From a9e6bfe09c82973aaf904ae99f51d88bf6b965b6 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Philip=20R=C3=B6ggla?= <philip.roeggla@onb.ac.at>
Date: Mon, 20 Jul 2020 09:27:32 +0200
Subject: [PATCH] a script to do the work

---
 travelogues_extraction/script/__init__.py |  0
 travelogues_extraction/script/script.py   | 44 +++++++++++++++++++++++
 2 files changed, 44 insertions(+)
 create mode 100644 travelogues_extraction/script/__init__.py
 create mode 100644 travelogues_extraction/script/script.py

diff --git a/travelogues_extraction/script/__init__.py b/travelogues_extraction/script/__init__.py
new file mode 100644
index 0000000..e69de29
diff --git a/travelogues_extraction/script/script.py b/travelogues_extraction/script/script.py
new file mode 100644
index 0000000..2d35fdf
--- /dev/null
+++ b/travelogues_extraction/script/script.py
@@ -0,0 +1,44 @@
+import asyncio
+import datetime
+import json
+import os
+import pathlib
+import re as regex
+
+import typing
+
+from travelogues_extraction.controller.main import FromAlmaOutputToExcel
+
+input_folder = './travelogues_extraction/script/input/'
+output_folder = './travelogues_extraction/script/output/'
+filter = 'D19'
+
+files = os.listdir(input_folder)
+files = [file for file in files if filter in file]
+files = [input_folder + file for file in files]
+
+
+async def extract(input_file: str, output_folder: str) -> typing.NoReturn:
+    input_path = pathlib.Path(input_file)
+    output_path = pathlib.Path(
+        output_folder + \
+        input_path.name.replace(
+            'ALMAoutput', 'script_output_' \
+                          + datetime.datetime.now().strftime("%Y-%m-%d-%H:%M:%S"))
+    )
+
+    converter = FromAlmaOutputToExcel(input_file, output_path.absolute(), slice(None, None, 1))
+    await converter.runasync()
+    await converter.close()
+    converter.write()
+    match = regex.search(r'[dD]1\d', input_file)
+    name = match.group(0) if match else 'VDXX'
+    converter.log.to_csv('log-' + name + datetime.datetime.now().strftime("-%Y-%m-%d-%H:%M:%S") + '.csv')
+    # to do unblock
+    with open('sub-log-' + name + datetime.datetime.now().strftime("-%Y-%m-%d-%H:%M:%S") + '.json', 'w') as file:
+        json.dump(converter.sub_log, file)
+
+loop = asyncio.new_event_loop()
+tasks = [loop.create_task(extract(file, output_folder)) for file in files]
+waiter = asyncio.wait(tasks)
+loop.run_until_complete(waiter)
-- 
GitLab