Mercurial > hg > cc > cirrus_home

annotate bin/doExtract.sh @ 33:4c23b1766692

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
towards sub-division of resulting tar files
author Henry S. Thompson <ht@inf.ed.ac.uk>
date Tue, 24 Mar 2020 17:53:35 +0000
parents 1e479e3ea9c4
children ec99b2d1d2fc
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
27
7a2bc060230d first cut at doing extraction here
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
1 #!/bin/bash
29
a95a1b31f5e9 try to fix multi-line lossage
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 27
diff changeset
2 # Usage doExtract.sh ccid jobid segid
27
7a2bc060230d first cut at doing extraction here
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
3 ccid=$1
29
a95a1b31f5e9 try to fix multi-line lossage
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 27
diff changeset
4 export jobid=$2
27
7a2bc060230d first cut at doing extraction here
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
5 segid=$3
7a2bc060230d first cut at doing extraction here
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
6 echo $(date) $(hostname) $jobid $segid
7a2bc060230d first cut at doing extraction here
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
7 mkdir -p $segid/logs
7a2bc060230d first cut at doing extraction here
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
8 cd $segid
29
a95a1b31f5e9 try to fix multi-line lossage
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 27
diff changeset
9 ls /beegfs/common_crawl/CC-MAIN-${ccid}/${segid}/CC-MAIN-*.warc.gz | \
33
4c23b1766692 towards sub-division of resulting tar files
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 30
diff changeset
10 parallel --joblog job_${jobid}.log -j 18 -N 1 'id=$(echo {} | cut -f 6 -d / | cut -f 3- -d - | cut -f 1 -d .) ; \
30
1e479e3ea9c4 better quoting
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 29
diff changeset
11 echo starting $id $(date) >> logs/${jobid}_{#}_log ; \
1e479e3ea9c4 better quoting
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 29
diff changeset
12 unpigz -dp 1 -c {} | $HOME/lib/valhalla/bin/warc.sh ${id} application/pdf 2>> logs/${jobid}_{#}_log ; \
1e479e3ea9c4 better quoting
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 29
diff changeset
13 echo finished ${id} $(date) >> logs/${jobid}_{#}_log'
29
a95a1b31f5e9 try to fix multi-line lossage
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 27
diff changeset
14 res=$?
a95a1b31f5e9 try to fix multi-line lossage
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 27
diff changeset
15 echo $(date) $(hostname) $jobid $segid $res
a95a1b31f5e9 try to fix multi-line lossage
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 27
diff changeset
16 if [ $res = 0 ]
a95a1b31f5e9 try to fix multi-line lossage
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 27
diff changeset
17 then
33
4c23b1766692 towards sub-division of resulting tar files
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 30
diff changeset
18 cat ../by11s.txt | while read
4c23b1766692 towards sub-division of resulting tar files
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 30
diff changeset
19 tar -cf /beegfs/common_crawl/CC-MAIN-${ccid}/${segid}/extracts.tar * &&
29
a95a1b31f5e9 try to fix multi-line lossage
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 27
diff changeset
20 echo $(date) $(hostname) $jobid /beegfs/common_crawl/CC-MAIN-${ccid}/${segid}/extracts.tar
a95a1b31f5e9 try to fix multi-line lossage
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 27
diff changeset
21 fi