Added statistics.py for more sophisticated statistics handling

git-svn-id: http://svn.savannah.nongnu.org/svn/rdiff-backup/trunk@104 2b77aa54-bcbc-44c9-a7ec-4f6cf2b41109

5 files changed, 323 insertions, 138 deletions

diff --git a/rdiff-backup/rdiff_backup/increment.py b/rdiff-backup/rdiff_backup/increment.py
index 4e96e59..499d709 100644
--- a/rdiff-backup/rdiff_backup/increment.py
+++ b/rdiff-backup/rdiff_backup/increment.py
@@ -1,4 +1,4 @@
-execfile("filename_mapping.py")
+execfile("statistics.py")
 
 #######################################################################
 #
@@ -108,7 +108,7 @@ class Inc:
 MakeStatic(Inc)
 
 
-class IncrementITR(IterTreeReducer):
+class IncrementITR(StatsITR):
 	"""Patch and increment iterator of increment triples
 
 	This has to be an ITR because directories that have files in them
@@ -127,13 +127,13 @@ class IncrementITR(IterTreeReducer):
 	Remember this object needs to be pickable.
 	
 	"""
-	directory, directory_replacement = None, None
+	mirror_isdirectory, directory_replacement = None, None
 	changed = None
 
 	def __init__(self, inc_rpath):
 		"""Set inc_rpath, an rpath of the base of the tree"""
 		self.inc_rpath = inc_rpath
-		IterTreeReducer.__init__(self, inc_rpath)
+		StatsITR.__init__(self, inc_rpath)
 
 	def start_process(self, index, diff_rorp, dsrp):
 		"""Initial processing of file
@@ -142,34 +142,15 @@ class IncrementITR(IterTreeReducer):
 		dsrp is the local file to be incremented
 
 		"""
-		self.init_statistics(diff_rorp, dsrp)
+		self.start_stats(dsrp)
 		incpref = self.inc_rpath.new_index(index)
 		if Globals.quoting_enabled: incpref.quote_path()
 		if dsrp.isdir():
 			self.init_dir(dsrp, diff_rorp, incpref)
-			self.setvals(diff_rorp, dsrp, incpref)
+			self.mirror_isdirectory = 1
 		else: self.init_non_dir(dsrp, diff_rorp, incpref)
-
-	def init_statistics(self, diff_rorp, dsrp):
-		"""Set initial values for various statistics
-
-		These refer to the old mirror or to new increment files.  Note
-		that changed_file_size could be bigger than total_file_size.
-		The other statistic, increment_file_size, is set later when we
-		have that information.
-
-		"""
-		if dsrp.lstat():
-			self.total_files = 1
-			self.total_file_size = dsrp.getsize()
-		else: self.total_files = self.total_file_size = 0
-		if diff_rorp:
-			self.changed_files = 1
-			if dsrp.lstat(): self.changed_file_size = dsrp.getsize()
-			else: self.changed_file_size = 0
-		else: self.changed_files = self.changed_file_size = 0
-		self.increment_file_size = 0
-
+		self.setvals(diff_rorp, dsrp, incpref)
+		
 	def override_changed(self):
 		"""Set changed flag to true
 
@@ -187,7 +168,6 @@ class IncrementITR(IterTreeReducer):
 		superclass.
 
 		"""
-		self.directory = 1
 		self.diff_rorp = diff_rorp
 		self.dsrp = dsrp
 		self.incpref = incpref
@@ -224,53 +204,32 @@ class IncrementITR(IterTreeReducer):
 			Robust.chain([Inc.Increment_action(diff_rorp, dsrp, incpref),
 						  RORPIter.patchonce_action(None, dsrp, diff_rorp)]
 						 ).execute()
-
-		self.increment_file_size += ((Inc._inc_file and Inc._inc_file.lstat()
-									  and Inc._inc_file.getsize()) or 0)
 		self.changed = 1
 
