repoman: fix popen unicode handling, bug #310789

1 files changed, 61 insertions, 27 deletions

diff --git a/bin/repoman b/bin/repoman
index d7c69b32f..1f4daede9 100755
--- a/bin/repoman
+++ b/bin/repoman
@@ -9,6 +9,7 @@
 from __future__ import print_function
 
 import calendar
+import codecs
 import copy
 import errno
 import formatter
@@ -741,6 +742,36 @@ else:
 def caterror(mycat):
 	err(mycat+" is not an official category.  Skipping QA checks in this directory.\nPlease ensure that you add "+catdir+" to "+repodir+"/profiles/categories\nif it is a new category.")
 
+class repoman_popen(portage.proxy.objectproxy.ObjectProxy):
+	"""
+	Implements an interface similar to os.popen(), but with customized
+	unicode handling (see bug #310789) and without the shell.
+	"""
+
+	__slots__ = ('_proc', '_stdout')
+
+	def __init__(self, cmd):
+		args = portage.util.shlex_split(cmd)
+		encoding = _encodings['fs']
+		if sys.hexversion < 0x3000000 or sys.hexversion >= 0x3020000:
+			# Python 3.1 does not support bytes in Popen args.
+			args = [_unicode_encode(x,
+				encoding=encoding, errors='strict') for x in args]
+		proc = subprocess.Popen(args, stdout=subprocess.PIPE)
+		object.__setattr__(self, '_proc', proc)
+		object.__setattr__(self, '_stdout',
+			codecs.getreader(encoding)(proc.stdout, 'strict'))
+
+	def _get_target(self):
+		return object.__getattribute__(self, '_stdout')
+
+	__enter__ = _get_target
+
+	def __exit__(self, exc_type, exc_value, traceback):
+		proc = object.__getattribute__(self, '_proc')
+		proc.wait()
+		proc.stdout.close()
+
 class ProfileDesc(object):
 	__slots__ = ('abs_path', 'arch', 'status', 'sub_path', 'tree_path',)
 	def __init__(self, arch, status, sub_path, tree_path):
@@ -1150,7 +1181,7 @@ if vcs == "cvs":
 		myremoved = cvstree.findremoved(mycvstree, recursive=1, basedir="./")
 
 elif vcs == "svn":
-	with os.popen("svn status") as f:
+	with repoman_popen("svn status") as f:
 		svnstatus = f.readlines()
 	mychanged = [ "./" + elem.split()[-1:][0] for elem in svnstatus if elem and elem[:1] in "MR" ]
 	mynew     = [ "./" + elem.split()[-1:][0] for elem in svnstatus if elem.startswith("A") ]
@@ -1158,23 +1189,23 @@ elif vcs == "svn":
 		myremoved = [ "./" + elem.split()[-1:][0] for elem in svnstatus if elem.startswith("D")]
 
 elif vcs == "git":
-	with os.popen("git diff-index --name-only "
+	with repoman_popen("git diff-index --name-only "
 		"--relative --diff-filter=M HEAD") as f:
 		mychanged = f.readlines()
 	mychanged = ["./" + elem[:-1] for elem in mychanged]
 
-	with os.popen("git diff-index --name-only "
+	with repoman_popen("git diff-index --name-only "
 		"--relative --diff-filter=A HEAD") as f:
 		mynew = f.readlines()
 	mynew = ["./" + elem[:-1] for elem in mynew]
 	if options.if_modified == "y":
-		with os.popen("git diff-index --name-only "
+		with repoman_popen("git diff-index --name-only "
 			"--relative --diff-filter=D HEAD") as f:
 			myremoved = f.readlines()
 		myremoved = ["./" + elem[:-1] for elem in myremoved]
 
