ocrd-galley/build

34 lines
2.2 KiB
Text
Raw Normal View History

#!/bin/bash
set -e
2019-10-31 15:22:12 +01:00
self=`realpath $0`
self_dir=`dirname "$self"`
2019-10-31 15:22:12 +01:00
2019-10-30 17:54:05 +01:00
DATA_SUBDIR=data
2019-10-31 15:22:12 +01:00
get_from_annex() {
2020-08-05 12:27:05 +02:00
annex_get 'calamari-models/GT4HistOCR/2019-07-22T15_49+0200/*.ckpt*'
2019-10-31 15:22:12 +01:00
annex_get 'tesseract-models/GT4HistOCR/*.traineddata'
annex_get 'textline_detection/*.h5'
annex_get 'mirror/github.com/tesseract-ocr/tessdata_best/archive/4.0.0-repacked.tar.gz'
annex_get 'sbb_binarization/*.h5'
2019-10-31 15:22:12 +01:00
}
get_from_web() {
2020-08-05 12:27:05 +02:00
download_to 'https://qurator-data.de/calamari-models/GT4HistOCR/model.tar.xz' 'calamari-models/GT4HistOCR/2019-07-22T15_49+0200'
download_to 'https://qurator-data.de/tesseract-models/GT4HistOCR/models.tar' 'tesseract-models/GT4HistOCR'
download_to 'https://qurator-data.de/sbb_textline_detector/models.tar.gz' 'textline_detection'
download_to --strip-components 1 'https://qurator-data.de/sbb_binarization/models.tar.gz' 'sbb_binarization'
download_to --no-unpack 'https://qurator-data.de/mirror/github.com/tesseract-ocr/tessdata_best/archive/4.0.0-repacked.tar.gz' 'mirror/github.com/tesseract-ocr/tessdata_best/archive/4.0.0-repacked.tar.gz'
2019-10-31 15:22:12 +01:00
}
. $self_dir/qurator_data_lib.sh
handle_data
2019-10-31 15:22:12 +01:00
docker build --cache-from=my_ocrd_workflow-core -t my_ocrd_workflow-core -f Dockerfile-core .
docker build --cache-from=my_ocrd_workflow-ocrd_calamari -t my_ocrd_workflow-ocrd_calamari -f Dockerfile-ocrd_calamari .
docker build --cache-from=my_ocrd_workflow-dinglehopper -t my_ocrd_workflow-dinglehopper -f Dockerfile-dinglehopper .
docker build --cache-from=my_ocrd_workflow-ocrd_olena -t my_ocrd_workflow-ocrd_olena -f Dockerfile-ocrd_olena .
docker build --cache-from=my_ocrd_workflow-ocrd_tesserocr -t my_ocrd_workflow-ocrd_tesserocr -f Dockerfile-ocrd_tesserocr .
docker build --cache-from=my_ocrd_workflow-sbb_textline_detector -t my_ocrd_workflow-sbb_textline_detector -f Dockerfile-sbb_textline_detector .
docker build --cache-from=my_ocrd_workflow-sbb_binarization -t my_ocrd_workflow-sbb_binarization -f Dockerfile-sbb_binarization .
2020-10-27 16:35:46 +01:00
docker build --cache-from=my_ocrd_workflow-ocrd_cis -t my_ocrd_workflow-ocrd_cis -f Dockerfile-ocrd_cis .