 	def end_process(self):
 		"""Do final work when leaving a tree (directory)"""
-		if not self.directory: return
 		diff_rorp, dsrp, incpref = self.diff_rorp, self.dsrp, self.incpref
-		if not diff_rorp and not self.changed: return
-
-		if self.directory_replacement:
-			tf = self.directory_replacement
-			Inc.Increment(tf, dsrp, incpref)
-			RORPIter.patchonce_action(None, dsrp, tf).execute()
-			tf.delete()
-		else:
-			Inc.Increment(diff_rorp, dsrp, incpref)
-			if diff_rorp:
-				RORPIter.patchonce_action(None, dsrp, diff_rorp).execute()
-
-		self.increment_file_size += ((Inc._inc_file and Inc._inc_file.lstat()
-									  and Inc._inc_file.getsize()) or 0)
-		self.write_statistics()
-
-	def write_statistics(self):
-		"""Write the accumulated totals into file in inc directory"""
-		if not self.incpref.isdir(): return # only write for directories
-		statrp = Inc.get_inc_ext(self.incpref.append("directory_statistics"),
-								 "data")
-		tf = TempFileManager.new(statrp)
-		def init_thunk():
-			fp = tf.open("w")
-			fp.write("TotalFiles %d\n" % self.total_files)
-			fp.write("TotalFileSize %d\n" % self.total_file_size)
-			fp.write("ChangedFiles %d\n" % self.changed_files)
-			fp.write("ChangedFileSize %d\n" % self.changed_file_size)
-			fp.write("IncrementFileSize %d\n" % self.increment_file_size)
-			fp.close()
-		Robust.make_tf_robustaction(init_thunk, (tf,), (statrp,)).execute()
+		if self.mirror_isdirectory:
+			if not diff_rorp and not self.changed: return
+
+			if self.directory_replacement:
+				tf = self.directory_replacement
+				Inc.Increment(tf, dsrp, incpref)
+				RORPIter.patchonce_action(None, dsrp, tf).execute()
+				tf.delete()
+			else:
+				Inc.Increment(diff_rorp, dsrp, incpref)
+				if diff_rorp:
+					RORPIter.patchonce_action(None, dsrp, diff_rorp).execute()
+
+		self.end_stats(diff_rorp, dsrp, Inc._inc_file)
+		if self.incpref.isdir() and (self.mirror_isdirectory or dsrp.isdir()):
+			self.write_stats_to_rp(Inc.get_inc_ext(
+				self.incpref.append("directory_statistics"), "data"))
 
 	def branch_process(self, subinstance):
 		"""Update statistics, and the has_changed flag if change in branch"""
 		if subinstance.changed: self.changed = 1	
+		self.add_file_stats(subinstance)
+
 
