noah/Taskfile.yml

# https://github.com/opencultureconsulting/openrefine-task-runner

version: '3'

includes:
  muenster: muenster
  siegen: siegen
  wuppertal: wuppertal

silent: true
output: prefixed

vars:
  DATE: '{{ now | date "2006-01-02"}}'

env:
  OPENREFINE:
    sh: readlink -m .openrefine/refine
  CLIENT:
    sh: readlink -m .openrefine/client

tasks:
  default:
    desc: execute all projects in parallel
    deps:
      - task: muenster:main
      - task: siegen:main
      - task: wuppertal:main

  install:
    desc: (re)install OpenRefine and openrefine-client into subdirectory .openrefine
    cmds:
      - | # delete existing install and recreate folder
        rm -rf .openrefine
        mkdir -p .openrefine
      - > # download OpenRefine archive
        wget --no-verbose -O openrefine.tar.gz
        https://github.com/OpenRefine/OpenRefine/releases/download/3.4.1/openrefine-linux-3.4.1.tar.gz
      - | # install OpenRefine into subdirectory .openrefine
        tar -xzf openrefine.tar.gz -C .openrefine --strip 1
        rm openrefine.tar.gz
      - | # optimize OpenRefine for batch processing
        sed -i 's/cd `dirname $0`/cd "$(dirname "$0")"/' ".openrefine/refine" # fix path issue in OpenRefine startup file
        sed -i '$ a JAVA_OPTIONS=-Drefine.headless=true' ".openrefine/refine.ini" # do not try to open OpenRefine in browser
        sed -i 's/#REFINE_AUTOSAVE_PERIOD=60/REFINE_AUTOSAVE_PERIOD=1440/' ".openrefine/refine.ini" # set autosave period from 5 minutes to 25 hours
      - > # download openrefine-client into subdirectory .openrefine
        wget --no-verbose -O .openrefine/client
        https://github.com/opencultureconsulting/openrefine-client/releases/download/v0.3.10/openrefine-client_0-3-10_linux
      - chmod +x .openrefine/client # make client executable

  start:
    dir: ./{{.PROJECT}}/refine
    cmds:
      - | # verify that OpenRefine is installed
        if [ ! -f "$OPENREFINE" ]; then
          echo 1>&2 "OpenRefine missing; try task install"; exit 1
        fi
      - | # delete temporary files and log file of previous run
        rm -rf ./*.project* workspace.json
        rm -rf "{{.PROJECT}}.log"
      - > # launch OpenRefine with specific data directory and redirect its output to a log file
        "$OPENREFINE" -v warn -p {{.PORT}} -m {{.RAM}}
        -d ../{{.PROJECT}}/refine
        >> "{{.PROJECT}}.log" 2>&1 &
      - | # wait until OpenRefine API is available
        timeout 30s bash -c "until
          wget -q -O - http://localhost:{{.PORT}} | cat | grep -q -o OpenRefine
          do sleep 1
        done"

  stop:
    dir: ./{{.PROJECT}}/refine
    cmds:
      - | # shut down OpenRefine gracefully
        PID=$(lsof -t -i:{{.PORT}})
        kill $PID
        while ps -p $PID > /dev/null; do sleep 1; done
      - > # archive the OpenRefine project
        tar cfz
        "{{.PROJECT}}.openrefine.tar.gz"
        -C $(grep -l "{{.PROJECT}}" *.project/metadata.json | cut -d '/' -f 1)
        .
      - rm -rf ./*.project* workspace.json # delete temporary files

  kill:
    dir: ./{{.PROJECT}}/refine
    cmds:
      - | # shut down OpenRefine immediately to save time and disk space
        PID=$(lsof -t -i:{{.PORT}})
        kill -9 $PID
        while ps -p $PID > /dev/null; do sleep 1; done
      - rm -rf ./*.project* workspace.json # delete temporary files

  check:
    dir: ./{{.PROJECT}}/refine
    cmds:
      - | # find log file(s) and check for "exception" or "error"
        if grep -i 'exception\|error' $(find . -name '*.log'); then
          echo 1>&2 "log contains warnings!"; exit 1
        fi
      - | # Prüfen, ob Mindestanzahl von Datensätzen generiert wurde
        if (( {{.MINIMUM}} > $(grep -c recordIdentifier {{.PROJECT}}.txt) )); then
           echo 1>&2 "Unerwartet geringe Anzahl an Datensätzen in $PWD/{{.PROJECT}}.txt!"; exit 1
        fi

