def sort_messages(moduleName, messages):
# sort messages so that sub-messages are defined first
# to avoid compile errors: XXX was not declared in this scope
dependencies = []
msg_dict = {}
for msg in messages:
msg_name = "." + moduleName + "." + msg.name
msg_dict[msg_name] = msg
deps = set()
for member in msg.field:
if member.type == TYPE["MESSAGE"]:
deps.add(member.type_name)
dependencies.append((msg_name, deps))
sorted_msg_names = order_dependencies(dependencies)
ordered_msgs = [msg_dict[n] for n in sorted_msg_names]
return ordered_msgs
def sort_messages(fileObject):
"""Return a sorted list of messages (sub-messages first).
This avoids compilation problems involving declaration order.
"""
dependencies = []
msgDict = {}
def visit(baseName, messages, parent=None):
"""Visitor for the message tree."""
for msg in messages:
# Build our type name (using the protocol buffer convention) and
# use it to register this message type object in our dictionary.
typeName = baseName + '.' + msg.name
msgDict[typeName] = msg
# If this is a nested message type, prepend our parent's name to
# our name for all future name lookups (via template expansion).
# This disambiguates nested message names so that two n-level
# messages can both have nested message types with the same name.
# This also matches the generated C++ code's naming convention.
if parent is not None:
msg.name = parent.name + '_' + msg.name
# If this message has nested message types, recurse.
if msg.nested_type:
visit(typeName, msg.nested_type, parent=msg)
# Generate the set of messages that this type is dependent upon.
deps = set([
field.type_name for field in msg.field
if field.type == TYPE['MESSAGE']
])
dependencies.append((typeName, deps))
# Start by visiting the file's top-level message types.
visit('.' + fileObject.package, fileObject.message_type)
sortedMsgNames = order_dependencies(dependencies)
return [msgDict[n] for n in sortedMsgNames]
def sort_messages(fileObject):
"""Return a sorted list of messages (sub-messages first).
This avoids compilation problems involving declaration order.
"""
dependencies = []
msgDict = {}
def visit(baseName, messages, parent=None):
"""Visitor for the message tree."""
for msg in messages:
# Build our type name (using the protocol buffer convention) and
# use it to register this message type object in our dictionary.
typeName = baseName + '.' + msg.name
msgDict[typeName] = msg
# If this is a nested message type, prepend our parent's name to
# our name for all future name lookups (via template expansion).
# This disambiguates nested message names so that two n-level
# messages can both have nested message types with the same name.
# This also matches the generated C++ code's naming convention.
if parent is not None:
msg.name = parent.name + '_' + msg.name
# If this message has nested message types, recurse.
if msg.nested_type:
visit(typeName, msg.nested_type, parent=msg)
# Generate the set of messages that this type is dependent upon.
deps = set([field.type_name for field in msg.field
if field.type == TYPE['MESSAGE']])
dependencies.append((typeName, deps))
# Start by visiting the file's top-level message types.
visit('.' + fileObject.package, fileObject.message_type)
sortedMsgNames = order_dependencies(dependencies)
return [msgDict[n] for n in sortedMsgNames]