-		self.total_files += subinstance.total_files
-		self.total_file_size += subinstance.total_file_size
-		self.changed_files += subinstance.changed_files
-		self.changed_file_size += subinstance.changed_file_size
-		self.increment_file_size += subinstance.increment_file_size
diff --git a/rdiff-backup/rdiff_backup/statistics.py b/rdiff-backup/rdiff_backup/statistics.py
new file mode 100644
index 0000000..a121591
--- /dev/null
+++ b/rdiff-backup/rdiff_backup/statistics.py
@@ -0,0 +1,133 @@
+execfile("filename_mapping.py")
+
+#######################################################################
+#
+# statistics - Generate and process aggregated backup information
+#
+
+class StatsException(Exception): pass
+
+class StatsObj:
+	"""Contains various statistics, provide string conversion functions"""
+
+	stat_file_attrs = ('SourceFiles', 'SourceFileSize',
+					   'MirrorFiles', 'MirrorFileSize',
+					   'NewFiles', 'NewFileSize',
+					   'DeletedFiles', 'DeletedFileSize',
+					   'ChangedFiles',
+					   'ChangedSourceSize', 'ChangedMirrorSize',
+					   'IncrementFileSize')
+	stat_time_attrs = ('StartTime', 'EndTime')
+	stat_attrs = stat_time_attrs + stat_file_attrs
+
+	# Set all stats to None, indicating info not available
+	for attr in stat_attrs: locals()[attr] = None
+
+	def get_stat(self, attribute):
+		"""Get a statistic"""
+		try: return self.__dict__[attribute]
+		except KeyError:
+			# this may be a hack, but seems no good way to get attrs in python
+			return eval("self.%s" % attribute)
+
+	def set_stat(self, attr, value):
+		"""Set attribute to given value"""
+		self.__dict__[attr] = value
+
+	def get_stats_string(self):
+		"""Return string printing out statistics"""
+		slist = ["%s %s" % (attr, self.get_stat(attr))
+				 for attr in self.stat_attrs
+				 if self.get_stat(attr) is not None]
+		return "\n".join(slist)
+
+	def init_stats_from_string(self, s):
+		"""Initialize attributes from string, return self for convenience"""
+		def error(line): raise StatsException("Bad line '%s'" % line)
+
+		for line in s.split("\n"):
+			if not line: continue
+			line_parts = line.split()
+			if len(line_parts) < 2: error(line)
+			attr, value_string = line_parts[:2]
+			if not attr in self.stat_attrs: error(line)
+			try: self.set_stat(attr, long(value_string))
+			except ValueError: error(line)
+		return self
+
+	def write_stats_to_rp(self, rp):
+		"""Write statistics string to given rpath"""
+		tf = TempFileManager.new(rp)
+		def init_thunk():
+			fp = tf.open("w")
+			fp.write(self.get_stats_string())
+			fp.close()
+		Robust.make_tf_robustaction(init_thunk, (tf,), (rp,)).execute()
+
+	def read_stats_from_rp(self, rp):
+		"""Set statistics from rpath, return self for convenience"""
+		fp = rp.open("r")
+		self.init_stats_from_string(fp.read())
+		fp.close()
+		return self
+
+	def stats_equal(self, s):
+		"""Return true if s has same statistics as self"""
+		assert isinstance(s, StatsObj)
+		for attr in self.stat_file_attrs:
+			if self.get_stat(attr) != s.get_stat(attr): return None
+		return 1
+
+
+class StatsITR(IterTreeReducer, StatsObj):
+	"""Keep track of per directory statistics
+
+	This is subclassed by the mirroring and incrementing ITRs.
+
+	"""
+	# zero out file statistics
+	for attr in StatsObj.stat_file_attrs: locals()[attr] = 0
+
+	def start_stats(self, mirror_dsrp):
+		"""Record status of mirror dsrp
+
+		This is called before the mirror is processed so we remember
+		the old state.
+
+		"""
+		if mirror_dsrp.lstat():
+			self.mirror_base_exists = 1
+			self.mirror_base_size = mirror_dsrp.getsize()
+		else: self.mirror_base_exists = None
+
+	def end_stats(self, diff_rorp, mirror_dsrp, inc_rp = None):
+		"""Set various statistics after mirror processed"""
+		if mirror_dsrp.lstat():
+			self.SourceFiles += 1
+			self.SourceFileSize += mirror_dsrp.getsize()
+			if self.mirror_base_exists:
+				self.MirrorFiles += 1
+				self.MirrorFileSize += self.mirror_base_size
+				if diff_rorp: # otherwise no change
+					self.ChangedFiles += 1
+					self.ChangedSourceSize += mirror_dsrp.getsize()
+					self.ChangedMirrorSize += self.mirror_base_size
+					self.IncrementFileSize += inc_rp and inc_rp.getsize() or 0
+			else: # new file was created
+				self.NewFiles += 1
+				self.NewFileSize += mirror_dsrp.getsize()
+				self.IncrementFileSize += inc_rp and inc_rp.getsize() or 0
+		else:
+			if self.mirror_base_exists: # file was deleted from mirror
+				self.MirrorFiles += 1
+				self.MirrorFileSize += self.mirror_base_size
+				self.DeletedFiles += 1
+				self.DeletedFileSize += self.mirror_base_size
+				self.IncrementFileSize += inc_rp and inc_rp.getsize() or 0
+			else: assert None # One of before and after should exist
+
+	def add_file_stats(self, subinstance):
+		"""Add all file statistics from subinstance to current totals"""
+		for attr in self.stat_file_attrs:
+			self.set_stat(attr,
+						  self.get_stat(attr) + subinstance.get_stat(attr))
diff --git a/rdiff-backup/src/Make b/rdiff-backup/src/Make
index b78ed95..2b79ffe 100755
--- a/rdiff-backup/src/Make
+++ b/rdiff-backup/src/Make
@@ -24,8 +24,9 @@ files = ["globals.py", "static.py", "lazy.py", "log.py", "ttime.py",
 		 "iterfile.py", "rdiff.py", "connection.py", "rpath.py",
 		 "hardlink.py", "robust.py", "rorpiter.py",
 		 "destructive_stepping.py", "selection.py",
-		 "filename_mapping.py", "increment.py", "restore.py",
-		 "manage.py", "highlevel.py", "setconnections.py", "main.py"]
+		 "filename_mapping.py", "statistics.py", "increment.py",
+		 "restore.py", "manage.py", "highlevel.py",
+		 "setconnections.py", "main.py"]
 
