From 6c4a642cb708033076af9bdfc7ec566ce6fefdb8 Mon Sep 17 00:00:00 2001
From: Patrick Jentsch <p.jentsch@uni-bielefeld.de>
Date: Mon, 3 Feb 2020 15:00:27 +0100
Subject: [PATCH] Add a switch for zip functionality

---
 ocr | 102 +++++++++++++++++++++++++++++++++---------------------------
 1 file changed, 56 insertions(+), 46 deletions(-)

diff --git a/ocr b/ocr
index 4a8c6bf..a25e4c2 100755
--- a/ocr
+++ b/ocr
@@ -65,6 +65,14 @@ def parse_arguments():
         required=False,
         type=int
     )
+    parser.add_argument(
+        '--zip',
+        action='store_true',
+        default=False,
+        dest='zip',
+        help='package result files in zip bundles',
+        required=False
+    )
     return parser.parse_args()
 
 
@@ -76,6 +84,7 @@ class OCRWorkflow(WorkflowRunner):
         self.lang = args.lang
         self.n_cores = args.n_cores
         self.output_dir = args.output_dir
+        self.zip = args.zip
 
     def workflow(self):
         if len(self.jobs) == 0:
@@ -386,59 +395,60 @@ class OCRWorkflow(WorkflowRunner):
                 )
             )
 
-        all_zip_jobs = []
-        all_zip_job_dependencies = (hocr_to_tei_jobs
-                                    + pdf_merge_jobs
-                                    + txt_merge_jobs)
-        cmd = 'cd "%s" && zip all.zip */*.{pdf,txt,xml} -x "pyflow.data*" && cd -' % (
-            self.output_dir
-        )
-        all_zip_jobs.append(
-            self.addTask(
-                command=cmd,
-                dependencies=all_zip_job_dependencies,
-                label='all_zip_job_-_%i' % (index)
+        if self.zip:
+            all_zip_jobs = []
+            all_zip_job_dependencies = (hocr_to_tei_jobs
+                                        + pdf_merge_jobs
+                                        + txt_merge_jobs)
+            cmd = 'cd "%s" && zip all.zip */*.{pdf,txt,xml} -x "pyflow.data*" && cd -' % (
+                self.output_dir
+            )
+            all_zip_jobs.append(
+                self.addTask(
+                    command=cmd,
+                    dependencies=all_zip_job_dependencies,
+                    label='all_zip_job'
+                )
             )
-        )
 
-        pdf_zip_jobs = []
-        pdf_zip_job_dependencies = pdf_merge_jobs
-        cmd = 'cd "%s" && zip pdf.zip */*.pdf -x "pyflow.data*" && cd -' % (
-            self.output_dir
-        )
-        pdf_zip_jobs.append(
-            self.addTask(
-                command=cmd,
-                dependencies=pdf_zip_job_dependencies,
-                label='pdf_zip_job_-_%i' % (index)
+            pdf_zip_jobs = []
+            pdf_zip_job_dependencies = all_zip_jobs
+            cmd = 'cd "%s" && zip -m pdf.zip */*.pdf -x "pyflow.data*" && cd -' % (
+                self.output_dir
+            )
+            pdf_zip_jobs.append(
+                self.addTask(
+                    command=cmd,
+                    dependencies=pdf_zip_job_dependencies,
+                    label='pdf_zip_job'
+                )
             )
-        )
 
-        txt_zip_jobs = []
-        txt_zip_job_dependencies = txt_merge_jobs
-        cmd = 'cd "%s" && zip txt.zip */*.txt -x "pyflow.data*" && cd -' % (
-            self.output_dir
-        )
-        txt_zip_jobs.append(
-            self.addTask(
-                command=cmd,
-                dependencies=txt_zip_job_dependencies,
-                label='txt_zip_job_-_%i' % (index)
+            txt_zip_jobs = []
+            txt_zip_job_dependencies = all_zip_jobs
+            cmd = 'cd "%s" && zip -m txt.zip */*.txt -x "pyflow.data*" && cd -' % (
+                self.output_dir
+            )
+            txt_zip_jobs.append(
+                self.addTask(
+                    command=cmd,
+                    dependencies=txt_zip_job_dependencies,
+                    label='txt_zip_job'
+                )
             )
-        )
 
-        xml_zip_jobs = []
-        xml_zip_job_dependencies = hocr_to_tei_jobs
-        cmd = 'cd "%s" && zip xml.zip */*.xml -x "pyflow.data*" && cd -' % (
-            self.output_dir
-        )
-        xml_zip_jobs.append(
-            self.addTask(
-                command=cmd,
-                dependencies=xml_zip_job_dependencies,
-                label='xml_zip_job_-_%i' % (index)
+            xml_zip_jobs = []
+            xml_zip_job_dependencies = all_zip_jobs
+            cmd = 'cd "%s" && zip -m xml.zip */*.xml -x "pyflow.data*" && cd -' % (
+                self.output_dir
+            )
+            xml_zip_jobs.append(
+                self.addTask(
+                    command=cmd,
+                    dependencies=xml_zip_job_dependencies,
+                    label='xml_zip_job'
+                )
             )
-        )
 
         '''
         ' ##################################################