CSV data¶
When a chart source is Direct Connection or From Service and returns CSV, the editor parses the file once and types each column before exposing it to the visual editor.
Date detection¶
A column is treated as a date axis (xaxis.type: date) when its values parse cleanly as ISO 8601 dates, YYYY-MM-DD, or common Plotly-friendly date strings. Mixed numeric/date columns fall back to category.
Numeric parsing¶
- Empty cells become
null(gaps in the trace, not zeros). - Strings with thousands separators (
1,234.5) are normalised before parsing. - Columns where every non-empty cell is a finite number are tagged numeric and shown in the Y picker; mixed columns appear only in the X picker.
Header handling¶
The first row is always treated as the header. Header names are used verbatim as trace name and axis title.text defaults — rename them in the source CSV rather than the editor when you can, so multiple charts stay consistent.