diff --git a/.drone.yml b/.drone.yml new file mode 100644 index 0000000..2bf0c22 --- /dev/null +++ b/.drone.yml @@ -0,0 +1,12 @@ +kind: pipeline +name: default + +steps: +- name: prepare data + image: alpine + commands: + - ./build-tmp-XXX +- name: build core + image: plugins/docker + dockerfile: Dockerfile-core + diff --git a/build-tmp-XXX b/build-tmp-XXX new file mode 100755 index 0000000..b8fcd6c --- /dev/null +++ b/build-tmp-XXX @@ -0,0 +1,17 @@ +#!/bin/bash +set -e + +self=`realpath $0` +self_dir=`dirname "$self"` + +DATA_SUBDIR=data +get_from_web() { + 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/calamari-models/GT4HistOCR/2019-12-11T11_10+0100/model.tar.xz' 'calamari-models/GT4HistOCR/2019-12-11T11_10+0100' + 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' +} +. $self_dir/qurator_data_lib.sh +handle_data