From dfc05be7db2afe1ed452936730ca808c56da9729 Mon Sep 17 00:00:00 2001
From: Patrick Jentsch <p.jentsch@uni-bielefeld.de>
Date: Mon, 20 Jan 2020 15:04:55 +0100
Subject: [PATCH] add zip creation of results

---
 Dockerfile |  1 +
 ocr        | 56 ++++++++++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 57 insertions(+)

diff --git a/Dockerfile b/Dockerfile
index b715aec..c848bad 100644
--- a/Dockerfile
+++ b/Dockerfile
@@ -19,6 +19,7 @@ RUN apt-get update \
       python2.7 \
       python3.5 \
       wget \
+      zip \
  && rm -rf /var/lib/apt/lists/*
 
 ENV OCROPY_VERSION 1.3.3
diff --git a/ocr b/ocr
index 28ecbd9..4a8c6bf 100755
--- a/ocr
+++ b/ocr
@@ -16,6 +16,7 @@ import os
 import re
 import sys
 from pyflow import WorkflowRunner
+from zipfile import ZipFile
 
 
 def parse_arguments():
@@ -74,6 +75,7 @@ class OCRWorkflow(WorkflowRunner):
         self.keep_intermediates = args.keep_intermediates
         self.lang = args.lang
         self.n_cores = args.n_cores
+        self.output_dir = args.output_dir
 
     def workflow(self):
         if len(self.jobs) == 0:
@@ -384,6 +386,60 @@ 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)
+            )
+        )
+
+        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)
+            )
+        )
+
+        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)
+            )
+        )
+
+        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)
+            )
+        )
+
         '''
         ' ##################################################
         ' # Cleanup                                        #