Transaction Parser

bench get-app https://github.com/resilient-tech/transaction-parser

[![Listed on Frappe Gems](https://frappegems.com/api/method/frappe_gems.seo.badge?app=resilient-tech%2Ftransaction-parser)](https://frappegems.com/gems/apps/resilient-tech/transaction-parser)

# Transaction Parser ## Overview Transaction Parser is an AI-powered add-on for ERPNext that automatically extracts data from PDFs and creates draft documents (Sales Order / Purchase Invoice) . It supports multiple document types and regions, making it easier to digitize and process business documents. ## Features **AI-Powered Extraction**: Uses advanced AI models (OpenAI, DeepSeek, Google Gemini, Anthropic) to extract structured data from PDFs * **Multi-Document Support**: Handles Sales Orders and Purchase Invoices (Expenses) * **Regional Support**: Special handling for India-specific requirements (GSTIN, PAN, HSN codes) * **Email Integration**: Automatically processes documents from incoming emails * **Customizable Schemas**: Flexible field mapping and custom schema support * **Smart Item Matching**: Automatically matches items from previous invoices ## Configuration 1\. Enable Transaction Parser Navigate to **Transaction Parser Settings** and configure: 1. **Enable**: Check to activate the app 2\. **Default AI Model**: Select from available models: * DeepSeek Chat * DeepSeek Reasoner * OpenAI gpt-4o * OpenAI gpt-4o-mini * OpenAI gpt-5 * OpenAI gpt-5-mini * Google Gemini Pro-2.5 * Google Gemini Flash-2.5 * Claude Haiku-4.5 3\. API Keys Setup Add your API keys for the AI services: | Service Provider | Models Supported | |:-----------------|:---------------------------------------| | OpenAI | gpt-4o, gpt-4o-mini, gpt-5, gpt-5-mini | | DeepSeek | deepseek-chat, deepseek-reasoner | | Google | gemini-2.5-pro, gemini-2.5-flash | | Anthropic | claude-haiku-4-5 | 4\. Email Configuration (Optional) To automatically process documents from emails: 1. **Parse Incoming Emails**: Enable email processing 2. **Incoming Email Accounts**: Configure which email accounts to monitor 3. **Party Emails**: Map email addresses to specific customers/suppliers 5\. Transaction Configuration * **Invoice Lookback Count**: Number of past invoices to consider for item matching (default: 5\) ## Usage ### Manual Document Processing 1. Navigate to Sales Order or Purchase Invoice list view 2. Click on **Actions → Parse Sales Order/Expense Invoice** 3. Upload your PDF file 4. Select: * **AI Model**: Choose the AI model to use * **Country**: Select India or Other * **Page Limit**: (Optional) Limit pages to process 5. Click **Submit** The system will: * Extract text from the PDF * Send it to the AI model for processing * Create a draft document with extracted data * Attach the original PDF to the created document ### Automatic Email Processing When enabled, the system automatically: 1. Monitors configured email accounts 2. Extracts PDF attachments from emails 3. Processes them based on sender and configuration 4. Creates draft documents ## Model Comparison | Model | Provider | Best For | Speed | Cost | |:------------------|:----------|:--------------------------------------------------------|:----------|:------------| | gpt-5 | OpenAI | State-of-the-art accuracy, complex multi-page documents | Medium | High | | gpt-5-mini | OpenAI | Efficient reasoning, cost-effective | Fast | Medium | | gpt-4o | OpenAI | Complex documents, high accuracy | Medium | Medium-High | | gpt-4o-mini | OpenAI | Cost-effective, good accuracy | Fast | Low | | gemini-2.5-pro | Google | Advanced reasoning, large context window | Medium | Medium | | gemini-2.5-flash | Google | Fast processing, bulk documents | Very Fast | Low | | deepseek-chat | DeepSeek | General purpose extraction | Fast | Low | | deepseek-reasoner | DeepSeek | Complex reasoning tasks | Slow | Medium | | claude-haiku-4-5 | Anthropic | Fast, lightweight tasks | Fast | Low | ## India-Specific Features The Transaction Parser app includes robust support for Indian business requirements through integration with the **India Compliance** app. These features enable automatic handling of GST regulations, Indian business identifiers, and region-specific validation requirements. ### Prerequisites * **India Compliance App**: Must be installed for India-specific features to work ### India-Specific AI Model Enhancements **Enhanced Data Extraction** - When processing documents with the India region selected, the AI models are enhanced to extract: 1. **GST Identification Numbers (GSTIN)** 2. **Permanent Account Numbers (PAN)** 3. **HSN/SAC Codes** 4. **Tax Components** ### Automatic Supplier Creation **GSTIN-Based Supplier Creation** When enabled in settings, the system can automatically create suppliers: 1. **Configuration** * Enable "Auto Create Supplier" in Transaction Parser Settings * Requires valid GSTIN in the invoice ## PDF Processor Setup * Transaction Parser supports three PDF processors for text extraction. * Only **PDFtoText** (the default) is installed as a required dependency. * The other two are optional. ### Installing Optional PDF Processors ```bash # Install OCRMyPDF env/bin/pip install -e "apps/transaction_parser[ocrmypdf]" # Install Docling env/bin/pip install -e "apps/transaction_parser[docling]" # Install all optional processors env/bin/pip install -e "apps/transaction_parser[all]" ``` ### 1. PDFtoText (Default) Layout-preserving text extraction using [pdftotext](https://github.com/jalan/pdftotext). > [!IMPORTANT] > Install OS dependencies **before** running `bench setup requirements` or `pip install`, otherwise the `pdftotext` Python package will fail to build. **OS Dependencies (Debian/Ubuntu):** ```bash sudo apt install build-essential libpoppler-cpp-dev pkg-config python3-dev ``` > For other operating systems, see [pdftotext OS dependencies](https://github.com/jalan/pdftotext#os-dependencies). ### 2. OCRMyPDF (Optional) OCR-based text extraction using [OCRmyPDF](https://github.com/ocrmypdf/OCRmyPDF). Useful for scanned or image-based PDFs. **OS Dependencies (Debian/Ubuntu):** ```bash sudo apt-get install -y tesseract-ocr ghostscript ``` ### 3. Docling (Optional) Advanced document understanding using [Docling](https://docling-project.github.io/docling/) with EasyOCR for OCR support. > See [Docling OCR engines](https://docling-project.github.io/docling/getting_started/installation/#ocr-engines) for more details. **Post-install fix for headless servers:** After installing the `docling` extra, replace `opencv-python` with the headless variant: ```bash bench pip uninstall opencv-python bench pip install opencv-python-headless ``` This is required because `opencv-python` depends on `libGL.so.1`, which is unavailable on headless servers: ```bash ImportError: libGL.so.1: cannot open shared object file: No such file or directory ``` ### Summary | Processor | Dependency Type | OS Packages Required | OCR | |:----------|:----------------|:------------------------------------------------------------|:----| | PDFtoText | **Required** | `build-essential libpoppler-cpp-dev pkg-config python3-dev` | No | | OCRMyPDF | Optional | `tesseract-ocr ghostscript` | Yes | | Docling | Optional | None | Yes | ## License [GNU General Public License (v3)](https://github.com/resilient-tech/transaction-parser/blob/version-15/license.txt)