From a7a39b3327fc6cf85e0d898ba6840e3a5ca5603e Mon Sep 17 00:00:00 2001
From: Timm Fitschen <t.fitschen@indiscale.com>
Date: Thu, 19 Sep 2019 23:52:51 +0200
Subject: [PATCH] WIP load tsv from file

---
 scripting/bin/xls_from_csv.py | 23 +++++++----------------
 1 file changed, 7 insertions(+), 16 deletions(-)

diff --git a/scripting/bin/xls_from_csv.py b/scripting/bin/xls_from_csv.py
index b64173a3..d830d862 100755
--- a/scripting/bin/xls_from_csv.py
+++ b/scripting/bin/xls_from_csv.py
@@ -31,27 +31,18 @@ import sys
 import pandas as pd
 
 
-def _parse_to_dataframe(csv_string):
-    """Attempts to create a valid dataframe from a CSV string.
+def _parse_to_dataframe(tsv_file):
+    """Attempts to create a valid dataframe from a TSV file.
 
-The CSV string typically starts with a header like this:
-
-```
-data:text/csv;charset=utf-8,colname1	colname2
-value1	value2
-...
-```
 Parameters
 ----------
-csv_string : The URL encoded CSV content, starts with `data:text/csv`.
+tsv_file : path to a tsv file.
 
 Returns
 -------
 out : The created dataframe.
     """
-    csv_string = csv_string.split(",")[1]
-    sio = io.StringIO(csv_string)
-    dataframe = pd.read_csv(sio, sep="\t")
+    dataframe = pd.read_csv(tsv_file, sep="\t")
     return dataframe
 
 
@@ -100,15 +91,15 @@ def _parse_arguments():
     tempdir = os.environ["SHARED_DIR"]
     parser.add_argument('-t', '--tempdir', required=False, default=tempdir,
                         help="Temporary dir for saving the result.")
-    parser.add_argument('-u', '--urlencoded', required=True,
-                        help="The URL encoded csv data.")
     parser.add_argument('-a', '--auth-token', required=False,
                         help="An authentication token (not needed, only for compatibility).")
+    parser.add_argument('tsv', required=True,
+                        help="The tsv file.")
     return parser.parse_args()
 
 def main():
   args = _parse_arguments()
-  dataframe = _parse_to_dataframe(args.urlencoded)
+  dataframe = _parse_to_dataframe(args.tsv)
   filename = _write_xls(dataframe, directory=args.tempdir)
   print(filename)
 
-- 
GitLab