def __replace_file(self, node, replacement_path):
"""
Replace the contents of a node.
@type node: SvnDumpNode
@param node: The node to replace the contents of.
@type replacement_path: string
@param replacement_path: The path to the replacement file.
@rtype: SvnDumpNode
@return: The converted node.
"""
if node.get_text_length() == -1:
# no text
self.__print(1, " no content to replace.")
return node
self.__print(1, " replace content with '%s'" % replacement_path)
if self.dry_run:
# do not replace with --dry-run
return node
# do the replacement
newnode = SvnDumpNode(node.get_path(), node.get_action(),
node.get_kind())
if node.has_copy_from():
newnode.set_copy_from(node.get_copy_from_path(),
node.get_copy_from_rev())
if node.has_properties():
newnode.set_properties(node.get_properties())
newnode.set_text_file(replacement_path)
return newnode
def __replace_file(self, node, replacement_path):
"""
Replace the contents of a node.
@type node: SvnDumpNode
@param node: The node to replace the contents of.
@type replacement_path: string
@param replacement_path: The path to the replacement file.
@rtype: SvnDumpNode
@return: The converted node.
"""
if node.get_text_length() == -1:
# no text
self.__print(1, " no content to replace.")
return node
self.__print(1, " replace content with '%s'" % replacement_path)
if self.dry_run:
# do not replace with --dry-run
return node
# do the replacement
newnode = SvnDumpNode(node.get_path(), node.get_action(),
node.get_kind())
if node.has_copy_from():
newnode.set_copy_from(node.get_copy_from_path(),
node.get_copy_from_rev())
if node.has_properties():
newnode.set_properties(node.get_properties())
newnode.set_text_file(replacement_path)
return newnode
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
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 node.has_text():
newNode.set_text_node(node)
return newNode
def copy_adding_git_ignore(srcfile, dstfile):
"""
Copy a dump file adding .gitignore files in all directories that have
an svn:ignore property.
@type srcfile: string
@param srcfile: Source filename.
@type dstfile: string
@param dstfile: Destination filename.
"""
# SvnDumpFile classes for reading/writing dumps
srcdmp = SvnDumpFile()
dstdmp = SvnDumpFile()
# open source file
srcdmp.open(srcfile)
gitignores = {}
# used to ensure that copyfrom-revs are correct after the copy. If
# there are any empty revision in the source dump file, the copyfrom-revs
# could be affected.
revmap = {}
hasrev = srcdmp.read_next_rev()
if hasrev:
# create the dump file
dstdmp.create_like(dstfile, srcdmp)
# now copy all the revisions
while hasrev:
dstdmp.add_rev(srcdmp.get_rev_props())
for node in srcdmp.get_nodes_iter():
if node.has_copy_from():
node.set_copy_from(node.get_copy_from_path(),
revmap[node.get_copy_from_rev()])
# add the original node
dstdmp.add_node(node)
if node.get_property("svn:ignore") is not None:
# find out what the change is and act on it appropriately
path = node.get_path() + "/.gitignore"
action = node.get_action()
if action == "change" or action == "add":
if path in gitignores:
# already saw this one - it is a change to the .gitignore file
newnode = SvnDumpNode(path, "change", "file")
else:
# haven't seen this one yet
newnode = SvnDumpNode(path, "add", "file")
gitignores[path] = True
f = open("gitignore", "wb")
f.write(node.get_property("svn:ignore"))
f.close()
newnode.set_text_file("gitignore")
dstdmp.add_node(newnode)
os.remove("gitignore")
elif action == "delete":
newnode = SvnDumpNode(path, "delete", "file")
dstdmp.add_node(newnode)
del gitignores[path]
else:
print("Unhandled action: '%s'" % action)
revmap[srcdmp.get_rev_nr()] = dstdmp.get_rev_nr()
hasrev = srcdmp.read_next_rev()
else:
print("no revisions in the source dump '%s' ???" % srcfile)
# cleanup
srcdmp.close()
dstdmp.close()
def read_next_rev( self ):
"""
Read the next revision.
@rtype: bool
@return: False if EOF occured.
