Split CSV files by megabyte and retain head row (row 1)

csv-split-by-lines.sh

#!/bin/bash
if [ ! -d "./backup" ]; then
	mkdir backup
fi

lines=2000
for i in *.csv; do
	split -b ${1:-$lines} $i ${i%.csv}-
	for j in ${i%.csv}-a*; do
		if [[ "$j" != *-aa ]]
			then
				echo $(head -n 1 $i)|cat - $j > /tmp/out && mv /tmp/out $j
		fi
		mv $j $j.csv
	done
	mv $i backup/$i
done

csv-split.sh

#!/bin/bash
if [ ! -d "./backup" ]; then
	mkdir backup
fi

megabyte=20
for i in *.csv; do
	split -b ${1:-$megabyte}m $i ${i%.csv}-
	for j in ${i%.csv}-a*; do
		if [[ "$j" != *-aa ]]
			then
				echo $(head -n 1 $i)|cat - $j > /tmp/out && mv /tmp/out $j
		fi
		mv $j $j.csv
	done
	mv $i backup/$i
done

Caveat: if your CSV contains newline characters within the actual records the point where you split the file may not split cleanly between two records but rather in the middle of one.

If you're just dealing with a few splitted files I found it easiest to just open each up and copy paste the correct data over from each other and delete the faulty rows.