[patch] merge-exclude.patch

merge-exclude.patch (text/plain), 6.95 KB, created by TGL on 2003-11-05 06:41:23 UTC

--- pym/portage.orig	2003-11-05 00:34:46.000000000 +0100
+++ pym/portage.py	2003-11-05 11:00:26.729303069 +0100
@@ -180,7 +180,8 @@
 	writemsg(red("*** Please add this user to the portage group if you wish to use portage.\n"))
 	writemsg("\n")
 
-incrementals=["USE","FEATURES","ACCEPT_KEYWORDS","ACCEPT_LICENSE","CONFIG_PROTECT_MASK","CONFIG_PROTECT","PRELINK_PATH","PRELINK_PATH_MASK"]
+#merge_exclude: $MERGE_EXCLUDE obviously should be incremental. Now, should it be sticky? I don't see why.
+incrementals=["USE","FEATURES","ACCEPT_KEYWORDS","ACCEPT_LICENSE","CONFIG_PROTECT_MASK","CONFIG_PROTECT","PRELINK_PATH","PRELINK_PATH_MASK","MERGE_EXCLUDE"]
 stickies=["KEYWORDS_ACCEPT","USE","CFLAGS","CXXFLAGS","MAKEOPTS","EXTRA_ECONF","EXTRA_EMAKE"]
 
 def getcwd():
@@ -4602,6 +4603,44 @@
 						masked=len(pmpath)
 		return (protected > masked)
 
+	#merge_exclude: create the global exclude paths list from config
+	# we append a "/" to all elements, including filenames, to avoid
+	# matching "/path/foobar" when "/path/foo" is the one to exclude
+	def updateexclude(self):
+		self.exclude=map((lambda p: os.path.normpath(self.myroot+"/"+p)+"/"), \
+				 string.split(settings["MERGE_EXCLUDE"]))
+
+	#merge_exclude: exclusion predicate
+	# for optimisation, we may use a partial exclusion list (optional third argument)
+	def isexcluded(self,obj,excl=None):
+		"""Checks if obj should be excluded from merge.
+		Optional parameter overides the global exclude lists.
+		Returns 1 if obj should be exclude, 0 otherwise."""
+		if excl is None:
+			excl=self.exclude
+		myobj=os.path.normpath(obj)+"/"
+		excluded=reduce((lambda r,p: (r or ((len(myobj) >= len(p)) \
+				                    and (myobj[0:len(p)]==p)))), \
+				excl, 0)
+		return excluded
+
+	#merge_exclude: this is an utility function for optimisation
+	# we will consider only paths that are relevant for the currently being merged directory,
+	def directsubpaths(self,pathslist,path):
+		"""Given a list of paths (normpathed with a trailing slash) l, and a path p,
+		returns the list of paths from l that are direct strict subpaths of p"""
+		mypath=os.path.normpath(path)
+		mynewpathslist=filter((lambda p: (mypath == os.path.split(p[:-1])[0])), pathslist)
+		return mynewpathslist
+	#merge_exclude: and this is another optimisation utility function
+	# Recursing through the directories to merge, we will only keep part of the exclude list
+	def subpaths(self,pathslist,path):
+		"""Given a list of paths (normpathed with a trailing slash) l, and a path p,
+		returns the list of paths from l that are strict subpaths of p"""
+		mypath=os.path.normpath(path)+"/"
+		mynewpathslist=filter((lambda p: ((len(p) > len(mypath)) and (p[0:len(mypath)]==mypath))), pathslist)
+		return mynewpathslist
+
 	def unmerge(self,pkgfiles=None,trimworld=1):
 		global dircache
 		dircache={}
@@ -4869,6 +4908,8 @@
 		outfile=open(inforoot+"/CONTENTS","w")
 
 		self.updateprotect()
+		#merge_exclude: set/update the global exclude paths list.
+		self.updateexclude()	
 
