Merge branch 'main' into example-deploy

.gitignore

@@ -174,4 +174,4 @@ cython_debug/
 src/controlflow/_version.py
 all_code.md 
 all_docs.md
-llm_guides.md
+llm_guides.md

docs/guides/configure-llms.mdx

@@ -47,6 +47,7 @@ At this time, supported providers for automatic configuration include:
 | Anthropic | `anthropic` | (included) |
 | Google   | `google`       | `langchain_google_genai` |
 | Groq      | `groq`       | `langchain_groq` |
+| Ollama    | `ollama`    | `langchain-ollama` |
 
 If the required dependencies are not installed, ControlFlow will be unable to load the model and will raise an error.
 

examples/asyncpg-memory.py

@@ -0,0 +1,55 @@
+import asyncio
+
+import controlflow as cf
+from controlflow.memory.async_memory import AsyncMemory
+from controlflow.memory.providers.postgres import AsyncPostgresMemory
+
+provider = AsyncPostgresMemory(
+    database_url="postgresql+psycopg://postgres:postgres@localhost:5432/database",
+    # embedding_dimension=1536,
+    # embedding_fn=OpenAIEmbeddings(),
+    table_name="vector_db_async",
+)
+
+# Create a memory module for user preferences
+user_preferences = AsyncMemory(
+    key="user_preferences",
+    instructions="Store and retrieve user preferences.",
+    provider=provider,
+)
+
+# Create an agent with access to the memory
+agent = cf.Agent(memories=[user_preferences])
+
+
+# Create a flow to ask for the user's favorite color
+@cf.flow
+async def remember_pet():
+    return await cf.run_async(
+        "Ask the user for their favorite animal and store it in memory",
+        agents=[agent],
+        interactive=True,
+    )
+
+
+# Create a flow to recall the user's favorite color
+@cf.flow
+async def recall_pet():
+    return await cf.run_async(
+        "What is the user's favorite animal?",
+        agents=[agent],
+    )
+
+
+async def main():
+    print("First flow:")
+    await remember_pet()
+
+    print("\nSecond flow:")
+    result = await recall_pet()
+    print(result)
+    return result
+
+
+if __name__ == "__main__":
+    asyncio.run(main())

examples/pg-memory.py

@@ -0,0 +1,47 @@
+import controlflow as cf
+from controlflow.memory.memory import Memory
+from controlflow.memory.providers.postgres import PostgresMemory
+
+provider = PostgresMemory(
+    database_url="postgresql://postgres:postgres@localhost:5432/database",
+    # embedding_dimension=1536,
+    # embedding_fn=OpenAIEmbeddings(),
+    table_name="vector_db",
+)
+# Create a memory module for user preferences
+user_preferences = cf.Memory(
+    key="user_preferences",
+    instructions="Store and retrieve user preferences.",
+    provider=provider,
+)
+
+# Create an agent with access to the memory
+agent = cf.Agent(memories=[user_preferences])
+
+
+# Create a flow to ask for the user's favorite color
+@cf.flow
+def remember_color():
+    return cf.run(
+        "Ask the user for their favorite color and store it in memory",
+        agents=[agent],
+        interactive=True,
+    )
+
+
+# Create a flow to recall the user's favorite color
+@cf.flow
+def recall_color():
+    return cf.run(
+        "What is the user's favorite color?",
+        agents=[agent],
+    )
+
+
+if __name__ == "__main__":
+    print("First flow:")
+    remember_color()
+
+    print("\nSecond flow:")
+    result = recall_color()
+    print(result)

pyproject.toml

@@ -12,8 +12,7 @@ dependencies = [
     "langchain_openai>=0.2",
     "langchain-anthropic>=0.2",
     "markdownify>=0.12.1",
-    "openai<1.47",
-    # 1.47.0 introduced a bug with attempting to reuse an async client that doesnt have an obvious solution
+    "openai>=1.55.3",
     "pydantic-settings>=2.2.1",
     "textual>=0.61.1",
     "tiktoken>=0.7.0",
@@ -52,6 +51,7 @@ tests = [
     "langchain_community",
     "langchain_google_genai",
     "langchain_groq",
+    "langchain-ollama',
     "pytest-asyncio>=0.18.2,!=0.22.0,<0.23.0",
     "pytest-env>=0.8,<2.0",
     "pytest-rerunfailures>=10,<14",

