Taskfile.yml

version: '3'

silent: true

tasks:
  default:
    desc: harvest and transform
    cmds:
      - task: harvest
      - task: transform

  reset:
    desc: delete cache; harvest and transform
    cmds:
      - rm -rf input/*
      - task: harvest
      - rm -rf output/*
      - task: transform

  install:
    desc: install requirements into subdirectories
    cmds:
      - task: install_openrefine
      - task: install_vufindharvest

  install_openrefine:
    cmds:
      - | # install OpenRefine into subdirectory .openrefine
        mkdir -p .openrefine
        wget --no-verbose -O openrefine.tar.gz https://github.com/OpenRefine/OpenRefine/releases/download/3.5.2/openrefine-linux-3.5.2.tar.gz
        tar -xzf openrefine.tar.gz -C .openrefine --strip 1 && rm openrefine.tar.gz
      - | # fix path issue in OpenRefine startup file
        sed -i 's/cd `dirname $0`/cd "$(dirname "$0")"/' ".openrefine/refine"
      - | # do not try to open OpenRefine in browser
        sed -i '$ a JAVA_OPTIONS=-Drefine.headless=true' ".openrefine/refine.ini"
      - | # set autosave period from 5 minutes to 25 hours
        sed -i 's/#REFINE_AUTOSAVE_PERIOD=60/REFINE_AUTOSAVE_PERIOD=1440/' ".openrefine/refine.ini"
      - | # install 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

  install_vufindharvest:
    cmds:
      - | # requirement php
        if [ -z "$(which php 2> /dev/null)" ]; then
          echo 1>&2 "This task needs PHP 7.3+"; exit 1
        fi
      - | # requirement composer
        if [ -z "$(which composer 2> /dev/null)" ]; then
          echo 1>&2 "This task needs Composer https://getcomposer.org"; exit 1
        fi
      - | # install VuFindHarvest into subdirectory .vufindharvest
        mkdir -p .vufindharvest
        wget --no-verbose -O vufindharvest.zip https://github.com/vufind-org/vufindharvest/archive/v4.1.0.zip
        unzip vufindharvest.zip && mv vufindharvest-*/* .vufindharvest && rm -r vufindharvest*
        (cd .vufindharvest && composer install)

  harvest:
    desc: harvest all records from OAI-PMH
    cmds:
      - | # requirement VuFindHarvest
        if [ ! -f .vufindharvest/bin/harvest_oai.php ]; then
          echo 1>&2 "VuFindHarvest missing; try task install_vufindharvest"; exit 1
        fi
      - | # prepare workspace
        mkdir -p input
      - | # harvest all journals
        (cd input && php ../.vufindharvest/bin/harvest_oai.php --ini ../config/harvest.ini)
      - | # rename harvested files
        (cd input &&
        find . -name '*_*.xml' |
        while IFS= read -r path; do
            f="${path##*/}"
            d="${path%/*}"
            n="${f%.*}"
            mv -f "$path" "${d}"/"${n:11}.xml"
        done)
      - | # handle deleted records
        (cd input &&
        find . -name '*.delete' |
        while IFS= read -r path; do
            f="${path##*/}"
            d="${path%/*}"
            n="${f%.*}"
            rm -f "${d}"/"${n:11}.xml"
            rm -f ../output/"${d}"/"${n:11}.xml"
            rm "$path"
        done)
      - | # check results
        minimum="2400"
        count="$(find input -name '*.xml' | wc -l)"
        if [ "$minimum" -le "$count" ]; then
          echo "Download enthält $count Datensätze"
        else
          echo 1>&2 "Anzahl der heruntergeladenen Datensätze (${count}) entspricht nicht der erwarteten Mindestanzahl (${minimum})! Bitte manuell prüfen."; exit 1
        fi

  transform:
    desc: transform data to mets/mods xml
    cmds:
      - task: transform_start
      - defer: { task: transform_stop } # will run even when one of the following commands fail
      - task: transform_client
      - task: transform_split
      - task: transform_validate
    sources:
      - input/*/*.xml
      - config/**
    generates:
      - output/**

