langroid 0.41.2__py3-none-any.whl → 0.41.4__py3-none-any.whl

langroid/agent/base.py

@@ -147,8 +147,7 @@ class Agent(ABC):
         self.llm_tools_usable: Set[str] = set()
         self.llm_tools_known: Set[str] = set()  # all known tools, handled/used or not
         # Indicates which tool-names are allowed to be inferred when
-        # the LLM "forgets" to include the request field in its
-        # tool-call.
+        # the LLM "forgets" to include the request field in its tool-call.
         self.enabled_requests_for_inference: Optional[Set[str]] = (
             None  # If None, we allow all
         )

langroid/agent/chat_agent.py

@@ -37,6 +37,7 @@ from langroid.utils.configuration import settings
 from langroid.utils.object_registry import ObjectRegistry
 from langroid.utils.output import status
 from langroid.utils.pydantic_utils import PydanticWrapper, get_pydantic_wrapper
+from langroid.utils.types import is_callable
 
 console = Console()
 
@@ -613,6 +614,8 @@ class ChatAgent(Agent):
                         return AgentDoneTool(
                             content=msg.content, tools=msg.tool_messages
                         )
+            elif is_callable(no_tool_option):
+                return no_tool_option(msg)
             # Otherwise just return `no_tool_option` as is:
             # This can be any string, such as a specific nudge/reminder to the LLM,
             # or even something like ResultTool etc.

langroid/embedding_models/models.py

@@ -450,9 +450,9 @@ class LlamaCppServerEmbeddings(EmbeddingModel):
 class GeminiEmbeddings(EmbeddingModel):
     def __init__(self, config: GeminiEmbeddingsConfig = GeminiEmbeddingsConfig()):
         try:
-            import google.generativeai as genai
+            from google import genai
         except ImportError as e:
-            raise LangroidImportError(extra="google-generativeai", error=str(e))
+            raise LangroidImportError(extra="google-genai", error=str(e))
         super().__init__()
         self.config = config
         load_dotenv()
@@ -464,29 +464,30 @@ class GeminiEmbeddings(EmbeddingModel):
                 GEMINI_API_KEY env variable must be set to use GeminiEmbeddings.
                 """
             )
-        genai.configure(api_key=self.config.api_key)  # type: ignore[attr-defined]
-        self.client = genai
+        self.client = genai.Client(api_key=self.config.api_key)
 
     def embedding_fn(self) -> Callable[[List[str]], Embeddings]:
         return EmbeddingFunctionCallable(self, self.config.batch_size)
 
-    def generate_embeddings(self, texts: List[str]) -> Embeddings:
-        all_embeddings = []  # More precise type hint
+    def generate_embeddings(self, texts: List[str]) -> List[List[float]]:
+        """Generates embeddings for a list of input texts."""
+        all_embeddings: List[List[float]] = []
+
         for batch in batched(texts, self.config.batch_size):
-            result = self.client.embed_content(  # type: ignore[attr-defined]
+            result = self.client.models.embed_content(  # type: ignore[attr-defined]
                 model=self.config.model_name,
-                content=batch,
-                task_type="RETRIEVAL_DOCUMENT",
+                contents=batch,
             )
 
-            embeddings = result["embedding"]
-            if not isinstance(embeddings, list):
-                raise ValueError("Unexpected format for embeddings: not a list")
+            if not hasattr(result, "embeddings") or not isinstance(
+                result.embeddings, list
+            ):
+                raise ValueError(
+                    "Unexpected format for embeddings: missing or incorrect type"
+                )
 