src/controlflow/__init__.py

@@ -13,6 +13,7 @@
 
 # functions, utilites, and decorators
 from .memory import Memory
+from .memory.async_memory import AsyncMemory
 from .instructions import instructions
 from .decorators import flow, task
 from .tools import tool

src/controlflow/agents/agent.py

@@ -32,6 +32,7 @@
 from controlflow.llm.models import get_model as get_model_from_string
 from controlflow.llm.rules import LLMRules
 from controlflow.memory import Memory
+from controlflow.memory.async_memory import AsyncMemory
 from controlflow.tools.tools import (
     Tool,
     as_lc_tools,
@@ -82,7 +83,7 @@ class Agent(ControlFlowModel, abc.ABC):
         default=False,
         description="If True, the agent is given tools for interacting with a human user.",
     )
-    memories: list[Memory] = Field(
+    memories: list[Memory] | list[AsyncMemory] = Field(
         default=[],
         description="A list of memory modules for the agent to use.",
     )

src/controlflow/defaults.py

@@ -6,6 +6,7 @@
 import controlflow.utilities
 import controlflow.utilities.logging
 from controlflow.llm.models import BaseChatModel
+from controlflow.memory.async_memory import AsyncMemoryProvider, get_memory_provider
 from controlflow.memory.memory import MemoryProvider, get_memory_provider
 from controlflow.utilities.general import ControlFlowModel
 
@@ -39,7 +40,9 @@ class Defaults(ControlFlowModel):
     model: Optional[Any]
     history: History
     agent: Agent
-    memory_provider: Optional[Union[MemoryProvider, str]]
+    memory_provider: (
+        Optional[Union[MemoryProvider, str]] | Optional[Union[AsyncMemoryProvider, str]]
+    )
 
     # add more defaults here
     def __repr__(self) -> str:

src/controlflow/llm/models.py

@@ -60,6 +60,14 @@ def get_model(
                 "To use Groq as an LLM provider, please install the `langchain_groq` package."
             )
         cls = ChatGroq
+    elif provider == "ollama":
+        try:
+            from langchain_ollama import ChatOllama
+        except ImportError:
+            raise ImportError(
+                "To use Ollama as an LLM provider, please install the `langchain-ollama` package."
+            )
+        cls = ChatOllama
     else:
         raise ValueError(
             f"Could not load provider `{provider}` automatically. Please provide the LLM class manually."

