#!/usr/bin/python3
# -*- coding: utf-8 -*-
# (c) 2020 Matt Martz <matt@sivel.net>
# GNU General Public License v3.0+
#     (see https://www.gnu.org/licenses/gpl-3.0.txt)

import argparse
import json
import re
import sys
from pathlib import Path
from urllib.request import urlopen

from git import Repo  # gitpython

import yaml  # pyyaml

IGNORE_PATHS = set((
    'test/sanity/ignore.txt',
    'docs/docsite/rst/porting_guides/porting_guide_2.10.rst',
    'lib/ansible/config/module_defaults.yml',
))

IGNORE_COLLECTIONS = set((
    'sivel.jinja2',
))

EMPTY_SHA256 = (
    "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855"
)

parser = argparse.ArgumentParser()
parser.add_argument('path', help='Path to ansible-base source', type=Path)
args = parser.parse_args()


repo = Repo(str(args.path))

core_files = set()


class JSONEncoder(json.JSONEncoder):
    def default(self, o):
        if isinstance(o, set):
            return list(o)
        return json.JSONEncoder.default(self, o)


def is_core_file(path):
    if path in core_files:
        return True

    full_path = args.path / path
    if full_path.exists():
        core_files.add(path)
        return True

    return False


def get_collections(pattern):
    f = urlopen(
        'https://sivel.eng.ansible.com/api/v1/collections/provides',
        data=json.dumps({'path': pattern.pattern}).encode()
    )
    data = json.load(f)
    collections = set()
    for name, info in data.items():
        if name in IGNORE_COLLECTIONS or name.startswith('testing.'):
            continue

        files = info['files']
        if (len(files) == 1 and files[0]['chksum_sha256'] == EMPTY_SHA256 and
                files[0]['name'].endswith('/__init__.py')):
            continue
        collections.add(name)
    return collections


def path_pattern(path):
    try:
        path = Path(path).relative_to('lib/ansible/')
    except ValueError:
        path = Path(path)
    parts = str(path).split('/')
    if parts[0] in ('modules', 'module_utils'):
        pattern = re.compile(
            r'plugins\/{}\/{}'.format(
                re.escape(parts[0]),
                re.escape(path.name)
            )
        )
    elif parts[0] == 'plugins':
        pattern = re.compile(
            re.escape(str(path))
        )
    elif parts[0] == 'test':
        parts[0] = 'tests'
        pattern = re.compile(
            re.escape('/'.join(parts))
        )
    else:
        raise ValueError
    return pattern


def get_changed_files(path):
    for commit, _ in repo.blame('HEAD', path):
        yield from commit.stats.files.keys()


moves = {}
fragment_dir = args.path / 'changelogs' / 'fragments'
for fragment in fragment_dir.iterdir():
    rel_frag = str(fragment.relative_to(args.path))
    changed_files = sorted(set(get_changed_files(rel_frag)))
    for filename in changed_files:
        if filename.startswith('changelogs/'):
            continue

        if filename in IGNORE_PATHS:
            continue

        if is_core_file(filename):
            collections = set(('ansible.builtin',))
            pattern = re.compile(
                'core - {}'.format(re.escape(filename))
            )
        else:
            try:
                pattern = path_pattern(filename)
            except ValueError:
                continue
            collections = get_collections(pattern)

        if collections:
            print(rel_frag, file=sys.stderr)
            print(
                '    {!r} - {}'.format(collections, pattern.pattern),
                file=sys.stderr
            )

            try:
                moves[rel_frag]['collections'].update(collections)
            except KeyError:
                clog_data = yaml.safe_load(fragment.read_text())
                moves[rel_frag] = {
                    'collections': collections,
                    'changelog': clog_data,
                }

print(file=sys.stderr)
print(json.dumps(moves, sort_keys=True, indent=4, cls=JSONEncoder))