  split:
    label: '{{.TASK}}-{{.PROJECT}}'
    dir: ./{{.PROJECT}}/split
    cmds:
      - test -n "{{.PROJECT}}"
      # in Einzeldateien aufteilen
      - csplit -s -z ../refine/{{.PROJECT}}.txt '/<mets:mets /' "{*}"
      # ggf. vorhandene XML-Dateien löschen
      - rm -f *.xml
      # Identifier als Dateinamen
      - for f in xx*; do mv "$f" "$(xmllint --xpath "//*[local-name(.) = 'recordIdentifier']/text()" "$f").xml"; done
    sources:
      - ../refine/{{.PROJECT}}.txt
    generates:
      - ./*.xml

  validate:
    label: '{{.TASK}}-{{.PROJECT}}'
    dir: ./{{.PROJECT}}/validate
    cmds:
      - test -n "{{.PROJECT}}"
      # Validierung gegen METS Schema
      - wget -q -nc https://www.loc.gov/standards/mets/mets.xsd
      - xmllint --schema mets.xsd --noout ../split/*.xml > validate.log 2>&1
    sources:
      - ../split/*.xml
    generates:
      - validate.log

  zip:
    label: '{{.TASK}}-{{.PROJECT}}'
    dir: ./{{.PROJECT}}/zip
    cmds:
      - test -n "{{.PROJECT}}"
      # ZIP-Archiv mit Zeitstempel erstellen
      - zip -q -FS -j {{.PROJECT}}_{{.DATE}}.zip ../split/*.xml
    sources:
      - ../split/*.xml
    generates:
      - '{{.PROJECT}}_{{.DATE}}.zip'

  diff:
    label: '{{.TASK}}-{{.PROJECT}}'
    dir: ./{{.PROJECT}}
    cmds:
      - test -n "{{.PROJECT}}"
      # Inhalt der beiden letzten ZIP-Archive vergleichen
      - if test -n "$(ls -t zip/*.zip | sed -n 2p)"; then unzip -q -d old $(ls -t zip/*.zip | sed -n 2p); unzip -q -d new $(ls -t zip/*.zip | sed -n 1p); fi
      - diff -d old new > diff.log || exit 0
      - rm -rf old new
      # Diff prüfen, ob es weniger als 500 Zeilen enthält
      - if (( 500 < $(wc -l <diff.log) )); then echo 1>&2 "Unerwartet große Änderungen in $PWD/diff.log!" && exit 1; fi
      # Diff archivieren
      - cp diff.log zip/{{.PROJECT}}_{{.DATE}}.diff
    sources:
      - split/*.xml
    generates:
      - diff.log

  linkcheck:
    label: '{{.TASK}}-{{.PROJECT}}'
    dir: ./{{.PROJECT}}
    cmds:
      - test -n "{{.PROJECT}}"
      # Links extrahieren
      - grep -o 'href="[^"]*"' split/*.xml | sed 's/:href=/\t/' | tr -d '"' | sort -k 2 --unique  > links.txt
      # http status code ermitteln
      - awk '{ print "url = " $2 "\noutput = /dev/null"; }' links.txt > curl.cfg
      - curl --silent --head --location --write-out "%{http_code}\t%{url_effective}\n" --config curl.cfg > curl.log
      # Tabelle mit status code, effektiver URL, Dateiname und start URL erstellen
      - paste curl.log links.txt > linkcheck.log
      - rm -rf curl.cfg curl.log links.txt
      # Logdatei auf status code != 2XX prüfen
      - if grep '^[^2]' linkcheck.log; then echo 1>&2 "Logdatei $PWD/linkcheck.log enthält problematische status codes!" && exit 1; fi
    sources:
      - split/*.xml
    generates:
      - linkcheck.log

  delete:
    label: '{{.TASK}}-{{.PROJECT}}'
    dir: ./{{.PROJECT}}
    cmds:
      - test -n "{{.PROJECT}}"
      - rm -rf harvest
      - rm -rf refine
      - rm -rf split
      - rm -rf validate
      - rm -f diff.log