If a transaction description on your CajaSur statement is long enough to wrap across two or three lines, it still needs to land in a single row once converted. When a transaction description is long enough to wrap across several lines on the statement page, the exported data can land in a single spreadsheet cell or fragment across multiple rows. The engine reads the entire description block, reassembles the wrapped text into one coherent record, and assigns it to the correct transaction row. No line is orphaned, and the narrative detail that a reviewer needs stays intact in the exported file.
When your CajaSur statement arrives as a scan or a photo of the page rather than a clean export, the numbers on it are just pixels until they are extracted. A scanned or photographed statement introduces a different class of risk: a smudge, a skewed angle or a slightly misread digit can alter a transaction amount or break the running balance. The engine treats every file as a series of numbers that must prove themselves. It extracts each transaction table row and immediately checks the printed running balance column, line by line, against the statement’s own closing figure. If a single 3 was read as an 8, the arithmetic fails at that row and the row is flagged for review. The rest of the file continues to process; you get every correctly read transaction exported to Excel, CSV or QBO while the one suspect row waits for a human check. No file is rejected wholesale because of a single scan flaw. The column mapping step preserves the statement’s original structure, so the date stays in the date column, the description in the description column and the amount in the amount column, exactly as your ledger expects, without manual reordering.