genlog

#!/usr/bin/env python

"""Generate CHANGELOG

Idea by Matt DeBoard
http://mattdeboard.net/2014/01/14/automatic-changelog-generation-with-git/
"""

import click
import datetime
import jinja2
import json
import os
import re
import subprocess


GIT_FORMAT = (
    '--pretty=format:{'
    'STRXSEPrevisionSTRXSEP: STRXSEP%hSTRXSEP, '
    'STRXSEPsubjectSTRXSEP: STRXSEP%sSTRXSEP, '
    'STRXSEPbodySTRXSEP: STRXSEP%bSTRXSEP}EOFL'
)

EXTRACT_PR = re.compile("Merge pull request #(\d+) ")

TEMPLATE = """
    {
        'version' : "{{ version }}",
        'date'    : "{{ date }}",
        'author'  : "{{ author }}",
        'changes' : [ {% for change in changes %}
            "{{ change.subject }}",
            [
                {% if change.uri %}"{{ change.uri }}",
                {% endif %}{% if change.body %}"{{ change.body }}",{% endif %}
            ],
            {% endfor %}
        ],
    },
"""


@click.command()
@click.option(
    '-m',
    '--merge',
    is_flag=True,
    help="Use this if you work with pull/merge-requests.",
)
@click.argument(
    'project_uri',
    nargs=1,
)
@click.argument(
    'version_file',
    nargs=1,
)
@click.argument(
    'start_commit',
    nargs=1,
)
@click.argument(
    'end_commit',
    nargs=1,
)
def genlog(merge, project_uri, version_file, start_commit, end_commit):
    """Generate CHANGELOG. Preprend the result to the CHANGELOG file and edit
    it as you wish.

    PROJECT_URI     Reference to a github project

    VERSION_FILE    File to read the version from

    START_COMMIT    Generate the log from this revision or tag

    END_COMMIT      Generate the log to this revision or tag
    """
    logs     = list(get_commit_msgs(merge, start_commit, end_commit))
    author   = get_author()
    date     = datetime.datetime.utcnow().strftime("%Y-%m-%dT%H:%M:%S+0000")
    version  = get_version(version_file)
    changes  = []
    template = jinja2.Template(TEMPLATE)
    for log in logs:
        subject   = log['subject']
        revision  = log['revision']
        match     = EXTRACT_PR.findall(subject)
        body      = log['body']
        bodysplit = body.split("LINEXSEP")
        bs_len    = len(bodysplit)
        newbody   = []
        for x, line in enumerate(bodysplit):
            if x < bs_len - 1:
                next_line = bodysplit[x + 1]
            else:
                next_line = None
            line = line.strip()
            fix = False
            if line.startswith("*"):
                fix = True
            elif line.startswith("-"):
                fix = True
            if fix:
                line = line[1:]
                line = line.strip()
                fix = False
                if next_line:
                    if next_line.startswith("*"):
                        fix = True
                    elif next_line.startswith("-"):
                        fix = True
                    if fix:
                        line = "%s." % line
            newbody.append(line)
        if match:
            pr = match[0]
            uri = "%s/pull/%s" % (project_uri, pr)
        else:
            uri = "%s/commit/%s" % (project_uri, revision)
        change = {
            'uri'      : uri,
            'subject'  : subject,
            'revision' : revision,
            'body'     : " ".join(newbody),
        }
        changes.append(change)
    print(template.render(
        version = version,
        author  = author,
        date    = date,
        changes = changes,
    ))


def get_version(version_file):
    """Get the version"""
    env = {}
    with open(version_file) as f:
        code = compile(f.read(), version_file, 'exec')
        exec(code, env)
    return env['__version__']


def get_commit_msgs(merge, start_commit, end_commit):
    """Get the commits in a parseable format"""
    args = []
    args.extend([
        "git",
        "log",
        GIT_FORMAT,
    ])
    if merge:
        args.append("--merges")
    args.append("%s...%s" % (start_commit, end_commit))
    res = subprocess.check_output(args).decode("UTF-8").strip()[:-1]
    # Escaping invalid json
    res = "\\\"".join(res.split("\""))
    res = "\"".join(res.split("STRXSEP"))
    res = "    ".join(res.split('\t'))
    lines = res.split("EOFL%s" % os.linesep)
    lines[-1] = lines[-1][:-3]
    for line in lines:
        if line:
            line = "LINEXSEP".join(line.split(os.linesep))
            # Clean all \n \r depending on platform we need one or the other
            line = "".join(line.split("\r"))
            line = "".join(line.split("\n"))
            yield json.loads("%s" % line)


def get_author():
    """Get the author of the changelog (current git user"""
    name = subprocess.check_output([
        "git",
        "config",
        "user.name",
    ]).decode("UTF-8").strip()
    email = subprocess.check_output([
        "git",
        "config",
        "user.email",
    ]).decode("UTF-8").strip()
    return "%s <%s>" % (name, email)


if __name__ == "__main__":
    genlog()