2020-07-13 12:42:14 +02:00
|
|
|
#!/bin/bash
|
|
|
|
# Scripte zur Transformation von Bibliotheca und Alephino nach PICA+
|
|
|
|
|
|
|
|
# ================================ ENVIRONMENT =============================== #
|
|
|
|
|
|
|
|
# make script executable from another directory
|
|
|
|
cd "${BASH_SOURCE%/*}/" || exit 1
|
|
|
|
|
|
|
|
# source the main script
|
|
|
|
source bash-refine.sh
|
|
|
|
|
|
|
|
# override default config
|
|
|
|
memory="8G"
|
2020-07-22 14:23:23 +02:00
|
|
|
endpoint="http://localhost:3334"
|
2020-07-13 12:42:14 +02:00
|
|
|
|
|
|
|
# check requirements, set trap, create workspace and tee to logfile
|
|
|
|
init
|
|
|
|
|
|
|
|
# ================================= WORKFLOW ================================= #
|
|
|
|
|
|
|
|
checkpoint "Bibliotheca Vorverarbeitung"; echo
|
|
|
|
source config/bibliotheca-01.sh
|
|
|
|
|
|
|
|
checkpoint "Bibliotheca Hauptverarbeitung"; echo
|
|
|
|
source config/bibliotheca-02.sh
|
|
|
|
|
2020-07-22 14:23:23 +02:00
|
|
|
checkpoint "PICA+ generieren"; echo
|
|
|
|
source config/ba-sachsen.sh
|
|
|
|
|
2020-07-13 12:42:14 +02:00
|
|
|
# ================================= STATS ================================= #
|
|
|
|
|
|
|
|
# calculate run time based on checkpoints
|
|
|
|
checkpoint_stats; echo
|
|
|
|
|
|
|
|
# word count on all files in workspace
|
|
|
|
count_output
|