# -*- Mode: Python; test-case-name: whipper.test.test_common_accurip -*-
# vi:si:et:sw=4:sts=4:ts=4

# Copyright (C) 2017 Samantha Baldwin
# Copyright (C) 2009 Thomas Vander Stichele

# This file is part of whipper.
#
# whipper is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
#
# whipper is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with whipper.  If not, see <http://www.gnu.org/licenses/>.

import struct
import whipper
from os import makedirs
from os.path import dirname, exists, join
from urllib.error import URLError, HTTPError
from urllib.request import urlopen, Request

from whipper.common import directory
from whipper.program.arc import accuraterip_checksum

import logging
logger = logging.getLogger(__name__)


ACCURATERIP_URL = "http://www.accuraterip.com/accuraterip/"
_CACHE_DIR = join(directory.cache_path(), 'accurip')


class EntryNotFound(Exception):
    pass


class _AccurateRipResponse:
    """
    An AccurateRip response contains a collection of metadata identifying a
    particular digital audio compact disc.

    For disc level metadata it contains the track count, two internal disc
    IDs, and the CDDB disc ID.

    A checksum and a confidence score is stored sequentially for each track in
    the disc index, which excludes any audio hidden in track pre-gaps (such as
    HTOA).

    The response is stored as a packed binary structure.
    """
    def __init__(self, data):
        """
        The checksums and confidences arrays are indexed by relative track
        position, so track 1 will have array index 0, track 2 will have array
        index 1, and so forth. HTOA and other hidden tracks are not included.
        """
        self.num_tracks = data[0]
        self.discId1 = "%08x" % struct.unpack("<L", data[1:5])[0]
        self.discId2 = "%08x" % struct.unpack("<L", data[5:9])[0]
        self.cddbDiscId = "%08x" % struct.unpack("<L", data[9:13])[0]

        self.confidences = []
        self.checksums = []
        pos = 13
        for _ in range(self.num_tracks):
            confidence = data[pos]
            checksum = "%08x" % struct.unpack("<L", data[pos + 1:pos + 5])[0]
            self.confidences.append(confidence)
            self.checksums.append(checksum)
            pos += 9

    def __eq__(self, other):
        return [
            self.num_tracks, self.discId1, self.discId2, self.cddbDiscId,
            self.confidences, self.checksums
        ] == [
            other.num_tracks, other.discId1, other.discId2, other.cddbDiscId,
            other.confidences, other.checksums
        ]


def _split_responses(raw_entry):
    responses = []
    while raw_entry:
        track_count = raw_entry[0]
        nbytes = 1 + 12 + track_count * (1 + 8)
        responses.append(_AccurateRipResponse(raw_entry[:nbytes]))
        raw_entry = raw_entry[nbytes:]
    return responses


def calculate_checksums(track_paths):
    """
    Return ARv1 and ARv2 checksums as two arrays of character strings in a
    dictionary: {'v1': ['deadbeef', ...], 'v2': [...]}

    Return None instead of checksum string for unchecksummable tracks.

    HTOA checksums are not included in the database and are not calculated.
    """
    track_count = len(track_paths)
    v1_checksums = []
    v2_checksums = []
    logger.debug('checksumming %d tracks', track_count)
    # This is done sequentially because it is very fast.
    for i, path in enumerate(track_paths):
        v1_sum, v2_sum = accuraterip_checksum(path, i+1, track_count)
        if v1_sum is None:
            logger.error('could not calculate AccurateRip v1 checksum '
                         'for track %d %r', i + 1, path)
            v1_checksums.append(None)
        else:
            v1_checksums.append("%08x" % v1_sum)
        if v2_sum is None:
            logger.error('could not calculate AccurateRip v2 checksum '
                         'for track %d %r', i + 1, path)
            v2_checksums.append(None)
        else:
            v2_checksums.append("%08x" % v2_sum)
    return {'v1': v1_checksums, 'v2': v2_checksums}


