diff --git a/src/utils/operandi_utils/hpc/nextflow_workflows/default_workflow.nf b/src/utils/operandi_utils/hpc/nextflow_workflows/default_workflow.nf index 087530e..d3ddb39 100755 --- a/src/utils/operandi_utils/hpc/nextflow_workflows/default_workflow.nf +++ b/src/utils/operandi_utils/hpc/nextflow_workflows/default_workflow.nf @@ -86,7 +86,7 @@ process ocrd_cis_ocropy_binarize_0 { script: """ - ${params.env_wrapper_cmd_step0} ocrd-cis-ocropy-binarize -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} + ${params.env_wrapper_cmd_step0} ocrd-cis-ocropy-binarize -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} """ } @@ -110,7 +110,7 @@ process ocrd_anybaseocr_crop_1 { script: """ - ${params.env_wrapper_cmd_step1} ocrd-anybaseocr-crop -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} + ${params.env_wrapper_cmd_step1} ocrd-anybaseocr-crop -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} """ } @@ -134,7 +134,7 @@ process ocrd_skimage_binarize_2 { script: """ - ${params.env_wrapper_cmd_step2} ocrd-skimage-binarize -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"method": "li"}' + ${params.env_wrapper_cmd_step2} ocrd-skimage-binarize -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"method": "li"}' """ } @@ -158,7 +158,7 @@ process ocrd_skimage_denoise_3 { script: """ - ${params.env_wrapper_cmd_step3} ocrd-skimage-denoise -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "page"}' + ${params.env_wrapper_cmd_step3} ocrd-skimage-denoise -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "page"}' """ } @@ -182,7 +182,7 @@ process ocrd_tesserocr_deskew_4 { script: """ - ${params.env_wrapper_cmd_step4} ocrd-tesserocr-deskew -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"operation_level": "page"}' + ${params.env_wrapper_cmd_step4} ocrd-tesserocr-deskew -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"operation_level": "page"}' """ } @@ -206,7 +206,7 @@ process ocrd_cis_ocropy_segment_5 { script: """ - ${params.env_wrapper_cmd_step5} ocrd-cis-ocropy-segment -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "page"}' + ${params.env_wrapper_cmd_step5} ocrd-cis-ocropy-segment -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "page"}' """ } @@ -230,7 +230,7 @@ process ocrd_cis_ocropy_dewarp_6 { script: """ - ${params.env_wrapper_cmd_step6} ocrd-cis-ocropy-dewarp -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} + ${params.env_wrapper_cmd_step6} ocrd-cis-ocropy-dewarp -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} """ } @@ -254,7 +254,7 @@ process ocrd_calamari_recognize_7 { script: """ - ${params.env_wrapper_cmd_step7} ocrd-calamari-recognize -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"checkpoint_dir": "qurator-gt4histocr-1.0"}' + ${params.env_wrapper_cmd_step7} ocrd-calamari-recognize -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"checkpoint_dir": "qurator-gt4histocr-1.0"}' """ } diff --git a/src/utils/operandi_utils/hpc/nextflow_workflows/default_workflow_with_MS.nf b/src/utils/operandi_utils/hpc/nextflow_workflows/default_workflow_with_MS.nf index 6315672..42198e0 100755 --- a/src/utils/operandi_utils/hpc/nextflow_workflows/default_workflow_with_MS.nf +++ b/src/utils/operandi_utils/hpc/nextflow_workflows/default_workflow_with_MS.nf @@ -86,7 +86,7 @@ process ocrd_cis_ocropy_binarize_0 { script: """ - ${params.env_wrapper_cmd_step0} ocrd-cis-ocropy-binarize -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} + ${params.env_wrapper_cmd_step0} ocrd-cis-ocropy-binarize -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} """ } @@ -110,7 +110,7 @@ process ocrd_anybaseocr_crop_1 { script: """ - ${params.env_wrapper_cmd_step1} ocrd-anybaseocr-crop -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} + ${params.env_wrapper_cmd_step1} ocrd-anybaseocr-crop -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} """ } @@ -134,7 +134,7 @@ process ocrd_skimage_binarize_2 { script: """ - ${params.env_wrapper_cmd_step2} ocrd-skimage-binarize -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"method": "li"}' + ${params.env_wrapper_cmd_step2} ocrd-skimage-binarize -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"method": "li"}' """ } @@ -158,7 +158,7 @@ process ocrd_skimage_denoise_3 { script: """ - ${params.env_wrapper_cmd_step3} ocrd-skimage-denoise -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "page"}' + ${params.env_wrapper_cmd_step3} ocrd-skimage-denoise -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "page"}' """ } @@ -182,7 +182,7 @@ process ocrd_tesserocr_deskew_4 { script: """ - ${params.env_wrapper_cmd_step4} ocrd-tesserocr-deskew -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"operation_level": "page"}' + ${params.env_wrapper_cmd_step4} ocrd-tesserocr-deskew -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"operation_level": "page"}' """ } @@ -206,7 +206,7 @@ process ocrd_cis_ocropy_segment_5 { script: """ - ${params.env_wrapper_cmd_step5} ocrd-cis-ocropy-segment -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "page"}' + ${params.env_wrapper_cmd_step5} ocrd-cis-ocropy-segment -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "page"}' """ } @@ -230,7 +230,7 @@ process ocrd_cis_ocropy_dewarp_6 { script: """ - ${params.env_wrapper_cmd_step6} ocrd-cis-ocropy-dewarp -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} + ${params.env_wrapper_cmd_step6} ocrd-cis-ocropy-dewarp -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} """ } @@ -254,7 +254,7 @@ process ocrd_calamari_recognize_7 { script: """ - ${params.env_wrapper_cmd_step7} ocrd-calamari-recognize -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"checkpoint_dir": "qurator-gt4histocr-1.0"}' + ${params.env_wrapper_cmd_step7} ocrd-calamari-recognize -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"checkpoint_dir": "qurator-gt4histocr-1.0"}' """ } diff --git a/src/utils/operandi_utils/hpc/nextflow_workflows/odem_workflow.nf b/src/utils/operandi_utils/hpc/nextflow_workflows/odem_workflow.nf index edc308f..c9589b8 100755 --- a/src/utils/operandi_utils/hpc/nextflow_workflows/odem_workflow.nf +++ b/src/utils/operandi_utils/hpc/nextflow_workflows/odem_workflow.nf @@ -90,7 +90,7 @@ process ocrd_cis_ocropy_binarize_0 { script: """ - ${params.env_wrapper_cmd_step0} ocrd-cis-ocropy-binarize -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"dpi": 300}' + ${params.env_wrapper_cmd_step0} ocrd-cis-ocropy-binarize -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"dpi": 300}' """ } @@ -114,7 +114,7 @@ process ocrd_anybaseocr_crop_1 { script: """ - ${params.env_wrapper_cmd_step1} ocrd-anybaseocr-crop -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"dpi": 300}' + ${params.env_wrapper_cmd_step1} ocrd-anybaseocr-crop -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"dpi": 300}' """ } @@ -138,7 +138,7 @@ process ocrd_cis_ocropy_denoise_2 { script: """ - ${params.env_wrapper_cmd_step2} ocrd-cis-ocropy-denoise -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"dpi": 300}' + ${params.env_wrapper_cmd_step2} ocrd-cis-ocropy-denoise -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"dpi": 300}' """ } @@ -162,7 +162,7 @@ process ocrd_cis_ocropy_deskew_3 { script: """ - ${params.env_wrapper_cmd_step3} ocrd-cis-ocropy-deskew -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "page"}' + ${params.env_wrapper_cmd_step3} ocrd-cis-ocropy-deskew -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "page"}' """ } @@ -186,7 +186,7 @@ process ocrd_tesserocr_segment_region_4 { script: """ - ${params.env_wrapper_cmd_step4} ocrd-tesserocr-segment-region -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"padding": 5.0, "find_tables": false, "dpi": 300}' + ${params.env_wrapper_cmd_step4} ocrd-tesserocr-segment-region -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"padding": 5.0, "find_tables": false, "dpi": 300}' """ } @@ -210,7 +210,7 @@ process ocrd_segment_repair_5 { script: """ - ${params.env_wrapper_cmd_step5} ocrd-segment-repair -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"plausibilize": true, "plausibilize_merge_min_overlap": 0.7}' + ${params.env_wrapper_cmd_step5} ocrd-segment-repair -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"plausibilize": true, "plausibilize_merge_min_overlap": 0.7}' """ } @@ -234,7 +234,7 @@ process ocrd_cis_ocropy_clip_6 { script: """ - ${params.env_wrapper_cmd_step6} ocrd-cis-ocropy-clip -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} + ${params.env_wrapper_cmd_step6} ocrd-cis-ocropy-clip -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} """ } @@ -258,7 +258,7 @@ process ocrd_cis_ocropy_segment_7 { script: """ - ${params.env_wrapper_cmd_step7} ocrd-cis-ocropy-segment -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"dpi": 300}' + ${params.env_wrapper_cmd_step7} ocrd-cis-ocropy-segment -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"dpi": 300}' """ } @@ -282,7 +282,7 @@ process ocrd_cis_ocropy_dewarp_8 { script: """ - ${params.env_wrapper_cmd_step8} ocrd-cis-ocropy-dewarp -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} + ${params.env_wrapper_cmd_step8} ocrd-cis-ocropy-dewarp -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} """ } @@ -306,7 +306,7 @@ process ocrd_tesserocr_recognize_9 { script: """ - ${params.env_wrapper_cmd_step9} ocrd-tesserocr-recognize -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"model": "Fraktur"}' + ${params.env_wrapper_cmd_step9} ocrd-tesserocr-recognize -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"model": "Fraktur"}' """ } diff --git a/src/utils/operandi_utils/hpc/nextflow_workflows/odem_workflow_with_MS.nf b/src/utils/operandi_utils/hpc/nextflow_workflows/odem_workflow_with_MS.nf index 39e5710..07ba186 100755 --- a/src/utils/operandi_utils/hpc/nextflow_workflows/odem_workflow_with_MS.nf +++ b/src/utils/operandi_utils/hpc/nextflow_workflows/odem_workflow_with_MS.nf @@ -90,7 +90,7 @@ process ocrd_cis_ocropy_binarize_0 { script: """ - ${params.env_wrapper_cmd_step0} ocrd-cis-ocropy-binarize -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"dpi": 300}' + ${params.env_wrapper_cmd_step0} ocrd-cis-ocropy-binarize -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"dpi": 300}' """ } @@ -114,7 +114,7 @@ process ocrd_anybaseocr_crop_1 { script: """ - ${params.env_wrapper_cmd_step1} ocrd-anybaseocr-crop -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"dpi": 300}' + ${params.env_wrapper_cmd_step1} ocrd-anybaseocr-crop -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"dpi": 300}' """ } @@ -138,7 +138,7 @@ process ocrd_cis_ocropy_denoise_2 { script: """ - ${params.env_wrapper_cmd_step2} ocrd-cis-ocropy-denoise -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"dpi": 300}' + ${params.env_wrapper_cmd_step2} ocrd-cis-ocropy-denoise -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"dpi": 300}' """ } @@ -162,7 +162,7 @@ process ocrd_cis_ocropy_deskew_3 { script: """ - ${params.env_wrapper_cmd_step3} ocrd-cis-ocropy-deskew -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "page"}' + ${params.env_wrapper_cmd_step3} ocrd-cis-ocropy-deskew -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "page"}' """ } @@ -186,7 +186,7 @@ process ocrd_tesserocr_segment_region_4 { script: """ - ${params.env_wrapper_cmd_step4} ocrd-tesserocr-segment-region -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"padding": 5.0, "find_tables": false, "dpi": 300}' + ${params.env_wrapper_cmd_step4} ocrd-tesserocr-segment-region -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"padding": 5.0, "find_tables": false, "dpi": 300}' """ } @@ -210,7 +210,7 @@ process ocrd_segment_repair_5 { script: """ - ${params.env_wrapper_cmd_step5} ocrd-segment-repair -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"plausibilize": true, "plausibilize_merge_min_overlap": 0.7}' + ${params.env_wrapper_cmd_step5} ocrd-segment-repair -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"plausibilize": true, "plausibilize_merge_min_overlap": 0.7}' """ } @@ -234,7 +234,7 @@ process ocrd_cis_ocropy_clip_6 { script: """ - ${params.env_wrapper_cmd_step6} ocrd-cis-ocropy-clip -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} + ${params.env_wrapper_cmd_step6} ocrd-cis-ocropy-clip -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} """ } @@ -258,7 +258,7 @@ process ocrd_cis_ocropy_segment_7 { script: """ - ${params.env_wrapper_cmd_step7} ocrd-cis-ocropy-segment -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"dpi": 300}' + ${params.env_wrapper_cmd_step7} ocrd-cis-ocropy-segment -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"dpi": 300}' """ } @@ -282,7 +282,7 @@ process ocrd_cis_ocropy_dewarp_8 { script: """ - ${params.env_wrapper_cmd_step8} ocrd-cis-ocropy-dewarp -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} + ${params.env_wrapper_cmd_step8} ocrd-cis-ocropy-dewarp -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} """ } @@ -306,7 +306,7 @@ process ocrd_tesserocr_recognize_9 { script: """ - ${params.env_wrapper_cmd_step9} ocrd-tesserocr-recognize -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"model": "Fraktur"}' + ${params.env_wrapper_cmd_step9} ocrd-tesserocr-recognize -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"model": "Fraktur"}' """ } diff --git a/src/utils/operandi_utils/hpc/nextflow_workflows/sbb_workflow.nf b/src/utils/operandi_utils/hpc/nextflow_workflows/sbb_workflow.nf index 881222a..4285177 100755 --- a/src/utils/operandi_utils/hpc/nextflow_workflows/sbb_workflow.nf +++ b/src/utils/operandi_utils/hpc/nextflow_workflows/sbb_workflow.nf @@ -72,7 +72,7 @@ process ocrd_tesserocr_recognize_0 { script: """ - ${params.env_wrapper_cmd_step0} ocrd-tesserocr-recognize -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"segmentation_level": "region", "textequiv_level": "word", "find_tables": true, "model": "deu"}' + ${params.env_wrapper_cmd_step0} ocrd-tesserocr-recognize -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"segmentation_level": "region", "textequiv_level": "word", "find_tables": true, "model": "deu"}' """ } diff --git a/src/utils/operandi_utils/hpc/nextflow_workflows/sbb_workflow_with_MS.nf b/src/utils/operandi_utils/hpc/nextflow_workflows/sbb_workflow_with_MS.nf index 75e1061..699b580 100755 --- a/src/utils/operandi_utils/hpc/nextflow_workflows/sbb_workflow_with_MS.nf +++ b/src/utils/operandi_utils/hpc/nextflow_workflows/sbb_workflow_with_MS.nf @@ -72,7 +72,7 @@ process ocrd_tesserocr_recognize_0 { script: """ - ${params.env_wrapper_cmd_step0} ocrd-tesserocr-recognize -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"segmentation_level": "region", "textequiv_level": "word", "find_tables": true, "model": "deu"}' + ${params.env_wrapper_cmd_step0} ocrd-tesserocr-recognize -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"segmentation_level": "region", "textequiv_level": "word", "find_tables": true, "model": "deu"}' """ } diff --git a/src/utils/operandi_utils/hpc/nextflow_workflows/template_workflow.nf b/src/utils/operandi_utils/hpc/nextflow_workflows/template_workflow.nf