[gentoo-commits] portage r13922 - in main/branches/prefix: . bin pym/_emerge pym/portage pym/portage/dbapi pym/portage/sets

[Fabian Groffen (grobian)]

Author: grobian
Date: 2009-08-05 18:10:06 +0000 (Wed, 05 Aug 2009)
New Revision: 13922

Modified:
   main/branches/prefix/bin/portageq
   main/branches/prefix/pym/_emerge/Scheduler.py
   main/branches/prefix/pym/_emerge/main.py
   main/branches/prefix/pym/portage/__init__.py
   main/branches/prefix/pym/portage/dbapi/vartree.py
   main/branches/prefix/pym/portage/dep.py
   main/branches/prefix/pym/portage/sets/files.py
   main/branches/prefix/pym/portage/util.py
   main/branches/prefix/subst-install.in
Log:
   Merged from trunk -r13895:13909

   | 13898   | Bug #280275 - Fix exception: AttributeError: 'config' object |
   | zmedico | has no attribute '_license_groups'.                          |
   
   | 13900   | Avoid UnicodeEncodeError in writemsg(). Thanks to Markos     |
   | zmedico | Chandras <hwoarang@g.o> for reporting.                |
   
   | 13901   | Open log files in text mode (to avoid UnicodeEncodeError).   |
   | zmedico | Thanks to Markos Chandras <hwoarang@g.o> for          |
   |         | reporting.                                                   |
   
   | 13902   | Open CONTENTS in text mode (to avoid UnicodeEncodeError).    |
   | zmedico | Thanks to Markos Chandras <hwoarang@g.o> for          |
   |         | reporting.                                                   |
   
   | 13903   | Bug #280269 - Fix StaticFileSet.multiBuilder() to handle     |
   | zmedico | unicode filenames.                                           |
   
   | 13904   | Bug #280269 - Decode commandline arguments to unicode when   |
   | zmedico | necessary.                                                   |
   
   | 13905   | Bug #280269 - Fix Atom.__str__ so that it doesn't try to     |
   | zmedico | encode a unicode string (resulting in UnicodeEncodeError).   |
   |         | If an Atom instance is passed into the constructor, just     |
   |         | return the given instance.                                   |
   
   | 13906   | Decode arguments to unicode if necessary (needed at least    |
   | zmedico | for unicode file names passed to the owners command).        |
   
   | 13907   | Always return unicode from vardbapi.aux_get().               |
   | zmedico |                                                              |
   
   | 13908   | Fix tar_contents() to open files in binary mode for py3k     |
   | zmedico | compatibility.                                               |
   
   | 13909   | Fix vardbapi._aux_get() to always return unicode.            |
   | zmedico |                                                              |


Modified: main/branches/prefix/bin/portageq
===================================================================
--- main/branches/prefix/bin/portageq	2009-08-05 18:07:45 UTC (rev 13921)
+++ main/branches/prefix/bin/portageq	2009-08-05 18:10:06 UTC (rev 13922)
@@ -606,10 +606,15 @@
 	sys.path.insert(0, pym_path)
 	import portage
 
+	args = sys.argv[2:]
+	if args and not isinstance(args[0], unicode):
+		for i in xrange(len(args)):
+			args[i] = unicode(args[i], encoding='utf_8', errors='replace')
+
 	try:
 		if uses_root:
-			sys.argv[2] = portage.settings["ROOT"]
-		retval = function(sys.argv[2:])
+			args[0] = portage.settings["ROOT"]
+		retval = function(args)
 		if retval:
 			sys.exit(retval)
 	except portage.exception.PermissionDenied, e:

Modified: main/branches/prefix/pym/_emerge/Scheduler.py
===================================================================
--- main/branches/prefix/pym/_emerge/Scheduler.py	2009-08-05 18:07:45 UTC (rev 13921)
+++ main/branches/prefix/pym/_emerge/Scheduler.py	2009-08-05 18:10:06 UTC (rev 13922)
@@ -2,6 +2,7 @@
 # Distributed under the terms of the GNU General Public License v2
 # $Id$
 
