[gentoo-commits] proj/tinderbox-cluster:master commit in: buildbot_gentoo_ci/logs/

["Magnus Granberg" <zorry@gentoo.org>]

commit:     57d8ef85943af7ddfeaa51d0cacdacba80a75a13
Author:     Magnus Granberg <zorry <AT> gentoo <DOT> org>
AuthorDate: Thu Feb  3 21:23:28 2022 +0000
Commit:     Magnus Granberg <zorry <AT> gentoo <DOT> org>
CommitDate: Thu Feb  3 21:23:28 2022 +0000
URL:        https://gitweb.gentoo.org/proj/tinderbox-cluster.git/commit/?id=57d8ef85

Fix UnicodeDecodeError in log parser

Signed-off-by: Magnus Granberg <zorry <AT> gentoo.org>

 buildbot_gentoo_ci/logs/log_parser.py | 11 ++++-------
 1 file changed, 4 insertions(+), 7 deletions(-)

diff --git a/buildbot_gentoo_ci/logs/log_parser.py b/buildbot_gentoo_ci/logs/log_parser.py
index 7531532..448192c 100644
--- a/buildbot_gentoo_ci/logs/log_parser.py
+++ b/buildbot_gentoo_ci/logs/log_parser.py
@@ -166,13 +166,10 @@ def runLogParser(args):
     # Is stored in a db instead of files.
     log_search_pattern_list = get_log_search_pattern(Session, args.uuid, config['default_uuid'])
     Session.close()
-    #FIXME: UnicodeDecodeError: 'utf-8' codec can't decode byte ... in some logs
-    with io.TextIOWrapper(io.BufferedReader(gzip.open(args.file, 'rb'))) as f:
-        for text_line in f:
-            logfile_text_dict[index] = text_line.strip('\n')
-            index = index + 1
-            max_text_lines = index
-        f.close()
+    for text_line in io.TextIOWrapper(io.BufferedReader(gzip.open(args.file)), encoding='utf8', errors='ignore'):
+        logfile_text_dict[index] = text_line.strip('\n')
+        index = index + 1
+        max_text_lines = index
     # run the parse patten on the line
     for tmp_index, text in logfile_text_dict.items():
         res = mp_pool.apply_async(search_buildlog, (log_search_pattern_list, logfile_text_dict, tmp_index, max_text_lines,))