Files
sim/apps
Waleed 74d0a47525 fix(trigger): fix Google Sheets trigger header detection and row index tracking (#4109)
* fix(trigger): auto-detect header row and rename lastKnownRowCount to lastIndexChecked

- Replace hardcoded !1:1 header fetch with detectHeaderRow(), which scans
  the first 10 rows and returns the first non-empty row as headers. This
  fixes row: null / headers: [] when a sheet has blank rows or a title row
  above the actual column headers (e.g. headers in row 3).
- Rename lastKnownRowCount → lastIndexChecked in GoogleSheetsWebhookConfig
  and all usage sites to clarify that the value is a row index pointer, not
  a total count.
- Remove config parameter from processRows() since it was unused after the
  includeHeaders flag was removed.

* fix(trigger): combine sheet state fetch, skip header/blank rows from data emission

- Replace separate getDataRowCount() + detectHeaderRow() with a single
  fetchSheetState() call that returns rowCount, headers, and headerRowIndex
  from one A:Z fetch. Saves one Sheets API round-trip per poll cycle when
  new rows are detected.
- Use headerRowIndex to compute adjustedStartRow, preventing the header row
  (and any blank rows above it) from being emitted as data events when
  lastIndexChecked was seeded from an empty sheet.
- Handle the edge case where the entire batch falls within the header/blank
  window by advancing the pointer and returning early without fetching rows.
- Skip empty rows (row.length === 0) in processRows rather than firing a
  workflow run with no meaningful data.

* fix(trigger): preserve lastModifiedTime when remaining rows exist after header skip

When all rows in a batch fall within the header/blank window (adjustedStartRow
> endRow), the early return was unconditionally updating lastModifiedTime to the
current value. If there were additional rows beyond the batch cap, the next
Drive pre-check would see an unchanged modifiedTime and skip polling entirely,
leaving those rows unprocessed. Mirror the hasRemainingOrFailed pattern from the
normal processing path.

* chore(trigger): remove verbose inline comments from google-sheets poller

* fix(trigger): revert to full-width A:Z fetch for correct row count and consistent column scope

* fix(trigger): don't count skipped empty rows as processed
2026-04-11 12:08:15 -07:00
..