 elif vcs == "bzr":
-	with os.popen("bzr status -S .") as f:
+	with repoman_popen("bzr status -S .") as f:
 		bzrstatus = f.readlines()
 	mychanged = [ "./" + elem.split()[-1:][0].split('/')[-1:][0] for elem in bzrstatus if elem and elem[1:2] == "M" ]
 	mynew     = [ "./" + elem.split()[-1:][0].split('/')[-1:][0] for elem in bzrstatus if elem and ( elem[1:2] == "NK" or elem[0:1] == "R" ) ]
@@ -1182,13 +1213,13 @@ elif vcs == "bzr":
 		myremoved = [ "./" + elem.split()[-3:-2][0].split('/')[-1:][0] for elem in bzrstatus if elem and ( elem[1:2] == "K" or elem[0:1] == "R" ) ]
 
 elif vcs == "hg":
-	with os.popen("hg status --no-status --modified .") as f:
+	with repoman_popen("hg status --no-status --modified .") as f:
 		mychanged = f.readlines()
 	mychanged = ["./" + elem.rstrip() for elem in mychanged]
-	mynew = os.popen("hg status --no-status --added .").readlines()
+	mynew = repoman_popen("hg status --no-status --added .").readlines()
 	mynew = ["./" + elem.rstrip() for elem in mynew]
 	if options.if_modified == "y":
-		with os.popen("hg status --no-status --removed .") as f:
+		with repoman_popen("hg status --no-status --removed .") as f:
 			myremoved = f.readlines()
 		myremoved = ["./" + elem.rstrip() for elem in myremoved]
 
@@ -1426,10 +1457,10 @@ for x in effective_scanlist:
 
 	if vcs in ("git", "hg") and check_ebuild_notadded:
 		if vcs == "git":
-			myf = os.popen("git ls-files --others %s" % \
+			myf = repoman_popen("git ls-files --others %s" % \
 				(portage._shell_quote(checkdir_relative),))
 		if vcs == "hg":
-			myf = os.popen("hg status --no-status --unknown %s" % \
+			myf = repoman_popen("hg status --no-status --unknown %s" % \
 				(portage._shell_quote(checkdir_relative),))
 		for l in myf:
 			if l[:-1][-7:] == ".ebuild":
@@ -1443,9 +1474,11 @@ for x in effective_scanlist:
 			if vcs == "cvs":
 				myf=open(checkdir+"/CVS/Entries","r")
 			if vcs == "svn":
-				myf = os.popen("svn status --depth=files --verbose " + checkdir)
+				myf = repoman_popen("svn status --depth=files --verbose " +
+					portage._shell_quote(checkdir))
 			if vcs == "bzr":
-				myf = os.popen("bzr ls -v --kind=file " + checkdir)
+				myf = repoman_popen("bzr ls -v --kind=file " +
+					portage._shell_quote(checkdir))
 			myl = myf.readlines()
 			myf.close()
 			for l in myl:
@@ -1473,7 +1506,8 @@ for x in effective_scanlist:
 					if l[-7:] == ".ebuild":
 						eadded.append(os.path.basename(l[:-7]))
 			if vcs == "svn":
-				myf = os.popen("svn status " + checkdir)
+				myf = repoman_popen("svn status " +
+					portage._shell_quote(checkdir))
 				myl=myf.readlines()
 				myf.close()
 				for l in myl:
@@ -2318,7 +2352,7 @@ else:
 			err("Error retrieving CVS tree; exiting.")
 	if vcs == "svn":
 		try:
-			with os.popen("svn status --no-ignore") as f:
+			with repoman_popen("svn status --no-ignore") as f:
 				svnstatus = f.readlines()
 			myunadded = [ "./"+elem.rstrip().split()[1] for elem in svnstatus if elem.startswith("?") or elem.startswith("I") ]
 		except SystemExit as e:
@@ -2327,12 +2361,12 @@ else:
 			err("Error retrieving SVN info; exiting.")
 	if vcs == "git":
 		# get list of files not under version control or missing
-		myf = os.popen("git ls-files --others")
+		myf = repoman_popen("git ls-files --others")
 		myunadded = [ "./" + elem[:-1] for elem in myf ]
 		myf.close()
 	if vcs == "bzr":
 		try:
-			with os.popen("bzr status -S .") as f:
+			with repoman_popen("bzr status -S .") as f:
 				bzrstatus = f.readlines()
 			myunadded = [ "./"+elem.rstrip().split()[1].split('/')[-1:][0] for elem in bzrstatus if elem.startswith("?") or elem[0:2] == " D" ]
 		except SystemExit as e:
@@ -2340,14 +2374,14 @@ else:
 		except:
 			err("Error retrieving bzr info; exiting.")
 	if vcs == "hg":
-		with os.popen("hg status --no-status --unknown .") as f:
+		with repoman_popen("hg status --no-status --unknown .") as f:
 			myunadded = f.readlines()
 		myunadded = ["./" + elem.rstrip() for elem in myunadded]
 		
