def execute(self, inputfile, outputfile):
"""
Fix a cvs2svn created dump file.
@type inputfile: string
@param inputfile: Name of the cvs2svn created dump file.
@type outputfile: string
@param outputfile: Name of the fixed dump file.
"""
indump = SvnDumpFile()
indump.open(inputfile)
has_rev = indump.read_next_rev()
outdump = SvnDumpFile()
outdump.create_like(outputfile, indump)
rc = 0
self.__history = {}
while has_rev and rc == 0:
outdump.add_rev(indump.get_rev_props())
for node in indump.get_nodes_iter():
msglist = self.__fix_node(indump.get_rev_nr(), node)
if msglist != None:
rc = 1
print "Error in r%d node %s:" % \
( indump.get_rev_nr(), node.get_path() )
for msg in msglist:
print " " + msg
break
outdump.add_node(node)
has_rev = indump.read_next_rev()
indump.close()
return rc
def execute( self, inputfile, outputfile ):
"""
Fix a cvs2svn created dump file.
@type inputfile: string
@param inputfile: Name of the cvs2svn created dump file.
@type outputfile: string
@param outputfile: Name of the fixed dump file.
"""
indump = SvnDumpFile()
indump.open( inputfile )
has_rev = indump.read_next_rev()
outdump = SvnDumpFile()
outdump.create_like( outputfile, indump )
rc = 0
self.__history = {}
while has_rev and rc == 0:
outdump.add_rev( indump.get_rev_props() )
for node in indump.get_nodes_iter():
msglist = self.__fix_node( indump.get_rev_nr(), node )
if msglist != None:
rc = 1
print "Error in r%d node %s:" % \
( indump.get_rev_nr(), node.get_path() )
for msg in msglist:
print " " + msg
break
outdump.add_node( node )
has_rev = indump.read_next_rev()
indump.close()
return rc
class SvnDumpMerge:
"""
A class for merging svn dump files.
"""
# handle copyfrom-rev !!!
def __init__(self):
"""
Initialize.
"""
# output file name
self.__out_file = ""
# date for revision 0
self.__out_r0_date = "9999-99-99T99:99:99.999999Z"
# list additional directories
self.__out_dirs = []
# log message for directory creating revision
self.__out_message = ""
# author for the additional revision
self.__out_author = "svndumpmerge"
# variables used for input dump files
# file names
self.__in_files = []
# path renames [ [ ( from, to ), ... ], ... ]
self.__in_renames = []
# path regex substitutions [ [ ( search, replace ), ... ], ... ]
self.__in_regex_subs = []
# mkdir excludes [ {}, ... ]
self.__in_excludes = []
# revision number mappings [ {}, ... ]
self.__in_rev_nr_maps = []
# dump files (class SvnDumpFile)
self.__in_dumps = []
# revision dates of the dumps
self.__in_rev_dates = []
def set_output_file(self, filename, startRev=0):
"""
Sets the output file name and optional start revision.
@type filename: string
@param filename: Name of the output dump file.
@type startRev: integer
@param startRev: Start revision number, default is 0.
"""
self.__out_file = filename
self.outStartRev = startRev
def add_input_file(self, filename):
"""
Adds an input file and returns it's index.
@type filename: string
@param filename: Name of a input dump file.
@rtype: integer
@return: Index of the input file.
"""
index = len(self.__in_files)
self.__in_files = self.__in_files + [filename]
self.__in_renames = self.__in_renames + [[]]
self.__in_regex_subs = self.__in_regex_subs + [[]]
self.__in_excludes = self.__in_excludes + [{}]
self.__in_rev_nr_maps = self.__in_rev_nr_maps + [{}]
return index
def add_rename(self, index, prefixFrom, prefixTo):
"""
Adds a path prefix reanme.
@type index: integer
@param index: Index of the dump file.
@type prefixFrom: string
@param prefixFrom: From-path prefix (directory).
@type prefixTo: string
@param prefixTo: To-path prefix (directory).
