From 57ec1d749cf75866f3b4da0864c436becd0e31b6 Mon Sep 17 00:00:00 2001
From: Martijn Vermaat <martijn@vermaat.name>
Date: Fri, 3 Jul 2015 15:13:33 +0200
Subject: [PATCH] Don't trust encoding auto-detection when decoding
---
mutalyzer/File.py | 18 +++++++++++++++---
mutalyzer/Retriever.py | 11 +++++++++--
2 files changed, 24 insertions(+), 5 deletions(-)
diff --git a/mutalyzer/File.py b/mutalyzer/File.py
index add6e5d4..772d2002 100644
--- a/mutalyzer/File.py
+++ b/mutalyzer/File.py
@@ -170,7 +170,13 @@ class File() :
handle = _UniversalNewlinesByteStreamIter(handle, encoding=encoding,
buffer_size=BUFFER_SIZE)
- buf = handle.read(BUFFER_SIZE)
+ try:
+ buf = handle.read(BUFFER_SIZE)
+ except UnicodeDecodeError:
+ self.__output.addMessage(__file__, 3, 'EBPARSE',
+ 'Could not decode file (using %s encoding).'
+ % encoding)
+ return None
# Default dialect
dialect = 'excel'
@@ -196,8 +202,14 @@ class File() :
reader = csv.reader(handle, dialect)
ret = []
- for i in reader:
- ret.append([c.decode('utf-8') for c in i])
+ try:
+ for i in reader:
+ ret.append([c.decode('utf-8') for c in i])
+ except UnicodeDecodeError:
+ self.__output.addMessage(__file__, 3, 'EBPARSE',
+ 'Could not decode file (using %s encoding).'
+ % encoding)
+ return None
return ret
#__parseCsvFile
diff --git a/mutalyzer/Retriever.py b/mutalyzer/Retriever.py
index 4cfaf642..e514ab96 100644
--- a/mutalyzer/Retriever.py
+++ b/mutalyzer/Retriever.py
@@ -114,7 +114,13 @@ class Retriever(object) :
encoding = 'utf-8'
if not util.is_utf8_alias(encoding):
- raw_data = raw_data.decode(encoding).encode('utf-8')
+ try:
+ raw_data = raw_data.decode(encoding).encode('utf-8')
+ except UnicodeDecodeError:
+ self._output.addMessage(__file__, 4, 'ENOPARSE',
+ 'Could not decode file (using %s encoding).'
+ % encoding)
+ return None
# Compress the data to save disk space.
comp = bz2.BZ2Compressor()
@@ -368,7 +374,8 @@ class GenBankRetriever(Retriever):
"number to reduce downloading overhead." % unicode(record.id))
#if
- self._write(raw_data, outfile)
+ if not self._write(raw_data, outfile):
+ return None
return outfile, GI
#write
--
GitLab