CSV and PDF Entry-List Import

Fills a participant preset from an existing entry list instead of typing it by hand. CSV is a local, instant parse: RaceTagger reads your column headers and maps them to race number, name, team, category and more. PDF sends the document to an AI edge function (parsePdfEntryList) that reads it visually — including scanned, image-only PDFs a text extractor can't decode — and returns each participant with a structured list of drivers. Both paths can either create a brand-new preset or merge into the preset you already have open, updating existing race numbers and adding new ones.

How to use it

CSV format

• Delimiter is comma only — semicolons and tabs are not read. Excel's Italian/ German/other European locales export semicolon-separated CSVs by default; save as CSV UTF-8 (Comma delimited) instead, or use Google Sheets' comma export.
• Headers are case-insensitive; a space is treated the same as an underscore (Bib Number = bib_number).
• Accepted headers (any one of each group), mapped in renderer/js/participants-manager.js (parseCSV) and mirrored in src/database-service.ts for the create-new path:
  • Race number (required) — Number, number, numero, race_number, bib, Bib
  • Name — Driver, driver, Person, person, Name, name, nome
  • Team — Team, team, squadra
  • Category — Category, category, class, Class, categoria
  • Plate — Plate_Number, plate_number, Plate, plate
  • Other columns read as-is: Car_Model, Nationality, Sponsors/sponsor, Metatag, Folder_1/Folder_2/Folder_3 (+ _Path variants)
  • Any unrecognised column is kept as a custom field, not discarded.
  • If not one row produces a race number after mapping, the import fails immediately with the list of headers it found, instead of silently creating a preset that can never match a photo.
• _Driver_Names (pipe-delimited, e.g. Kaya, Mustafa Mehmet|John Doe) is the unambiguous way to give a driver a name that itself contains a comma. When present it is preferred over splitting the plain Driver/nome cell. When absent, the legacy rule still applies: a comma inside Driver/nome means two drivers ("Max Mustermann, John Doe" → two people). _Driver_IDs, _Driver_Metatags and _Driver_Nationalities are the matching hidden pipe-delimited columns for round-tripping per-driver metadata.
• Export as CSV always writes all of the above columns, including the hidden _Driver_* ones, so an export → re-import round-trip is lossless — comma-names included.

PDF import

• Drop or browse a .pdf file, 20 MB maximum (checked client-side before upload; larger files are rejected with a notification and never sent).
• The renderer base64-encodes the file and calls the supabase-parse-pdf-entry-list IPC handler, which invokes the parsePdfEntryList Supabase Edge Function — the PDF is read visually rather than text-extracted, so column-heavy layouts (numbers detaching from names) and scanned/image-only lists both work. Supports official FIA, FIM and race-organizer entry lists per the modal's own copy.
• If the document isn't recognised as an entry/start list, the modal shows a Document Not Recognized state with the detected type and confidence instead of a generic error.
• The Edge Function returns a structured drivers[] array per participant — each element is one whole driver — so a Lastname, Firstname name (common in German lists) is never mis-split the way a plain comma-joined name would be. The preview table and the actual import both consume this structured array first, falling back to splitting the joined name only when the array is absent.

Limits and known behaviour

• CSV preview shows only the first 5 rows and PDF preview only the first 10 participants — display only; the full file is parsed and imported on confirm.
• CSV delimiter is comma-only; a semicolon-delimited export from a European-locale Excel silently fails to find any recognised header.
• A plain Driver/nome CSV cell always splits on comma into multiple drivers — a single driver whose name contains a comma needs the _Driver_Names column or a PDF import to stay one person.
• PDF import size cap is 20 MB per file; page/row counts on the AI side are not enforced or capped in this codebase (the edge function itself lives outside this repo).
• Merge mode (Update participants / merging CSV into an open preset) never deletes anything until "Save Preset" is clicked; "Deactivate" is the default action for participants missing from the new file, keeping their existing photos and data.