Change all open() calls to use Unicode.

We are using the following import from portage:
from portage import _encodings, _unicode_decode, _unicode_encode

A generalized call using the definitions from portage looks like:
with open(_unicode_encode(path), encoding=_encodings['fs']) as open_file

The portage code has been in place since 2013 and using the definitions
from portage ensures we maintain compatibility if portage changes.

All portage versions in the tree contain the above code.

4 files changed, 18 insertions, 9 deletions

diff --git a/pym/gentoolkit/revdep_rebuild/analyse.py b/pym/gentoolkit/revdep_rebuild/analyse.py
index c0e7231..0f89b03 100644
--- a/pym/gentoolkit/revdep_rebuild/analyse.py
+++ b/pym/gentoolkit/revdep_rebuild/analyse.py
@@ -8,6 +8,7 @@ import os
 import re
 import time
 
+from portage import _encodings, _unicode_decode, _unicode_encode
 from portage.output import bold, blue, yellow, green
 
 from .stuff import scan
@@ -82,7 +83,7 @@ def extract_dependencies_from_la(la, libraries, to_check, logger):
 		if not os.path.exists(_file):
 			continue
 
-		for line in open(_file, 'r').readlines():
+		for line in open(_unicode_encode(_file), encoding=_encodings['fs'], mode='r').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 3d925d8..31ee2c9 100644
--- a/pym/gentoolkit/revdep_rebuild/cache.py
+++ b/pym/gentoolkit/revdep_rebuild/cache.py
@@ -8,6 +8,7 @@ from __future__ import print_function
 import os
 import time
 
+from portage import _encodings, _unicode_decode, _unicode_encode
 from portage.output import red
 from .settings import DEFAULTS
 
@@ -29,7 +30,8 @@ def read_cache(temp_path=DEFAULTS['DEFAULT_TMP_DIR']):
 		}
 	try:
 		for key,val in ret.items():
-			_file = open(os.path.join(temp_path, key))
+			_file = open(_unicode_encode(os.path.join(temp_path, key)),
+				encoding=_encodings['fs'])
 			for line in _file.readlines():
 				val.add(line.strip())
 			#libraries.remove('\n')
@@ -52,12 +54,14 @@ def save_cache(logger, to_save={}, temp_path=DEFAULTS['DEFAULT_TMP_DIR']):
 		os.makedirs(temp_path)
 
 	try:
-		_file = open(os.path.join(temp_path, 'timestamp'), 'w')
+		_file = open(_unicode_encode(os.path.join(temp_path, 'timestamp')),
+			encoding=_encodings['fs'], mode='w')
 		_file.write(str(int(time.time())))
 		_file.close()
 
 		for key,val in to_save.items():
-			_file = open(os.path.join(temp_path, key), 'w')
+			_file = open(_unicode_encode(os.path.join(temp_path, key)),
+				encoding=_encodings['fs'], mode='w')
 			for line in val:
 				_file.write(line + '\n')
 			_file.close()
@@ -85,7 +89,7 @@ def check_temp_files(temp_path=DEFAULTS['DEFAULT_TMP_DIR'], max_delay=3600,
 		return False
 
 	try:
-		_file = open(timestamp_path)
+		_file = open(_unicode_encode(timestamp_path), encoding=_encodings['fs'])
 		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 2a431cb..758bcf7 100644
--- a/pym/gentoolkit/revdep_rebuild/collect.py
+++ b/pym/gentoolkit/revdep_rebuild/collect.py
@@ -11,6 +11,7 @@ import stat
 import sys
 
 import portage
+from portage import _encodings, _unicode_decode, _unicode_encode
 from portage.output import blue, yellow
 from .settings import parse_revdep_config
 
@@ -34,7 +35,7 @@ def parse_conf(conf_file, visited=None, logger=None):
 
 	for conf in conf_file:
 		try:
-			with open(conf) as _file:
+			with open(_unicode_encode(conf), encoding=_encodings['fs']) as _file:
 				for line in _file.readlines():
 					line = line.strip()
 					if line.startswith('#'):
@@ -74,8 +75,9 @@ def prepare_search_dirs(logger, settings):
 	lib_dirs = set(['/lib', '/usr/lib', ])
 
 	#try:
-	with open(os.path.join(
-		portage.root, settings['DEFAULT_ENV_FILE']), 'r') as _file:
+	with open(_unicode_encode(os.path.join(
+		portage.root, settings['DEFAULT_ENV_FILE'])),
+		encoding=_encodings['fs'], mode='r') 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 08220f8..257bd3a 100644
--- a/pym/gentoolkit/revdep_rebuild/settings.py
+++ b/pym/gentoolkit/revdep_rebuild/settings.py
@@ -11,6 +11,7 @@ import re
 import glob
 
 import portage
+from portage import _encodings, _unicode_decode, _unicode_encode
 
 DEFAULTS = {
 		'DEFAULT_LD_FILE': os.path.join(portage.root, 'etc/ld.so.conf'),
@@ -136,7 +137,8 @@ def parse_revdep_config(revdep_confdir):
 	masked_files = os.environ.get('LD_LIBRARY_MASK', '')
 
 	for _file in os.listdir(revdep_confdir):
-		for line in open(os.path.join(revdep_confdir, _file)):
+		for line in open(_unicode_encode(os.path.join(revdep_confdir, _file)),
+				encoding=_encodings['fs']):
 			line = line.strip()
 			#first check for comment, we do not want to regex all lines
 			if not line.startswith('#'):