From: Debian Science Maintainers <debian-science-maintainers@lists.alioth.debian.org>
Date: Sat, 2 Mar 2019 14:59:35 +0000 (+0000)
Subject: Don't use np.genfromtxt on non-ASCII data
X-Git-Tag: archive/raspbian/0.11.1-2+rpi1~2^2^2~4
X-Git-Url: https://dgit.raspbian.org/?a=commitdiff_plain;h=cb6772395d51e6e216ba8443fe52d664f72c7fed;p=statsmodels.git

Don't use np.genfromtxt on non-ASCII data

Fixes UnicodeDecodeError on file e6.dat

Author: Peter "thequackdaddy" Quackenbush, Rebecca N. Palmer <rebecca_palmer@zoho.com>
Origin: upstream
Forwarded: not-needed


Gbp-Pq: Name lutkepohl_data_load.patch
---

diff --git a/statsmodels/tsa/vector_ar/util.py b/statsmodels/tsa/vector_ar/util.py
index 6cc8a9b..ef00f6e 100644
--- a/statsmodels/tsa/vector_ar/util.py
+++ b/statsmodels/tsa/vector_ar/util.py
@@ -134,7 +134,7 @@ def parse_lutkepohl_data(path): # pragma: no cover
             year, freq, start_point = m.groups()
             break
 
-    data = np.genfromtxt(path, names=True, skip_header=to_skip+1)
+    data = pandas.read_csv(path, delimiter=r"\s+", header=to_skip+1).to_records(index=False)
 
     n = len(data)