  transform_start:
    cmds:
      - | # requirement OpenRefine
        if [ ! -f .openrefine/refine ]; then
          echo 1>&2 "OpenRefine missing; try task install_openrefine"; exit 1
        fi
      - | # delete temporary files of previous run
        rm -rf .openrefine/data .openrefine/log.txt
      - | # launch OpenRefine with specific data directory and redirect its output to a log file
        .openrefine/refine -v warn -p 3333 -m 5120M -d data >> .openrefine/log.txt 2>&1 &
      - | # wait until OpenRefine API is available
        timeout 30s bash -c "until wget -q -O - -o /dev/null http://localhost:3333 | cat | grep -q -o OpenRefine; do sleep 1; done"

  transform_client:
    cmds:
      - | # prepare workspace
        mkdir -p output
      - | # create temporary zip archive with new files (modified less than 1 day ago)
        zip -r -q input/tmp.zip config/schema.xml -@ <<< $(find input -name '*.xml' -mtime -1)
      - | # import (requires absolute path)
        .openrefine/client \
        --create "$(readlink -m input/tmp.zip)" \
        --format xml \
        --recordPath oai_dc:dc \
        --storeEmptyStrings false --trimStrings true \
        --projectName rub-journals \
        > >(tee -a .openrefine/log.txt) 2>&1
      - | # delete temporary zip archive
        rm input/tmp.zip
      - | # apply transformation rules
        for f in config/*.json; do
          .openrefine/client rub-journals --apply "$f" > >(tee -a .openrefine/log.txt) 2>&1
        done
      - | # templating export to METS:MODS
        .openrefine/client rub-journals \
        --template "$(< config/template.txt)" \
        --rowSeparator '' \
        --output "$(readlink -m output/rub-journals.txt)" \
        > >(tee -a .openrefine/log.txt) 2>&1

  transform_stop:
    cmds:
      - | # kill OpenRefine gracefully and print stats
        PID="$(lsof -t -i:3333)"
        echo "used $(($(ps --no-headers -o rss -p "$PID") / 1024)) MB RAM" > >(tee -a .openrefine/log.txt) 2>&1
        echo "used $(ps --no-headers -o cputime -p "$PID") CPU time" > >(tee -a .openrefine/log.txt) 2>&1
        kill $PID; while ps -p $PID > /dev/null; do sleep 1; done
      - | # archive OpenRefine projects
        for p in .openrefine/data/*.project/
            do tar -czf .openrefine/data/"${p:17:13}.openrefine.tar.gz" -C $p .
        done
      - | # check log file
        if grep -i 'exception\|error' .openrefine/log.txt
          then echo 1>&2 "log contains warnings!"; echo; cat .openrefine/log.txt; exit 1
        fi

  transform_split:
    cmds:
      - | # split into one file per record
        (cd output &&
        csplit -s -z rub-journals.txt '/<mets:mets /' "{*}" &&
        rm rub-journals.txt)
      - | # rename files
        (cd output &&
        for f in xx*
            do mv "$f" "$(xmllint --xpath "//*[local-name(.) = 'recordIdentifier']/text()" "$f").xml"
        done)

  transform_validate:
    desc: check output and validate against METS schema
    cmds:
      - | # requirement xmllint
        if [ -z "$(which xmllint 2> /dev/null)" ]; then
          echo 1>&2 "This task needs xmllint"; exit 1
        fi
      - | # check results
        minimum="2400"
        count="$(find output -name '*.xml' | wc -l)"
        if [ "$minimum" -le "$count" ]; then
          echo "Es wurden $count Datensätze generiert"
        else
          echo 1>&2 "Anzahl der generierten Datensätze (${count}) entspricht nicht der erwarteten Mindestanzahl (${minimum})! Bitte manuell prüfen."
        fi
      - | # validate against METS schema
        xmllint --schema config/mets.xsd --noout output/*.xml

  git:
    desc: commit and push if something changed
    cmds:
      - git add -A
      - git commit -m "latest change $(date -u)" || exit 0
      - git push