txn-msg-parser 0.1.1

A Python package for parsing transaction SMS messages into structured transaction dictionaries

Transaction SMS Parser

A Python package for parsing transaction SMS messages into structured transaction dictionaries using AI.

Features

• Parse transaction SMS messages from various banks
• Extract transaction details like amount, type, payee/payer, category
• Configurable accounts and categories
• Uses AI (Ollama) for intelligent parsing

Installation

From PyPI (once published)

pip install txn-msg-parser

From source

git clone https://github.com/yourusername/txn_msg_parser.git
cd txn_msg_parser
pip install -e .

Requirements

• Python 3.8+
• Ollama installed and running locally (see Ollama installation)
• Ollama model: deepseek-r1:8b (or configure your own)

The package automatically installs the Python ollama library as a dependency.

Usage

Basic Example

from txn_msg_parser import TextParser, TxnText

# Initialize the parser with your bank accounts
accounts = ["HDFCBK", "SBIUPI"]
parser = TextParser(accounts=accounts)

# Create a transaction text object
txn_text = TxnText()
txn_text.sender = "HDFCBK-S(smsft)"
txn_text.text = "Sent Rs.437.00 From HDFC Bank A/C *5232 To BLINKIT COMMERCE PRIVATE LIMITED On 21/10/25"
txn_text.date = "2025-10-21 21:08:15"
txn_text.type = "sms"
txn_text.id = "unique-id"

# Parse the transaction
txn = parser.parse_text(txn_text)

# Access transaction details
print(f"Account: {txn.account}")
print(f"Amount: {txn.amount}")  # Amount in paise (multiplied by 100)
print(f"Type: {txn.txn_type}")  # 'debit' or 'credit'
print(f"Payee: {txn.payee}")
print(f"Category: {txn.category}")

Batch Processing

from txn_msg_parser import TextParser, TxnText

accounts = ["HDFCBK", "SBIUPI"]
parser = TextParser(accounts=accounts)

# Parse multiple messages
texts = [txn_text1, txn_text2, txn_text3]
transactions = parser.parse_texts(texts)

for txn in transactions:
    print(f"{txn.date}: {txn.amount} - {txn.category}")

Custom Categories

from txn_msg_parser import TextParser

custom_categories = ["Salary", "Rent", "Groceries", "Entertainment", "Utilities"]
accounts = ["HDFCBK", "SBIUPI"]

parser = TextParser(accounts=accounts, categories=custom_categories)

Configuration

Default Model

The package uses deepseek-r1:8b by default. You can change this:

parser = TextParser(accounts=accounts, model="your-ollama-local-model")

Default Categories

Default categories include: ["Salary", "EMI", "Food", "Travel", "Bills", "Shopping"]

Output Format

The Txn object contains:

• account: Bank account identifier
• amount: Transaction amount in paise (cents)
• txn_type: "debit" or "credit"
• payee: Recipient of payment (for debit transactions)
• payer: Sender of payment (for credit transactions)
• category: Transaction category
• full_text: Original SMS text
• date: Transaction date
• id: Unique transaction identifier

Development

Setup Development Environment

git clone https://github.com/yourusername/txn_msg_parser.git
cd txn_msg_parser
python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate
pip install -e .

Running Tests

python test.py

License

MIT License

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.