| 
									
										
										
										
											2020-02-10 15:12:27 +01:00
										 |  |  | #!/bin/bash | 
					
						
							| 
									
										
										
										
											2020-02-10 19:23:17 +01:00
										 |  |  | set -e | 
					
						
							| 
									
										
										
										
											2019-10-31 15:22:12 +01:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2020-02-10 19:23:17 +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*' | 
					
						
							| 
									
										
										
										
											2020-11-17 10:00:38 +01:00
										 |  |  |   annex_get 'calamari-models/GT4HistOCR/2019-12-11T11_10+0100/*.ckpt*' | 
					
						
							| 
									
										
										
										
											2019-10-31 15:22:12 +01:00
										 |  |  |   annex_get 'tesseract-models/GT4HistOCR/*.traineddata' | 
					
						
							|  |  |  |   annex_get 'textline_detection/*.h5' | 
					
						
							| 
									
										
										
										
											2020-08-05 16:03:17 +02:00
										 |  |  |   annex_get 'mirror/github.com/tesseract-ocr/tessdata_best/archive/4.0.0-repacked.tar.gz' | 
					
						
							| 
									
										
										
										
											2020-10-22 21:08:13 +02:00
										 |  |  |   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' | 
					
						
							| 
									
										
										
										
											2020-11-19 17:27:31 +01:00
										 |  |  |   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' | 
					
						
							| 
									
										
										
										
											2020-08-05 12:27:05 +02:00
										 |  |  |   download_to 'https://qurator-data.de/tesseract-models/GT4HistOCR/models.tar'  'tesseract-models/GT4HistOCR' | 
					
						
							| 
									
										
										
										
											2020-02-10 15:46:55 +01:00
										 |  |  |   download_to 'https://qurator-data.de/sbb_textline_detector/models.tar.gz'     'textline_detection' | 
					
						
							| 
									
										
										
										
											2020-10-22 21:08:13 +02:00
										 |  |  |   download_to --strip-components 1 'https://qurator-data.de/sbb_binarization/models.tar.gz' 'sbb_binarization' | 
					
						
							| 
									
										
										
										
											2020-08-06 13:08:46 +02:00
										 |  |  |   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
										 |  |  | } | 
					
						
							| 
									
										
										
										
											2020-02-10 19:23:17 +01:00
										 |  |  | . $self_dir/qurator_data_lib.sh | 
					
						
							|  |  |  | handle_data | 
					
						
							| 
									
										
										
										
											2019-10-31 15:22:12 +01:00
										 |  |  | 
 | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2020-09-01 13:09:37 +02:00
										 |  |  | 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 . | 
					
						
							| 
									
										
										
										
											2020-10-22 21:08:13 +02:00
										 |  |  | 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              . | 
					
						
							| 
									
										
										
										
											2020-11-13 17:47:52 +01:00
										 |  |  | docker build --cache-from=my_ocrd_workflow-ocrd_fileformat       -t my_ocrd_workflow-ocrd_fileformat       -f Dockerfile-ocrd_fileformat       . | 
					
						
							| 
									
										
										
										
											2020-11-17 10:00:38 +01:00
										 |  |  | 
 | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | # XXX | 
					
						
							|  |  |  | docker build --cache-from=my_ocrd_workflow-ocrd_calamari-feat-update-calamari1 -t my_ocrd_workflow-ocrd_calamari-feat-update-calamari1 -f Dockerfile-ocrd_calamari-feat-update-calamari1 . |