From eb5ccf4e2159c29bbbb43dfde0dbd661bebd2a6e Mon Sep 17 00:00:00 2001 From: stephan Date: Tue, 18 Feb 2020 10:16:24 +0100 Subject: [PATCH] Add ocr to filenames --- ocr | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) diff --git a/ocr b/ocr index 167fae1..cf3dae1 100755 --- a/ocr +++ b/ocr @@ -75,7 +75,7 @@ def parse_arguments(): default='ocr-result-files', dest='zip', type=str, - help='package result files in zip bundles', + help='package result files in zip bundles and asign an filename prefix', required=False ) return parser.parse_args() @@ -405,7 +405,7 @@ class OCRWorkflow(WorkflowRunner): all_zip_job_dependencies = (hocr_to_tei_jobs + pdf_merge_jobs + txt_merge_jobs) - cmd = 'cd "%s" && zip "%s"-all-files.zip */*.{pdf,txt,xml} -x "pyflow.data*" && cd -' % ( + cmd = 'cd "%s" && zip "%s"-all-ocr-files.zip */*.{pdf,txt,xml} -x "pyflow.data*" && cd -' % ( self.output_dir, self.zip ) @@ -419,7 +419,7 @@ class OCRWorkflow(WorkflowRunner): pdf_zip_jobs = [] pdf_zip_job_dependencies = all_zip_jobs - cmd = 'cd "%s" && zip -m "%s"-pdf.zip */*.pdf -x "pyflow.data*" && cd -' % ( + cmd = 'cd "%s" && zip -m "%s"-ocr-pdf.zip */*.pdf -x "pyflow.data*" && cd -' % ( self.output_dir, self.zip ) @@ -433,7 +433,7 @@ class OCRWorkflow(WorkflowRunner): txt_zip_jobs = [] txt_zip_job_dependencies = all_zip_jobs - cmd = 'cd "%s" && zip -m "%s"-txt.zip */*.txt -x "pyflow.data*" && cd -' % ( + cmd = 'cd "%s" && zip -m "%s"-ocr-txt.zip */*.txt -x "pyflow.data*" && cd -' % ( self.output_dir, self.zip ) @@ -447,7 +447,7 @@ class OCRWorkflow(WorkflowRunner): xml_zip_jobs = [] xml_zip_job_dependencies = all_zip_jobs - cmd = 'cd "%s" && zip -m "%s"-xml.zip */*.xml -x "pyflow.data*" && cd -' % ( + cmd = 'cd "%s" && zip -m "%s"-ocr-xml.zip */*.xml -x "pyflow.data*" && cd -' % ( self.output_dir, self.zip )