XML to CSV and Excel: export without losses
The main fork when converting XML to a flat format is not technical but a matter of meaning: you must decide what a row is. The number of rows and the correctness of every sum depend on it.
Step one: choose the level
In any reasonably complex XML the data sits in levels: a document contains records, a record contains sub-records, a sub-record contains a list. A flat table can represent only one level.
So the first question is which level this particular task needs. Reconciling the number of orders has one answer, recounting line items another. A good tool shows which elements repeat and how many times, and lets you switch without reopening the file.
Step two: decide what to do with nested lists
If a repeating element remains inside the chosen row, there are three options:
- Join the values with a separator — fine when there are two or three and you just want to see them in the row;
- Count them — when the fact and the number matter, not the content;
- Change the level and make the nested element the row — the right answer when you will actually work with that data.
What not to do is flatten everything into one table at once. You get a Cartesian product in which top-level data is duplicated and every total over it is inflated. The Levels button opens line items as a linked table instead: each row carries its parent’s fields and each parent gets an item counter.
Step three: types and encoding
In XML every value is text. On export, numbers should become numbers, dates dates, and codes should stay text. This matters most for identifiers such as account numbers, postal codes and phone numbers: turned into numbers, they lose leading zeros and precision — the same mechanism as in CSV.
The encoding of the source XML is stated in its declaration, but it does not carry over to the export automatically. For CSV, choose UTF-8 with a BOM if the file is going to Excel, and a legacy single-byte encoding only if an old system expects it.
When a viewer is the wrong tool
If the conversion repeats regularly on a schedule, automate it: an XSLT transform or a script with streaming parsing. A tool with an interface is good for a one-off job and for investigation, not for a pipeline.
The line runs along repetition: look once and export — use an interface; the same thing every week — write a script.
FAQ
Are attributes kept on export?
Yes, they become separate columns, with a marker so they are not confused with nested elements of the same name.
What happens to very long text values?
In CSV and XLSX they are exported in full. A limit appears only on export to DBF, where long text goes into a memo field.
Can I export several levels at once?
As separate files, yes: switch the level and export again. Merging them into one table is not advisable.
Free for personal use. Your file is not uploaded to a server. Windows version — 3.5 MB, no installation: details. Organizations: license.