infimum-core 1.2.20

Infimum core library: base, engine, database, AI (LLM/VLM/speech), utils

infimum-core

Infimum core library: base entities, engine (DI/context/startup), database layer (PostgreSQL, Milvus, Qdrant, MongoDB), AI (LLM, VLM, speech, embeddings), and utilities.

Documentation: From repo root run pip install -e ./core && pip install -r docs/requirements.txt && mkdocs serve, then open http://127.0.0.1:8000. Or run mkdocs build to output the static site to site/.

Install

From PyPI:

pip install infimum-core

From GitHub (default branch):

pip install "git+https://github.com/inf-codebase/infimum.git#subdirectory=infimum"

From a GitHub release: use the release tag (e.g. v1.2.20) with @tag before #subdirectory:

pip install "git+https://github.com/inf-codebase/infimum.git@v1.2.20#subdirectory=infimum"

With optional extras:

# All database backends
pip install "infimum-core[all-db]"

# LLM/Agent stack
pip install "infimum-core[llm]"

# Security features
pip install "infimum-core[security]"

# All extras
pip install "infimum-core[all-db,security,llm]"

Local development: pip install -e . from the infimum directory. With uv: uv sync from the infimum directory.

Optional extras

Extra	Description
mongo	MongoDB support (pymongo, motor)
milvus	Milvus vector database
qdrant	Qdrant vector database
all-db	All database backends (mongo, milvus, qdrant)
security	JWT, password hashing (bcrypt, python-jose)
llm	LLM/agent stack (LangChain, LangGraph)

Usage

Basic Imports

from infimum import Engine
from infimum.base.entity import BaseEntity, Document
from infimum.database import DatabaseManager, VectorIndexConfig
from infimum.engine import context, startup
from infimum.utils import string_utils, auto_config

Example 1: Define Entities

from infimum.base.entity import BaseEntity

class User(BaseEntity):
    """User entity with base fields."""
    name: str
    email: str
    age: int = None

# Create instance
user = User(name="John Doe", email="[e-mail]", age=30)
print(user.model_dump())

Example 2: Initialize Engine with Dependency Injection

from infimum.engine import Engine

# Create engine instance
engine = Engine()

# Register services
engine.register("database", DatabaseManager(...))
engine.register("embedder", EmbeddingProvider(...))

# Retrieve services
db = engine.get("database")

Example 3: Database Operations

from infimum.database import DatabaseManager, VectorIndexConfig

# Create database manager
db = DatabaseManager(
    db_type="postgres",
    connection_string="postgresql://user:password@localhost/dbname"
)

# Create a table/collection
db.create_collection("users", User)

# Insert document
user = User(name="Alice", email="[e-mail]", age=28)
db.insert("users", user)

# Query
results = db.query("users", {"name": "Alice"})

Example 4: Vector Database (Embeddings)

from infimum.database import VectorIndexConfig
from infimum.ai.embeddings import OpenAIEmbeddingProvider

# Create embeddings
embedder = OpenAIEmbeddingProvider(api_key="sk-...")

# Create vector index
vector_config = VectorIndexConfig(
    collection_name="documents",
    dimension=1536,
    metric_type="cosine"
)

# Embed and search
text = "What is artificial intelligence?"
embedding = embedder.embed(text)

# Search similar vectors
results = db.search_vectors("documents", embedding, top_k=5)

Example 5: LLM Integration

from infimum.ai.llm import LLMProvider
from langchain_openai import ChatOpenAI

# Initialize LLM
llm = ChatOpenAI(model="gpt-4", api_key="sk-...")

# Simple generation
response = llm.invoke("What is the capital of France?")
print(response.content)

# With context from database
context = db.query("documents", {"topic": "France"})
prompt = f"Based on {context}, answer: What is the capital of France?"
response = llm.invoke(prompt)

Example 6: Speech Processing

from infimum.ai.speech.providers import MedASRProvider

# Initialize speech provider
speech_provider = MedASRProvider(api_key="...")

# Transcribe audio
audio_file = "speech.wav"
transcript = speech_provider.transcribe(audio_file)
print(f"Transcription: {transcript}")

Example 7: Auto Configuration

from infimum.utils.auto_config import AutoConfig

# Auto-load configuration from environment or config file
config = AutoConfig.from_env()

# Access config values
db_url = config.get("DATABASE_URL")
api_key = config.get("OPENAI_API_KEY")

Example 8: Context Management

from infimum.engine import context

# Set context values (useful for multi-tenant apps)
context.set("user_id", "user_123")
context.set("tenant_id", "tenant_456")

# Retrieve context values
user_id = context.get("user_id")

Example 9: Complete Application

from infimum import Engine
from infimum.base.entity import BaseEntity
from infimum.database import DatabaseManager
from infimum.ai.llm import LLMProvider
from langchain_openai import ChatOpenAI

# Define entity
class Article(BaseEntity):
    title: str
    content: str
    author: str

# Initialize components
engine = Engine()
db = DatabaseManager(db_type="postgres", connection_string="...")
llm = ChatOpenAI(model="gpt-4", api_key="...")

# Register with engine
engine.register("database", db)
engine.register("llm", llm)

# Use throughout app
db_service = engine.get("database")
db_service.insert("articles", Article(title="AI Guide", content="...", author="John"))

llm_service = engine.get("llm")
summary = llm_service.invoke("Summarize this article: ...")

Subpackages

• infimum.base — Base entities, registry, repository interfaces
• infimum.engine — Dependency injection, context management, startup hooks, security (optional)
• infimum.database — Database managers and interfaces (PostgreSQL, Milvus, Qdrant, MongoDB)
• infimum.ai — LLM, VLM, speech, embeddings, data loaders, preprocessing
• infimum.utils — Configuration, validation, helpers, Redis client, etc.

Configuration

Set environment variables for database connections:

export DATABASE_URL="postgresql://user:password@localhost/dbname"
export OPENAI_API_KEY="sk-..."
export MILVUS_URL="http://localhost:19530"

Then load with auto_config:

from infimum.utils.auto_config import AutoConfig

config = AutoConfig.from_env()
db_url = config.get("DATABASE_URL")