index c2ae497..4dfa4a6 100755 --- a/src/utils/operandi_utils/hpc/nextflow_workflows/template_workflow.nf +++ b/src/utils/operandi_utils/hpc/nextflow_workflows/template_workflow.nf @@ -72,7 +72,7 @@ process ocrd_cis_ocropy_binarize_0 { script: """ - ${params.env_wrapper_cmd_step0} ocrd-cis-ocropy-binarize -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} + ${params.env_wrapper_cmd_step0} ocrd-cis-ocropy-binarize -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} """ } diff --git a/src/utils/operandi_utils/hpc/nextflow_workflows/template_workflow_with_MS.nf b/src/utils/operandi_utils/hpc/nextflow_workflows/template_workflow_with_MS.nf index 90d5d61..e95640e 100755 --- a/src/utils/operandi_utils/hpc/nextflow_workflows/template_workflow_with_MS.nf +++ b/src/utils/operandi_utils/hpc/nextflow_workflows/template_workflow_with_MS.nf @@ -72,7 +72,7 @@ process ocrd_cis_ocropy_binarize_0 { script: """ - ${params.env_wrapper_cmd_step0} ocrd-cis-ocropy-binarize -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} + ${params.env_wrapper_cmd_step0} ocrd-cis-ocropy-binarize -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} """ } diff --git a/src/utils/operandi_utils/oton/process_call_arguments.py b/src/utils/operandi_utils/oton/process_call_arguments.py index e1f7600..def5aa4 100644 --- a/src/utils/operandi_utils/oton/process_call_arguments.py +++ b/src/utils/operandi_utils/oton/process_call_arguments.py @@ -2,7 +2,7 @@ from logging import getLevelName, getLogger from typing import Optional from operandi_utils.oton.constants import ( - BS, CONST_DIR_IN, CONST_DIR_OUT, CONST_WORKSPACE_DIR, CONST_METS_PATH, CONST_METS_SOCKET_PATH, + BS, CONST_DIR_IN, CONST_DIR_OUT, CONST_WORKSPACE_DIR, CONST_METS_PATH, CONST_PAGE_RANGE, CONST_METS_SOCKET_PATH, OCRD_ALL_JSON, OTON_LOG_LEVEL ) @@ -58,6 +58,7 @@ def dump_bash_form_with_placeholders(self): dump += f' -U ${BS[0]}{CONST_METS_SOCKET_PATH}{BS[1]}' dump += f' -w ${BS[0]}{CONST_WORKSPACE_DIR}{BS[1]}' dump += f' -m ${BS[0]}{CONST_METS_PATH}{BS[1]}' + dump += f' --page-id ${BS[0]}{CONST_PAGE_RANGE}{BS[1]}' dump += f' -I ${BS[0]}{CONST_DIR_IN}{BS[1]}' dump += f' -O ${BS[0]}{CONST_DIR_OUT}{BS[1]}' if self.parameters: diff --git a/tests/assets/oton/test_output_nextflow1_apptainer.nf b/tests/assets/oton/test_output_nextflow1_apptainer.nf index 087530e..d3ddb39 100644 --- a/tests/assets/oton/test_output_nextflow1_apptainer.nf +++ b/tests/assets/oton/test_output_nextflow1_apptainer.nf @@ -86,7 +86,7 @@ process ocrd_cis_ocropy_binarize_0 { script: """ - ${params.env_wrapper_cmd_step0} ocrd-cis-ocropy-binarize -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} + ${params.env_wrapper_cmd_step0} ocrd-cis-ocropy-binarize -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} """ } @@ -110,7 +110,7 @@ process ocrd_anybaseocr_crop_1 { script: """ - ${params.env_wrapper_cmd_step1} ocrd-anybaseocr-crop -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} + ${params.env_wrapper_cmd_step1} ocrd-anybaseocr-crop -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} """ } @@ -134,7 +134,7 @@ process ocrd_skimage_binarize_2 { script: """ - ${params.env_wrapper_cmd_step2} ocrd-skimage-binarize -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"method": "li"}' + ${params.env_wrapper_cmd_step2} ocrd-skimage-binarize -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"method": "li"}' """ } @@ -158,7 +158,7 @@ process ocrd_skimage_denoise_3 { script: """ - ${params.env_wrapper_cmd_step3} ocrd-skimage-denoise -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "page"}' + ${params.env_wrapper_cmd_step3} ocrd-skimage-denoise -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "page"}' """ } @@ -182,7 +182,7 @@ process ocrd_tesserocr_deskew_4 { script: """ - ${params.env_wrapper_cmd_step4} ocrd-tesserocr-deskew -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"operation_level": "page"}' + ${params.env_wrapper_cmd_step4} ocrd-tesserocr-deskew -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"operation_level": "page"}' """ } @@ -206,7 +206,7 @@ process ocrd_cis_ocropy_segment_5 { script: """ - ${params.env_wrapper_cmd_step5} ocrd-cis-ocropy-segment -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "page"}' + ${params.env_wrapper_cmd_step5} ocrd-cis-ocropy-segment -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "page"}' """ } @@ -230,7 +230,7 @@ process ocrd_cis_ocropy_dewarp_6 { script: """ - ${params.env_wrapper_cmd_step6} ocrd-cis-ocropy-dewarp -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} + ${params.env_wrapper_cmd_step6} ocrd-cis-ocropy-dewarp -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} """ } @@ -254,7 +254,7 @@ process ocrd_calamari_recognize_7 { script: """ - ${params.env_wrapper_cmd_step7} ocrd-calamari-recognize -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"checkpoint_dir": "qurator-gt4histocr-1.0"}' + ${params.env_wrapper_cmd_step7} ocrd-calamari-recognize -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"checkpoint_dir": "qurator-gt4histocr-1.0"}' """ } diff --git a/tests/assets/oton/test_output_nextflow1_apptainer_with_MS.nf b/tests/assets/oton/test_output_nextflow1_apptainer_with_MS.nf index 6315672..42198e0 100644 --- a/tests/assets/oton/test_output_nextflow1_apptainer_with_MS.nf +++ b/tests/assets/oton/test_output_nextflow1_apptainer_with_MS.nf @@ -86,7 +86,7 @@ process ocrd_cis_ocropy_binarize_0 { script: """ - ${params.env_wrapper_cmd_step0} ocrd-cis-ocropy-binarize -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} + ${params.env_wrapper_cmd_step0} ocrd-cis-ocropy-binarize -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} """ } @@ -110,7 +110,7 @@ process ocrd_anybaseocr_crop_1 { script: """ - ${params.env_wrapper_cmd_step1} ocrd-anybaseocr-crop -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} + ${params.env_wrapper_cmd_step1} ocrd-anybaseocr-crop -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} """ } @@ -134,7 +134,7 @@ process ocrd_skimage_binarize_2 { script: """ - ${params.env_wrapper_cmd_step2} ocrd-skimage-binarize -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"method": "li"}' + ${params.env_wrapper_cmd_step2} ocrd-skimage-binarize -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"method": "li"}' """ } @@ -158,7 +158,7 @@ process ocrd_skimage_denoise_3 { script: """ - ${params.env_wrapper_cmd_step3} ocrd-skimage-denoise -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "page"}' + ${params.env_wrapper_cmd_step3} ocrd-skimage-denoise -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "page"}' """ } @@ -182,7 +182,7 @@ process ocrd_tesserocr_deskew_4 { script: """ - ${params.env_wrapper_cmd_step4} ocrd-tesserocr-deskew -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"operation_level": "page"}' + ${params.env_wrapper_cmd_step4} ocrd-tesserocr-deskew -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"operation_level": "page"}' """ } @@ -206,7 +206,7 @@ process ocrd_cis_ocropy_segment_5 { script: """ - ${params.env_wrapper_cmd_step5} ocrd-cis-ocropy-segment -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "page"}' + ${params.env_wrapper_cmd_step5} ocrd-cis-ocropy-segment -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "page"}' """ } @@ -230,7 +230,7 @@ process ocrd_cis_ocropy_dewarp_6 { script: """ - ${params.env_wrapper_cmd_step6} ocrd-cis-ocropy-dewarp -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} + ${params.env_wrapper_cmd_step6} ocrd-cis-ocropy-dewarp -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} """ } @@ -254,7 +254,7 @@ process ocrd_calamari_recognize_7 { script: """ - ${params.env_wrapper_cmd_step7} ocrd-calamari-recognize -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"checkpoint_dir": "qurator-gt4histocr-1.0"}' + ${params.env_wrapper_cmd_step7} ocrd-calamari-recognize -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"checkpoint_dir": "qurator-gt4histocr-1.0"}' """ } diff --git a/tests/assets/oton/test_output_nextflow1_docker.nf b/tests/assets/oton/test_output_nextflow1_docker.nf index c1dc0c3..7598d3d 100644 --- a/tests/assets/oton/test_output_nextflow1_docker.nf +++ b/tests/assets/oton/test_output_nextflow1_docker.nf @@ -74,7 +74,7 @@ process ocrd_cis_ocropy_binarize_0 { script: """ - ${params.env_wrapper_cmd_step0} ocrd-cis-ocropy-binarize -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} + ${params.env_wrapper_cmd_step0} ocrd-cis-ocropy-binarize -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} """ } @@ -96,7 +96,7 @@ process ocrd_anybaseocr_crop_1 { script: """ - ${params.env_wrapper_cmd_step1} ocrd-anybaseocr-crop -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} + ${params.env_wrapper_cmd_step1} ocrd-anybaseocr-crop -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} """ } @@ -118,7 +118,7 @@ process ocrd_skimage_binarize_2 { script: """ - ${params.env_wrapper_cmd_step2} ocrd-skimage-binarize -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"method": "li"}' + ${params.env_wrapper_cmd_step2} ocrd-skimage-binarize -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"method": "li"}' """ } @@ -140,7 +140,7 @@ process ocrd_skimage_denoise_3 { script: """ - ${params.env_wrapper_cmd_step3} ocrd-skimage-denoise -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "page"}' + ${params.env_wrapper_cmd_step3} ocrd-skimage-denoise -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "page"}' """ } @@ -162,7 +162,7 @@ process ocrd_tesserocr_deskew_4 { script: """ - ${params.env_wrapper_cmd_step4} ocrd-tesserocr-deskew -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"operation_level": "page"}' + ${params.env_wrapper_cmd_step4} ocrd-tesserocr-deskew -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"operation_level": "page"}' """ } @@ -184,7 +184,7 @@ process ocrd_cis_ocropy_segment_5 { script: """ - ${params.env_wrapper_cmd_step5} ocrd-cis-ocropy-segment -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "page"}' + ${params.env_wrapper_cmd_step5} ocrd-cis-ocropy-segment -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "page"}' """ } @@ -206,7 +206,7 @@ process ocrd_cis_ocropy_dewarp_6 { script: """ - ${params.env_wrapper_cmd_step6} ocrd-cis-ocropy-dewarp -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} + ${params.env_wrapper_cmd_step6} ocrd-cis-ocropy-dewarp -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} """ } @@ -228,7 +228,7 @@ process ocrd_calamari_recognize_7 { script: """ - ${params.env_wrapper_cmd_step7} ocrd-calamari-recognize -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"checkpoint_dir": "qurator-gt4histocr-1.0"}' + ${params.env_wrapper_cmd_step7} ocrd-calamari-recognize -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"checkpoint_dir": "qurator-gt4histocr-1.0"}' """ } diff --git a/tests/assets/oton/test_output_nextflow1_docker_with_MS.nf b/tests/assets/oton/test_output_nextflow1_docker_with_MS.nf index fd80771..49f0940 100644 --- a/tests/assets/oton/test_output_nextflow1_docker_with_MS.nf +++ b/tests/assets/oton/test_output_nextflow1_docker_with_MS.nf @@ -74,7 +74,7 @@ process ocrd_cis_ocropy_binarize_0 { script: """ - ${params.env_wrapper_cmd_step0} ocrd-cis-ocropy-binarize -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} + ${params.env_wrapper_cmd_step0} ocrd-cis-ocropy-binarize -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} """ } @@ -96,7 +96,7 @@ process ocrd_anybaseocr_crop_1 { script: """ - ${params.env_wrapper_cmd_step1} ocrd-anybaseocr-crop -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} + ${params.env_wrapper_cmd_step1} ocrd-anybaseocr-crop -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} """ } @@ -118,7 +118,7 @@ process ocrd_skimage_binarize_2 { script: """ - ${params.env_wrapper_cmd_step2} ocrd-skimage-binarize -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"method": "li"}' + ${params.env_wrapper_cmd_step2} ocrd-skimage-binarize -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"method": "li"}' """ } @@ -140,7 +140,7 @@ process ocrd_skimage_denoise_3 { script: """ - ${params.env_wrapper_cmd_step3} ocrd-skimage-denoise -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "page"}' + ${params.env_wrapper_cmd_step3} ocrd-skimage-denoise -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "page"}' """ } @@ -162,7 +162,7 @@ process ocrd_tesserocr_deskew_4 { script: """ - ${params.env_wrapper_cmd_step4} ocrd-tesserocr-deskew -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"operation_level": "page"}' + ${params.env_wrapper_cmd_step4} ocrd-tesserocr-deskew -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"operation_level": "page"}' """ } @@ -184,7 +184,7 @@ process ocrd_cis_ocropy_segment_5 { script: """ - ${params.env_wrapper_cmd_step5} ocrd-cis-ocropy-segment -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "page"}' + ${params.env_wrapper_cmd_step5} ocrd-cis-ocropy-segment -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "page"}' """ } @@ -206,7 +206,7 @@ process ocrd_cis_ocropy_dewarp_6 { script: """ - ${params.env_wrapper_cmd_step6} ocrd-cis-ocropy-dewarp -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} + ${params.env_wrapper_cmd_step6} ocrd-cis-ocropy-dewarp -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} """ } @@ -228,7 +228,7 @@ process ocrd_calamari_recognize_7 { script: """ - ${params.env_wrapper_cmd_step7} ocrd-calamari-recognize -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"checkpoint_dir": "qurator-gt4histocr-1.0"}' + ${params.env_wrapper_cmd_step7} ocrd-calamari-recognize -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"checkpoint_dir": "qurator-gt4histocr-1.0"}' """ } diff --git a/tests/assets/oton/test_output_nextflow1_local.nf b/tests/assets/oton/test_output_nextflow1_local.nf index e2e846d..726402e 100644 --- a/tests/assets/oton/test_output_nextflow1_local.nf +++ b/tests/assets/oton/test_output_nextflow1_local.nf @@ -72,7 +72,7 @@ process ocrd_cis_ocropy_binarize_0 { script: """ - ocrd-cis-ocropy-binarize -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} + ocrd-cis-ocropy-binarize -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} """ } @@ -94,7 +94,7 @@ process ocrd_anybaseocr_crop_1 { script: """ - ocrd-anybaseocr-crop -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} + ocrd-anybaseocr-crop -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} """ } @@ -116,7 +116,7 @@ process ocrd_skimage_binarize_2 { script: """ - ocrd-skimage-binarize -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"method": "li"}' + ocrd-skimage-binarize -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"method": "li"}' """ } @@ -138,7 +138,7 @@ process ocrd_skimage_denoise_3 { script: """ - ocrd-skimage-denoise -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "page"}' + ocrd-skimage-denoise -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "page"}' """ } @@ -160,7 +160,7 @@ process ocrd_tesserocr_deskew_4 { script: """ - ocrd-tesserocr-deskew -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"operation_level": "page"}' + ocrd-tesserocr-deskew -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"operation_level": "page"}' """ } @@ -182,7 +182,7 @@ process ocrd_cis_ocropy_segment_5 { script: """ - ocrd-cis-ocropy-segment -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "page"}' + ocrd-cis-ocropy-segment -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "page"}' """ } @@ -204,7 +204,7 @@ process ocrd_cis_ocropy_dewarp_6 { script: """ - ocrd-cis-ocropy-dewarp -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} + ocrd-cis-ocropy-dewarp -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} """ } @@ -226,7 +226,7 @@ process ocrd_calamari_recognize_7 { script: """ - ocrd-calamari-recognize -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"checkpoint_dir": "qurator-gt4histocr-1.0"}' + ocrd-calamari-recognize -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"checkpoint_dir": "qurator-gt4histocr-1.0"}' """ } diff --git a/tests/assets/oton/test_output_nextflow1_local_with_MS.nf b/tests/assets/oton/test_output_nextflow1_local_with_MS.nf index 89bac4c..2985355 100644 --- a/tests/assets/oton/test_output_nextflow1_local_with_MS.nf +++ b/tests/assets/oton/test_output_nextflow1_local_with_MS.nf @@ -72,7 +72,7 @@ process ocrd_cis_ocropy_binarize_0 { script: """ - ocrd-cis-ocropy-binarize -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} + ocrd-cis-ocropy-binarize -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} """ } @@ -94,7 +94,7 @@ process ocrd_anybaseocr_crop_1 { script: """ - ocrd-anybaseocr-crop -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} + ocrd-anybaseocr-crop -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} """ } @@ -116,7 +116,7 @@ process ocrd_skimage_binarize_2 { script: """ - ocrd-skimage-binarize -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"method": "li"}' + ocrd-skimage-binarize -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"method": "li"}' """ } @@ -138,7 +138,7 @@ process ocrd_skimage_denoise_3 { script: """ - ocrd-skimage-denoise -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "page"}' + ocrd-skimage-denoise -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "page"}' """ } @@ -160,7 +160,7 @@ process ocrd_tesserocr_deskew_4 { script: """ - ocrd-tesserocr-deskew -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"operation_level": "page"}' + ocrd-tesserocr-deskew -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"operation_level": "page"}' """ } @@ -182,7 +182,7 @@ process ocrd_cis_ocropy_segment_5 { script: """ - ocrd-cis-ocropy-segment -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "page"}' + ocrd-cis-ocropy-segment -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "page"}' """ } @@ -204,7 +204,7 @@ process ocrd_cis_ocropy_dewarp_6 { script: """ - ocrd-cis-ocropy-dewarp -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} + ocrd-cis-ocropy-dewarp -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} """ } @@ -226,7 +226,7 @@ process ocrd_calamari_recognize_7 { script: """ - ocrd-calamari-recognize -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"checkpoint_dir": "qurator-gt4histocr-1.0"}' + ocrd-calamari-recognize -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"checkpoint_dir": "qurator-gt4histocr-1.0"}' """ } diff --git a/tests/assets/oton/test_output_nextflow2.nf b/tests/assets/oton/test_output_nextflow2.nf index 83b8b84..62b4c6b 100644 --- a/tests/assets/oton/test_output_nextflow2.nf +++ b/tests/assets/oton/test_output_nextflow2.nf @@ -70,7 +70,7 @@ process ocrd_cis_ocropy_binarize_0 { script: """ - ocrd-cis-ocropy-binarize -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} + ocrd-cis-ocropy-binarize -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} """ } @@ -92,7 +92,7 @@ process ocrd_anybaseocr_crop_1 { script: """ - ocrd-anybaseocr-crop -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} + ocrd-anybaseocr-crop -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} """ } @@ -114,7 +114,7 @@ process ocrd_skimage_denoise_2 { script: """ - ocrd-skimage-denoise -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "page"}' + ocrd-skimage-denoise -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "page"}' """ } @@ -136,7 +136,7 @@ process ocrd_tesserocr_deskew_3 { script: """ - ocrd-tesserocr-deskew -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"operation_level": "page"}' + ocrd-tesserocr-deskew -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"operation_level": "page"}' """ } @@ -158,7 +158,7 @@ process ocrd_tesserocr_segment_4 { script: """ - ocrd-tesserocr-segment -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"shrink_polygons": true}' + ocrd-tesserocr-segment -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"shrink_polygons": true}' """ } @@ -180,7 +180,7 @@ process ocrd_cis_ocropy_dewarp_5 { script: """ - ocrd-cis-ocropy-dewarp -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} + ocrd-cis-ocropy-dewarp -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} """ } @@ -202,7 +202,7 @@ process ocrd_tesserocr_recognize_6 { script: """ - ocrd-tesserocr-recognize -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"textequiv_level": "glyph", "overwrite_segments": true, "model": "GT4HistOCR_50000000.997_191951"}' + ocrd-tesserocr-recognize -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"textequiv_level": "glyph", "overwrite_segments": true, "model": "GT4HistOCR_50000000.997_191951"}' """ } diff --git a/tests/assets/oton/test_output_nextflow3.nf b/tests/assets/oton/test_output_nextflow3.nf index 71f670c..5d2c84b 100644 --- a/tests/assets/oton/test_output_nextflow3.nf +++ b/tests/assets/oton/test_output_nextflow3.nf @@ -62,7 +62,7 @@ process ocrd_dinglehopper_0 { script: """ - ocrd-dinglehopper -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} + ocrd-dinglehopper -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} """ } @@ -84,7 +84,7 @@ process ocrd_dinglehopper_1 { script: """ - ocrd-dinglehopper -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} + ocrd-dinglehopper -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} """ } @@ -106,7 +106,7 @@ process ocrd_dinglehopper_2 { script: """ - ocrd-dinglehopper -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} + ocrd-dinglehopper -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} """ } diff --git a/tests/assets/oton/test_output_nextflow4.nf b/tests/assets/oton/test_output_nextflow4.nf index b8bc5f5..e7eb76b 100644 --- a/tests/assets/oton/test_output_nextflow4.nf +++ b/tests/assets/oton/test_output_nextflow4.nf @@ -82,7 +82,7 @@ process ocrd_olena_binarize_0 { script: """ - ocrd-olena-binarize -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"impl": "sauvola"}' + ocrd-olena-binarize -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"impl": "sauvola"}' """ } @@ -104,7 +104,7 @@ process ocrd_anybaseocr_crop_1 { script: """ - ocrd-anybaseocr-crop -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} + ocrd-anybaseocr-crop -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} """ } @@ -126,7 +126,7 @@ process ocrd_olena_binarize_2 { script: """ - ocrd-olena-binarize -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"impl": "kim"}' + ocrd-olena-binarize -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"impl": "kim"}' """ } @@ -148,7 +148,7 @@ process ocrd_cis_ocropy_denoise_3 { script: """ - ocrd-cis-ocropy-denoise -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "page"}' + ocrd-cis-ocropy-denoise -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "page"}' """ } @@ -170,7 +170,7 @@ process ocrd_cis_ocropy_deskew_4 { script: """ - ocrd-cis-ocropy-deskew -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "page"}' + ocrd-cis-ocropy-deskew -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "page"}' """ } @@ -192,7 +192,7 @@ process ocrd_tesserocr_segment_region_5 { script: """ - ocrd-tesserocr-segment-region -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} + ocrd-tesserocr-segment-region -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} """ } @@ -214,7 +214,7 @@ process ocrd_segment_repair_6 { script: """ - ocrd-segment-repair -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"plausibilize": true}' + ocrd-segment-repair -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"plausibilize": true}' """ } @@ -236,7 +236,7 @@ process ocrd_cis_ocropy_deskew_7 { script: """ - ocrd-cis-ocropy-deskew -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "region"}' + ocrd-cis-ocropy-deskew -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "region"}' """ } @@ -258,7 +258,7 @@ process ocrd_cis_ocropy_clip_8 { script: """ - ocrd-cis-ocropy-clip -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "region"}' + ocrd-cis-ocropy-clip -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"level-of-operation": "region"}' """ } @@ -280,7 +280,7 @@ process ocrd_tesserocr_segment_line_9 { script: """ - ocrd-tesserocr-segment-line -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} + ocrd-tesserocr-segment-line -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} """ } @@ -302,7 +302,7 @@ process ocrd_segment_repair_10 { script: """ - ocrd-segment-repair -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"sanitize": true}' + ocrd-segment-repair -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"sanitize": true}' """ } @@ -324,7 +324,7 @@ process ocrd_cis_ocropy_dewarp_11 { script: """ - ocrd-cis-ocropy-dewarp -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} + ocrd-cis-ocropy-dewarp -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} """ } @@ -346,7 +346,7 @@ process ocrd_calamari_recognize_12 { script: """ - ocrd-calamari-recognize -w ${workspace_dir} -m ${mets_path} -I ${input_group} -O ${output_group} -p '{"checkpoint_dir": "qurator-gt4histocr-1.0"}' + ocrd-calamari-recognize -w ${workspace_dir} -m ${mets_path} --page-id ${page_range} -I ${input_group} -O ${output_group} -p '{"checkpoint_dir": "qurator-gt4histocr-1.0"}' """ }