lfx-nightly 0.1.11.dev0__py3-none-any.whl

lfx/interface/initialize/loading.py

@@ -0,0 +1,224 @@
+from __future__ import annotations
+
+import inspect
+import os
+import warnings
+from typing import TYPE_CHECKING, Any
+
+import orjson
+from pydantic import PydanticDeprecatedSince20
+
+from lfx.custom.eval import eval_custom_component_code
+from lfx.log.logger import logger
+from lfx.schema.artifact import get_artifact_type, post_process_raw
+from lfx.schema.data import Data
+from lfx.services.deps import get_settings_service, session_scope
+from lfx.services.session import NoopSession
+
+if TYPE_CHECKING:
+    from lfx.custom.custom_component.component import Component
+    from lfx.custom.custom_component.custom_component import CustomComponent
+    from lfx.graph.vertex.base import Vertex
+
+    # This is forward declared to avoid circular import
+    class EventManager:
+        pass
+
+
+def instantiate_class(
+    vertex: Vertex,
+    user_id=None,
+    event_manager: EventManager | None = None,
+) -> Any:
+    """Instantiate class from module type and key, and params."""
+    vertex_type = vertex.vertex_type
+    base_type = vertex.base_type
+    logger.debug(f"Instantiating {vertex_type} of type {base_type}")
+
+    if not base_type:
+        msg = "No base type provided for vertex"
+        raise ValueError(msg)
+
+    custom_params = get_params(vertex.params)
+    code = custom_params.pop("code")
+    class_object: type[CustomComponent | Component] = eval_custom_component_code(code)
+    custom_component: CustomComponent | Component = class_object(
+        _user_id=user_id,
+        _parameters=custom_params,
+        _vertex=vertex,
+        _tracing_service=None,
+        _id=vertex.id,
+    )
+    if hasattr(custom_component, "set_event_manager"):
+        custom_component.set_event_manager(event_manager)
+    return custom_component, custom_params
+
+
+async def get_instance_results(
+    custom_component,
+    custom_params: dict,
+    vertex: Vertex,
+    *,
+    fallback_to_env_vars: bool = False,
+    base_type: str = "component",
+):
+    custom_params = await update_params_with_load_from_db_fields(
+        custom_component,
+        custom_params,
+        vertex.load_from_db_fields,
+        fallback_to_env_vars=fallback_to_env_vars,
+    )
+    with warnings.catch_warnings():
+        warnings.filterwarnings("ignore", category=PydanticDeprecatedSince20)
+        if base_type == "custom_components":
+            return await build_custom_component(params=custom_params, custom_component=custom_component)
+        if base_type == "component":
+            return await build_component(params=custom_params, custom_component=custom_component)
+        msg = f"Base type {base_type} not found."
+        raise ValueError(msg)
+
+
+def get_params(vertex_params):
+    params = vertex_params
+    params = convert_params_to_sets(params)
+    params = convert_kwargs(params)
+    return params.copy()
+
+
+def convert_params_to_sets(params):
+    """Convert certain params to sets."""
+    if "allowed_special" in params:
+        params["allowed_special"] = set(params["allowed_special"])
+    if "disallowed_special" in params:
+        params["disallowed_special"] = set(params["disallowed_special"])
+    return params
+
+
+def convert_kwargs(params):
+    # Loop through items to avoid repeated lookups
+    items_to_remove = []
+    for key, value in params.items():
+        if ("kwargs" in key or "config" in key) and isinstance(value, str):
+            try:
+                params[key] = orjson.loads(value)
+            except orjson.JSONDecodeError:
+                items_to_remove.append(key)
+
+    # Remove invalid keys outside the loop to avoid modifying dict during iteration
+    for key in items_to_remove:
+        params.pop(key, None)
+
+    return params
+
+
+def load_from_env_vars(params, load_from_db_fields):
+    for field in load_from_db_fields:
+        if field not in params or not params[field]:
+            continue
+        key = os.getenv(params[field])
+        if key:
+            logger.info(f"Using environment variable {params[field]} for {field}")
+        else:
+            logger.error(f"Environment variable {params[field]} is not set.")
+        params[field] = key if key is not None else None
+        if key is None:
+            logger.warning(f"Could not get value for {field}. Setting it to None.")
+    return params
+
+
+async def update_params_with_load_from_db_fields(
+    custom_component: CustomComponent,
+    params,
+    load_from_db_fields,
+    *,
+    fallback_to_env_vars=False,
+):
+    async with session_scope() as session:
+        settings_service = get_settings_service()
+        is_noop_session = isinstance(session, NoopSession) or (
+            settings_service and settings_service.settings.use_noop_database
+        )
+        if is_noop_session:
+            logger.debug("Loading variables from environment variables because database is not available.")
+            return load_from_env_vars(params, load_from_db_fields)
+        for field in load_from_db_fields:
+            if field not in params or not params[field]:
+                continue
+
+            try:
+                key = await custom_component.get_variable(name=params[field], field=field, session=session)
+            except ValueError as e:
+                if any(reason in str(e) for reason in ["User id is not set", "variable not found."]):
+                    raise
+                logger.debug(str(e))
+                key = None
+
+            if fallback_to_env_vars and key is None:
+                key = os.getenv(params[field])
+                if key:
+                    logger.info(f"Using environment variable {params[field]} for {field}")
+                else:
+                    logger.error(f"Environment variable {params[field]} is not set.")
+
+            params[field] = key if key is not None else None
+            if key is None:
+                logger.warning(f"Could not get value for {field}. Setting it to None.")
+
+        return params
+
+
+async def build_component(
+    params: dict,
+    custom_component: Component,
+):
+    # Now set the params as attributes of the custom_component
+    custom_component.set_attributes(params)
+    build_results, artifacts = await custom_component.build_results()
+
+    return custom_component, build_results, artifacts
+
+
+async def build_custom_component(params: dict, custom_component: CustomComponent):
+    if "retriever" in params and hasattr(params["retriever"], "as_retriever"):
+        params["retriever"] = params["retriever"].as_retriever()
+
+    # Determine if the build method is asynchronous
+    is_async = inspect.iscoroutinefunction(custom_component.build)
+
+    # New feature: the component has a list of outputs and we have
+    # to check the vertex.edges to see which is connected (coulb be multiple)
+    # and then we'll get the output which has the name of the method we should call.
+    # the methods don't require any params because they are already set in the custom_component
+    # so we can just call them
+
+    if is_async:
+        # Await the build method directly if it's async
+        build_result = await custom_component.build(**params)
+    else:
+        # Call the build method directly if it's sync
+        build_result = custom_component.build(**params)
+    custom_repr = custom_component.custom_repr()
+    if custom_repr is None and isinstance(build_result, dict | Data | str):
+        custom_repr = build_result
+    if not isinstance(custom_repr, str):
+        custom_repr = str(custom_repr)
+    raw = custom_component.repr_value
+    if hasattr(raw, "data") and raw is not None:
+        raw = raw.data
+
+    elif hasattr(raw, "model_dump") and raw is not None:
+        raw = raw.model_dump()
+    if raw is None and isinstance(build_result, dict | Data | str):
+        raw = build_result.data if isinstance(build_result, Data) else build_result
+
+    artifact_type = get_artifact_type(custom_component.repr_value or raw, build_result)
+    raw = post_process_raw(raw, artifact_type)
+    artifact = {"repr": custom_repr, "raw": raw, "type": artifact_type}
+
+    if custom_component.get_vertex() is not None:
+        custom_component.set_artifacts({custom_component.get_vertex().outputs[0].get("name"): artifact})
+        custom_component.set_results({custom_component.get_vertex().outputs[0].get("name"): build_result})
+        return custom_component, build_result, artifact
+
+    msg = "Custom component does not have a vertex"
+    raise ValueError(msg)