 os.system("cp header.py rdiff-backup")
 
diff --git a/rdiff-backup/src/increment.py b/rdiff-backup/src/increment.py
index 4e96e59..499d709 100644
--- a/rdiff-backup/src/increment.py
+++ b/rdiff-backup/src/increment.py
@@ -1,4 +1,4 @@
-execfile("filename_mapping.py")
+execfile("statistics.py")
 
 #######################################################################
 #
@@ -108,7 +108,7 @@ class Inc:
 MakeStatic(Inc)
 
 
-class IncrementITR(IterTreeReducer):
+class IncrementITR(StatsITR):
 	"""Patch and increment iterator of increment triples
 
 	This has to be an ITR because directories that have files in them
@@ -127,13 +127,13 @@ class IncrementITR(IterTreeReducer):
 	Remember this object needs to be pickable.
 	
 	"""
-	directory, directory_replacement = None, None
+	mirror_isdirectory, directory_replacement = None, None
 	changed = None
 
 	def __init__(self, inc_rpath):
 		"""Set inc_rpath, an rpath of the base of the tree"""
 		self.inc_rpath = inc_rpath
-		IterTreeReducer.__init__(self, inc_rpath)
+		StatsITR.__init__(self, inc_rpath)
 
 	def start_process(self, index, diff_rorp, dsrp):
 		"""Initial processing of file
@@ -142,34 +142,15 @@ class IncrementITR(IterTreeReducer):
 		dsrp is the local file to be incremented
 
 		"""
-		self.init_statistics(diff_rorp, dsrp)
+		self.start_stats(dsrp)
 		incpref = self.inc_rpath.new_index(index)
 		if Globals.quoting_enabled: incpref.quote_path()
 		if dsrp.isdir():
 			self.init_dir(dsrp, diff_rorp, incpref)
-			self.setvals(diff_rorp, dsrp, incpref)
+			self.mirror_isdirectory = 1
 		else: self.init_non_dir(dsrp, diff_rorp, incpref)
-
-	def init_statistics(self, diff_rorp, dsrp):
-		"""Set initial values for various statistics
-
-		These refer to the old mirror or to new increment files.  Note
-		that changed_file_size could be bigger than total_file_size.
-		The other statistic, increment_file_size, is set later when we
-		have that information.
-
-		"""
-		if dsrp.lstat():
-			self.total_files = 1
-			self.total_file_size = dsrp.getsize()
-		else: self.total_files = self.total_file_size = 0
-		if diff_rorp:
-			self.changed_files = 1
-			if dsrp.lstat(): self.changed_file_size = dsrp.getsize()
-			else: self.changed_file_size = 0
-		else: self.changed_files = self.changed_file_size = 0
-		self.increment_file_size = 0
-
+		self.setvals(diff_rorp, dsrp, incpref)
+		
 	def override_changed(self):
 		"""Set changed flag to true
 
@@ -187,7 +168,6 @@ class IncrementITR(IterTreeReducer):
 		superclass.
 
