From 2a934756dd05992b409bfcd229bc7a6c06bff929 Mon Sep 17 00:00:00 2001
From: bescoto <bescoto@2b77aa54-bcbc-44c9-a7ec-4f6cf2b41109>
Date: Wed, 14 Dec 2005 03:13:57 +0000
Subject: Fix for KeyError while incrementing in v1.1.3

git-svn-id: http://svn.savannah.nongnu.org/svn/rdiff-backup/trunk@706 2b77aa54-bcbc-44c9-a7ec-4f6cf2b41109
---
 rdiff-backup/CHANGELOG              |  4 +++-
 rdiff-backup/rdiff_backup/backup.py | 43 ++++++++++++++++++++++++++++++++++---
 rdiff-backup/testing/finaltest.py   | 14 ++++++++++++
 3 files changed, 57 insertions(+), 4 deletions(-)

diff --git a/rdiff-backup/CHANGELOG b/rdiff-backup/CHANGELOG
index a425b50..faa7c09 100644
--- a/rdiff-backup/CHANGELOG
+++ b/rdiff-backup/CHANGELOG
@@ -1,4 +1,4 @@
-New in v1.1.4 (????/??/??)
+New in v1.1.4 (2005/12/13)
 --------------------------
 
 Quoting should be enabled only as needed between case-sensitive and
@@ -7,6 +7,8 @@ non-case-sensitive systems (thanks for Andrew Ferguson for report).
 Files with ACLs will not be unnecessarily marked as changed (bug
 report by Carsten Lorenz).
 
+Fix for common KeyError bug introduced in v1.1.3.
+
 
 New in v1.1.3 (2005/11/25)
 --------------------------
diff --git a/rdiff-backup/rdiff_backup/backup.py b/rdiff-backup/rdiff_backup/backup.py
index 3041bd9..819ae91 100644
--- a/rdiff-backup/rdiff_backup/backup.py
+++ b/rdiff-backup/rdiff_backup/backup.py
@@ -300,6 +300,10 @@ class CacheCollatedPostProcess:
 		# computed once.
 		self.inode_digest_dict = {}
 
+		# Contains list of (index, (source_rorp, diff_rorp)) pairs for
+		# the parent directories of the last item in the cache.
+		self.parent_list = []
+
 	def __iter__(self): return self
 
 	def next(self):
@@ -354,6 +358,30 @@ class CacheCollatedPostProcess:
 		self.post_process(old_source_rorp, old_dest_rorp,
 						  changed_flag, success_flag, inc)
 		if self.dir_perms_list: self.reset_dir_perms(first_index)
+		self.update_parent_list(first_index, old_source_rorp, old_dest_rorp)
+
+	def update_parent_list(self, index, src_rorp, dest_rorp):
+		"""Update the parent cache with the recently expired main cache entry
+
+		This method keeps parent directories in the secondary parent
+		cache until all their children have expired from the main
+		cache.  This is necessary because we may realize we need a
+		parent directory's information after we have processed many
+		subfiles.
+
+		"""
+		if not (src_rorp and src_rorp.isdir() or
+				dest_rorp and dest_rorp.isdir()): return # neither is directory
+		if not self.parent_list: assert index == (), index
+		else:
+			last_parent_index = self.parent_list[-1][0]
+			lp_index, li = len(last_parent_index), len(index)
+			if li > lp_index: # descended into previous parent
+				assert li == lp_index + 1, (index, last_parent_index)
+			else: # In new directory
+				assert last_parent_index[:li-1] == index[:-1], index
+				self.parent_list = self.parent_list[:li]
+		self.parent_list.append((index, (src_rorp, dest_rorp)))
 
 	def post_process(self, source_rorp, dest_rorp, changed, success, inc):
 		"""Post process source_rorp and dest_rorp.
@@ -411,20 +439,29 @@ class CacheCollatedPostProcess:
 		"""Set the increment of the current file"""
 		self.cache_dict[index][4] = inc
 
+	def get_parent_rorps(self, index):
+		"""Retrieve (src_rorp, dest_rorp) pair from parent cache"""
+		for parent_index, pair in self.parent_list:
+			if parent_index == index: return pair
+		raise KeyError(index)
+
 	def get_rorps(self, index):
 		"""Retrieve (source_rorp, dest_rorp) from cache"""
-		return self.cache_dict[index][:2]
+		try: return self.cache_dict[index][:2]
+		except KeyError: return self.get_parent_rorps(index)
 
 	def get_source_rorp(self, index):
 		"""Retrieve source_rorp with given index from cache"""
 		assert index >= self.cache_indicies[0], \
 			   ("CCPP index out of order: %s %s" %
 				(repr(index), repr(self.cache_indicies[0])))
-		return self.cache_dict[index][0]
+		try: return self.cache_dict[index][0]
+		except KeyError: return self.get_parent_rorps(index)[0]
 
 	def get_mirror_rorp(self, index):
 		"""Retrieve mirror_rorp with given index from cache"""
-		return self.cache_dict[index][1]
+		try: return self.cache_dict[index][1]
+		except KeyError: return self.get_parent_rorps(index)[1]
 
 	def update_hash(self, index, sha1sum):
 		"""Update the source rorp's SHA1 hash"""
diff --git a/rdiff-backup/testing/finaltest.py b/rdiff-backup/testing/finaltest.py
index 6dc73f2..d773c21 100644
--- a/rdiff-backup/testing/finaltest.py
+++ b/rdiff-backup/testing/finaltest.py
@@ -711,5 +711,19 @@ class FinalBugs(PathSetter):
 		rp1_d_f.setdata()
 		assert rp1_d_f.isreg(), 'File %s corrupted' % (rp1_d_f.path,)
 
+	def test_CCPP_keyerror(self):
+		"""Test when no change until middle of a directory
+
+		This tests CCPP, to make sure it isn't asked to provide rorps
+		for indicies that are out of the cache.
+
+		"""
+		self.delete_tmpdirs()
+		rdiff_backup(1, 1, 'testfiles/bigdir', 'testfiles/output')
+		rp = rpath.RPath(Globals.local_connection,
+						 'testfiles/bigdir/subdir3/subdir49/file49')
+		assert rp.isreg(), rp
+		rp.touch()
+		rdiff_backup(1, 1, 'testfiles/bigdir', 'testfiles/output')
 		
 if __name__ == "__main__": unittest.main()
-- 
cgit v1.2.1