#!/usr/bin/python3
# generate debian/copyright from debian/copyright.template and bower_components
# Author: Martin Pitt <mpitt@debian.org>
#
# bower.json license: format uses https://spdx.org/licenses/ identifiers, which
# are mostly compatible with
# https://www.debian.org/doc/packaging-manuals/copyright-format/1.0/#license-specification

import os
import json
import re
import subprocess
from glob import glob

TEMPLATE = 'debian/copyright.template'


def template_licenses(template):
    '''Return set of existing License: short names'''

    ids = set()
    for l in template.splitlines():
        if l.startswith('License:'):
            ids.add(l.split(None, 1)[1])
    return ids


def module_license(moddir):
    '''Return License: short name for given module'''

    # First check if bower.json has a "license" field
    try:
        with open(os.path.join(moddir, 'bower.json')) as f:
            l = json.load(f)['license']
            if 'and MIT' in l:
                # https://github.com/requirejs/requirejs/issues/1645
                return 'MIT'
            return l
    except (FileNotFoundError, KeyError):
        pass

    # *LICENSE*/COPYING/README
    if os.path.exists(os.path.join(moddir, 'MIT-LICENSE.txt')):
        return 'MIT'
    try:
        with open((glob(os.path.join(moddir, 'LICENSE*')) +
                   glob(os.path.join(moddir, 'COPYING')) +
                   glob(os.path.join(moddir, 'README.md')))[0]) as f:
            text = f.read()
        if 'Permission is hereby granted,' in text and 'THE SOFTWARE IS PROVIDED "AS IS"' in text:
            return 'MIT'
        if 'Redistribution and use in source and binary forms' in text and 'THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT' in text:
            if '4. Neither' in text:
                return 'BSD-4-clause'
            else:
                return 'BSD-3-clause'
        if re.search('Apache License.*2.0', text):
            return 'Apache-2.0'
        if 'GNU LESSER GENERAL PUBLIC LICENSE' in text and 'Version 2.1' in text:
            return 'LGPL-2.1'
    except IndexError:
        pass

    # missing licenses
    mod = os.path.basename(moddir)
    if mod == 'kubernetes-object-describer':
        # upstream says "same as what we use for origin and origin-web-console which is Apache"
        # https://github.com/kubernetes-ui/object-describer/issues/31
        return 'Apache-2.0'
    if mod == 'redux':
        # our tarball only ships the minified version, original source is at
        # https://github.com/reactjs/redux; as it's "MIT" we don't legally
        # require to ship the preferred form of modification
        return 'MIT'

    raise SystemError('Could not determine license from %s' % moddir)


def module_copyright(moddir):
    '''Return Copyrights for given module'''

    copyrights = set()
    try:
        out = subprocess.check_output(['env', '-u', 'LANGUAGE', 'LC_ALL=C.UTF-8', 'grep', '-hri', 'copyright.*[1-9][0-9]\+'],
                                      cwd=moddir, universal_newlines=True)
        for l in out.splitlines():
            # ignore compiled .css.map files
            if l.startswith('{"version":') or '*//*!' in l:
                continue
            # weed out some noise
            if 'grunt.template.today' in l or 'copyright-mark' in l or '@font-face' in l or 'Binary file' in l:
                continue
            l = l.replace('<copyright>', '').replace('</copyright>', '')
            l = l.replace('&copy;', '(c)').replace('copyright = ', '')
            l = l.strip(' \t*/\'|')
            if l.startswith("u'") or l.startswith('u"'):  # Python unicode prefix
                l = l[2:]
            copyrights.add(' ' + l)  # space prefix for debian/copyright RFC822 format
    except subprocess.CalledProcessError:
        pass

    # missing copyrights
    mod = os.path.basename(moddir)
    if mod == 'angular-bootstrap':
        # only committer is Wesley Cho, so assume his copyright
        # https://github.com/angular-ui/bootstrap-bower/issues/74
        copyrights.add('Copyright: (C) 2015-2017 Wesley Cho')
    if mod == 'kubernetes-object-describer':
        # only committer is Jessica Forrester, working for Red Hat
        # https://github.com/kubernetes-ui/object-describer/issues/35
        copyrights.add('Copyright: (C) 2015-2017 Red Hat, Inc.')
    if mod == 'redux':
        # our tarball only ships the minified version, original source is at
        # https://github.com/reactjs/redux
        copyrights.add('Copyright (c) 2015-present Dan Abramov')

    if not copyrights:
        raise SystemError('Did not find any copyrights in %s' % moddir)
    return copyrights

#
# main
#

with open(TEMPLATE) as f:
    template = f.read()

license_ids = template_licenses(template)
paragraphs = []
for module in os.listdir('bower_components'):
    if module == 'README':
        continue
    moddir = os.path.join('bower_components', module)
    license = module_license(moddir)
    if license not in license_ids:
        raise KeyError('%s has license %s which is not in %s' % (module, license, TEMPLATE))
    copyrights = module_copyright(moddir)
    paragraphs.append('''Files: %s/*
Copyright:%s
License: %s''' % (moddir, '\n'.join(sorted(copyrights)), license))

with open('debian/copyright', 'w') as f:
    for l in template.splitlines():
        if '#BOWER' in l:
            f.write('\n\n'.join(paragraphs))
        else:
            f.write(l + '\n')