 		"""
-		self.directory = 1
 		self.diff_rorp = diff_rorp
 		self.dsrp = dsrp
 		self.incpref = incpref
@@ -224,53 +204,32 @@ class IncrementITR(IterTreeReducer):
 			Robust.chain([Inc.Increment_action(diff_rorp, dsrp, incpref),
 						  RORPIter.patchonce_action(None, dsrp, diff_rorp)]
 						 ).execute()
-
-		self.increment_file_size += ((Inc._inc_file and Inc._inc_file.lstat()
-									  and Inc._inc_file.getsize()) or 0)
 		self.changed = 1
 
 	def end_process(self):
 		"""Do final work when leaving a tree (directory)"""
-		if not self.directory: return
 		diff_rorp, dsrp, incpref = self.diff_rorp, self.dsrp, self.incpref
-		if not diff_rorp and not self.changed: return
-
-		if self.directory_replacement:
-			tf = self.directory_replacement
-			Inc.Increment(tf, dsrp, incpref)
-			RORPIter.patchonce_action(None, dsrp, tf).execute()
-			tf.delete()
-		else:
-			Inc.Increment(diff_rorp, dsrp, incpref)
-			if diff_rorp:
-				RORPIter.patchonce_action(None, dsrp, diff_rorp).execute()
-
-		self.increment_file_size += ((Inc._inc_file and Inc._inc_file.lstat()
-									  and Inc._inc_file.getsize()) or 0)
-		self.write_statistics()
-
-	def write_statistics(self):
-		"""Write the accumulated totals into file in inc directory"""
-		if not self.incpref.isdir(): return # only write for directories
-		statrp = Inc.get_inc_ext(self.incpref.append("directory_statistics"),
-								 "data")
-		tf = TempFileManager.new(statrp)
-		def init_thunk():
-			fp = tf.open("w")
-			fp.write("TotalFiles %d\n" % self.total_files)
-			fp.write("TotalFileSize %d\n" % self.total_file_size)
-			fp.write("ChangedFiles %d\n" % self.changed_files)
-			fp.write("ChangedFileSize %d\n" % self.changed_file_size)
-			fp.write("IncrementFileSize %d\n" % self.increment_file_size)
-			fp.close()
-		Robust.make_tf_robustaction(init_thunk, (tf,), (statrp,)).execute()
+		if self.mirror_isdirectory:
+			if not diff_rorp and not self.changed: return
+
+			if self.directory_replacement:
+				tf = self.directory_replacement
+				Inc.Increment(tf, dsrp, incpref)
+				RORPIter.patchonce_action(None, dsrp, tf).execute()
+				tf.delete()
+			else:
+				Inc.Increment(diff_rorp, dsrp, incpref)
+				if diff_rorp:
+					RORPIter.patchonce_action(None, dsrp, diff_rorp).execute()
+
+		self.end_stats(diff_rorp, dsrp, Inc._inc_file)
+		if self.incpref.isdir() and (self.mirror_isdirectory or dsrp.isdir()):
+			self.write_stats_to_rp(Inc.get_inc_ext(
+				self.incpref.append("directory_statistics"), "data"))
 
 	def branch_process(self, subinstance):
 		"""Update statistics, and the has_changed flag if change in branch"""
 		if subinstance.changed: self.changed = 1	
+		self.add_file_stats(subinstance)
+
 
