The C based gRPC (C++, Python, Ruby, Objective-C, PHP, C#) https://grpc.io/
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

159 lines
4.9 KiB

#!/usr/bin/env python3
# Copyright 2021 gRPC authors.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# Eliminate the kind of redundant namespace qualifiers that tend to
# creep in when converting C to C++.
import collections
import os
import re
import sys
def find_closing_mustache(contents, initial_depth):
"""Find the closing mustache for a given number of open mustaches."""
depth = initial_depth
start_len = len(contents)
while contents:
# Skip over strings.
if contents[0] == '"':
contents = contents[1:]
while contents[0] != '"':
if contents.startswith('\\\\'):
contents = contents[2:]
elif contents.startswith('\\"'):
contents = contents[2:]
else:
contents = contents[1:]
contents = contents[1:]
# And characters that might confuse us.
elif contents.startswith("'{'") or contents.startswith(
"'\"'") or contents.startswith("'}'"):
contents = contents[3:]
# Skip over comments.
elif contents.startswith("//"):
contents = contents[contents.find('\n'):]
elif contents.startswith("/*"):
contents = contents[contents.find('*/') + 2:]
# Count up or down if we see a mustache.
elif contents[0] == '{':
contents = contents[1:]
depth += 1
elif contents[0] == '}':
contents = contents[1:]
depth -= 1
if depth == 0:
return start_len - len(contents)
# Skip over everything else.
else:
contents = contents[1:]
return None
def update_file(contents, namespaces):
"""Scan the contents of a file, and for top-level namespaces in namespaces remove redundant usages."""
output = ''
while contents:
m = re.search(r'namespace ([a-zA-Z0-9_]*) {', contents)
if not m:
output += contents
break
output += contents[:m.end()]
contents = contents[m.end():]
end = find_closing_mustache(contents, 1)
if end is None:
print('Failed to find closing mustache for namespace {}'.format(
m.group(1)))
print('Remaining text:')
print(contents)
sys.exit(1)
body = contents[:end]
namespace = m.group(1)
if namespace in namespaces:
while body:
# Find instances of 'namespace::'
m = re.search(r'\b' + namespace + r'::\b', body)
if not m:
break
# But ignore instances of '::namespace::' -- these are usually meant to be there.
if m.start() >= 2 and body[m.start() - 2:].startswith('::'):
output += body[:m.end()]
else:
output += body[:m.start()]
body = body[m.end():]
output += body
contents = contents[end:]
return output
# self check before doing anything
_TEST = """
namespace bar {
namespace baz {
}
}
namespace foo {}
namespace foo {
foo::a;
}
"""
_TEST_EXPECTED = """
namespace bar {
namespace baz {
}
}
namespace foo {}
namespace foo {
a;
}
"""
output = update_file(_TEST, ['foo'])
if output != _TEST_EXPECTED:
import difflib
print('FAILED: self check')
print('\n'.join(
difflib.ndiff(_TEST_EXPECTED.splitlines(1), output.splitlines(1))))
sys.exit(1)
# Main loop.
Config = collections.namedtuple('Config', ['dirs', 'namespaces'])
_CONFIGURATION = (Config(['src/core', 'test/core'], ['grpc_core']),)
changed = []
for config in _CONFIGURATION:
for dir in config.dirs:
for root, dirs, files in os.walk(dir):
for file in files:
if file.endswith('.cc') or file.endswith('.h'):
path = os.path.join(root, file)
try:
with open(path) as f:
contents = f.read()
except IOError:
continue
updated = update_file(contents, config.namespaces)
if updated != contents:
changed.append(path)
with open(os.path.join(root, file), 'w') as f:
f.write(updated)
if changed:
print('The following files were changed:')
for path in changed:
print(' ' + path)
sys.exit(1)