[sitka/sitka-tools.git] / marc_export_custom / stats.sh

#!/bin/bash
basedir=$(readlink -f $(dirname $0))
files=$(ls *.marc.bz2 *.marc 2>/dev/null |grep -v split |sed 's,\.bz2,,g' | sort | uniq)
for i in $files; do 
	s=${i}.stats
	j=${i}.bz2
	[ -f $s ] && continue
	echo === $i
	if [ -f $j -a ! -f $i ]; then
		bunzip2 -k $j
	fi
	if [ -f $i -a ! -f $j ]; then
		pbzip2 -9k -p4 $i
	fi
	if [ ! -f $i -o ! -f $j ]; then
		echo "Missing $i or $j"
		continue
	fi
	echo -n '' >$s
	$basedir/stats-distribution.pl $i | tee $s-holding-distribution.csv
	ls -la $i $j | tee -a $s
	sha1sum $i $j | tee -a $s
	marcdump --stats --quiet --noprint $i | tee -a ${s}
	[ "$DELETE" != "0" ] && rm -f $i
done
Commit	Line	Data
8ec81273	1	#!/bin/bash
e38e56a8	2	basedir=$(readlink -f $(dirname $0))
7716c3a7	3	files=$(ls .marc.bz2 .marc 2>/dev/null \|grep -v split \|sed 's,\.bz2,,g' \| sort \| uniq)
8ec81273	4	for i in $files; do
c91db487 RJ	5	s=${i}.stats
c91db487 RJ	6	j=${i}.bz2
8ec81273 RJ	7	[ -f $s ] && continue
	8	echo === $i
	9	if [ -f $j -a ! -f $i ]; then
	10	bunzip2 -k $j
	11	fi
	12	if [ -f $i -a ! -f $j ]; then
2ec1922a	13	pbzip2 -9k -p4 $i
8ec81273 RJ	14	fi
	15	if [ ! -f $i -o ! -f $j ]; then
	16	echo "Missing $i or $j"
	17	continue
	18	fi
7716c3a7 RJ	19	echo -n '' >$s
	20	$basedir/stats-distribution.pl $i \| tee $s-holding-distribution.csv
	21	ls -la $i $j \| tee -a $s
8ec81273 RJ	22	sha1sum $i $j \| tee -a $s
8ec81273 RJ	23	marcdump --stats --quiet --noprint $i \| tee -a ${s}
c91db487	24	[ "$DELETE" != "0" ] && rm -f $i
8ec81273	25	done