Change open function to use UTF-8 encoding for content.

Uses io.open for Python 2 and built-in open for Python 3
All data from files is treated as Unicode and this should fix
most UnicodeDecodeErrors.

4 files changed, 23 insertions, 8 deletions

diff --git a/pym/gentoolkit/revdep_rebuild/analyse.py b/pym/gentoolkit/revdep_rebuild/analyse.py
index 59240b4..9f018b5 100644
--- a/pym/gentoolkit/revdep_rebuild/analyse.py
+++ b/pym/gentoolkit/revdep_rebuild/analyse.py
@@ -7,6 +7,9 @@ from __future__ import print_function
 import os
 import re
 import time
+import sys
+if sys.hexversion < 0x3000000:
+	from io import open
 
 from portage import _encodings, _unicode_decode, _unicode_encode
 from portage.output import bold, blue, yellow, green
@@ -83,7 +86,8 @@ def extract_dependencies_from_la(la, libraries, to_check, logger):
 		if not os.path.exists(_file):
 			continue
 
-		for line in open(_unicode_encode(_file, encoding=_encodings['fs']), mode='r').readlines():
+		for line in open(_unicode_encode(_file, encoding=_encodings['fs']), mode='r',
+			encoding=_encodings['content']).readlines():
 			line = line.strip()
 			if line.startswith('dependency_libs='):
 				match = re.match("dependency_libs='([^']+)'", line)
diff --git a/pym/gentoolkit/revdep_rebuild/cache.py b/pym/gentoolkit/revdep_rebuild/cache.py
index 6d1a1a3..7359d05 100644
--- a/pym/gentoolkit/revdep_rebuild/cache.py
+++ b/pym/gentoolkit/revdep_rebuild/cache.py
@@ -7,6 +7,9 @@ from __future__ import print_function
 
 import os
 import time
+import sys
+if sys.hexversion < 0x3000000:
+	from io import open
 
 from portage import _encodings, _unicode_decode, _unicode_encode
 from portage.output import red
@@ -31,7 +34,7 @@ def read_cache(temp_path=DEFAULTS['DEFAULT_TMP_DIR']):
 	try:
 		for key,val in ret.items():
 			_file = open(_unicode_encode(os.path.join(temp_path, key),
-				encoding=_encodings['fs']))
+				encoding=_encodings['fs']), encoding=_encodings['content'])
 			for line in _file.readlines():
 				val.add(line.strip())
 			#libraries.remove('\n')
@@ -55,13 +58,14 @@ def save_cache(logger, to_save={}, temp_path=DEFAULTS['DEFAULT_TMP_DIR']):
 
 	try:
 		_file = open(_unicode_encode(os.path.join(temp_path, 'timestamp'),
-			encoding=_encodings['fs']), mode='w')
+			encoding=_encodings['fs']), mode='w', encoding=_encodings['content'])
 		_file.write(str(int(time.time())))
 		_file.close()
 
 		for key,val in to_save.items():
 			_file = open(_unicode_encode(os.path.join(temp_path, key),
-				encoding=_encodings['fs']), mode='w')
+				encoding=_encodings['fs']), mode='w',
+				encoding=_encodings['content'])
 			for line in val:
 				_file.write(line + '\n')
 			_file.close()
@@ -89,7 +93,8 @@ def check_temp_files(temp_path=DEFAULTS['DEFAULT_TMP_DIR'], max_delay=3600,
 		return False
 
 	try:
-		_file = open(_unicode_encode(timestamp_path, encoding=_encodings['fs']))
+		_file = open(_unicode_encode(timestamp_path, encoding=_encodings['fs']),
+			encoding=_encodings['content'])
 		timestamp = int(_file.readline())
 		_file .close()
 	except Exception as ex:
diff --git a/pym/gentoolkit/revdep_rebuild/collect.py b/pym/gentoolkit/revdep_rebuild/collect.py
index 1f34f1c..ab3ef97 100644
--- a/pym/gentoolkit/revdep_rebuild/collect.py
+++ b/pym/gentoolkit/revdep_rebuild/collect.py
@@ -9,6 +9,8 @@ import os
 import glob
 import stat
 import sys
+if sys.hexversion < 0x3000000:
+	from io import open
 
 import portage
 from portage import _encodings, _unicode_decode, _unicode_encode
@@ -35,7 +37,8 @@ def parse_conf(conf_file, visited=None, logger=None):
 
 	for conf in conf_file:
 		try:
-			with open(_unicode_encode(conf, encoding=_encodings['fs'])) as _file:
+			with open(_unicode_encode(conf, encoding=_encodings['fs']),
+					encoding=_encodings['content']) as _file:
 				for line in _file.readlines():
 					line = line.strip()
 					if line.startswith('#'):
@@ -77,7 +80,8 @@ def prepare_search_dirs(logger, settings):
 	#try:
 	with open(_unicode_encode(os.path.join(
 		portage.root, settings['DEFAULT_ENV_FILE']),
-		encoding=_encodings['fs']), mode='r') as _file:
+		encoding=_encodings['fs']), mode='r',
+		encoding=_encodings['content']) as _file:
 		for line in _file.readlines():
 			line = line.strip()
 			match = re.match("^export (ROOT)?PATH='([^']+)'", line)
diff --git a/pym/gentoolkit/revdep_rebuild/settings.py b/pym/gentoolkit/revdep_rebuild/settings.py
index 14f5bc8..589ea29 100644
--- a/pym/gentoolkit/revdep_rebuild/settings.py
+++ b/pym/gentoolkit/revdep_rebuild/settings.py
@@ -7,6 +7,8 @@ from __future__ import print_function
 import argparse
 import os
 import sys
+if sys.hexversion < 0x3000000:
+	from io import open
 import re
 import glob
 
@@ -138,7 +140,7 @@ def parse_revdep_config(revdep_confdir):
 
 	for _file in os.listdir(revdep_confdir):
 		for line in open(_unicode_encode(os.path.join(revdep_confdir, _file),
-				encoding=_encodings['fs'])):
+				encoding=_encodings['fs']), encoding=_encodings['content']):
 			line = line.strip()
 			#first check for comment, we do not want to regex all lines
 			if not line.startswith('#'):