mirror of
https://github.com/yuzu-emu/mbedtls.git
synced 2024-11-23 23:45:41 +01:00
82ebaa4d36
Parse the existing dependencies. For now, just write them back. Subsequent commits will implement the dependency processing that is the goal of this program. Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
133 lines
5.2 KiB
Python
Executable File
133 lines
5.2 KiB
Python
Executable File
#!/usr/bin/env python3
|
|
|
|
"""Edit test cases to use PSA dependencies instead of classic dependencies.
|
|
"""
|
|
|
|
# Copyright The Mbed TLS Contributors
|
|
# SPDX-License-Identifier: Apache-2.0
|
|
#
|
|
# 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.
|
|
|
|
import os
|
|
import re
|
|
import sys
|
|
|
|
def updated_dependencies(file_name, function_name, arguments, dependencies):
|
|
"""Rework the list of dependencies into PSA_WANT_xxx.
|
|
|
|
Remove classic crypto dependencies such as MBEDTLS_RSA_C,
|
|
MBEDTLS_PKCS1_V15, etc.
|
|
|
|
Add systematic PSA_WANT_xxx dependencies based on the called function and
|
|
its arguments, replacing existing PSA_WANT_xxx dependencies.
|
|
"""
|
|
return dependencies #TODO
|
|
|
|
def process_data_stanza(stanza, file_name, test_case_number):
|
|
"""Update PSA crypto dependencies in one Mbed TLS test case.
|
|
|
|
stanza is the test case text (including the description, the dependencies,
|
|
the line with the function and arguments, and optionally comments). Return
|
|
a new stanza with an updated dependency line, preserving everything else
|
|
(description, comments, arguments, etc.).
|
|
"""
|
|
if not stanza.lstrip('\n'):
|
|
# Just blank lines
|
|
return stanza
|
|
# Expect 2 or 3 non-comment lines: description, optional dependencies,
|
|
# function-and-arguments.
|
|
content_matches = list(re.finditer(r'^[\t ]*([^\t #].*)$', stanza, re.M))
|
|
if len(content_matches) < 2:
|
|
raise Exception('Not enough content lines in paragraph {} in {}'
|
|
.format(test_case_number, file_name))
|
|
if len(content_matches) > 3:
|
|
raise Exception('Too many content lines in paragraph {} in {}'
|
|
.format(test_case_number, file_name))
|
|
arguments = content_matches[-1].group(0).split(':')
|
|
function_name = arguments.pop(0)
|
|
if len(content_matches) == 2:
|
|
# Insert a line for the dependencies. If it turns out that there are
|
|
# no dependencies, we'll remove that empty line below.
|
|
dependencies_location = content_matches[-1].start()
|
|
text_before = stanza[:dependencies_location]
|
|
text_after = '\n' + stanza[dependencies_location:]
|
|
old_dependencies = []
|
|
dependencies_leader = 'depends_on:'
|
|
else:
|
|
dependencies_match = content_matches[-2]
|
|
text_before = stanza[:dependencies_match.start()]
|
|
text_after = stanza[dependencies_match.end():]
|
|
old_dependencies = dependencies_match.group(0).split(':')
|
|
dependencies_leader = old_dependencies.pop(0) + ':'
|
|
if dependencies_leader != 'depends_on:':
|
|
raise Exception('Next-to-last line does not start with "depends_on:"'
|
|
' in paragraph {} in {}'
|
|
.format(test_case_number, file_name))
|
|
new_dependencies = updated_dependencies(file_name, function_name, arguments,
|
|
old_dependencies)
|
|
if new_dependencies:
|
|
stanza = (text_before +
|
|
dependencies_leader + ':'.join(new_dependencies) +
|
|
text_after)
|
|
else:
|
|
# The dependencies have become empty. Remove the depends_on: line.
|
|
assert text_after[0] == '\n'
|
|
stanza = text_before + text_after[1:]
|
|
return stanza
|
|
|
|
def process_data_file(file_name, old_content):
|
|
"""Update PSA crypto dependencies in an Mbed TLS test suite data file.
|
|
|
|
Process old_content (the old content of the file) and return the new content.
|
|
"""
|
|
old_stanzas = old_content.split('\n\n')
|
|
new_stanzas = [process_data_stanza(stanza, file_name, n)
|
|
for n, stanza in enumerate(old_stanzas, start=1)]
|
|
return '\n\n'.join(new_stanzas)
|
|
|
|
def update_file(file_name, old_content, new_content):
|
|
"""Update the given file with the given new content.
|
|
|
|
Replace the existing file. The previous version is renamed to *.bak.
|
|
Don't modify the file if the content was unchanged.
|
|
"""
|
|
if new_content == old_content:
|
|
return
|
|
backup = file_name + '.bak'
|
|
tmp = file_name + '.tmp'
|
|
with open(tmp, 'w', encoding='utf-8') as new_file:
|
|
new_file.write(new_content)
|
|
os.replace(file_name, backup)
|
|
os.replace(tmp, file_name)
|
|
|
|
def process_file(file_name):
|
|
"""Update PSA crypto dependencies in an Mbed TLS test suite data file.
|
|
|
|
Replace the existing file. The previous version is renamed to *.bak.
|
|
Don't modify the file if the content was unchanged.
|
|
"""
|
|
old_content = open(file_name, encoding='utf-8').read()
|
|
if file_name.endswith('.data'):
|
|
new_content = process_data_file(file_name, old_content)
|
|
else:
|
|
raise Exception('File type not recognized: {}'
|
|
.format(file_name))
|
|
update_file(file_name, old_content, new_content)
|
|
|
|
def main(args):
|
|
for file_name in args:
|
|
process_file(file_name)
|
|
|
|
if __name__ == '__main__':
|
|
main(sys.argv[1:])
|