"""
# check state
if self.__state != self.ST_READ:
raise SvnDumpException, "invalid state %d (should be %d)" % \
( self.__state, self.ST_READ )
# check for end of file
if self.__file_eof:
self.__state = self.ST_EOF
return False
# go to start of revision
if self.__rev_start_offset != self.__file.tell():
self.__file.seek( self.__rev_start_offset )
# get rev tags
tags = self.__get_tag_list()
self.__rev_nr = int( tags["Revision-number:"] )
# read revision properties
self.__rev_props = self.__get_prop_list()
self.__skip_empty_line()
if not self.__rev_props.has_key("svn:log"):
self.__rev_props["svn:log"] = ""
if not self.__rev_props.has_key("svn:author"):
self.__rev_props["svn:author"] = ""
if self.__rev_props.has_key("svn:date"):
self.set_rev_date(self.__rev_props["svn:date"] )
else:
self.set_rev_date( "" )
# read nodes (files, dirs)
self.__nodes.clear()
#self.nodeList = []
tags = self.__get_tag_list()
while len(tags) != 0:
# check that it's not the next revision
if tags.has_key( "Revision-number:" ):
# go back to start of tag list
self.__file.seek( self.__tag_start_offset )
self.__line_nr = self.__tag_start_line_nr
break
# get node properties
if tags.has_key( "Prop-content-length:" ):
properties = self.__get_prop_list()
else:
properties = None
# skip node data
if tags.has_key( "Text-content-length:" ):
tags["Text-content-length:"] = int( tags["Text-content-length:"] )
offset = self.__file.tell()
self.__skip_bin( tags["Text-content-length:"] )
self.__skip_empty_line()
else:
offset = 0
# add node
path = tags["Node-path:"].lstrip('/')
action = tags["Node-action:"]
kind = ""
if tags.has_key( "Node-kind:" ):
kind = tags["Node-kind:"]
node = SvnDumpNode( path, action, kind )
if properties != None:
node.set_properties( properties )
if tags.has_key( "Node-copyfrom-path:" ):
node.set_copy_from( tags["Node-copyfrom-path:"].lstrip('/'),
int(tags["Node-copyfrom-rev:"]) )
if tags.has_key( "Text-content-md5:" ):
md5 = tags["Text-content-md5:"]
else:
md5 = ""
if tags.has_key( "Text-content-length:" ):
node.set_text_fileobj( self.__file, offset,
int(tags["Text-content-length:"]),
md5 )
upath = ( action[0].upper(), path )
self.__nodes[upath] = node
# next one...
tags = self.__get_tag_list()
self.__rev_start_offset = self.__file.tell()
return True
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 read_next_rev(self):
"""
Read the next revision.
@rtype: bool
@return: False if EOF occured.
"""
# check state
if self.__state != self.ST_READ:
raise SvnDumpException("invalid state %d (should be %d)" % \
(self.__state, self.ST_READ))
# check for end of file
if self.__file_eof:
self.__state = self.ST_EOF
return False
# go to start of revision
if self.__rev_start_offset != self.__file.tell():
self.__file.seek(self.__rev_start_offset)
# get rev tags
tags = self.__get_tag_list()
self.__rev_nr = int(tags["Revision-number:"])
# read revision properties
self.__rev_props = self.__get_prop_list()
self.__skip_empty_line()
if not self.__rev_props.has_key("svn:log"):
self.__rev_props["svn:log"] = ""
if not self.__rev_props.has_key("svn:author"):
self.__rev_props["svn:author"] = ""
if self.__rev_props.has_key("svn:date"):
self.set_rev_date(self.__rev_props["svn:date"])
else:
self.set_rev_date("")
# read nodes (files, dirs)
self.__nodes.clear()
# self.nodeList = []
tags = self.__get_tag_list()
while len(tags) != 0:
# check that it's not the next revision
if tags.has_key("Revision-number:"):
# go back to start of tag list
self.__file.seek(self.__tag_start_offset)
self.__line_nr = self.__tag_start_line_nr
break
# get node properties
if tags.has_key("Prop-content-length:"):
properties = self.__get_prop_list()
else:
properties = None
# skip node data
if tags.has_key("Text-content-length:"):
tags["Text-content-length:"] = int(tags["Text-content-length:"])
offset = self.__file.tell()
self.__skip_bin(tags["Text-content-length:"])
self.__skip_empty_line()
else:
offset = 0
# add node
path = tags["Node-path:"].lstrip('/')
action = tags["Node-action:"]
kind = ""
if tags.has_key("Node-kind:"):
kind = tags["Node-kind:"]
node = SvnDumpNode(path, action, kind)
if properties is not None:
node.set_properties(properties)
if tags.has_key("Node-copyfrom-path:"):
node.set_copy_from(tags["Node-copyfrom-path:"].lstrip('/'),
int(tags["Node-copyfrom-rev:"]))
if tags.has_key("Text-content-md5:"):
md5 = tags["Text-content-md5:"]
else:
md5 = ""
if tags.has_key("Text-content-sha1:"):
sha1 = tags["Text-content-sha1:"]
else:
sha1 = ""
if tags.has_key("Text-content-length:"):
node.set_text_fileobj(self.__file, offset,
int(tags["Text-content-length:"]),
md5,
sha1)
upath = (action[0].upper(), path)
self.__nodes[upath] = node
# next one...
tags = self.__get_tag_list()
self.__rev_start_offset = self.__file.tell()
return True
def copy_adding_git_ignore(srcfile, dstfile):
"""
Copy a dump file adding .gitignore files in all directories that have
an svn:ignore property.
@type srcfile: string
@param srcfile: Source filename.
