# https://taskfile.dev version: '3' output: 'group' vars: DATE: '{{ now | date "20060102_150405"}}' env: REFINE_MEMORY: 8g REFINE_ENDPOINT: http://localhost:3334 tasks: default: desc: Generierung PICA+ deps: [bibliotheca, alephino] cmds: - tasks/03-ba-sachsen.sh "output/02-bibliotheca-main" sources: - tasks/03-ba-sachsen.sh # - output/02-alephino-main/alephino.csv - output/02-bibliotheca-main/bibliotheca.csv generates: - output/03-ba-sachsen/ba-sachsen.pic - output/03-ba-sachsen/ba-sachsen.openrefine.tar.gz env: REFINE_WORKDIR: output/03-ba-sachsen REFINE_LOGFILE: log/03-ba-sachsen/{{.DATE}}.log alephino: desc: Alephino Hauptverarbeitung deps: [leipzig, riesa] cmds: - tasks/02-alephino-main.sh "output/01-alephino-pre" sources: - tasks/02-alephino-main.sh - output/01-alephino-pre/*.tsv generates: # - output/02-alephino-main/alephino.csv - output/02-alephino-main/alephino.openrefine.tar.gz env: REFINE_ENDPOINT: http://localhost:3334 REFINE_WORKDIR: output/02-alephino-main REFINE_LOGFILE: log/02-alephino-main/{{.DATE}}.log bibliotheca: desc: Bibliotheca Hauptverarbeitung # deps: [bautzen, breitenbrunn, dresden, glauchau, plauen] cmds: - task: bautzen - task: breitenbrunn - task: dresden - task: glauchau # - task: plauen - tasks/02-bibliotheca-main.sh "output/01-bibliotheca-pre" sources: - tasks/01-bibliotheca-pre.sh - tasks/02-bibliotheca-main.sh - output/01-bibliotheca-pre/*.tsv generates: - output/02-bibliotheca-main/bibliotheca.csv - output/02-bibliotheca-main/bibliotheca.openrefine.tar.gz env: REFINE_ENDPOINT: http://localhost:3335 REFINE_WORKDIR: output/02-bibliotheca-main REFINE_LOGFILE: log/02-bibliotheca-main/{{.DATE}}.log bautzen: desc: Bibliotheca Vorverarbeitung cmds: - tasks/01-bibliotheca-pre.sh "{{.INPUT}}" sources: - tasks/01-bibliotheca-pre.sh - '{{.INPUT}}' generates: - output/01-bibliotheca-pre/bautzen.tsv vars: INPUT: '{{.INPUT | default "input/bautzen.imp"}}' env: REFINE_MEMORY: '{{.REFINE_MEMORY | default "6G"}}' REFINE_ENDPOINT: http://localhost:3334 REFINE_WORKDIR: output/01-bibliotheca-pre REFINE_LOGFILE: log/01-bibliotheca-pre/{{.DATE}}_bautzen.log breitenbrunn: desc: Bibliotheca Vorverarbeitung cmds: - tasks/01-bibliotheca-pre.sh "{{.INPUT}}" sources: - tasks/01-bibliotheca-pre.sh - '{{.INPUT}}' generates: - output/01-bibliotheca-pre/breitenbrunn.tsv vars: INPUT: '{{.INPUT | default "input/breitenbrunn.imp"}}' env: REFINE_MEMORY: '{{.REFINE_MEMORY | default "4G"}}' REFINE_ENDPOINT: http://localhost:3335 REFINE_WORKDIR: output/01-bibliotheca-pre REFINE_LOGFILE: log/01-bibliotheca-pre/{{.DATE}}_breitenbrunn.log dresden: desc: Bibliotheca Vorverarbeitung cmds: - tasks/01-bibliotheca-pre.sh "{{.INPUT}}" sources: - tasks/01-bibliotheca-pre.sh - '{{.INPUT}}' generates: - output/01-bibliotheca-pre/dresden.tsv vars: INPUT: '{{.INPUT | default "input/dresden.imp"}}' env: REFINE_MEMORY: '{{.REFINE_MEMORY | default "7G"}}' REFINE_ENDPOINT: http://localhost:3336 REFINE_WORKDIR: output/01-bibliotheca-pre REFINE_LOGFILE: log/01-bibliotheca-pre/{{.DATE}}_dresden.log leipzig: desc: Alephino Vorverarbeitung cmds: - tasks/01-alephino-pre.sh "{{.TITEL}}" "{{.EXEMPLARE}}" sources: - tasks/01-alephino-pre.sh - '{{.TITEL}}' - '{{.EXEMPLARE}}' generates: - output/01-alephino-pre/leipzig.tsv vars: TITEL: '{{.TITEL | default "input/leipzig-titel.txt"}}' EXEMPLARE: '{{.EXEMPLARE | default "input/leipzig-exemplare.txt"}}' env: REFINE_MEMORY: '{{.REFINE_MEMORY | default "7G"}}' REFINE_ENDPOINT: http://localhost:3337 REFINE_WORKDIR: output/01-alephino-pre REFINE_LOGFILE: log/01-alephino-pre/{{.DATE}}_leipzig.log glauchau: desc: Bibliotheca Vorverarbeitung cmds: - tasks/01-bibliotheca-pre.sh "{{.INPUT}}" sources: - tasks/01-bibliotheca-pre.sh - '{{.INPUT}}' generates: - output/01-bibliotheca-pre/glauchau.tsv vars: INPUT: '{{.INPUT | default "input/glauchau.imp"}}' env: REFINE_MEMORY: '{{.REFINE_MEMORY | default "4G"}}' REFINE_ENDPOINT: http://localhost:3338 REFINE_WORKDIR: output/01-bibliotheca-pre REFINE_LOGFILE: log/01-bibliotheca-pre/{{.DATE}}_glauchau.log plauen: desc: Bibliotheca Vorverarbeitung cmds: - tasks/01-bibliotheca-pre.sh "input/plauen.imp" sources: - tasks/01-bibliotheca-pre.sh - input/plauen.imp generates: - output/01-bibliotheca-pre/plauen.tsv env: REFINE_MEMORY: '{{.REFINE_MEMORY | default "2G"}}' REFINE_ENDPOINT: http://localhost:3339 REFINE_WORKDIR: output/01-bibliotheca-pre REFINE_LOGFILE: log/01-bibliotheca-pre/{{.DATE}}_plauen.log riesa: desc: Alephino Vorverarbeitung cmds: - tasks/01-alephino-pre.sh "{{.TITEL}}" "{{.EXEMPLARE}}" sources: - tasks/01-alephino-pre.sh - '{{.TITEL}}' - '{{.EXEMPLARE}}' generates: - output/01-alephino-pre/riesa.tsv vars: TITEL: '{{.TITEL | default "input/riesa-titel.txt"}}' EXEMPLARE: '{{.EXEMPLARE | default "input/riesa-exemplare.txt"}}' env: REFINE_MEMORY: '{{.REFINE_MEMORY | default "7G"}}' REFINE_ENDPOINT: http://localhost:3340 REFINE_WORKDIR: output/01-alephino-pre REFINE_LOGFILE: log/01-alephino-pre/{{.DATE}}_riesa.log clean: desc: Alle Daten löschen (reset auf Ausgangszustand) cmds: - rm -r lib log output mkdir: desc: Ordner erstellen cmds: - mkdir -p output/01-alephino-pre log/01-alephino-pre - mkdir -p output/01-bibliotheca-pre log/01-bibliotheca-pre - mkdir -p output/02-alephino-main log/02-alephino-main - mkdir -p output/02-bibliotheca-main log/02-bibliotheca-main - mkdir -p output/03-ba-sachsen log/03-ba-sachsen