Convert legacy reports and fixed-width text to clean Excel

A plain-text file where each field sits at a fixed range of character positions instead of being separated by commas or tabs. Printed reports and data exports from IBM i (AS/400), mainframes, COBOL programs and older ERP, accounting and payroll systems usually come this way, as .txt, .prn, .rpt, .lis, .out or .dat files.

Drop the file onto the converter above, or paste the text. The columns are found for you, page headers, dividers and subtotals are removed, and you download a clean .xlsx or .csv file. There are no formulas to write and no column widths to set.

Both make you set the column breaks yourself, and both treat a report as plain data: page headers, repeated column titles, subtotals and blank lines come in as rows, and amounts like 1,440.00- arrive as text. This converter finds the columns, removes the report clutter and fixes the values for you, the way report-mining software does, right in your browser.

.prn and spool files are reports printed to a file, laid out in fixed-width columns with page headers on every page. Excel opens them as one long column of text. Drop the file here instead: it's split into columns, the page headers and printer codes are removed, and you can download it as Excel.

Yes. It reads spool files and data exports from IBM i (AS/400) and z/OS, including EBCDIC files, ASA carriage-control characters, records with no line breaks, JES job logs around the report and CYYMMDD dates. Signed amounts such as 1,250.00CR or COBOL zoned decimals become real negative numbers.

Yes. Add the .cpy copybook along with the data file and each record is read to its exact layout: PIC X and 9 fields, implied decimals (V99), signed zoned decimals, OCCURS, REDEFINES, and packed decimal (COMP-3) and binary (COMP) fields decoded straight from the file's bytes. Header and trailer records are set aside, and the detail records become the table.

Yes. SAP lists drawn with | bars, SQL Server “results to text”, Oracle SQL*Plus, DB2, MySQL and PostgreSQL output are all read column by column. Messages such as “4 rows selected.” are left out, NULL becomes a blank cell, and a file holding several result sets gives one table per result.

That's the point. Nothing is deleted silently: every line is either in the table or in a review list where you can restore it. Negative formats, dates and leading zeros are converted by fixed rules, not retyped by AI, and rows that may sit off their columns are flagged so you can check them against the source.

A chatbot can tidy up a short sample, but a full report is often more than it reads in one go, it may retype a number wrong, and the whole file goes to the provider. Here the conversion is done by code that slices every line the same way, so every value comes straight from your file, and AI only sees a short masked sample to help find the columns.

No. Your file is read, cleaned and converted entirely in your browser, and the full file never leaves it. To help find the columns, AI receives a sample of up to 70 lines. Long numbers in it are masked by our code in your browser before it is sent, not by the AI; other text in those lines is sent as-is. AI only suggests where the columns are and never changes your values. We don't store your file or the sample, and we don't use your data to train AI models. The AI provider handles the sample under its own terms, which our Privacy Policy explains.

Up to 50 MB. A 150,000-line report converts in a few seconds, and work runs in the background so the page stays responsive. If the result has more rows than one Excel sheet holds (1,048,576), it continues on further sheets.

Click “Columns look off? Adjust them manually” below the table to drag, add or remove column boundaries. The table and your download update instantly.

No. There's no account to create and nothing to install. Open the page and convert.