"""
Free embedding service for transaction categorization using Sentence Transformers
"""
from typing import List, Optional, Tuple
from decimal import Decimal
from datetime import datetime
try:
import numpy as np
from sentence_transformers import SentenceTransformer
EMBEDDINGS_AVAILABLE = True
except ImportError:
EMBEDDINGS_AVAILABLE = False
np = None
SentenceTransformer = None
class FreeEmbeddingService:
"""Service for generating embeddings using free local models"""
def __init__(self):
"""Initialize with a free, efficient model"""
if not EMBEDDINGS_AVAILABLE:
raise ImportError("numpy and sentence-transformers are required for embeddings. Install with: pip install numpy sentence-transformers")
# This model is small (90MB) and works well for similarity search
self.model = SentenceTransformer('all-MiniLM-L6-v2')
self.embedding_dimension = 384 # This model outputs 384 dimensions
def format_transaction_text(
self,
date: datetime,
amount: Decimal,
merchant: str,
description: Optional[str] = None,
category: Optional[str] = None
) -> str:
"""
Format transaction data into a standardized text representation
Format: "YYYY-MM-DD│₹amount│merchant│description│category"
"""
# Format date
date_str = date.strftime("%Y-%m-%d")
# Format amount with currency symbol
amount_str = f"₹{float(amount)}"
# Clean merchant name
merchant_str = merchant.strip() if merchant else "Unknown"
# Include description if available
desc_str = description.strip() if description else ""
# Build the text representation
parts = [date_str, amount_str, merchant_str]
if desc_str:
parts.append(desc_str)
# Add category if provided (for training data)
if category:
parts.append(f"→ {category}")
return "│".join(parts)
async def generate_embedding(self, text: str) -> List[float]:
"""
Generate embedding vector for the given text using local model
Args:
text: The text to embed
Returns:
List of floats representing the embedding vector
"""
try:
# Generate embedding (this runs locally, no API needed)
embedding = self.model.encode(text)
return embedding.tolist()
except Exception as e:
raise Exception(f"Failed to generate embedding: {str(e)}")
def calculate_similarity(self, embedding1: List[float], embedding2: List[float]) -> float:
"""
Calculate cosine similarity between two embeddings
Returns:
Similarity score between 0 and 1 (1 being identical)
"""
vec1 = np.array(embedding1)
vec2 = np.array(embedding2)
# Cosine similarity
norm1 = np.linalg.norm(vec1)
norm2 = np.linalg.norm(vec2)
if norm1 == 0 or norm2 == 0:
return 0.0
dot_product = np.dot(vec1, vec2)
similarity = dot_product / (norm1 * norm2)
return float(similarity)
def extract_transaction_details(self, transaction_text: str) -> dict:
"""
Extract transaction details from formatted text
Returns dict with: date, amount, merchant, description
"""
parts = transaction_text.split("│")
details = {
"date": parts[0] if len(parts) > 0 else None,
"amount": parts[1].replace("₹", "") if len(parts) > 1 else None,
"merchant": parts[2] if len(parts) > 2 else None,
"description": parts[3] if len(parts) > 3 and "→" not in parts[3] else None,
}
# Extract category if present
for part in parts:
if "→" in part:
details["category"] = part.split("→")[1].strip()
break
return details