-            if embeddings and isinstance(embeddings[0], list):
-                all_embeddings.extend(embeddings)
-            else:
-                all_embeddings.append(embeddings)
+            # Extract .values from ContentEmbedding objects
+            all_embeddings.extend([emb.values for emb in result.embeddings])
 
         return all_embeddings
 

langroid/language_models/base.py

@@ -19,7 +19,7 @@ from typing import (
 
 from langroid.cachedb.base import CacheDBConfig
 from langroid.cachedb.redis_cachedb import RedisCacheConfig
-from langroid.language_models.model_info import get_model_info
+from langroid.language_models.model_info import ModelInfo, get_model_info
 from langroid.parsing.agent_chats import parse_message
 from langroid.parsing.parse_json import parse_imperfect_json, top_level_json_field
 from langroid.prompts.dialog import collate_chat_history
@@ -425,6 +425,7 @@ class LanguageModel(ABC):
 
     def __init__(self, config: LLMConfig = LLMConfig()):
         self.config = config
+        self.chat_model_orig = config.chat_model
 
     @staticmethod
     def create(config: Optional[LLMConfig]) -> Optional["LanguageModel"]:
@@ -595,6 +596,34 @@ class LanguageModel(ABC):
     def __call__(self, prompt: str, max_tokens: int) -> LLMResponse:
         return self.generate(prompt, max_tokens)
 
+    def info(self) -> ModelInfo:
+        """Info of relevant chat model"""
+        model = (
+            self.config.completion_model
+            if self.config.use_completion_for_chat
+            else self.config.chat_model
+        )
+        orig_model = (
+            self.config.completion_model
+            if self.config.use_completion_for_chat
+            else self.chat_model_orig
+        )
+        return get_model_info(orig_model, model)
+
+    def completion_info(self) -> ModelInfo:
+        """Info of relevant completion model"""
+        model = (
+            self.config.chat_model
+            if self.config.use_chat_for_completion
+            else self.config.completion_model
+        )
+        orig_model = (
+            self.chat_model_orig
+            if self.config.use_chat_for_completion
+            else self.config.completion_model
+        )
+        return get_model_info(orig_model, model)
+
     def chat_context_length(self) -> int:
         return self.config.chat_context_length
 

langroid/language_models/openai_gpt.py

@@ -50,7 +50,6 @@ from langroid.language_models.model_info import (
     DeepSeekModel,
     GeminiModel,
     OpenAI_API_ParamInfo,
-    get_model_info,
 )
 from langroid.language_models.model_info import (
     OpenAIChatModel as OpenAIChatModel,
@@ -233,6 +232,7 @@ class OpenAIGPTConfig(LLMConfig):
     params: OpenAICallParams | None = None
     # these can be any model name that is served at an OpenAI-compatible API end point
     chat_model: str = default_openai_chat_model
+    chat_model_orig: str = default_openai_chat_model
     completion_model: str = default_openai_completion_model
     run_on_first_use: Callable[[], None] = noop
     parallel_tool_calls: Optional[bool] = None
@@ -372,7 +372,8 @@ class OpenAIGPT(LanguageModel):
         super().__init__(config)
         self.config: OpenAIGPTConfig = config
         # save original model name such as `provider/model` before
-        # we strip out the `provider`
+        # we strip out the `provider` - we retain the original in
+        # case some params are specific to a provider.
         self.chat_model_orig = self.config.chat_model
 
         # Run the first time the model is used
@@ -467,11 +468,7 @@ class OpenAIGPT(LanguageModel):
             # these features (with JSON schema restricted to a limited set of models)
             self.supports_strict_tools = self.api_base is None
             self.supports_json_schema = (
-                self.api_base is None
-                and get_model_info(  # look for family/provider-specific then generic
-                    self.chat_model_orig,  # e.g. "gemini/gemini-2.0-flash"
-                    self.config.chat_model,  # e.g. "gemini-2.0-flash"
-                ).has_structured_output
+                self.api_base is None and self.info().has_structured_output
             )
 
         if settings.chat_model != "":
@@ -621,13 +618,7 @@ class OpenAIGPT(LanguageModel):
         return self.config.chat_model in openai_chat_models
 
     def supports_functions_or_tools(self) -> bool:
-        return (
-            self.is_openai_chat_model()
-            and get_model_info(
-                self.chat_model_orig,
-                self.config.chat_model,
-            ).has_tools
-        )
+        return self.is_openai_chat_model() and self.info().has_tools
 
