|
|
|
#!/usr/bin/env python
|
|
|
|
|
|
|
|
# Licensed to the Apache Software Foundation (ASF) under one or more
|
|
|
|
# contributor license agreements. See the NOTICE file distributed with
|
|
|
|
# this work for additional information regarding copyright ownership.
|
|
|
|
# The ASF licenses this file to You 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.
|
|
|
|
|
|
|
|
"""Usage: release_notes.py <version> > RELEASE_NOTES.html
|
|
|
|
|
|
|
|
Generates release notes for a Kafka release by generating an HTML doc containing some introductory information about the
|
|
|
|
release with links to the Kafka docs followed by a list of issues resolved in the release. The script will fail if it finds
|
|
|
|
any unresolved issues still marked with the target release. You should run this script after either resolving all issues or
|
|
|
|
moving outstanding issues to a later release.
|
|
|
|
|
|
|
|
"""
|
|
|
|
|
|
|
|
from jira import JIRA
|
|
|
|
import itertools, sys
|
|
|
|
|
|
|
|
if len(sys.argv) < 2:
|
|
|
|
print >>sys.stderr, "Usage: release_notes.py <version>"
|
|
|
|
sys.exit(1)
|
|
|
|
|
|
|
|
version = sys.argv[1]
|
|
|
|
minor_version_dotless = "".join(version.split(".")[:3]) # i.e., 0.10.0 if version == 0.10.0.1
|
|
|
|
|
|
|
|
JIRA_BASE_URL = 'https://issues.apache.org/jira'
|
|
|
|
MAX_RESULTS = 100 # This is constrained for cloud instances so we need to fix this value
|
|
|
|
|
|
|
|
def get_issues(jira, query, **kwargs):
|
|
|
|
"""
|
|
|
|
Get all issues matching the JQL query from the JIRA instance. This handles expanding paginated results for you. Any additional keyword arguments are forwarded to the JIRA.search_issues call.
|
|
|
|
"""
|
|
|
|
results = []
|
|
|
|
startAt = 0
|
|
|
|
new_results = None
|
|
|
|
while new_results == None or len(new_results) == MAX_RESULTS:
|
|
|
|
new_results = jira.search_issues(query, startAt=startAt, maxResults=MAX_RESULTS, **kwargs)
|
|
|
|
results += new_results
|
|
|
|
startAt += len(new_results)
|
|
|
|
return results
|
|
|
|
|
|
|
|
def issue_link(issue):
|
|
|
|
return "%s/browse/%s" % (JIRA_BASE_URL, issue.key)
|
|
|
|
|
|
|
|
|
|
|
|
if __name__ == "__main__":
|
|
|
|
apache = JIRA(JIRA_BASE_URL)
|
|
|
|
issues = get_issues(apache, 'project=KAFKA and fixVersion=%s' % version)
|
|
|
|
if not issues:
|
|
|
|
print >>sys.stderr, "Didn't find any issues for the target fix version"
|
|
|
|
sys.exit(1)
|
|
|
|
|
|
|
|
unresolved_issues = [issue for issue in issues if issue.fields.resolution is None]
|
|
|
|
if unresolved_issues:
|
|
|
|
print >>sys.stderr, "The release is not completed since unresolved issues were found still tagged with this release as the fix version:"
|
|
|
|
for issue in unresolved_issues:
|
|
|
|
print >>sys.stderr, "Unresolved issue: %s %s" % (issue.key, issue_link(issue))
|
|
|
|
sys.exit(1)
|
|
|
|
|
|
|
|
# Get list of (issue type, [issues]) sorted by the issue ID type, with each subset of issues sorted by their key so they
|
|
|
|
# are in increasing order of bug #. To get a nice ordering of the issue types we customize the key used to sort by issue
|
|
|
|
# type a bit to ensure features and improvements end up first.
|
|
|
|
def issue_type_key(issue):
|
|
|
|
if issue.fields.issuetype.name == 'New Feature':
|
|
|
|
return -2
|
|
|
|
if issue.fields.issuetype.name == 'Improvement':
|
|
|
|
return -1
|
|
|
|
return issue.fields.issuetype.id
|
|
|
|
by_group = [(k,sorted(g, key=lambda issue: issue.id)) for k,g in itertools.groupby(sorted(issues, key=issue_type_key), lambda issue: issue.fields.issuetype.name)]
|
|
|
|
|
|
|
|
print "<h1>Release Notes - Kafka - Version %s</h1>" % version
|
|
|
|
print """<p>Below is a summary of the JIRA issues addressed in the %(version)s release of Kafka. For full documentation of the
|
|
|
|
release, a guide to get started, and information about the project, see the <a href="http://kafka.apache.org/">Kafka
|
|
|
|
project site</a>.</p>
|
|
|
|
|
|
|
|
<p><b>Note about upgrades:</b> Please carefully review the
|
|
|
|
<a href="http://kafka.apache.org/%(minor)s/documentation.html#upgrade">upgrade documentation</a> for this release thoroughly
|
|
|
|
before upgrading your cluster. The upgrade notes discuss any critical information about incompatibilities and breaking
|
|
|
|
changes, performance changes, and any other changes that might impact your production deployment of Kafka.</p>
|
|
|
|
|
|
|
|
<p>The documentation for the most recent release can be found at
|
|
|
|
<a href="http://kafka.apache.org/documentation.html">http://kafka.apache.org/documentation.html</a>.</p>""" % { 'version': version, 'minor': minor_version_dotless }
|
|
|
|
for itype, issues in by_group:
|
|
|
|
print "<h2>%s</h2>" % itype
|
|
|
|
print "<ul>"
|
|
|
|
for issue in issues:
|
|
|
|
print '<li>[<a href="%(link)s">%(key)s</a>] - %(summary)s</li>' % {'key': issue.key, 'link': issue_link(issue), 'summary': issue.fields.summary}
|
|
|
|
print "</ul>"
|