"""
# make sure that prefixFrom starts and ends with a /
if prefixFrom[0:1] == "/":
prefixFrom = prefixFrom[1:]
if prefixFrom[len(prefixFrom) - 1:] != "/":
prefixFrom = prefixFrom + "/"
# make sure that prefixTo starts and ends with a /
if prefixTo[0:1] == "/":
prefixTo = prefixTo[1:]
if prefixTo[len(prefixTo) - 1:] != "/":
prefixTo = prefixTo + "/"
# add the rename
self.__in_renames[index] = self.__in_renames[index] + \
[(prefixFrom, prefixTo)]
def add_regex_sub(self, index, reSearch, reReplace):
"""
Adds a path prefix rename.
@type index: integer
@param index: Index of the dump file.
@type reSearch: string
@param reSearch: Search regular expression.
@type reReplace: string
@param reReplace: Replace regular expression.
"""
# compile regex object
reSearch = re.compile(reSearch)
# add the rename
self.__in_regex_subs[index] = self.__in_regex_subs[index] + \
[(reSearch, reReplace)]
def add_mkdir_exclude(self, index, dirName):
"""
Adds a mkdir exclude.
@type index: integer
@param index: Index of the dump file.
@type dirName: string
@param dirName: Name of the directory.
"""
# add the mkdir exclude
self.__in_excludes[index][dirName] = None
def add_directory(self, dirName):
"""
Adds an additional directory ('mkdir').
@type dirName: string
@param dirName: Name of the directory.
"""
if dirName[0:1] == "/":
dirName = dirName[1:]
if dirName[-1:] == "/":
dirName = dirName[:-1]
self.__out_dirs = self.__out_dirs + [dirName]
def set_log_message(self, msg):
"""
Set log message for additional dirs revision.
@type msg: string
@param msg: Log message.
"""
self.__out_message = msg
def merge(self):
"""
Executes the merge.
"""
if len(self.__in_files) == 0:
print("merge: no input files specified")
return
if len(self.__out_file) == 0:
print("merge: no output file specified")
return
# open input dump files
for inFile in self.__in_files:
inDump = SvnDumpFile()
inDump.open(inFile)
inDump.read_next_rev()
self.__in_dumps = self.__in_dumps + [inDump]
if inDump.get_rev_date_str() < self.__out_r0_date:
self.__out_r0_date = inDump.get_rev_date_str()
# remove empty dumps
dumpCount = self.__remove_empty_dumps()
if dumpCount == 0:
return
# open output file
self.outDump = SvnDumpFile()
if self.outStartRev == 0:
self.outDump.create_with_rev_0(self.__out_file,
self.__in_dumps[0].get_uuid(), self.__out_r0_date)
else:
self.outDump.create_with_rev_n(self.__out_file,
self.__in_dumps[0].get_uuid(), self.outStartRev)
# skip revision 0 of all dumps
for inDump in self.__in_dumps:
if inDump.get_rev_nr() == 0:
# +++ what about r0 revprops?
inDump.read_next_rev()
# remove empty dumps
dumpCount = self.__remove_empty_dumps()
if dumpCount == 0:
self.outDump.close()
return
# get revision dates
oldest = None
oldestStr = ""
for index in range(len(self.__in_dumps)):
revDat = self.__in_dumps[index].get_rev_date()
self.__in_rev_dates.append(revDat)
if oldest is None or revDat < oldest:
oldest = revDat
oldestStr = self.__in_dumps[index].get_rev_date_str()
# add additional directories
if len(self.__out_dirs) > 0:
self.outDump.add_rev({"svn:log": self.__out_message,
"svn:author": self.__out_author,
"svn:date": oldestStr})
for dirName in self.__out_dirs:
node = SvnDumpNode(dirName, "add", "dir")
self.outDump.add_node(node)
# loop over all revisions
while dumpCount > 0:
# find index of the oldest revision
oldestIndex = 0
for index in range(1, dumpCount):
if self.__in_rev_dates[index] < self.__in_rev_dates[oldestIndex]:
oldestIndex = index
# copy revision
self.__copy_revision(oldestIndex)
print("Revision: %-8d from r%-8d %s" % (self.outDump.get_rev_nr(),
self.__in_dumps[oldestIndex].get_rev_nr(),
self.__in_files[oldestIndex]))
# read next revision
srcDump = self.__in_dumps[oldestIndex]
if srcDump.read_next_rev():
self.__in_rev_dates[oldestIndex] = srcDump.get_rev_date()
else:
dumpCount = self.__remove_empty_dumps()
# close output
print("created %d revisions" % self.outDump.get_rev_nr())
self.outDump.close()
def __copy_revision(self, dumpIndex):
"""
Copies a revision from inDump[dumpIndex] to outDump.
