With black magic I’d imagine this workflow. Full page IDs (canvas-xxxxxxxxxx) are available through Coda API. So Zapier/Integromat/API etc can first insert a canvas ID into a cell, then we could have a reconstructed button that would render that particular canvas into HTML into a cell (with API we can click buttons), and then we could grab the text at that button.
There may be race conditions though, i.e. the doc wouldn’t recalculate with the new canvas ID and the button click wouldn’t render the proper page. You could potentially avoid this by:
- have the API call insert a row into some “export requests” table
- have Coda automation watch on new rows, and insert HTML into a column of that newly created row
- have another zap / polling API watch over that freshly rendered HTMLs and react independently on each row.