Bank Statement Converter

Problem: Mapping Statements to Import Templates

Bank statements arrive as MT940, domestic Excel, or PDF, while target systems such as BIPV5, EAS_YXH, FINGARD, NSTC, and YYNCC expect different field names, column orders, and file layouts. Manual cleanup often requires renaming columns, separating debit/credit amounts, normalizing currency, and aligning dates, which becomes error-prone in batch workflows.

How It Works: Parse, Map, Export

• Multi-format parsing: supports MT940, Excel, and PDF; PDF tables are reconstructed from PyMuPDF coordinates and handle wrapped headers.
• Field matching: uses cosine similarity to recognize near-match column names, reducing manual mapping work.
• Target-system output: emits templates for BIPV5, EAS_YXH, FINGARD, NSTC, YYNCC, and other mapped systems.
• Batch merging: combines multiple statements in a directory into merged_{target}.xlsx, or converts a single file to {original_name}_{target}.xlsx.

Limits to Check

• Encrypted PDFs, scanned PDFs, or unusual mixed layouts may need conversion to a text-based PDF first.
• Missing required fields, malformed dates or amounts, and invalid mapping configs produce explicit error codes, so verify the source file and mapping JSON.