Fix CSV import problems without damaging values

Inspect before transforming

Start with CSV viewer and select the export’s delimiter. A semicolon file opened as comma-separated may appear as a single column without being malformed. Choose the file encoding before reading it. UTF-8 is the default; UTF-16 and Windows-1252 are available when the source system explicitly uses them.

Keep identifiers as text. Opening a file in a spreadsheet and saving it again may turn 00123 into 123 before another converter ever sees it.

A valid quoted record

id,name,note
00123,Ada,"Hello, world"
00456,Sam,"two
lines"

This has two data records and three columns. The comma in “Hello, world” is part of a value. The newline inside Sam’s quoted note is also part of a value. An internal quote must be doubled: “She said ""hello""” when represented inside a quoted CSV field.

Correct structural errors explicitly

If a record has fewer fields, add the missing empty cell or correct its quoting at the source. Do not silently discard it. Headers must be nonempty and unique when header mode is enabled. A UTF-8 BOM is removed from the start of parsing; it is not part of the first header.

Use delimiter conversion rather than global replacement. Use the column editor for name/order mismatches. Merge requires the same headers in the same order; split works by complete records, not physical lines.

Export for the receiving application

CSV-to-JSON preserves strings by default. Enable type conversion only when exact numbers, booleans and null values should change type. Spreadsheet-safe CSV export prefixes formula-like cells with an apostrophe; this changes representation and is not a universal security guarantee for every spreadsheet.

The preview shows at most 50 records, but the download contains all accepted records. Check the record count and reopen a synthetic output in the intended importer before processing a larger batch.