src/controlflow/memory/__init__.py

@@ -1 +1,2 @@
 from .memory import Memory
+from .async_memory import AsyncMemory

src/controlflow/memory/async_memory.py

@@ -0,0 +1,149 @@
+import abc
+import re
+from typing import Dict, List, Optional, Union
+
+from pydantic import Field, field_validator, model_validator
+
+import controlflow
+from controlflow.tools.tools import Tool
+from controlflow.utilities.general import ControlFlowModel, unwrap
+from controlflow.utilities.logging import get_logger
+
+logger = get_logger("controlflow.memory")
+
+
+def sanitize_memory_key(key: str) -> str:
+    # Remove any characters that are not alphanumeric or underscore
+    return re.sub(r"[^a-zA-Z0-9_]", "", key)
+
+
+class AsyncMemoryProvider(ControlFlowModel, abc.ABC):
+    async def configure(self, memory_key: str) -> None:
+        """Configure the provider for a specific memory."""
+        pass
+
+    @abc.abstractmethod
+    async def add(self, memory_key: str, content: str) -> str:
+        """Create a new memory and return its ID."""
+        pass
+
+    @abc.abstractmethod
+    async def delete(self, memory_key: str, memory_id: str) -> None:
+        """Delete a memory by its ID."""
+        pass
+
+    @abc.abstractmethod
+    async def search(self, memory_key: str, query: str, n: int = 20) -> Dict[str, str]:
+        """Search for n memories using a string query."""
+        pass
+
+
+class AsyncMemory(ControlFlowModel):
+    """
+    A memory module is a partitioned collection of memories that are stored in a
+    vector database, configured by a MemoryProvider.
+    """
+
+    key: str
+    instructions: str = Field(
+        description="Explain what this memory is for and how it should be used."
+    )
+    provider: AsyncMemoryProvider = Field(
+        default_factory=lambda: controlflow.defaults.memory_provider,
+        validate_default=True,
+    )
+
+    def __hash__(self) -> int:
+        return id(self)
+
+    @field_validator("provider", mode="before")
+    @classmethod
+    def validate_provider(
+        cls, v: Optional[Union[AsyncMemoryProvider, str]]
+    ) -> AsyncMemoryProvider:
+        if isinstance(v, str):
+            return get_memory_provider(v)
+        if v is None:
+            raise ValueError(
+                unwrap(
+                    """
+                    Memory modules require a MemoryProvider to configure the
+                    underlying vector database. No provider was passed as an
+                    argument, and no default value has been configured. 
+                    
+                    For more information on configuring a memory provider, see
+                    the [Memory
+                    documentation](https://controlflow.ai/patterns/memory), and
+                    please review the [default provider
+                    guide](https://controlflow.ai/guides/default-memory) for
+                    information on configuring a default provider.
+                    
+                    Please note that if you are using ControlFlow for the first
+                    time, this error is expected because ControlFlow does not include
+                    vector dependencies by default.
+                    """
+                )
+            )
+        return v
+
+    @field_validator("key")
+    @classmethod
+    def validate_key(cls, v: str) -> str:
+        sanitized = sanitize_memory_key(v)
+        if sanitized != v:
+            raise ValueError(
+                "Memory key must contain only alphanumeric characters and underscores"
+            )
+        return sanitized
+
+    async def _configure_provider(self):
+        await self.provider.configure(self.key)
+        return self
+
+    async def add(self, content: str) -> str:
+        return await self.provider.add(self.key, content)
+
+    async def delete(self, memory_id: str) -> None:
+        await self.provider.delete(self.key, memory_id)
+
+    async def search(self, query: str, n: int = 20) -> Dict[str, str]:
+        return await self.provider.search(self.key, query, n)
+
+    def get_tools(self) -> List[Tool]:
+        return [
+            Tool.from_function(
+                self.add,
+                name=f"store_memory_{self.key}",
+                description=f'Create a new memory in Memory: "{self.key}".',
+            ),
+            Tool.from_function(
+                self.delete,
+                name=f"delete_memory_{self.key}",
+                description=f'Delete a memory by its ID from Memory: "{self.key}".',
+            ),
+            Tool.from_function(
+                self.search,
+                name=f"search_memories_{self.key}",
+                description=f'Search for memories relevant to a string query in Memory: "{self.key}". Returns a dictionary of memory IDs and their contents.',
+            ),
+        ]
+
+
+def get_memory_provider(provider: str) -> AsyncMemoryProvider:
+    logger.debug(f"Loading memory provider: {provider}")
+
+    # --- async postgres ---
+
+    if provider.startswith("async-postgres"):
+        try:
+            import sqlalchemy
+        except ImportError:
+            raise ImportError(
+                """To use async Postgres as a memory provider, please install the `sqlalchemy, `psycopg-pool`,
+                    `psycopg-binary`, and `psycopg` packages."""
+            )
+
+        import controlflow.memory.providers.postgres as postgres_providers
+
+        return postgres_providers.AsyncPostgresMemory()
+    raise ValueError(f'Memory provider "{provider}" could not be loaded from a string.')

src/controlflow/memory/memory.py

@@ -166,4 +166,18 @@ def get_memory_provider(provider: str) -> MemoryProvider:
 
         return lance_providers.LanceMemory()
 
+    # --- Postgres ---
+    elif provider.startswith("postgres"):
+        try:
+            import sqlalchemy
+        except ImportError:
+            raise ImportError(
+                """To use Postgres as a memory provider, please install the `sqlalchemy, `psycopg-pool`,
+                    `psycopg-binary`, and `psycopg` `psycopg2-binary` packages."""
+            )
+
+        import controlflow.memory.providers.postgres as postgres_providers
+
+        return postgres_providers.PostgresMemory()
+
     raise ValueError(f'Memory provider "{provider}" could not be loaded from a string.')