n3world commented on a change in pull request #10790:
URL: https://github.com/apache/arrow/pull/10790#discussion_r700563231
##########
File path: cpp/src/arrow/csv/parser.cc
##########
@@ -324,9 +324,29 @@ class BlockParserImpl {
if (*(end - 1) == '\r') {
--end;
}
- return MismatchingColumns(batch_.num_cols_, num_cols,
- first_row_ < 0 ? -1 : first_row_ +
batch_.num_rows_,
- util::string_view(start, end - start));
+ int32_t batch_row = batch_.num_rows_ + batch_.num_skipped_rows();
+ InvalidRow row{batch_.num_cols_, num_cols,
+ first_row_ < 0 ? -1 : first_row_ + batch_row,
+ util::string_view(start, end - start)};
+
+ if (options_.invalid_row_handler) {
+ if (options_.invalid_row_handler.value()(row) ==
InvalidRowResult::Skip) {
+ values_writer->RollbackLine();
+ parsed_writer->RollbackLine();
+ auto last_skip = batch_.skipped_rows_.rbegin();
+ if (last_skip == batch_.skipped_rows_.rend() ||
+ last_skip->second + 1 != batch_row) {
+ batch_.skipped_rows_.emplace_back(batch_row, batch_row);
+ } else {
+ last_skip->second = batch_row;
+ }
Review comment:
Just trying to reduce the potential overhead of a lot of skipped rows,
but if this is over engineered I can just go to a simple list of skipped rows
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