@type dstfile: string
@param dstfile: Destination filename.
"""
# SvnDumpFile classes for reading/writing dumps
srcdmp = SvnDumpFile()
dstdmp = SvnDumpFile()
# open source file
srcdmp.open(srcfile)
gitignores = {}
# used to ensure that copyfrom-revs are correct after the copy. If
# there are any empty revision in the source dump file, the copyfrom-revs
# could be affected.
revmap = {}
hasrev = srcdmp.read_next_rev()
if hasrev:
# create the dump file
dstdmp.create_like(dstfile, srcdmp)
# now copy all the revisions
while hasrev:
dstdmp.add_rev(srcdmp.get_rev_props())
for node in srcdmp.get_nodes_iter():
if node.has_copy_from():
node.set_copy_from(node.get_copy_from_path(), revmap[node.get_copy_from_rev()])
# add the original node
dstdmp.add_node(node)
if node.get_property("svn:ignore") is not None:
# find out what the change is and act on it appropriately
path = node.get_path() + "/.gitignore"
action = node.get_action()
if action == "change" or action == "add":
if path in gitignores:
# already saw this one - it is a change to the .gitignore file
newnode = SvnDumpNode(path, "change", "file")
else:
# haven't seen this one yet
newnode = SvnDumpNode(path, "add", "file")
gitignores[path] = True
f = open("gitignore", "wb")
f.write(node.get_property("svn:ignore"))
f.close()
newnode.set_text_file("gitignore")
dstdmp.add_node(newnode)
os.remove("gitignore")
elif action == "delete":
newnode = SvnDumpNode(path, "delete", "file")
dstdmp.add_node(newnode)
del gitignores[path]
else:
print("Unhandled action: '%s'" % action)
revmap[srcdmp.get_rev_nr()] = dstdmp.get_rev_nr()
hasrev = srcdmp.read_next_rev()
else:
print("no revisions in the source dump '%s' ???" % srcfile)
# cleanup
srcdmp.close()
dstdmp.close()
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 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 __convert_eol(self, node, revnr):
"""
Convert EOL of a node.
@type node: SvnDumpNode
@param node: The node to convert.
@type revnr: integer
@param revnr: The current revision number.
@rtype: SvnDumpNode
@return: The converted node.
"""
if not node.has_text():
# no text
print(" selected file, no text changes")
return node
# search for CR
need_conv = False
handle = node.text_open()
data = node.text_read(handle)
while len(data) > 0:
if data.find("\r") != -1:
# found one, need to convert the file
need_conv = True
break
data = node.text_read(handle)
fix = self.__fix
if need_conv:
# special fix option for rev/file ?
key = (revnr, node.get_path())
if self.__fix_rev_path.has_key(key):
fix = self.__fix_rev_path[key]
print(" selected file, convert (fix option %d)" % fix)
if self.__dry_run or fix == 0:
# do not convert with --dry-run or when there's nothing to fix
need_conv = False
else:
print(" selected file, no conversion required")
if need_conv:
# do the conversion
node.text_reopen(handle)
outfilename = self.__temp_file_name()
outfile = open(outfilename, "wb")
outlen = 0
md = sdt_md5()
data = node.text_read(handle)
carry = ""
warning_printed = False
while len(data) > 0:
if len(carry) != 0:
data = carry + data
n = len(data) - 1
carry = data[n]
if carry == "\r":
data = data[:n]
else:
carry = ""
if fix & 1 != 0:
data = data.replace("\r\n", "\n")
if fix & 2 != 0:
data = data.replace("\r", "\n")
if fix & 4 != 0:
data = data.replace("\r", "")
if not warning_printed and data.find("\r") >= 0:
warning_printed = True
print(" WARNING: file still contains CR")
print(" file: '%s'" % node.get_path())
if self.__warning_file is not None:
self.__warning_file.write(
"# WARNING: file still contains CR\n")
file = node.get_path()
while file[0] == "/":
file = file[1:]
tmpfile = file.replace("/", "__")
cmd = '$SVN cat -r %d "$REPOS/%s" > "%s"\n' % \
(revnr, node.get_path(), tmpfile)
self.__warning_file.write(cmd)
self.__warning_count += 1
md.update(data)
outfile.write(data)
outlen = outlen + len(data)
data = node.text_read(handle)
if len(carry) != 0:
if fix & 2 != 0:
carry = "\n"
elif fix & 4 != 0:
carry = ""
outfile.write(carry)
md.update(carry)
outlen += len(carry)
outfile.close()
newnode = SvnDumpNode(node.get_path(), node.get_action(),
node.get_kind())
if node.has_copy_from():
newnode.set_copy_from(node.get_copy_from_path(),
node.get_copy_from_rev())
if node.has_properties():
newnode.set_properties(node.get_properties())
newnode.set_text_file(outfilename, outlen, md.hexdigest())
else:
newnode = node
node.text_close(handle)
return newnode
