaml")
-            resolve_snippets: Whether to resolve MkDocs snippets in file content
-        """
-        self.name = name
-        self.file_extension = file_extension
-        self.resolve_snippets = resolve_snippets
-        self._path_cache: Path | None = None
-
-    def get_path(self) -> Path:
-        """Get the path to this resource directory.
-
-        Returns:
-            Path to the resource directory, trying installed package first,
-            then falling back to development path.
-        """
-        if self._path_cache is not None:
-            return self._path_cache
-
-        try:
-            # Try to get from installed package
-            resource_root = files("qtype") / self.name
-            # Check if it exists by trying to iterate
-            list(resource_root.iterdir())
-            self._path_cache = Path(str(resource_root))
-        except (FileNotFoundError, AttributeError, TypeError):
-            # Fall back to development path
-            self._path_cache = Path(__file__).parent.parent.parent / self.name
-
-        return self._path_cache
-
-    def get_file(self, file_path: str) -> str:
-        """Get the content of a specific file.
-
-        Args:
-            file_path: Relative path to the file from the resource root.
-
-        Returns:
-            The full content of the file.
-
-        Raises:
-            FileNotFoundError: If the specified file doesn't exist.
-            ValueError: If the path tries to access files outside the directory.
-        """
-        resource_path = self.get_path()
-
-        # Resolve the requested file path
-        requested_file = (resource_path / file_path).resolve()
-
-        # Security check: ensure the resolved path is within resource directory
-        try:
-            requested_file.relative_to(resource_path.resolve())
-        except ValueError:
-            raise ValueError(
-                f"Invalid path: '{file_path}' is outside {self.name} directory"
-            )
-
-        if not requested_file.exists():
-            raise FileNotFoundError(
-                f"{self.name.capitalize()} file not found: '{file_path}'. "
-                f"Use list_{self.name} to see available files."
-            )
-
-        if not requested_file.is_file():
-            raise ValueError(f"Path is not a file: '{file_path}'")
-
-        content = requested_file.read_text(encoding="utf-8")
-
-        # Apply snippet resolution if enabled
-        if self.resolve_snippets:
-            content = _resolve_snippets(content, requested_file)
-
-        return content
-
-    def list_files(self) -> list[str]:
-        """List all files in this resource directory.
-
-        Returns:
-            Sorted list of relative paths to all files with the configured extension.
-
-        Raises:
-            FileNotFoundError: If the resource directory doesn't exist.
-        """
-        resource_path = self.get_path()
-
-        if not resource_path.exists():
-            raise FileNotFoundError(
-                f"{self.name.capitalize()} directory not found: {resource_path}"
-            )
-
-        # Find all files with the configured extension
-        pattern = f"*{self.file_extension}"
-        files_list = []
-        for file in resource_path.rglob(pattern):
-            # Get relative path from resource root
-            rel_path = file.relative_to(resource_path)
-            files_list.append(str(rel_path))
-
-        return sorted(files_list)
-
-
-# Initialize resource directories
-_docs_resource = ResourceDirectory("docs", ".md", resolve_snippets=True)
-_examples_resource = ResourceDirectory("examples", ".yaml")
+# Get resource directories from base layer
+_docs_resource = get_docs_resource()
+_examples_resource = get_examples_resource()
 
 
 # ============================================================================
@@ -171,40 +52,6 @@ def _load_schema() -> dict[str, Any]:
             return json.load(f)
 
 
-def _resolve_snippets(content: str, base_path: Path) -> str:
-    """
-    Recursively finds and replaces MkDocs snippets in markdown content.
-    Mimics the behavior of pymdownx.snippets.
-
-    Args:
-        content: The markdown content to process
-        base_path: Path to the file being processed (used to resolve relative paths)
-    """
-    docs_root = _docs_resource.get_path()
-    project_root = docs_root.parent
-
-    def replace_match(match):
-        snippet_path = match.group(1)
-
-        # pymdownx logic: try relative to current file, then relative to docs, then project root
-        candidates = [
-            base_path.parent / snippet_path,  # Relative to the doc file
-            docs_root / snippet_path,  # Relative to docs root
-            project_root / snippet_path,  # Relative to project root
-        ]
-
-        for candidate in candidates:
-            if candidate.exists() and candidate.is_file():
-                # Recursively resolve snippets inside the included file
-                return _resolve_snippets(
-                    candidate.read_text(encoding="utf-8"), candidate
-                )
-
-        return f"> [!WARNING] Could not resolve snippet: {snippet_path}"
-
-    return SNIPPET_REGEX.sub(replace_match, content)
-
-
 @lru_cache(maxsize=1)
 def _build_search_index() -> tantivy.Index:
     """Build and cache a Tantivy search index for docs and examples.
@@ -268,13 +115,19 @@ def _build_search_index() -> tantivy.Index:
                 return line[2:].strip()
         return file_path.stem
 
