explicit encoding for text file operations

Fixes: https://pagure.io/koji/issue/2641

koji/__init__.py

@@ -29,6 +29,7 @@ import datetime
 import errno
 import hashlib
 import imp
+import json
 import logging
 import logging.handlers
 import optparse
@@ -1282,9 +1283,8 @@ def parse_pom(path=None, contents=None):
     values = {}
     handler = POMHandler(values, fields)
     if path:
-        fd = open(path)
-        contents = fd.read()
-        fd.close()
+        with open(path, encoding='utf-8') as fd:
+            contents = fd.read()
 
     if not contents:
         raise GenericError(
@@ -1353,6 +1353,16 @@ def hex_string(s):
     return ''.join(['%02x' % _ord(x) for x in s])
 
 
+def load_json(filepath):
+    """Loads json from file"""
+    return json.load(open(filepath, 'rt', encoding='utf-8'))
+
+
+def dump_json(filepath, data, indent=4, sort_keys=False):
+    """Write json to file"""
+    json.dump(data, open(filepath, 'wt', encoding='utf-8'), indent=indent, sort_keys=sort_keys)
+
+
 def make_groups_spec(grplist, name='buildsys-build', buildgroup=None):
     """Return specfile contents representing the group"""
     if buildgroup is None:
@@ -1611,9 +1621,7 @@ def genMockConfig(name, arch, managed=False, repoid=None, tag_name=None, **opts)
     if opts.get('use_host_resolv', False) and os.path.exists('/etc/hosts'):
         # if we're setting up DNS,
         # also copy /etc/hosts from the host
-        etc_hosts = open('/etc/hosts')
-        files['etc/hosts'] = etc_hosts.read()
-        etc_hosts.close()
+        files['etc/hosts'] = open('/etc/hosts', 'rt', encoding='utf-8').read()
     mavenrc = ''
     if opts.get('maven_opts'):
         mavenrc = 'export MAVEN_OPTS="%s"\n' % ' '.join(opts['maven_opts'])
@@ -1832,7 +1840,7 @@ def openRemoteFile(relpath, topurl=None, topdir=None, tempdir=None):
         downloadFile(url, path=relpath, fo=fo)
     elif topdir:
         fn = "%s/%s" % (topdir, relpath)
-        fo = open(fn)
+        fo = open(fn, 'rb')
     else:
         raise GenericError("No access method for remote file: %s" % relpath)
     return fo

koji/arch.py

@@ -237,7 +237,7 @@ def _try_read_cpuinfo():
     """ Try to read /proc/cpuinfo ... if we can't ignore errors (ie. proc not
         mounted). """
     try:
-        return open("/proc/cpuinfo", "r")
+        return open("/proc/cpuinfo", "rt")
     except Exception:
         return []
 
@@ -383,9 +383,8 @@ def getCanonX86_64Arch(arch):
 def getCanonArch(skipRpmPlatform=0):
     if not skipRpmPlatform and os.access("/etc/rpm/platform", os.R_OK):
         try:
-            f = open("/etc/rpm/platform", "r")
-            line = f.readline()
-            f.close()
+            with open("/etc/rpm/platform", "rt", encoding='utf-8') as f:
+                line = f.readline()
             (arch, vendor, opersys) = line.split("-", 2)
             return arch
         except Exception:

koji/daemon.py

@@ -790,7 +790,7 @@ class TaskManager(object):
             fn = "%s/%s" % (configdir, f)
             if not os.path.isfile(fn):
                 continue
-            fo = open(fn, 'r')
+            fo = open(fn, 'rt', encoding='utf-8')
             id = None
             name = None
             for n in range(10):
@@ -1089,7 +1089,7 @@ class TaskManager(object):
             proc_path = '/proc/%i/stat' % pid
             if not os.path.isfile(proc_path):
                 return None
-            proc_file = open(proc_path)
+            proc_file = open(proc_path, 'rt', encoding='utf-8')
             procstats = [not field.isdigit() and field or int(field)
                          for field in proc_file.read().split()]
             proc_file.close()
@@ -1097,7 +1097,7 @@ class TaskManager(object):
             cmd_path = '/proc/%i/cmdline' % pid
             if not os.path.isfile(cmd_path):
                 return None
-            cmd_file = open(cmd_path)
+            cmd_file = open(cmd_path, 'rt', encoding='utf-8')
             procstats[1] = cmd_file.read().replace('\0', ' ').strip()
             cmd_file.close()
             if not procstats[1]: