ai-search-match-framework 0.5.0

Lightweight framework for AI-powered document analysis and search workflows with multi-provider support

AI Search Match Framework (ASMF)

Lightweight Python framework for building AI-powered document analysis applications.

Clean abstractions for AI providers, document parsing, and custom analyzers. Build domain-specific tools like patent analyzers, grant finders, or contract reviewers.

Quick Start

pip install ai-search-match-framework

from asmf.providers import AIProviderFactory
from asmf.parsers import PDFPatentParser
from asmf.analyzers import BaseAnalyzer

# AI Provider with automatic fallback (Gemini → Ollama)
provider = AIProviderFactory.create_provider()
response = provider.analyze_text("Analyze this document...")

# Parse documents
parser = PDFPatentParser()
doc = parser.parse("document.pdf")

# Create custom analyzer
class MyAnalyzer(BaseAnalyzer):
    def analyze(self, data):
        return self.provider.analyze_text(f"Evaluate: {data}")

analyzer = MyAnalyzer(provider)
results = analyzer.batch_analyze([doc1, doc2])

Features

• 🤖 Multi-provider AI - Unified interface for Gemini, Ollama with automatic fallback
• 📄 Document parsing - Extract structured data from PDFs and other formats
• 🔧 Extensible analyzers - Base classes for custom domain logic
• 🔍 PR Review Webhook - Automated code review using Ollama with anti-pattern detection
• 🎯 Type-safe - Full type hints for excellent IDE support
• ✅ Well-tested - 78%+ coverage, production-ready
• 🐳 Docker support - Containerized testing and deployment

Core Components

Component	Purpose
providers	AI provider interface (Gemini, Ollama) with failover
parsers	Document parsing (PDF, etc.) with structured output
analyzers	Base classes for domain-specific analysis logic
llm	Task-specific model selection with VRAM detection
utils	Configuration, logging, common utilities

Configuration

Quick Start with Ollama (Local, Free)

# 1. Install Ollama: https://ollama.ai/download
# 2. Pull a model
ollama pull qwen2.5:14b-q4

# 3. Configure ASMF
echo "OLLAMA_BASE_URL=http://localhost:11434" >> .env
echo "PREFER_LOCAL=true" >> .env

Or use the automated setup:

python scripts/setup_ollama.py

Cloud Provider Setup

Set environment variables:

GEMINI_API_KEY=your_key_here  # For Gemini
PREFER_LOCAL=false  # Cloud-first (default)

Advanced Configuration

# Local-first with fallback
from asmf.providers import AIProviderFactory
provider = AIProviderFactory.create_provider(prefer_local=True)

# Direct provider usage
from asmf.providers import GeminiProvider, OllamaProvider
gemini = GeminiProvider(api_key="...", model="gemini-1.5-pro")
ollama = OllamaProvider(model="qwen2.5:14b-q4", base_url="http://localhost:11434")

# Task-specific model selection (NEW!)
from asmf.llm import ModelSelector, TaskType
selector = ModelSelector()  # Auto-detects GPU
model = selector.select_model(TaskType.CODE_REVIEW)  # Best model for code review
ollama = OllamaProvider(model=model)

See docs/OLLAMA_SETUP.md for comprehensive Ollama setup guide and task-specific model recommendations.

Use Cases

Not a general LLM framework - Use LangChain/LlamaIndex for RAG/chatbots.
ASMF is for linear document analysis workflows:

• ✅ Patent analysis and prior art search
• ✅ Grant/RFP matching and evaluation
• ✅ Contract review and risk assessment
• ✅ Resume screening and candidate matching
• ✅ Research paper analysis and summarization

Examples

See examples/job_finder/ for a complete application.

Documentation

• Full Documentation - Complete guide
• API Reference - Detailed docs
• Configuration - Setup guide
• Usage Patterns - Best practices
• Ollama Setup - Local LLM setup
• Webhook Server - Automated PR review with Ollama

Requirements

• Python 3.10+
• Dependencies: requests, beautifulsoup4, google-generativeai, httpx, pypdf, pdfplumber

License

MIT License - See LICENSE

Contributing

Contributions welcome! See CONTRIBUTING.md for guidelines.

AI Contributions: We track AI-generated commits with [AI] prefix for accessibility advocacy and productivity metrics. See commit conventions for details.

---

Why ASMF vs others?

Framework	Focus	Best For
ASMF	Linear document analysis	Domain-specific evaluation tools
LangChain	RAG, agents, chains	Chatbots, conversational AI
LlamaIndex	Data indexing, retrieval	Knowledge base search

Built with ❤️ using AI assistance (GitHub Copilot, Gemini, Ollama)