@type dumpIndex: integer
@param dumpIndex: Index of the input dump file.
"""
srcDump = self.__in_dumps[dumpIndex]
# add revision and revprops
self.outDump.add_rev(srcDump.get_rev_props())
# add nodes
index = 0
nodeCount = srcDump.get_node_count()
while index < nodeCount:
node = srcDump.get_node(index)
newNode = self.__change_node(dumpIndex, node)
if newNode is not None:
self.outDump.add_node(newNode)
index = index + 1
# add revision info
self.__in_rev_nr_maps[dumpIndex][srcDump.get_rev_nr()] = \
self.outDump.get_rev_nr()
def __change_node(self, dumpIndex, node):
"""
Creates a new node if the path changed, else returns the old node.
@type dumpIndex: integer
@param dumpIndex: Index of the input dump file.
@type node: SvnDumpNode
@param node: A node.
"""
path = node.get_path()
# mkdir exclude check
if node.get_kind() == "dir" and node.get_action() == "add":
if path in self.__in_excludes[dumpIndex]:
return None
fromPath = ""
fromRev = 0
if node.has_copy_from():
fromPath = node.get_copy_from_path()
fromRev = node.get_copy_from_rev()
change = 0
newPath = self.__rename_path(path, dumpIndex)
newFromPath = fromPath
newFromRev = fromRev
if path != newPath:
change = 1
if fromRev > 0:
newFromPath = self.__rename_path(fromPath, dumpIndex)
if fromPath != newFromPath:
change = 1
newFromRev = self.__in_rev_nr_maps[dumpIndex][fromRev]
if fromRev != newFromRev:
change = 1
newMergeInfo = ""
if node.has_properties():
properties = node.get_properties()
if properties.has_key('svn:mergeinfo'):
mergeInfo = properties['svn:mergeinfo']
for line in mergeInfo.split('\n'):
m = re.match('^(.*):(.*)', line)
if m is not None:
mergePath = m.group(1)
revPart = m.group(2)
newMergePath = self.__rename_path(mergePath, dumpIndex)
if not newMergePath.startswith("/"):
newMergePath = "/" + newMergePath
if len(newMergeInfo) != 0:
newMergeInfo = newMergeInfo + "\n"
newMergeInfo = newMergeInfo + newMergePath + ":"
revSep = ""
for rm in re.finditer('(\d+)(?:-(\d+))?,?', revPart):
mergeFrom = int(rm.group(1))
newMergeFrom = self.__in_rev_nr_maps[dumpIndex][mergeFrom]
newMergeInfo = newMergeInfo + revSep + str(newMergeFrom)
revSep = ","
if rm.group(2) is not None:
mergeTo = int(rm.group(2))
newMergeTo = self.__in_rev_nr_maps[dumpIndex][mergeTo]
newMergeInfo = newMergeInfo + "-" + str(newMergeTo)
if mergeInfo != newMergeInfo:
change = 1
if not change:
# no change needed
return node
# do the rename
newNode = SvnDumpNode(newPath, node.get_action(), node.get_kind())
if node.has_copy_from():
newNode.set_copy_from(newFromPath, newFromRev)
if node.has_properties():
newNode.set_properties(node.get_properties())
if len(newMergeInfo) > 0:
newNode.set_property('svn:mergeinfo', newMergeInfo)
if node.has_text():
newNode.set_text_node(node)
return newNode
def __rename_path(self, path, dumpIndex):
"""
Applies the renames to the path and returns the new path.
@type path: string
@param path: A path.
@type renames: list( ( string, string ) )
@param renames: List of rename tuples.
@rtype: string
@return Renamed path.