     def is_openai_completion_model(self) -> bool:
         openai_completion_models = [e.value for e in OpenAICompletionModel]
@@ -650,11 +641,7 @@ class OpenAIGPT(LanguageModel):
         """
         List of params that are not supported by the current model
         """
-        model_info = get_model_info(
-            self.chat_model_orig,
-            self.config.chat_model,
-        )
-        unsupported = set(model_info.unsupported_params)
+        unsupported = set(self.info().unsupported_params)
         for param, model_list in OpenAI_API_ParamInfo().params.items():
             if (
                 self.config.chat_model not in model_list
@@ -668,44 +655,21 @@ class OpenAIGPT(LanguageModel):
         Map of param name -> new name for specific models.
         Currently main troublemaker is o1* series.
         """
-        return get_model_info(
-            self.chat_model_orig,
-            self.config.chat_model,
-        ).rename_params
+        return self.info().rename_params
 
     def chat_context_length(self) -> int:
         """
         Context-length for chat-completion models/endpoints
         Get it from the dict, otherwise fail-over to general method
         """
-        model = (
-            self.config.completion_model
-            if self.config.use_completion_for_chat
-            else self.config.chat_model
-        )
-        orig_model = (
-            self.config.completion_model
-            if self.config.use_completion_for_chat
-            else self.chat_model_orig
-        )
-        return get_model_info(orig_model, model).context_length
+        return self.info().context_length
 
     def completion_context_length(self) -> int:
         """
         Context-length for completion models/endpoints
         Get it from the dict, otherwise fail-over to general method
         """
-        model = (
-            self.config.chat_model
-            if self.config.use_chat_for_completion
-            else self.config.completion_model
-        )
-        orig_model = (
-            self.chat_model_orig
-            if self.config.use_chat_for_completion
-            else self.config.completion_model
-        )
-        return get_model_info(orig_model, model).context_length
+        return self.completion_info().context_length
 
     def chat_cost(self) -> Tuple[float, float]:
         """
@@ -713,7 +677,7 @@ class OpenAIGPT(LanguageModel):
         models/endpoints.
         Get it from the dict, otherwise fail-over to general method
         """
-        info = get_model_info(self.config.chat_model)
+        info = self.info()
         return (info.input_cost_per_million / 1000, info.output_cost_per_million / 1000)
 
     def set_stream(self, stream: bool) -> bool:
@@ -731,10 +695,7 @@ class OpenAIGPT(LanguageModel):
         return (
             self.config.stream
             and settings.stream
-            and get_model_info(
-                self.chat_model_orig,
-                self.config.chat_model,
-            ).allows_streaming
+            and self.info().allows_streaming
             and not settings.quiet
         )
 
@@ -1744,11 +1705,7 @@ class OpenAIGPT(LanguageModel):
         args: Dict[str, Any] = dict(
             model=chat_model,
             messages=[
-                m.api_dict(
-                    has_system_role=get_model_info(
-                        self.config.chat_model
-                    ).allows_system_message
-                )
+                m.api_dict(has_system_role=self.info().allows_system_message)
                 for m in (llm_messages)
             ],
             max_tokens=max_tokens,

langroid/utils/types.py

@@ -1,5 +1,6 @@
 import json
 import logging
+from inspect import signature
 from typing import Any, Optional, Type, TypeVar, Union, get_args, get_origin
 
 from langroid.pydantic_v1 import BaseModel
@@ -91,3 +92,22 @@ def from_string(
         return s
     else:
         return None
+
+
+def is_callable(obj: Any, k: int = 1) -> bool:
+    """Check if object is callable and accepts exactly k args.
+
+    Args:
+        obj: Object to check
+
+    Returns:
+        bool: True if object is callable with k args, False otherwise
+    """
+    if not callable(obj):
+        return False
+    try:
+        sig = signature(obj)
+        params = list(sig.parameters.values())
+        return len(params) == k
+    except ValueError:
+        return False