 		# Mercurial doesn't handle manually deleted files as removed from
 		# the repository, so the user need to remove them before commit,
 		# using "hg remove [FILES]"
-		with os.popen("hg status --no-status --deleted .") as f:
+		with repoman_popen("hg status --no-status --deleted .") as f:
 			mydeleted = f.readlines()
 		mydeleted = ["./" + elem.rstrip() for elem in mydeleted]
 
@@ -2393,36 +2427,36 @@ else:
 
 
 	if vcs == "svn":
-		with os.popen("svn status") as f:
+		with repoman_popen("svn status") as f:
 			svnstatus = f.readlines()
 		mychanged = [ "./" + elem.split()[-1:][0] for elem in svnstatus if (elem[:1] in "MR" or elem[1:2] in "M")]
 		mynew     = [ "./" + elem.split()[-1:][0] for elem in svnstatus if elem.startswith("A")]
 		myremoved = [ "./" + elem.split()[-1:][0] for elem in svnstatus if elem.startswith("D")]
 
 		# Subversion expands keywords specified in svn:keywords properties.
-		with os.popen("svn propget -R svn:keywords") as f:
+		with repoman_popen("svn propget -R svn:keywords") as f:
 			props = f.readlines()
 		expansion = dict(("./" + prop.split(" - ")[0], prop.split(" - ")[1].split()) \
 			for prop in props if " - " in prop)
 
 	elif vcs == "git":
-		with os.popen("git diff-index --name-only "
+		with repoman_popen("git diff-index --name-only "
 			"--relative --diff-filter=M HEAD") as f:
 			mychanged = f.readlines()
 		mychanged = ["./" + elem[:-1] for elem in mychanged]
 
-		with os.popen("git diff-index --name-only "
+		with repoman_popen("git diff-index --name-only "
 			"--relative --diff-filter=A HEAD") as f:
 			mynew = f.readlines()
 		mynew = ["./" + elem[:-1] for elem in mynew]
 
-		with os.popen("git diff-index --name-only "
+		with repoman_popen("git diff-index --name-only "
 			"--relative --diff-filter=D HEAD") as f:
 			myremoved = f.readlines()
 		myremoved = ["./" + elem[:-1] for elem in myremoved]
 
 	if vcs == "bzr":
-		with os.popen("bzr status -S .") as f:
+		with repoman_popen("bzr status -S .") as f:
 			bzrstatus = f.readlines()
 		mychanged = [ "./" + elem.split()[-1:][0].split('/')[-1:][0] for elem in bzrstatus if elem and elem[1:2] == "M" ]
 		mynew     = [ "./" + elem.split()[-1:][0].split('/')[-1:][0] for elem in bzrstatus if elem and ( elem[1:2] in "NK" or elem[0:1] == "R" ) ]
@@ -2431,15 +2465,15 @@ else:
 		# Bazaar expands nothing.
 
 	if vcs == "hg":
-		with os.popen("hg status --no-status --modified .") as f:
+		with repoman_popen("hg status --no-status --modified .") as f:
 			mychanged = f.readlines()
 		mychanged = ["./" + elem.rstrip() for elem in mychanged]
 
-		with os.popen("hg status --no-status --added .") as f:
+		with repoman_popen("hg status --no-status --added .") as f:
 			mynew = f.readlines()
 		mynew = ["./" + elem.rstrip() for elem in mynew]
 
-		with os.popen("hg status --no-status --removed .") as f:
+		with repoman_popen("hg status --no-status --removed .") as f:
 			myremoved = f.readlines()
 		myremoved = ["./" + elem.rstrip() for elem in myremoved]