lfx/interface/listing.py

@@ -0,0 +1,26 @@
+from typing_extensions import override
+
+from lfx.services.deps import get_settings_service
+from lfx.utils.lazy_load import LazyLoadDictBase
+
+
+class AllTypesDict(LazyLoadDictBase):
+    def __init__(self) -> None:
+        self._all_types_dict = None
+
+    def _build_dict(self):
+        langchain_types_dict = self.get_type_dict()
+        return {
+            **langchain_types_dict,
+            "Custom": ["Custom Tool", "Python Function"],
+        }
+
+    @override
+    def get_type_dict(self):
+        from lfx.custom.utils import get_all_types_dict
+
+        settings_service = get_settings_service()
+        return get_all_types_dict(settings_service.settings.components_path)
+
+
+lazy_load_dict = AllTypesDict()

lfx/interface/run.py

@@ -0,0 +1,16 @@
+def get_memory_key(langchain_object):
+    """Get the memory key from the LangChain object's memory attribute.
+
+    Given a LangChain object, this function retrieves the current memory key from the object's memory attribute.
+    It then checks if the key exists in a dictionary of known memory keys and returns the corresponding key,
+    or None if the current key is not recognized.
+    """
+    mem_key_dict = {
+        "chat_history": "history",
+        "history": "chat_history",
+    }
+    # Check if memory_key attribute exists
+    if hasattr(langchain_object.memory, "memory_key"):
+        memory_key = langchain_object.memory.memory_key
+        return mem_key_dict.get(memory_key)
+    return None  # or some other default value or action