{langroid-0.41.2.dist-info → langroid-0.41.4.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: langroid
-Version: 0.41.2
+Version: 0.41.4
 Summary: Harness LLMs with Multi-Agent Programming
 Author-email: Prasad Chalasani <pchalasani@gmail.com>
 License: MIT
@@ -21,7 +21,7 @@ Requires-Dist: fakeredis<3.0.0,>=2.12.1
 Requires-Dist: fire<1.0.0,>=0.5.0
 Requires-Dist: gitpython<4.0.0,>=3.1.43
 Requires-Dist: google-api-python-client<3.0.0,>=2.95.0
-Requires-Dist: google-generativeai<1.0.0,>=0.5.2
+Requires-Dist: google-genai>=1.0.0
 Requires-Dist: groq<1.0.0,>=0.13.0
 Requires-Dist: grpcio<2.0.0,>=1.62.1
 Requires-Dist: halo<1.0.0,>=0.0.31
@@ -112,8 +112,10 @@ Provides-Extra: exa
 Requires-Dist: exa-py>=1.8.7; extra == 'exa'
 Provides-Extra: fastembed
 Requires-Dist: fastembed<0.4.0,>=0.3.1; extra == 'fastembed'
+Provides-Extra: google-genai
+Requires-Dist: google-genai>=1.0.0; extra == 'google-genai'
 Provides-Extra: google-generativeai
-Requires-Dist: google-generativeai>=0.8.4; extra == 'google-generativeai'
+Requires-Dist: google-genai>=1.0.0; extra == 'google-generativeai'
 Provides-Extra: hf-embeddings
 Requires-Dist: sentence-transformers<3.0.0,>=2.2.2; extra == 'hf-embeddings'
 Requires-Dist: torch<3.0.0,>=2.0.0; extra == 'hf-embeddings'
@@ -319,7 +321,7 @@ teacher_task.run()
 - **Jan 2025:**
   - [0.36.0](https://github.com/langroid/langroid/releases/tag/0.36.0): Weaviate vector-db support (thanks @abab-dev).
   - [0.35.0](https://github.com/langroid/langroid/releases/tag/0.35.0): Capture/Stream reasoning content from 
-    Reasoning LLMs (e.g. DeepSeek, OpenAI o1) in addition to final answer.
+    Reasoning LLMs (e.g. DeepSeek-R1, OpenAI o1) in addition to final answer.
   - [0.34.0](https://github.com/langroid/langroid/releases/tag/0.34.0): DocChatAgent 
     chunk enrichment to improve retrieval. (collaboration with @dfm88). 
   - [0.33.0](https://github.com/langroid/langroid/releases/tag/0.33.3) Move from Poetry to uv! (thanks @abab-dev).
@@ -627,7 +629,7 @@ For many practical scenarios, you may need additional optional dependencies:
 - For "chat with databases", use the `db` extra:
     ```bash
     pip install "langroid[db]"
-    ``
+    ```
 - You can specify multiple extras by separating them with commas, e.g.:
     ```bash
     pip install "langroid[doc-chat,db]"

{langroid-0.41.2.dist-info → langroid-0.41.4.dist-info}/RECORD

@@ -3,9 +3,9 @@ langroid/exceptions.py,sha256=OPjece_8cwg94DLPcOGA1ddzy5bGh65pxzcHMnssTz8,2995