"""
# ensure that path does not have a leading slash
if len(path) > 1 and path[0:1] == "/":
path = path[1:]
sPath = path + "/"
for sPfx, dPfx in self.__in_renames[dumpIndex]:
sLen = len(sPfx)
if sPfx == "/":
return dPfx + path
elif sPath[:sLen] == sPfx:
if len(path) <= len(sPfx):
# it's the full path
return dPfx[0:len(dPfx) - 1]
else:
# there's a suffix
return dPfx + path[sLen:]
for reSearch, sReplace in self.__in_regex_subs[dumpIndex]:
path = reSearch.sub(sReplace, path, count=1)
return path
def __remove_empty_dumps(self):
"""
Removes dump files which reached EOF and returns the count of dumps.
@rtype: integer
@return: Count of remaining input dump files.
"""
index = 0
while index < len(self.__in_dumps):
inDump = self.__in_dumps[index]
if inDump.has_revision():
index = index + 1
else:
inDump.close()
eidx = index + 1
self.__in_files[index:eidx] = []
self.__in_renames[index:eidx] = []
self.__in_regex_subs[index:eidx] = []
self.__in_excludes[index:eidx] = []
self.__in_rev_nr_maps[index:eidx] = []
self.__in_dumps[index:eidx] = []
self.__in_rev_dates[index:eidx] = []
return index
class SvnDumpMerge:
"""
A class for merging svn dump files.
"""
# handle copyfrom-rev !!!
def __init__(self):
"""
Initialize.
"""
# output file name
self.__out_file = ""
# date for revision 0
self.__out_r0_date = "9999-99-99T99:99:99.999999Z"
# list additional directories
self.__out_dirs = []
# log message for directory creating revision
self.__out_message = ""
# author for the additional revision
self.__out_author = "svndumpmerge"
# variables used for input dump files
# file names
self.__in_files = []
# path renames [ [ ( from, to ), ... ], ... ]
self.__in_renames = []
# path regex substitutions [ [ ( search, replace ), ... ], ... ]
self.__in_regex_subs = []
# mkdir excludes [ {}, ... ]
self.__in_excludes = []
# revision number mappings [ {}, ... ]
self.__in_rev_nr_maps = []
# dump files (class SvnDumpFile)
self.__in_dumps = []
# revision dates of the dumps
self.__in_rev_dates = []
def set_output_file(self, filename, startRev=0):
"""
Sets the output file name and optional start revision.
@type filename: string
@param filename: Name of the output dump file.
@type startRev: integer
@param startRev: Start revision number, default is 0.
"""
self.__out_file = filename
self.outStartRev = startRev
def add_input_file(self, filename):
"""
Adds an input file and returns it's index.
@type filename: string
@param filename: Name of a input dump file.
@rtype: integer
@return: Index of the input file.
"""
index = len(self.__in_files)
self.__in_files = self.__in_files + [filename]
self.__in_renames = self.__in_renames + [[]]
self.__in_regex_subs = self.__in_regex_subs + [[]]
self.__in_excludes = self.__in_excludes + [{}]
self.__in_rev_nr_maps = self.__in_rev_nr_maps + [{}]
return index
def add_rename(self, index, prefixFrom, prefixTo):
"""
Adds a path prefix reanme.
@type index: integer
@param index: Index of the dump file.
@type prefixFrom: string
@param prefixFrom: From-path prefix (directory).
@type prefixTo: string
@param prefixTo: To-path prefix (directory).
"""
# make sure that prefixFrom starts and ends with a /
if prefixFrom[0:1] == "/":
prefixFrom = prefixFrom[1:]
if prefixFrom[len(prefixFrom) - 1:] != "/":
prefixFrom = prefixFrom + "/"
# make sure that prefixTo starts and ends with a /
if prefixTo[0:1] == "/":
prefixTo = prefixTo[1:]
if prefixTo[len(prefixTo) - 1:] != "/":
prefixTo = prefixTo + "/"
# add the rename
self.__in_renames[index] = self.__in_renames[index] + \
[ (prefixFrom, prefixTo ) ]
def add_regex_sub(self, index, reSearch, reReplace):
"""
Adds a path prefix rename.
@type index: integer
@param index: Index of the dump file.
@type reSearch: string
@param reSearch: Search regular expression.
@type reReplace: string
@param reReplace: Replace regular expression.
