treerag 0.1.1

Lightweight structural RAG library — index documents as trees, query without a vector DB.

treerag 🌳

RAG without vector databases.

A lightweight structural RAG library that indexes documents as hierarchical trees and retrieves answers using structure instead of embeddings.

Works with any LangChain-compatible LLM: OpenAI, Anthropic, Gemini, Ollama.

---

🚀 Why treerag?

• ❌ No vector database required
• ⚡ Fast hierarchical retrieval
• 🧠 Uses document structure instead of embeddings
• 🔌 Works with any LLM
• 🪶 Lightweight and easy to integrate

---

📦 Install

# Base
pip install treerag

# With OpenAI
pip install "treerag[openai]"

# With Anthropic
pip install "treerag[anthropic]"

# Everything
pip install "treerag[all]"

---

⚡ Quick Start

from treerag import index_document, make_summarizer, ask, make_retriever
from langchain_openai import ChatOpenAI

llm = ChatOpenAI(model="gpt-4o")

# 1. Index document
doc = index_document("my_doc.md", summarizer=make_summarizer(llm))

# 2. Ask question
result = ask("What does this document cover?", doc, make_retriever(llm))

print(result.content)            # answer
print(result.references)         # sections used
print(result.response_metadata)  # token usage

---

🧠 How It Works

Indexing

File / URL
    ↓ read_file()
    ↓ parse_sections()
    ↓ make_summarizer()
    ↓ build_hierarchy()
    ↓ flatten_tree()
    ↓ save_registry()

Querying

Query
    ↓ tree search (LLM selects relevant nodes)
    ↓ fetch content
    ↓ answer generation
    ↓ AIMessage with references

---

📂 Supported Inputs

# Files
index_document("file.md")
index_document("report.pdf")
index_document("document.docx")
index_document("notes.txt")

# URLs
index_document("https://docs.example.com")

---

📊 Response Formats

# Default (LangChain AIMessage)
result = ask("What is this?", doc, retriever)
print(result.content)
print(result.references)

# Plain dict
result = ask("What is this?", doc, retriever, return_raw=False)
print(result["answer"])

# Streaming
for chunk in ask("What is this?", doc, retriever, stream=True):
    if isinstance(chunk, dict):
        print(chunk["__references__"])
    else:
        print(chunk, end="")

---

⚡ Async Support

from treerag import aask, make_async_retriever

retriever = make_async_retriever(llm)
result = await aask("What is this?", doc, retriever)
print(result.content)

---

📚 Multi-Document Q&A

from treerag import ask_multi, get_document_by_id

doc1 = get_document_by_id("uuid-1")
doc2 = get_document_by_id("uuid-2")

result = ask_multi("What is the budget cap?", [doc1, doc2], retriever)
print(result.content)
print(result.references)

---

🗂️ Registry Management

from treerag import list_documents, get_document_by_id, delete_document

for doc in list_documents():
    print(doc["name"], doc["doc_id"])

doc = get_document_by_id("uuid")
delete_document("uuid")

---

🎯 Custom Prompts

summarizer = make_summarizer(
    llm,
    system_prompt="You are a legal expert. Summarize clauses and obligations."
)

retriever = make_retriever(
    llm,
    answer_system_prompt=(
        "You are a legal assistant. "
        "Answer using ONLY the provided context."
    )
)

result = ask(
    "What are the key obligations?",
    doc,
    retriever,
    extra_context="This is a legal agreement."
)

---

🔌 Supported Providers

from langchain_openai import ChatOpenAI
from langchain_anthropic import ChatAnthropic
from langchain_google_genai import ChatGoogleGenerativeAI
from langchain_ollama import ChatOllama

# OpenAI
make_retriever(ChatOpenAI(model="gpt-4o"))

# Anthropic
make_retriever(ChatAnthropic(model="claude-haiku"))

# Gemini
make_retriever(ChatGoogleGenerativeAI(model="gemini-2.0-flash"))

# Ollama (local)
make_retriever(ChatOllama(model="llama3"))

---

🏗️ Production Usage

doc = index_document("file.md", persist=False)

# store externally
db.save(doc["doc_id"], doc)

# load later
doc = db.get("id")

result = ask("Your question?", doc, retriever)

---

📌 Example Use Cases

• 📄 Documentation Q&A
• 📚 Internal knowledge base
• 🤖 AI assistants without vector DB
• 🧾 Legal / contract analysis

---

⚔️ treerag vs Traditional RAG

Feature	treerag	Vector RAG
Setup	Simple	Complex
DB required	❌	✅
Cost	Low	High
Retrieval	Structure-based	Embeddings

---

📄 License

MIT