+import codecs
 import logging
 import os
 import sys
@@ -478,7 +479,8 @@
 		return self._pkg(cpv, type_name, root_config, installed=installed)
 
 	def _append_to_log_path(self, log_path, msg):
-		f = open(log_path, 'a')
+		f = codecs.open(log_path, mode='a',
+			encoding='utf_8', errors='replace')
 		try:
 			f.write(msg)
 		finally:
@@ -492,7 +494,8 @@
 		background = self._background
 
 		if background and log_path is not None:
-			log_file = open(log_path, 'a')
+			log_file = codecs.open(log_path, mode='a',
+				encoding='utf_8', errors='replace')
 			out = log_file
 
 		try:
@@ -867,7 +870,8 @@
 			log_path = self._locate_failure_log(failed_pkg)
 			if log_path is not None:
 				try:
-					log_file = open(log_path)
+					log_file = codecs.open(log_path, mode='r',
+						encoding='utf_8', errors='replace')
 				except IOError:
 					pass
 

Modified: main/branches/prefix/pym/_emerge/main.py
===================================================================
--- main/branches/prefix/pym/_emerge/main.py	2009-08-05 18:07:45 UTC (rev 13921)
+++ main/branches/prefix/pym/_emerge/main.py	2009-08-05 18:10:06 UTC (rev 13922)
@@ -760,6 +760,10 @@
 	if myaction is None and myoptions.deselect is True:
 		myaction = 'deselect'
 
+	if myargs and not isinstance(myargs[0], unicode):
+		for i in xrange(len(myargs)):
+			myargs[i] = unicode(myargs[i], encoding='utf_8', errors='replace')
+
 	myfiles += myargs
 
 	return myaction, myopts, myfiles

Modified: main/branches/prefix/pym/portage/__init__.py
===================================================================
--- main/branches/prefix/pym/portage/__init__.py	2009-08-05 18:07:45 UTC (rev 13921)
+++ main/branches/prefix/pym/portage/__init__.py	2009-08-05 18:10:06 UTC (rev 13922)
@@ -1212,6 +1212,7 @@
 		self._accept_chost_re = None
 		self._accept_license = None
 		self._accept_license_str = None
+		self._license_groups = {}
 
 		self.virtuals = {}
 		self.virts_p = {}
@@ -1297,6 +1298,7 @@
 
 			self._accept_license = copy.deepcopy(clone._accept_license)
 			self._plicensedict = copy.deepcopy(clone._plicensedict)
+			self._license_groups = copy.deepcopy(clone._license_groups)
 		else:
 
 			def check_var_directory(varname, var):
@@ -1793,7 +1795,6 @@
 					self.pprovideddict[mycatpkg]=[x]
 
 			# parse licensegroups
-			self._license_groups = {}
 			for x in locations:
 				self._license_groups.update(
 					grabdict(os.path.join(x, "license_groups")))

Modified: main/branches/prefix/pym/portage/dbapi/vartree.py
===================================================================
--- main/branches/prefix/pym/portage/dbapi/vartree.py	2009-08-05 18:07:45 UTC (rev 13921)
+++ main/branches/prefix/pym/portage/dbapi/vartree.py	2009-08-05 18:10:06 UTC (rev 13922)
@@ -38,6 +38,7 @@
 
 from portage.cache.mappings import slot_dict_class
 
+import codecs
 import os, re, shutil, stat, errno, copy, subprocess
 import logging
 import shlex
@@ -1866,6 +1867,12 @@
 			cache_mtime, metadata = pkg_data
 			cache_valid = cache_mtime == mydir_mtime
 		if cache_valid:
+			for k, v in metadata.iteritems():
+				if not isinstance(v, unicode):
+					# Migrate old metadata to unicode.
+					metadata[k] = unicode(v,
+						encoding='utf_8', errors='replace')
+
 			mydata.update(metadata)
 			pull_me.difference_update(mydata)
 
@@ -1886,7 +1893,7 @@
 		if not mydata['SLOT']:
 			# Empty slot triggers InvalidAtom exceptions when generating slot
 			# atoms for packages, so translate it to '0' here.
-			mydata['SLOT'] = '0'
+			mydata['SLOT'] = u'0'
 		return [mydata[x] for x in wants]
 
 	def _aux_get(self, mycpv, wants, st=None):
@@ -1909,7 +1916,8 @@
 				results.append(long(st.st_mtime))
 				continue
 			try:
-				myf = open(os.path.join(mydir, x), "r")
+				myf = codecs.open(os.path.join(mydir, x),
+					mode='r', encoding='utf_8', errors='replace')
 				try:
 					myd = myf.read()
 				finally:
@@ -1919,9 +1927,9 @@
 				if self._aux_multi_line_re.match(x) is None:
 					myd = " ".join(myd.split())
 			except IOError:
-				myd = ""
+				myd = u''
 			if x == "EAPI" and not myd:
-				results.append("0")
+				results.append(u'0')
 			else:
 				results.append(myd)
 		return results
@@ -2558,7 +2566,8 @@
 			return self.contentscache
 		pkgfiles = {}
 		try:
-			myc = open(contents_file,"r")
+			myc = codecs.open(contents_file, mode='r',
+				encoding='utf_8', errors='replace')
 		except EnvironmentError, e:
 			if e.errno != errno.ENOENT:
 				raise
@@ -4141,7 +4150,8 @@
 		lcfile.close()
 