 langroid/mytypes.py,sha256=FXSH62MUCeMCJP-66RVmbNaHCDLMxllEShZ-xEeTn9A,2833
 langroid/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 langroid/agent/__init__.py,sha256=ll0Cubd2DZ-fsCMl7e10hf9ZjFGKzphfBco396IKITY,786
-langroid/agent/base.py,sha256=CVPvy-bLI_6wHZxYf_spo4eq-utR373Ur7T3zKpe18U,78222
+langroid/agent/base.py,sha256=k5kJGTpo2CcnVl2cEM0luBQ__a7C9put1aH5-wd3hQ8,78212
 langroid/agent/batch.py,sha256=vi1r5i1-vN80WfqHDSwjEym_KfGsqPGUtwktmiK1nuk,20635
-langroid/agent/chat_agent.py,sha256=q9J5NQzyZZxWX8y-u1xMITJObFN4GlDFS3CEMl39taY,84077
+langroid/agent/chat_agent.py,sha256=hUu13nYhhr6ph01Sln8y_WuOIpcd38icN6p22h6IiDY,84211
 langroid/agent/chat_document.py,sha256=xzMtrPbaW-Y-BnF7kuhr2dorsD-D5rMWzfOqJ8HAoo8,17885
 langroid/agent/openai_assistant.py,sha256=JkAcs02bIrgPNVvUWVR06VCthc5-ulla2QMBzux_q6o,34340
 langroid/agent/task.py,sha256=XrXUbSoiFasvpIsZPn_cBpdWaTCKljJPRimtLMrSZrs,90347
@@ -59,7 +59,7 @@ langroid/cachedb/momento_cachedb.py,sha256=YEOJ62hEcV6iIeMr5aGgRYgWQqFYaej9gEDEc
 langroid/cachedb/redis_cachedb.py,sha256=7kgnbf4b5CKsCrlL97mHWKvdvlLt8zgn7lc528jEpiE,5141
 langroid/embedding_models/__init__.py,sha256=KyYxR3jDFUCfYjSuCL86qjAmrq6mXXjOT4lFNOKVj6Y,955
 langroid/embedding_models/base.py,sha256=Ml7oA6PzQm0wZmIYn3fhF7dvZCi-amviWUwOeBegH3A,2562
-langroid/embedding_models/models.py,sha256=fKQBiaaG7uYoeELDbAiNxwLdn-CWN8dyiVEZcdk_bjI,18959
+langroid/embedding_models/models.py,sha256=93rQjvcltGLDYbgT1TIW8PZiAbNPAvPAqxCLlLS9GAo,18955
 langroid/embedding_models/remote_embeds.py,sha256=6_kjXByVbqhY9cGwl9R83ZcYC2km-nGieNNAo1McHaY,5151
 langroid/embedding_models/protoc/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 langroid/embedding_models/protoc/embeddings.proto,sha256=_O-SgFpTaylQeOTgSpxhEJ7CUw7PeCQQJLaPqpPYKJg,321
@@ -68,11 +68,11 @@ langroid/embedding_models/protoc/embeddings_pb2.pyi,sha256=UkNy7BrNsmQm0vLb3NtGX
 langroid/embedding_models/protoc/embeddings_pb2_grpc.py,sha256=9dYQqkW3JPyBpSEjeGXTNpSqAkC-6FPtBHyteVob2Y8,2452
 langroid/language_models/__init__.py,sha256=3aD2qC1lz8v12HX4B-dilv27gNxYdGdeu1QvDlkqqHs,1095
 langroid/language_models/azure_openai.py,sha256=zNQzzsERxNestq-hFfQZbvTzK43G2vjRWnTV3ktm1DQ,5845
-langroid/language_models/base.py,sha256=mN6HAjLgF2xpHObz5uPZ3JDID7jdTiRLEkoGgGrqLM8,25177
+langroid/language_models/base.py,sha256=is4l3x858tdPHbrJU2jxJXe2j9PCGb9kk_c5nyfShxs,26150
 langroid/language_models/config.py,sha256=9Q8wk5a7RQr8LGMT_0WkpjY8S4ywK06SalVRjXlfCiI,378
