#!/usr/bin/env python2.7 # # Copyright 2016, Rackspace US, Inc. # # 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. # # (c) 2016, Jesse Pretorius # """Read ansible-role-requirements.yml content from the CLI and output yaml content to stdout to be used when submitting release requests.""" from __future__ import print_function from cStringIO import StringIO try: from urllib.parse import urlparse except ImportError: from urlparse import urlparse import argparse import re import yaml # To ensure that the dicts are always output in the same order # we setup a representation for dict objects and register it # with the yaml class. def represent_dict(self, data): def key_function(elem): key = elem[0] # Prioritizes certain keys when sorting. prio = {"version": 0, "projects": 1, "repo": 2, "hash": 3}.get(key, 99) return (prio, key) items = data.items() items.sort(key=key_function) return self.represent_mapping(u'tag:yaml.org,2002:map', items) yaml.add_representer(dict, represent_dict) # sourced from # http://stackoverflow.com/questions/25108581/python-yaml-dump-bad-indentation def yaml_dump(dump, indentSize=2): stream = StringIO(dump) out = StringIO() pat = re.compile(r'(\s*)([^:]*)(:*)') last = None prefix = 0 for s in stream: indent, key, colon = pat.match(s).groups() if indent == "" and key[0] != '-': prefix = 0 if last: if len(last[0]) == len(indent) and last[2] == ':': if all([ not last[1].startswith('-'), s.strip().startswith('-')]): prefix += indentSize out.write(" " * prefix + s) last = indent, key, colon return out.getvalue() def main(): """Run the main application.""" # Setup argument parsing parser = argparse.ArgumentParser( description='release.yml producer', epilog='Licensed "Apache 2.0"') parser.add_argument( '-f', '--file', help=' ansible-role-requirements.yml file location', default='ansible-role-requirements.yml' ) parser.add_argument( '-v', '--version', help=' The release version to include in the output', required=True ) # Parse arguments args = parser.parse_args() # Read the ansible-role-requirements.yml file into memory with open(args.file, "r") as role_req_file: reqs = yaml.safe_load(role_req_file) # Prepare the vars for output version = args.version projects = [] # Prepare the regex match regex = re.compile('^.*openstack/(ansible-hardening|openstack-ansible.*)$') # Loop through the list of roles for role_data in reqs: # Only add OpenStack repositories to the release if regex.match(role_data['src']): # Prepare the repo release dict repo_release = {} # Figure out the repo from the git source repo = urlparse(role_data['src']).path.lstrip('/') # Assemble the dict repo_release['repo'] = repo repo_release['hash'] = role_data['version'] # Add the dict to the projects list projects.append(repo_release.copy()) # Put the yaml content together releases = {'releases': [{'version': version, 'projects': projects}]} # Product the YAML output for the resulting releases data output = yaml.dump(releases, default_flow_style=False) # Print the output, formatted as expected print(yaml_dump(output)) if __name__ == "__main__": main()