-		self.total_files += subinstance.total_files
-		self.total_file_size += subinstance.total_file_size
-		self.changed_files += subinstance.changed_files
-		self.changed_file_size += subinstance.changed_file_size
-		self.increment_file_size += subinstance.increment_file_size
diff --git a/rdiff-backup/src/statistics.py b/rdiff-backup/src/statistics.py
new file mode 100644
index 0000000..a121591
--- /dev/null
+++ b/rdiff-backup/src/statistics.py
@@ -0,0 +1,133 @@
+execfile("filename_mapping.py")
+
+#######################################################################
+#
+# statistics - Generate and process aggregated backup information
+#
+
+class StatsException(Exception): pass
+
+class StatsObj:
+	"""Contains various statistics, provide string conversion functions"""
+
+	stat_file_attrs = ('SourceFiles', 'SourceFileSize',
+					   'MirrorFiles', 'MirrorFileSize',
+					   'NewFiles', 'NewFileSize',
+					   'DeletedFiles', 'DeletedFileSize',
+					   'ChangedFiles',
+					   'ChangedSourceSize', 'ChangedMirrorSize',
+					   'IncrementFileSize')
+	stat_time_attrs = ('StartTime', 'EndTime')
+	stat_attrs = stat_time_attrs + stat_file_attrs
+
+	# Set all stats to None, indicating info not available
+	for attr in stat_attrs: locals()[attr] = None
+
+	def get_stat(self, attribute):
+		"""Get a statistic"""
+		try: return self.__dict__[attribute]
+		except KeyError:
+			# this may be a hack, but seems no good way to get attrs in python
+			return eval("self.%s" % attribute)
+
+	def set_stat(self, attr, value):
+		"""Set attribute to given value"""
+		self.__dict__[attr] = value
+
+	def get_stats_string(self):
+		"""Return string printing out statistics"""
+		slist = ["%s %s" % (attr, self.get_stat(attr))
+				 for attr in self.stat_attrs
+				 if self.get_stat(attr) is not None]
+		return "\n".join(slist)
+
+	def init_stats_from_string(self, s):
+		"""Initialize attributes from string, return self for convenience"""
+		def error(line): raise StatsException("Bad line '%s'" % line)
+
+		for line in s.split("\n"):
+			if not line: continue
+			line_parts = line.split()
+			if len(line_parts) < 2: error(line)
+			attr, value_string = line_parts[:2]
+			if not attr in self.stat_attrs: error(line)
+			try: self.set_stat(attr, long(value_string))
+			except ValueError: error(line)
+		return self
+
+	def write_stats_to_rp(self, rp):
+		"""Write statistics string to given rpath"""
+		tf = TempFileManager.new(rp)
+		def init_thunk():
+			fp = tf.open("w")
+			fp.write(self.get_stats_string())
+			fp.close()
+		Robust.make_tf_robustaction(init_thunk, (tf,), (rp,)).execute()
+
+	def read_stats_from_rp(self, rp):
+		"""Set statistics from rpath, return self for convenience"""
+		fp = rp.open("r")
+		self.init_stats_from_string(fp.read())
+		fp.close()
+		return self
+
+	def stats_equal(self, s):
+		"""Return true if s has same statistics as self"""
+		assert isinstance(s, StatsObj)
+		for attr in self.stat_file_attrs:
+			if self.get_stat(attr) != s.get_stat(attr): return None
+		return 1
+
+
+class StatsITR(IterTreeReducer, StatsObj):
+	"""Keep track of per directory statistics
+
+	This is subclassed by the mirroring and incrementing ITRs.
+
+	"""
+	# zero out file statistics
+	for attr in StatsObj.stat_file_attrs: locals()[attr] = 0
+
+	def start_stats(self, mirror_dsrp):
+		"""Record status of mirror dsrp
+
+		This is called before the mirror is processed so we remember
+		the old state.
+
+		"""
+		if mirror_dsrp.lstat():
+			self.mirror_base_exists = 1
+			self.mirror_base_size = mirror_dsrp.getsize()
+		else: self.mirror_base_exists = None
+
+	def end_stats(self, diff_rorp, mirror_dsrp, inc_rp = None):
+		"""Set various statistics after mirror processed"""
+		if mirror_dsrp.lstat():
+			self.SourceFiles += 1
+			self.SourceFileSize += mirror_dsrp.getsize()
+			if self.mirror_base_exists:
+				self.MirrorFiles += 1
+				self.MirrorFileSize += self.mirror_base_size
+				if diff_rorp: # otherwise no change
+					self.ChangedFiles += 1
+					self.ChangedSourceSize += mirror_dsrp.getsize()
+					self.ChangedMirrorSize += self.mirror_base_size
+					self.IncrementFileSize += inc_rp and inc_rp.getsize() or 0
+			else: # new file was created
+				self.NewFiles += 1
+				self.NewFileSize += mirror_dsrp.getsize()
+				self.IncrementFileSize += inc_rp and inc_rp.getsize() or 0
+		else:
+			if self.mirror_base_exists: # file was deleted from mirror
+				self.MirrorFiles += 1
+				self.MirrorFileSize += self.mirror_base_size
+				self.DeletedFiles += 1
+				self.DeletedFileSize += self.mirror_base_size
+				self.IncrementFileSize += inc_rp and inc_rp.getsize() or 0
+			else: assert None # One of before and after should exist
+
+	def add_file_stats(self, subinstance):
+		"""Add all file statistics from subinstance to current totals"""
+		for attr in self.stat_file_attrs:
+			self.set_stat(attr,
+						  self.get_stat(attr) + subinstance.get_stat(attr))