 langroid/language_models/mock_lm.py,sha256=5BgHKDVRWFbUwDT_PFgTZXz9-k8wJSA2e3PZmyDgQ1k,4022
 langroid/language_models/model_info.py,sha256=yKAaKoCanPoqaoHCzRVNPjg-M9a4S2Vm2AJGnwMeO-M,11360
-langroid/language_models/openai_gpt.py,sha256=hC6knKQ9s9-bNixPWduxLRhLxs3y1zyfW7hU8oWWtHQ,78681
+langroid/language_models/openai_gpt.py,sha256=aajZ3ZvGkwI-3QdsNWgJ4QSyGpnyXJ5n4p2fYGUmdo4,77317
 langroid/language_models/utils.py,sha256=L4_CbihDMTGcsg0TOG1Yd5JFEto46--h7CX_14m89sQ,5016
 langroid/language_models/prompt_formatter/__init__.py,sha256=2-5cdE24XoFDhifOLl8yiscohil1ogbP1ECkYdBlBsk,372
 langroid/language_models/prompt_formatter/base.py,sha256=eDS1sgRNZVnoajwV_ZIha6cba5Dt8xjgzdRbPITwx3Q,1221
@@ -111,7 +111,7 @@ langroid/utils/object_registry.py,sha256=iPz9GHzvmCeVoidB3JdAMEKcxJEqTdUr0otQEex
 langroid/utils/pandas_utils.py,sha256=UctS986Jtl_MvU5rA7-GfrjEHXP7MNu8ePhepv0bTn0,755
 langroid/utils/pydantic_utils.py,sha256=R7Ps8VP56-eSo-LYHWllFo-SJ2zDmdItuuYpUq2gGJ8,20854
 langroid/utils/system.py,sha256=cJqDgOf9mM82l1GyUeQQdEYAwepYXQwtpJU8Xrz0-MA,8453
-langroid/utils/types.py,sha256=4GrOnU3HLWh-UwaUPp7LlB3V413q3K5OSzc0ggDoQ6A,2510
+langroid/utils/types.py,sha256=-BvyIf_LmAJ5jR9NC7S4CSVNEr3XayAaxJ5o0TiIej0,2992
 langroid/utils/algorithms/__init__.py,sha256=WylYoZymA0fnzpB4vrsH_0n7WsoLhmuZq8qxsOCjUpM,41
 langroid/utils/algorithms/graph.py,sha256=JbdpPnUOhw4-D6O7ou101JLA3xPCD0Lr3qaPoFCaRfo,2866
 langroid/utils/output/__init__.py,sha256=7P0f--4IZneNsTxXY5fd6d6iW-CeVe-KSsl-87sbBPc,340
@@ -128,7 +128,7 @@ langroid/vector_store/pineconedb.py,sha256=otxXZNaBKb9f_H75HTaU3lMHiaR2NUp5MqwLZ
 langroid/vector_store/postgres.py,sha256=DQHd6dt-OcV_QVNm-ymn28rlTfhI6hqgcpLTPCsm0jI,15990
 langroid/vector_store/qdrantdb.py,sha256=v7TAsIoj_vxeKDYS9tpwJLBZA8fuTweTYxHo0X_uawM,17949
 langroid/vector_store/weaviatedb.py,sha256=ONEr2iGS0Ii73oMe7tRk6bB-BEXQUa70fYSrdI8d3yo,11481
-langroid-0.41.2.dist-info/METADATA,sha256=FWd7a2bi9c79wldH9KrVvq2U7Jtu_XeEWI9XcF17FpE,61259
-langroid-0.41.2.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-langroid-0.41.2.dist-info/licenses/LICENSE,sha256=EgVbvA6VSYgUlvC3RvPKehSg7MFaxWDsFuzLOsPPfJg,1065
-langroid-0.41.2.dist-info/RECORD,,
+langroid-0.41.4.dist-info/METADATA,sha256=hU062qh537keZYiNQZhdKWffOldsn_lz5BWYew1TQbg,61331
+langroid-0.41.4.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+langroid-0.41.4.dist-info/licenses/LICENSE,sha256=EgVbvA6VSYgUlvC3RvPKehSg7MFaxWDsFuzLOsPPfJg,1065
+langroid-0.41.4.dist-info/RECORD,,