"""
# compile regex object
reSearch = re.compile(reSearch)
# add the rename
self.__in_regex_subs[index] = self.__in_regex_subs[index] + \
[ (reSearch, reReplace ) ]
def add_mkdir_exclude(self, index, dirName):
"""
Adds a mkdir exclude.
@type index: integer
@param index: Index of the dump file.
@type dirName: string
@param dirName: Name of the directory.
"""
# add the mkdir exclude
self.__in_excludes[index][dirName] = None
def add_directory(self, dirName):
"""
Adds an additional directory ('mkdir').
@type dirName: string
@param dirName: Name of the directory.
"""
if dirName[0:1] == "/":
dirName = dirName[1:]
if dirName[-1:] == "/":
dirName = dirName[:-1]
self.__out_dirs = self.__out_dirs + [dirName]
def set_log_message(self, msg):
"""
Set log message for additional dirs revision.
@type msg: string
@param msg: Log message.
"""
self.__out_message = msg
def merge(self):
"""
Executes the merge.
"""
if len(self.__in_files) == 0:
print "merge: no input files specified"
return
if len(self.__out_file) == 0:
print "merge: no output file specified"
return
# open input dump files
for inFile in self.__in_files:
inDump = SvnDumpFile()
inDump.open(inFile)
inDump.read_next_rev()
self.__in_dumps = self.__in_dumps + [inDump]
if inDump.get_rev_date_str() < self.__out_r0_date:
self.__out_r0_date = inDump.get_rev_date_str()
# remove empty dumps
dumpCount = self.__remove_empty_dumps()
if dumpCount == 0:
return
# open output file
self.outDump = SvnDumpFile()
if self.outStartRev == 0:
self.outDump.create_with_rev_0(self.__out_file,
self.__in_dumps[0].get_uuid(),
self.__out_r0_date)
else:
self.outDump.create_with_rev_n(self.__out_file,
self.__in_dumps[0].get_uuid(),
self.outStartRev)
# skip revision 0 of all dumps
for inDump in self.__in_dumps:
if inDump.get_rev_nr() == 0:
# +++ what about r0 revprops?
inDump.read_next_rev()
# remove empty dumps
dumpCount = self.__remove_empty_dumps()
if dumpCount == 0:
self.outDump.close()
return
# get revision dates
oldest = None
oldestStr = ""
for index in range(len(self.__in_dumps)):
revDat = self.__in_dumps[index].get_rev_date()
self.__in_rev_dates.append(revDat)
if oldest == None or revDat < oldest:
oldest = revDat
oldestStr = self.__in_dumps[index].get_rev_date_str()
# add additional directories
if len(self.__out_dirs) > 0:
self.outDump.add_rev({
"svn:log": self.__out_message,
"svn:author": self.__out_author,
"svn:date": oldestStr
})
for dirName in self.__out_dirs:
node = SvnDumpNode(dirName, "add", "dir")
self.outDump.add_node(node)
# loop over all revisions
while dumpCount > 0:
# find index of the oldest revision
oldestIndex = 0
for index in range(1, dumpCount):
if self.__in_rev_dates[index] < self.__in_rev_dates[
oldestIndex]:
oldestIndex = index
# copy revision
self.__copy_revision(oldestIndex)
print "Revision: %-8d from r%-8d %s" % (
self.outDump.get_rev_nr(),
self.__in_dumps[oldestIndex].get_rev_nr(),
self.__in_files[oldestIndex])
# read next revision
srcDump = self.__in_dumps[oldestIndex]
if srcDump.read_next_rev():
self.__in_rev_dates[oldestIndex] = srcDump.get_rev_date()
else:
dumpCount = self.__remove_empty_dumps()
# close output
print "created %d revisions" % self.outDump.get_rev_nr()
self.outDump.close()
def __copy_revision(self, dumpIndex):
"""
Copies a revision from inDump[dumpIndex] to outDump.
@type dumpIndex: integer
@param dumpIndex: Index of the input dump file.
"""
srcDump = self.__in_dumps[dumpIndex]
# add revision and revprops
self.outDump.add_rev(srcDump.get_rev_props())
# add nodes
index = 0
nodeCount = srcDump.get_node_count()
while index < nodeCount:
node = srcDump.get_node(index)
newNode = self.__change_node(dumpIndex, node)
if newNode != None:
self.outDump.add_node(newNode)
index = index + 1
# add revision info
self.__in_rev_nr_maps[dumpIndex][srcDump.get_rev_nr()] = \
self.outDump.get_rev_nr()
def __change_node(self, dumpIndex, node):
"""
Creates a new node if the path changed, else returns the old node.
