Generate release notes from git commit messages

change_log.py

#!/usr/bin/python

# A script to automate changelog generation from Git commit messages
#
# For use with a git-flow workflow, it will take changes from the last tagged release
# where commit messages contain NEW, FIXED, and IMPROVED keywords and sort and fromat
# them into a Markdown release note list.
#
# Forked from https://gist.github.com/ttscoff/17fbce4f229609082b45681bba7a9967


from git import Repo
import re
from collections import OrderedDict


class LogEntry:
    def __init__(self, log_hash, date, title, msg_lines, *args, **kwargs):
        super(LogEntry, self).__init__(*args, **kwargs)
        self.log_hash = log_hash
        self.date = date
        self.title = title
        self.msg_lines = msg_lines

    def __str__(self):
        return self.log_hash


class ChangeLoger:

    def __init__(self, repo, *args, **kwargs):
        super(ChangeLoger, self).__init__(*args, **kwargs)
        sections = OrderedDict([
            ('new', {'title': "NEW", 'rx': r'(NEW|ADD(ED)?)'}),
            ('improved', {'title': "IMPROVED", 'rx': r'IMPROV(MENT|ED)?'}),
            ('fixed', {'title': "FIXED", 'rx': r'FIX(ED)?'}),
        ])
        for k, v in sections.items():
            sections[k]['rx'] = re.compile(
                r'^([\s]*(-)?[\s]*)' +
                sections[k]['rx'] +
                r'[\s]*:?[\s]*(?P<msg>.*)'
            )
        self.sections = sections
        log_entries = self.gitlog(repo)
        self.changes = self.find_changes(log_entries)

    def gitlog(self, repo):
        git = repo.git
        last_tag_hash = git.rev_list(tags=True, max_count=1)
        tag_time = git.show(last_tag_hash, format='%ad', s=True)
        log = git.log(pretty="format:===%h%n%ci%n%s%n%b", since=tag_time)

        if log and len(log) > 0:
            log_entries = [
                ChangeLoger.split_gitlog(entry.strip())
                for entry in log.split('===')
                if entry.strip() != ""
            ]
            return log_entries

        raise Exception("error reading log items")

    @staticmethod
    def split_gitlog(entry):
        '''Split a raw log entry into a clean Change object'''
        lines = entry.split("\n")
        log_hash = lines.pop(0)
        date = lines.pop(0)
        title = lines.pop(0)
        changes = [l.strip() for l in lines if not l.isspace()]
        return LogEntry(log_hash, date, title, changes)

    def find_changes(self, log_entries):
        '''Split the log_entries into changes'''
        changes = {k: [] for k, v in self.sections.items()}

        msg_lines = [line for entry in log_entries for line in entry.msg_lines]
        for line in msg_lines:
            for k, v in self.sections.items():
                match = v['rx'].match(line)
                if match:
                    changes[k].append(match.group('msg'))

        return changes

    def __str__(self):
        '''Build String representation'''
        output = ''
        for key, changes in self.changes.items():
            if len(changes) > 0:
                output = '{old_output}#### {title}\n{changes}\n'.format(
                    old_output=output,
                    title=self.sections[key]['title'],
                    changes='\n'.join(['- ' + line for line in changes])
                )

        if output == '':
            output = "Cool changes under the hood!"

        return output


if __name__ == "__main__":
    cl = ChangeLoger(Repo())
    print(cl)