|
- Replace the CSV parsing function (arrayForCSV:) with a new SPCSVParser class
- Make speed improvements to SPCSVParser to achieve 1.9x faster parsing than the old arrayForCSV: function
- Rewrite CSV imports to be performed as a streaming import, keeping memory usage much much lower
- CSV field mapping preview is now shown very early on in the import process, as soon as the first hundred rwos are available for a preview
- Progress bars are more consistent and accurate
- CSV rows are grouped into batches of up to 50 (depending on line length) for import, falling back to one-query-per-row if errors occur. The current error reporting level is therefore maintained, but imports of non-erroring data are much much faster.
- Improve processing speed slightly
- Fix some odd edge cases in CSV parsing
This addresses issue #389.
|