14
|
1 #!/usr/bin/bash
|
|
2 hn=$1
|
|
3 jn=$2
|
|
4 tfn=$3
|
|
5
|
|
6 mkdir -p /dev/shm/x$hn/${tfn}
|
|
7 cd /dev/shm/x$hn/${tfn}
|
|
8 tar -xf /beegfs/common_crawl/CC-MAIN-2019-35/pdfs/${tfn}.tar '*.hdr'
|
|
9 echo $(date) $hn $(pwd) untarred ${tfn}.tar for job $jn 1>&2
|
|
10 ls *.hdr | sort --field-separator=_ -k1,1 -k2n,2 | cat -n |\
|
|
11 while read n f; do echo "$n\t"$(egrep -a '^X-HST-Target-URI: ' $f|cut -f 2 -d ' ')"\t"$(egrep -ia '^Last-Modified: ' $f|cut -f 2 -d ' '); done >> hdrs_$tfn.tsv
|
|
12 echo $(date) $hn moving $(ls hdrs_*.tsv|wc -l) results from job $jn for $tfn in $(pwd) 1>&2
|
|
13 mv hdrs_$tfn.tsv /beegfs/common_crawl/CC-MAIN-2019-35/pdfs/hdrs
|
|
14 echo $(date) $(pwd) rm $(ls -lt hdrs_*.tsv) 1>&2
|
|
15 rm *
|
|
16 cd ..
|
|
17 echo $(date) $(pwd) rmdir ${tfn} 1>&2
|
|
18 rmdir ${tfn}
|
|
19 echo $(date) $hn finished job ${jn} for ${tfn}
|