+    # For snippet resolution in search indexing
+    def resolve_for_indexing(content: str, file_path: Path) -> str:
+        from qtype.base.resources import _resolve_snippets
+
+        return _resolve_snippets(content, file_path, _docs_resource)
+
     # Index documentation and examples
     index_files(
         docs_path,
         "*.md",
         "documentation",
         "docs",
-        process_content=_resolve_snippets,
+        process_content=resolve_for_indexing,
         extract_title=extract_md_title,
     )
     index_files(examples_path, "*.yaml", "example", "examples")

qtype/semantic/visualize.py

@@ -11,7 +11,6 @@ from typing import Any
 
 from pydantic import BaseModel
 
-from qtype.dsl.model import Index
 from qtype.semantic.model import (
     Agent,
     APITool,
@@ -21,6 +20,8 @@ from qtype.semantic.model import (
     DocumentIndex,
     DocumentSearch,
     Flow,
+    Index,
+    IndexUpsert,
     InvokeFlow,
     InvokeTool,
     LLMInference,
@@ -238,6 +239,12 @@ def _generate_step_node(
         )
         index_id = f"INDEX_{_sanitize_id(step.index.id)}"
         external_connections.append(f"    {node_id} -.-> {index_id}")
+    elif isinstance(step, IndexUpsert):
+        lines.append(
+            f'        {node_id}@{{shape: rect, label: "💾 {step.id}"}}'
+        )
+        index_id = f"INDEX_{_sanitize_id(step.index.id)}"
+        external_connections.append(f"    {node_id} -.->|writes| {index_id}")
     else:
         # Generic step
         lines.append(
@@ -382,7 +389,7 @@ def _generate_shared_resources(app: Application) -> list[str]:
             index_id = f"INDEX_{_sanitize_id(index.id)}"
             if isinstance(index, VectorIndex):
                 lines.append(
-                    f'        {index_id}@{{shape: cyl, label: "🗂️ {index.id}"}}'
+                    f'        {index_id}@{{shape: cyl, label: "�️ {index.id}"}}'
                 )
                 # Connect to embedding model
                 emb_model_id = f"EMB_{_sanitize_id(index.embedding_model.id)}"
@@ -396,7 +403,7 @@ def _generate_shared_resources(app: Application) -> list[str]:
                 )
             else:
                 lines.append(
-                    f'        {index_id}@{{shape: cyl, label: "🗂️ {index.id}"}}'
+                    f'        {index_id}@{{shape: cyl, label: "�️ {index.id}"}}'
                 )
 
             if index.auth:

{qtype-0.1.13.dist-info → qtype-0.1.14.dist-info}/METADATA

@@ -1,12 +1,13 @@
 Metadata-Version: 2.4
 Name: qtype
-Version: 0.1.13
+Version: 0.1.14
 Summary: DSL for Generative AI Prototyping
 Project-URL: Homepage, https://github.com/bazaarvoice/qtype
 Author-email: Lou Kratz <lou.kratz+qtype@bazaarvoice.com>
 License-Expression: Apache-2.0
 License-File: LICENSE
 Requires-Python: >=3.10
+Requires-Dist: cachetools>=6.2.1
 Requires-Dist: fsspec>=2025.5.1
 Requires-Dist: google-cloud-aiplatform>=1.120.0
 Requires-Dist: jsonschema>=4.24.0
@@ -51,7 +52,6 @@ Requires-Dist: s3fs>=2025.7.0; extra == 'interpreter'
 Requires-Dist: sqlalchemy>=2.0.42; extra == 'interpreter'
 Requires-Dist: uvicorn[standard]>=0.35.0; extra == 'interpreter'
 Provides-Extra: mcp
-Requires-Dist: cachetools>=6.2.1; extra == 'mcp'
 Requires-Dist: httpx>=0.28.1; extra == 'mcp'
 Requires-Dist: mcp[cli]>=1.25.0; extra == 'mcp'
 Requires-Dist: tantivy>=0.25.1; extra == 'mcp'