Calc - Import CSV - improvements needed

Hi!

I am a first time LibreOffice user. I tried to import some CSV files into Calc to do some analysis. I thought I’d share a fresh look from the outside on this topic.

  1. I didn’t find an option (checkbox) to use first row as the column names in the import window. If you ask “why would you need that”? For the following reasons:
  • The AutoFilter should be automatically applied when this option is checked.
  • Calc would ignore the header row when inferring data types (e.g., a column header “Year” wouldn’t interfere with numeric detection).
  • In the future a distinct header style (bold/shaded) would be automatically applied to that row.
  1. The data type inference should be done automatically by default. Most users expect their data to be “live” numbers or dates immediately upon import without having to perform extra steps.
  2. Rename the “Detect special numbers” to “Detect data types” or “Detect number/date formats” and enable it by default. “Special numbers” has no meaning, really.
  3. In the “Column type” dropdown in the import dialog is missing common numeric types like Percent, Currency, Scientific, Fraction, and Boolean, all of which are present in the main toolbar.
  4. The “Character set” should default to UTF-8, not ISO-8859-1. UTF-8 is the modern standard, compatible with ASCII. Defaulting to a legacy encoding may result in “mojibake” (garbled text) and may force a user to repeat the import.
  5. The “From row” setting should reset to 1 on every import. Currently if I set “From row” to 2 for one file, Calc remembers that for the next file (even after a restart). This is a trap! Most imports are unrelated. It may result in silent data loss that goes unnoticed by the user. Please consider resetting this to 1 by default or adding a “Remember for this session” toggle next to it.

The above items make working with CSV import require much more work than necessary, and it very often may be a deal breaker, especially if the CSV file contains dozens / hundreds of columns.

Thank you in advance!

EDIT: This is 26.2.2.2 on x86_64 (Ubuntu Linux 26.04)

1 Like

That option you can find, when you import from Calc to Base.

This kind of automatic but wrong imports where one of my main reasons to avoid using M$-Office and use OpenOffice.
I assume a lot of users think this is an easy task, because they never had to sort out garbled data (and may even think their way to write dates is the only option).

Make useful suggestions or be not surprised of the outcome

That seems to be the difference between format and type

sounds not to wrong for a legacy-system like csv where format is an adventure-game for nerds.

Most requests here are monthly statements from banks etc, so have common properties. So maybe you are wrong a bit…
The whole idea of not showing the import filter at all came from this users…

In that case, set the field From Row to 2
.
CSV includes any layout of data, you need to do some interpretation, sometimes using information that is not necessarily supplied within the csv, or you can get garbled data

Dependency tree for Bug 109236