update_changelog_diffs.py

update_changelog_diffs.py

#!/usr/bin/env python3

"""Given the path to a Markdown-formatted change log, this script will update
the diff links for each version at the bottom of the document. For an example
of the diff links, see: https://keepachangelog.com

Currently only supports GitHub, GitLab, and Bitbucket diff links.
"""


import argparse
import os
import re
import sys
from distutils.version import LooseVersion
from urllib import parse

__author__ = "Elliot Jordan"
__version__ = "1.0"


def build_argument_parser():
    """Build and return the argument parser."""

    parser = argparse.ArgumentParser(
        description=__doc__, formatter_class=argparse.RawDescriptionHelpFormatter
    )
    parser.add_argument(
        "changelog",
        help="Path to the Markdown-formatted change log to update.",
    )
    parser.add_argument(
        "--prefix",
        default="v",
        help="The character or string that is added to the beginning of each tag. Defaults to 'v'.",
    )
    return parser


def main():
    """Main process."""

    # Parse command line arguments.
    argparser = build_argument_parser()
    args = argparser.parse_args()

    # Bail out if the file doesn't exist.
    if not os.path.isfile(args.changelog):
        sys.exit("ERROR: No such file: %s" % args.changelog)

    # Read the lines of the change log.
    with open(args.changelog, "r") as openfile:
        content = openfile.readlines()

    # Create a list to store versions.
    version_pattern = r"## \[?([\d\.]+)\]? - "
    found_versions = []

    # Keep track of where the diff links start.
    diff_pattern = r"^\[Unreleased\]: (https?:\/\/\S+\/\S+)"
    diff_start_index = -1

    # Need to know in order to create the diff link.
    repo_info = {}

    # Iterate through change log content, keeping track of the H2 headings
    # with versions we find along the way. Stop when we get to the diff links.
    for idx, line in enumerate(content):
        version_m = re.match(version_pattern, line)
        if version_m:
            found_versions.append(version_m.group(1))

        diff_m = re.match(diff_pattern, line)
        if diff_m:
            diff_start_index = idx
            repo_info["found_url"] = diff_m.group(1)
            # No need to read any further lines, since we'll be
            # regenerating them anyway.
            break

    # Bail out if any of these errors prevent continuing.
    if diff_start_index < 0:
        sys.exit("ERROR: No existing diff links found.")
    if not found_versions:
        sys.exit("ERROR: No H2 version headings found in change log.")
    if "found_url" not in repo_info:
        sys.exit("ERROR: No Git URL found in existing diff links.")

    # Link templates based on Git host.
    link_templates = {
        "github.com": "[%s]: https://github.com/%s/compare/%s...%s\n",
        "gitlab.com": "[%s]: https://gitlab.com/%s/-/compare/%s...%s\n",
        "bitbucket.org": "[%s]: https://bitbucket.org/%s/branches/compare/%s..%s\n",
    }

    # Based on Git URL, extrapolate information about the project that we can
    # use to create diff URLs.
    parsed_url = parse.urlparse(repo_info["found_url"])
    repo_info["scheme"] = parsed_url.scheme
    repo_info["hostname"] = parsed_url.hostname
    repo_info["project"] = "/".join(parsed_url.path.split("/")[1:3])

    # Bail out if we don't know how to create a diff URL for this host.
    if repo_info["hostname"] not in link_templates:
        sys.exit("ERROR: %s diff links are not yet supported." % repo_info["hostname"])

    # Sort versions in reverse.
    found_versions = sorted(found_versions, key=LooseVersion, reverse=True)

    # Start diff output with an Unreleased link comparing most recent version to HEAD.
    diff_output = [
        link_templates[repo_info["hostname"]]
        % ("Unreleased", repo_info["project"], args.prefix + found_versions[0], "HEAD")
    ]

    # Iterate (backwards) through the versions and add each link to the output.
    found_versions = sorted(found_versions, key=LooseVersion, reverse=True)
    for idx, vers in enumerate(found_versions):

        # For initial (highest index) version, there's nothing to compare to.
        if idx == len(found_versions) - 1:
            continue

        # For other versions, use diff template.
        diff_output.append(
            link_templates[repo_info["hostname"]]
            % (
                vers,
                repo_info["project"],
                args.prefix + found_versions[idx + 1],
                args.prefix + vers,
            )
        )

    # Write the updated change log.
    new_content = content[:diff_start_index] + diff_output
    with open(args.changelog, "w") as openfile:
        openfile.write("".join(new_content))
    print("Updated: %s" % args.changelog)


if __name__ == "__main__":
    main()