 		# open CONTENTS file (possibly overwriting old one) for recording
-		outfile = open(os.path.join(self.dbtmpdir, "CONTENTS"),"w")
+		outfile = codecs.open(os.path.join(self.dbtmpdir, 'CONTENTS'),
+			mode='w', encoding='utf_8', errors='replace')
 
 		self.updateprotect()
 
@@ -4846,7 +4856,7 @@
 				tarinfo.size = 0
 				tar.addfile(tarinfo)
 			else:
-				f = open(path)
+				f = open(path, 'rb')
 				try:
 					tar.addfile(tarinfo, f)
 				finally:

Modified: main/branches/prefix/pym/portage/dep.py
===================================================================
--- main/branches/prefix/pym/portage/dep.py	2009-08-05 18:07:45 UTC (rev 13921)
+++ main/branches/prefix/pym/portage/dep.py	2009-08-05 18:10:06 UTC (rev 13922)
@@ -494,6 +494,8 @@
 	identical instances when available.
 	"""
 	def __call__(cls, s):
+		if isinstance(s, Atom):
+			return s
 		instance = cls._atoms.get(s)
 		if instance is None:
 			instance = super(_AtomCache, cls).__call__(s)
@@ -620,7 +622,7 @@
 		return repr(self._str)
 
 	def __str__(self):
-		return str(self._str)
+		return self._str
 
 	def endswith(self, *pargs, **kargs):
 		return self._str.endswith(*pargs, **kargs)

Modified: main/branches/prefix/pym/portage/sets/files.py
===================================================================
--- main/branches/prefix/pym/portage/sets/files.py	2009-08-05 18:07:45 UTC (rev 13921)
+++ main/branches/prefix/pym/portage/sets/files.py	2009-08-05 18:10:06 UTC (rev 13922)
@@ -131,6 +131,9 @@
 					if d[:1] == '.':
 						dirs.remove(d)
 				for filename in files:
+					if not isinstance(filename, unicode):
+						filename = unicode(filename,
+							encoding='utf_8', errors='replace')
 					if filename[:1] == '.':
 						continue
 					if filename.endswith(".metadata"):

Modified: main/branches/prefix/pym/portage/util.py
===================================================================
--- main/branches/prefix/pym/portage/util.py	2009-08-05 18:07:45 UTC (rev 13921)
+++ main/branches/prefix/pym/portage/util.py	2009-08-05 18:10:06 UTC (rev 13922)
@@ -57,6 +57,9 @@
 	if fd is None:
 		fd = sys.stderr
 	if noiselevel <= noiselimit:
+		if sys.hexversion < 0x3000000 and isinstance(mystr, unicode):
+			# avoid potential UnicodeEncodeError
+			mystr = mystr.encode('utf_8', 'replace')
 		fd.write(mystr)
 		fd.flush()
 

Modified: main/branches/prefix/subst-install.in
===================================================================
--- main/branches/prefix/subst-install.in	2009-08-05 18:07:45 UTC (rev 13921)
+++ main/branches/prefix/subst-install.in	2009-08-05 18:10:06 UTC (rev 13922)
@@ -4,15 +4,17 @@
 prefix="@prefix@"
 exec_prefix="@exec_prefix@"
 
-# for bug #279550 we have to do some nasty trick to make sure that sed
+# For bug #279550 we have to do some nasty trick to make sure that sed
 # doesn't strip the backslash in the replacement value (because it can
-# be a backreference) and hence escape those.
+# be a backreference) and hence escape those.  Eventually in strings we
+# need to escape the backslash too, such that the single backslash
+# doesn't get lost when considered an invalid escape
 rootuser="@rootuser@"
 portagegroup="@portagegroup@"
 portageuser="@portageuser@"
 rootuser=${rootuser//\\/\\\\}
-portagegroup=${portagegroup//\\/\\\\}
-portageuser=${portageuser//\\/\\\\}
+portagegroup=${portagegroup//\\/\\\\\\\\}
+portageuser=${portageuser//\\/\\\\\\\\}
 
 # there are many ways to do this all dynamic, but we only care for raw
 # speed here, so let configure fill in this list and be done with it