Mercurial > hg > cc > cirrus_home

annotate bin/doExtract.sh @ 27:7a2bc060230d

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
first cut at doing extraction here
author Henry S. Thompson <ht@inf.ed.ac.uk>
date Wed, 18 Mar 2020 15:23:53 +0000
parents
children a95a1b31f5e9
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
7a2bc060230d first cut at doing extraction here
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
2 # Usage extract.sh ccid jobid segid
7a2bc060230d first cut at doing extraction here
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
3 ccid=$1
7a2bc060230d first cut at doing extraction here
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
4 jobid=$2
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 $segid
7a2bc060230d first cut at doing extraction here
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
9 ls /beegfs/common-crawl/CC-MAIN-${ccid}/${segid}/CC-MAIN-*.warc.gz | \
7a2bc060230d first cut at doing extraction here
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
10 parallel -j 12 -N 1 '{
7a2bc060230d first cut at doing extraction here
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
11 id=$(echo {} | cut -f 6 -d / | cut -f 3- -d - | cut -f 1 -d .)
7a2bc060230d first cut at doing extraction here
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
12 echo starting $id $(date) >> logs/${jobid}_{#}_log
7a2bc060230d first cut at doing extraction here
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
13 unpigz -dp 1 -c {} | $HOME/lib/valhalla/bin/warc.sh ${id} application/pdf 2>> logs/${jobid}_{#}_log
7a2bc060230d first cut at doing extraction here
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
14 echo finished ${id} $(date) >> logs/${jobid}_{#}_log
7a2bc060230d first cut at doing extraction here
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
15 }'
7a2bc060230d first cut at doing extraction here
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
16 echo $(date) $(hostname) $jobid $segid $?
7a2bc060230d first cut at doing extraction here
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
17 tar -cf /beegfs/common-crawl/CC-MAIN-${ccid}/${segid}/extracts.tar *
7a2bc060230d first cut at doing extraction here
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
18 echo $(date) $(hostname) $jobid /beegfs/common-crawl/CC-MAIN-${ccid}/${segid}/extracts.tar