lfx/interface/utils.py

@@ -0,0 +1,111 @@
+import base64
+import json
+import os
+from io import BytesIO
+from pathlib import Path
+from string import Formatter
+
+import yaml
+from langchain_core.language_models import BaseLanguageModel
+from PIL.Image import Image
+
+from lfx.log.logger import logger
+from lfx.services.chat.config import ChatConfig
+from lfx.services.deps import get_settings_service
+
+
+def load_file_into_dict(file_path: str) -> dict:
+    file_path_ = Path(file_path)
+    if not file_path_.exists():
+        msg = f"File not found: {file_path}"
+        raise FileNotFoundError(msg)
+
+    # Files names are UUID, so we can't find the extension
+    with file_path_.open(encoding="utf-8") as file:
+        try:
+            data = json.load(file)
+        except json.JSONDecodeError:
+            file.seek(0)
+            data = yaml.safe_load(file)
+        except ValueError as exc:
+            msg = "Invalid file type. Expected .json or .yaml."
+            raise ValueError(msg) from exc
+    return data
+
+
+def pil_to_base64(image: Image) -> str:
+    buffered = BytesIO()
+    image.save(buffered, format="PNG")
+    img_str = base64.b64encode(buffered.getvalue())
+    return img_str.decode("utf-8")
+
+
+def try_setting_streaming_options(langchain_object):
+    """Try setting streaming options on LangChain objects."""
+    # If the LLM type is OpenAI or ChatOpenAI, set streaming to True
+    # First we need to find the LLM
+    llm = None
+    if hasattr(langchain_object, "llm"):
+        llm = langchain_object.llm
+    elif hasattr(langchain_object, "llm_chain") and hasattr(langchain_object.llm_chain, "llm"):
+        llm = langchain_object.llm_chain.llm
+
+    if isinstance(llm, BaseLanguageModel):
+        if hasattr(llm, "streaming") and isinstance(llm.streaming, bool):
+            llm.streaming = ChatConfig.streaming
+        elif hasattr(llm, "stream") and isinstance(llm.stream, bool):
+            llm.stream = ChatConfig.streaming
+
+    return langchain_object
+
+
+def extract_input_variables_from_prompt(prompt: str) -> list[str]:
+    """Extract variable names from a prompt string using Python's built-in string formatter.
+
+    Uses the same convention as Python's .format() method:
+    - Single braces {name} are variable placeholders
+    - Double braces {{name}} are escape sequences that render as literal {name}
+    """
+    formatter = Formatter()
+    variables: list[str] = []
+    seen: set[str] = set()
+
+    # Use local bindings for micro-optimization
+    variables_append = variables.append
+    seen_add = seen.add
+    seen_contains = seen.__contains__
+
+    for _, field_name, _, _ in formatter.parse(prompt):
+        if field_name and not seen_contains(field_name):
+            variables_append(field_name)
+            seen_add(field_name)
+
+    return variables
+
+
+def setup_llm_caching() -> None:
+    """Setup LLM caching."""
+    settings_service = get_settings_service()
+    try:
+        set_langchain_cache(settings_service.settings)
+    except ImportError:
+        logger.warning(f"Could not import {settings_service.settings.cache_type}. ")
+    except Exception:  # noqa: BLE001
+        logger.warning("Could not setup LLM caching.")
+
+
+def set_langchain_cache(settings) -> None:
+    from langchain.globals import set_llm_cache
+    from langflow.interface.importing.utils import import_class
+
+    if cache_type := os.getenv("LANGFLOW_LANGCHAIN_CACHE"):
+        try:
+            cache_class = import_class(f"langchain_community.cache.{cache_type or settings.LANGCHAIN_CACHE}")
+
+            logger.debug(f"Setting up LLM caching with {cache_class.__name__}")
+            set_llm_cache(cache_class())
+            logger.info(f"LLM caching setup with {cache_class.__name__}")
+        except ImportError:
+            logger.warning(f"Could not import {cache_type}. ")
+    else:
+        logger.debug("No LLM cache set.")

lfx/io/__init__.py

@@ -0,0 +1,63 @@
+from lfx.inputs import (
+    BoolInput,
+    CodeInput,
+    DataFrameInput,
+    DataInput,
+    DefaultPromptField,
+    DictInput,
+    DropdownInput,
+    FileInput,
+    FloatInput,
+    HandleInput,
+    IntInput,
+    LinkInput,
+    McpInput,
+    MessageInput,
+    MessageTextInput,
+    MultilineInput,
+    MultilineSecretInput,
+    MultiselectInput,
+    NestedDictInput,
+    PromptInput,
+    QueryInput,
+    SecretStrInput,
+    SliderInput,
+    StrInput,
+    TabInput,
+    TableInput,
+    ToolsInput,
+)
+from lfx.template import Output
+
+__all__ = [
+    "BoolInput",
+    "CodeInput",
+    "DataFrameInput",
+    "DataInput",
+    "DefaultPromptField",
+    "DefaultPromptField",
+    "DictInput",
+    "DropdownInput",
+    "FileInput",
+    "FloatInput",
+    "HandleInput",
+    "IntInput",
+    "LinkInput",
+    "LinkInput",
+    "McpInput",
+    "MessageInput",
+    "MessageTextInput",
+    "MultilineInput",
+    "MultilineSecretInput",
+    "MultiselectInput",
+    "NestedDictInput",
+    "Output",
+    "PromptInput",
+    "QueryInput",
+    "SecretStrInput",
+    "SliderInput",
+    "StrInput",
+    "TabInput",
+    "TableInput",
+    "ToolsInput",
+]