2nd patch from debian

rdiff_backup/Hardlink.py

@@ -95,7 +95,13 @@ def rorp_eq(src_rorp, dest_rorp):
 		src_rorp.getnumlinks() == dest_rorp.getnumlinks() == 1):
 		return 1 # Hard links don't apply
 
-	if src_rorp.getnumlinks() < dest_rorp.getnumlinks(): return 0
+	"""The sha1 of linked files is only stored in the metadata of the first 
+	linked file on the dest side.  If the first linked file on the src side is
+	deleted, then the sha1 will also be deleted on the dest side, so we test for this
+	& report not equal so that another sha1 will be stored with the next linked
+	file on the dest side"""
+	if (not islinked(src_rorp) and not dest_rorp.has_sha1()): return 0
+	if src_rorp.getnumlinks() != dest_rorp.getnumlinks(): return 0
 	src_key = get_inode_key(src_rorp)
 	index, remaining, dest_key, digest = _inode_index[src_key]
 	if dest_key == "NA":

rdiff_backup/compare.py

@@ -25,7 +25,7 @@ compare.  This uses elements of the backup and restore modules.
 """
 
 from __future__ import generators
-import Globals, restore, rorpiter, log, backup, static, rpath, hash, robust
+import Globals, restore, rorpiter, log, backup, static, rpath, hash, robust, Hardlink
 
 def Compare(src_rp, mirror_rp, inc_rp, compare_time):
 	"""Compares metadata in src_rp dir with metadata in mirror_rp at time"""
@@ -80,14 +80,15 @@ def Verify(mirror_rp, inc_rp, verify_time):
 	bad_files = 0
 	for repo_rorp in repo_iter:
 		if not repo_rorp.isreg(): continue
-		if not repo_rorp.has_sha1():
+		verify_sha1 = get_hash(repo_rorp)
+		if not verify_sha1:
 			log.Log("Warning: Cannot find SHA1 digest for file %s,\n"
 					"perhaps because this feature was added in v1.1.1"
 					% (repo_rorp.get_indexpath(),), 2)
 			continue
 		fp = RepoSide.rf_cache.get_fp(base_index + repo_rorp.index, repo_rorp)
 		computed_hash = hash.compute_sha1_fp(fp)
-		if computed_hash == repo_rorp.get_sha1():
+		if computed_hash == verify_sha1:
 			log.Log("Verified SHA1 digest of " + repo_rorp.get_indexpath(), 5)
 		else:
 			bad_files += 1
@@ -95,11 +96,24 @@ def Verify(mirror_rp, inc_rp, verify_time):
 					"doesn't match recorded digest of\n   %s\n"
 					"Your backup repository may be corrupted!" %
 					(repo_rorp.get_indexpath(), computed_hash,
-					 repo_rorp.get_sha1()), 2)
+					 verify_sha1), 2)
 	RepoSide.close_rf_cache()
 	if not bad_files: log.Log("Every file verified successfully.", 3)
 	return bad_files
 
+def get_hash (repo_rorp):
+	""" Try to get a sha1 digest from the repository.  If hardlinks 
+	are saved in the metadata, get the sha1 from the first hardlink """
+	Hardlink.add_rorp(repo_rorp)
+	if Hardlink.islinked(repo_rorp):
+		verify_sha1 = Hardlink.get_sha1(repo_rorp)
+	elif repo_rorp.has_sha1():
+		verify_sha1 = repo_rorp.get_sha1()
+	else:
+		verify_sha1 = None
+	Hardlink.del_rorp(repo_rorp)
+	return verify_sha1
+
 def print_reports(report_iter):
 	"""Given an iter of CompareReport objects, print them to screen"""
 	assert not Globals.server
@@ -199,12 +213,13 @@ class DataSide(backup.SourceStruct):
 		"""Like above, but also compare sha1 sums of any regular files"""
 		def hashes_changed(src_rp, mir_rorp):
 			"""Return 0 if their data hashes same, 1 otherwise"""
-			if not mir_rorp.has_sha1():
+			verify_sha1 = get_hash(mir_rorp)
+			if not verify_sha1:
 				log.Log("Warning: Metadata file has no digest for %s, "
 						"unable to compare." % (mir_rorp.get_indexpath(),), 2)
 				return 0
 			elif (src_rp.getsize() == mir_rorp.getsize() and
-				  hash.compute_sha1(src_rp) == mir_rorp.get_sha1()):
+				  hash.compute_sha1(src_rp) == verify_sha1):
 				return 0
 			return 1