 		#if we have a file containing previously-merged config file md5sums, grab it.
 		if os.path.exists(destroot+"/var/cache/edb/config"):
@@ -4895,7 +4936,8 @@
 			# clear the thirdhand.	Anything from our second hand that couldn't get merged will be
 			# added to thirdhand.
 			thirdhand=[]
-			self.mergeme(srcroot,destroot,outfile,thirdhand,secondhand,cfgfiledict,mymtime)
+			#merge_exclude: their should not be anything to exclude here.
+			self.mergeme(srcroot,destroot,outfile,thirdhand,secondhand,cfgfiledict,mymtime,[])
 			#swap hands
 			lastlen=len(secondhand)
 			# our thirdhand now becomes our secondhand.  It's ok to throw away secondhand since 
@@ -4903,8 +4945,11 @@
 			secondhand=thirdhand
 		if len(secondhand):
 			# force merge of remaining symlinks (broken or circular; oh well)
-			self.mergeme(srcroot,destroot,outfile,None,secondhand,cfgfiledict,mymtime)
-		
+			#merge_exclude: their should not be anything to exclude here.
+			# Note that symlinks to excluded files may be merged, despite they are broken.
+			# I see this as an acceptable behavior, but I could add some checks for this case in mergeme if needed.
+			self.mergeme(srcroot,destroot,outfile,None,secondhand,cfgfiledict,mymtime,[])
+
 		#restore umask
 		os.umask(prevmask)
 		#if we opened it, close it	
@@ -5014,8 +5059,14 @@
 			return new_pfile
 		else:
 			return (new_pfile, old_pfile)
-		
-	def mergeme(self,srcroot,destroot,outfile,secondhand,stufftomerge,cfgfiledict,thismtime):
+	
+	#merge_exclude: added an optional argument which is the list of paths to exclude.
+	# At first, I was using this for an other optimisation, which no more makes sense with "directsubpaths".
+	# I keep it here only for giving an empty list in case of secondhand merge.
+	def mergeme(self,srcroot,destroot,outfile,secondhand,stufftomerge,cfgfiledict,thismtime,excl=None):
+		#merge_exclude: give its default value to the exclude paths list if needed
+		if excl is None:
+			excl=self.exclude
 		srcroot=os.path.normpath("///"+srcroot)+"/"
 		destroot=os.path.normpath("///"+destroot)+"/"
 		# this is supposed to merge a list of files.  There will be 2 forms of argument passing.
@@ -5028,7 +5079,12 @@
 			# Trailing / ensures that protects/masks with trailing /'s match.
 			mytruncpath="/"+offset+"/"
 			myppath=self.isprotected(mytruncpath)
+			#merge_exclude: we reduce or exclude paths list to direct subpaths of current directory
+			myexclude=self.directsubpaths(excl,mytruncpath)
 		else:
+			#merge_exclude: this can only occur for "secondhand" merges, right?
+			# so there are no more optimisation to do, secondhand comes with excl=[].
+			myexclude=excl
 			mergelist=stufftomerge
 			offset=""
 		for x in mergelist:
@@ -5036,6 +5090,17 @@
 			mydest=os.path.normpath("///"+destroot+offset+x)
 			# myrealdest is mydest without the $ROOT prefix (makes a difference if ROOT!="/")
 			myrealdest="/"+offset+x
+
+			#merge_exclude: If a destination path is to be excluded, from a syntactical point of view, 
+			# then exclude it. That's all we will do for now.
+			#   Another option would be to take care of symlinks, ie. to forbid merging into existing 
+			# symlinks pointing to existing excluded dir for instance... Not sure it how useful it
+			# would, will do it later if I find a good reason to.
+			#   Anyway, in both cases, output will be "### /path/to/obj" 
+			if self.isexcluded(mydest,myexclude):
+				print "###",mydest
+				continue
+
 			# stat file once, test using S_* macros many times (faster that way)
 			try:
 				mystat=os.lstat(mysrc)
@@ -5143,7 +5208,9 @@
 					print ">>>",mydest+"/"
 				outfile.write("dir "+myrealdest+"\n")
 				# recurse and merge this directory
-				if self.mergeme(srcroot,destroot,outfile,secondhand,offset+x+"/",cfgfiledict,thismtime):
+				#merge_exclude: for this subdir, we only need part of the exclude list
+				mysubexcl=self.subpaths(excl,mydest)
+				if self.mergeme(srcroot,destroot,outfile,secondhand,offset+x+"/",cfgfiledict,thismtime,mysubexcl):
 					return 1
 			elif S_ISREG(mymode):
 				# we are merging a regular file