@type dumpIndex: integer
@param dumpIndex: Index of the input dump file.
@type node: SvnDumpNode
@param node: A node.
"""
path = node.get_path()
# mkdir exclude check
if node.get_kind() == "dir" and node.get_action() == "add":
if path in self.__in_excludes[dumpIndex]:
return None
fromPath = ""
fromRev = 0
if node.has_copy_from():
fromPath = node.get_copy_from_path()
fromRev = node.get_copy_from_rev()
change = 0
newPath = self.__rename_path(path, dumpIndex)
newFromPath = fromPath
newFromRev = fromRev
if path != newPath:
change = 1
if fromRev > 0:
newFromPath = self.__rename_path(fromPath, dumpIndex)
if fromPath != newFromPath:
change = 1
newFromRev = self.__in_rev_nr_maps[dumpIndex][fromRev]
if fromRev != newFromRev:
change = 1
newMergeInfo = ""
if node.has_properties():
properties = node.get_properties()
if properties.has_key('svn:mergeinfo'):
mergeInfo = properties['svn:mergeinfo']
for line in mergeInfo.split('\n'):
m = re.match('^(.*):(\d+)-(\d+)', line)
if m != None:
mergePath = m.group(1)
mergeFrom = int(m.group(2))
mergeTo = int(m.group(3))
newMergePath = self.__rename_path(mergePath, dumpIndex)
newMergeFrom = self.__in_rev_nr_maps[dumpIndex][
mergeFrom]
newMergeTo = self.__in_rev_nr_maps[dumpIndex][mergeTo]
if len(newMergeInfo) != 0:
newMergeInfo = newMergeInfo + "\n"
newMergeInfo = newMergeInfo + newMergePath + ":" + str(
newMergeFrom) + "-" + str(newMergeTo)
if mergeInfo != newMergeInfo:
change = 1
if not change:
# no change needed
return node
# do the rename
newNode = SvnDumpNode(newPath, node.get_action(), node.get_kind())
if node.has_copy_from():
newNode.set_copy_from(newFromPath, newFromRev)
if node.has_properties():
newNode.set_properties(node.get_properties())
if len(newMergeInfo) > 0:
newNode.set_property('svn:mergeinfo', newMergeInfo)
if node.has_text():
newNode.set_text_node(node)
return newNode
def __rename_path(self, path, dumpIndex):
"""
Applies the renames to the path and returns the new path.
@type path: string
@param path: A path.
@type renames: list( ( string, string ) )
@param renames: List of rename tuples.
@rtype: string
@return Renamed path.
"""
# ensure that path does not have a leading slash
if len(path) > 1 and path[0:1] == "/":
path = path[1:]
sPath = path + "/"
for sPfx, dPfx in self.__in_renames[dumpIndex]:
sLen = len(sPfx)
if sPfx == "/":
return dPfx + path
elif sPath[:sLen] == sPfx:
if len(path) <= len(sPfx):
# it's the full path
return dPfx[0:len(dPfx) - 1]
else:
# there's a suffix
return dPfx + path[sLen:]
for reSearch, sReplace in self.__in_regex_subs[dumpIndex]:
path = reSearch.sub(sReplace, path, count=1)
return path
def __remove_empty_dumps(self):
"""
Removes dump files which reached EOF and returns the count of dumps.
@rtype: integer
@return: Count of remaining input dump files.
"""
index = 0
while index < len(self.__in_dumps):
inDump = self.__in_dumps[index]
if inDump.has_revision():
index = index + 1
else:
inDump.close()
eidx = index + 1
self.__in_files[index:eidx] = []
self.__in_renames[index:eidx] = []
self.__in_regex_subs[index:eidx] = []
self.__in_excludes[index:eidx] = []
self.__in_rev_nr_maps[index:eidx] = []
self.__in_dumps[index:eidx] = []
self.__in_rev_dates[index:eidx] = []
return index