def _download_entry(path):
    url = ACCURATERIP_URL + path
    UA = "whipper/%s ( https://github.com/whipper-team/whipper )" % whipper.__version__  # noqa: E501
    logger.debug('downloading AccurateRip entry from %s', url)
    try:
        with urlopen(Request(url, headers={'User-Agent': UA})) as resp:
            return resp.read()
    except (URLError, HTTPError) as e:
        logger.error('error retrieving AccurateRip entry: %s', e)


def _save_entry(raw_entry, path):
    logger.debug('saving AccurateRip entry to %s', path)
    try:
        makedirs(dirname(path), exist_ok=True)
    except OSError as e:
        logger.error('could not save entry to %s: %s', path, e)
        return
    with open(path, 'wb') as f:
        f.write(raw_entry)


def get_db_entry(path):
    """
    Retrieve cached AccurateRip disc entry as array of _AccurateRipResponses.
    Downloads entry from accuraterip.com on cache fault.

    `path' is in the format of the output of table.accuraterip_path().
    """
    cached_path = join(_CACHE_DIR, path)
    if exists(cached_path):
        logger.debug('found accuraterip entry at %s', cached_path)
        with open(cached_path, 'rb') as f:
            raw_entry = f.read()
    else:
        raw_entry = _download_entry(path)
        if raw_entry:
            _save_entry(raw_entry, cached_path)
    if not raw_entry:
        logger.warning('entry not found in AccurateRip database')
        raise EntryNotFound
    return _split_responses(raw_entry)


def _assign_checksums_and_confidences(tracks, checksums, responses):
    for i, track in enumerate(tracks):
        for v in ('v1', 'v2'):
            track.AR[v]['CRC'] = checksums[v][i]
        track.AR['DBMaxConfidence'], track.AR['DBMaxConfidenceCRC'] = max(
            [(r.confidences[i], r.checksums[i]) for r in responses],
            key=lambda t: t[0]
        )


def _match_responses(tracks, responses):
    """
    Match and save track accuraterip response checksums against
    all non-hidden tracks.

    Returns True if every track has a match for every entry for either
    AccurateRip version.
    """
    for r in responses:
        for i, track in enumerate(tracks):
            for v in ('v1', 'v2'):
                if track.AR[v]['CRC'] == r.checksums[i]:
                    if (track.AR[v]['DBConfidence'] is None or
                            r.confidences[i] > track.AR[v]['DBConfidence']):
                        track.AR[v]['DBCRC'] = r.checksums[i]
                        track.AR[v]['DBConfidence'] = r.confidences[i]
                        logger.debug(
                            'track %d matched response %s in AccurateRip'
                            ' database: %s crc %s confidence %s',
                            i, r.cddbDiscId, v, track.AR[v]['DBCRC'],
                            track.AR[v]['DBConfidence'])
    return any((
        all([t.AR['v1']['DBCRC'] for t in tracks]),
        all([t.AR['v2']['DBCRC'] for t in tracks])
    ))


def verify_result(result, responses, checksums):
    """
    Verify track AccurateRip checksums against database responses.
    Stores track checksums and database values on result.
    """
    if not (result and responses and checksums):
        return False
    # exclude HTOA from AccurateRip verification
    # NOTE: if pre-gap hidden audio support is expanded to include
    # tracks other than HTOA, this is invalid.
    tracks = [t for t in result.tracks if t.number != 0]
    if not tracks:
        return False
    _assign_checksums_and_confidences(tracks, checksums, responses)
    return _match_responses(tracks, responses)


def print_report(result):
    """
    Print AccurateRip verification results.
    """
    for _, track in enumerate(result.tracks):
        status = 'rip NOT accurate'
        conf = '(not found)'
        db = 'notfound'
        if track.AR['DBMaxConfidence'] is not None:
            db = track.AR['DBMaxConfidenceCRC']
            conf = '(max confidence    %3d)' % track.AR['DBMaxConfidence']
            if track.AR['v1']['DBCRC'] or track.AR['v2']['DBCRC']:
                status = 'rip accurate'
                db = ', '.join([_f for _f in (
                    track.AR['v1']['DBCRC'],
                    track.AR['v2']['DBCRC']
                ) if _f])
            max_conf = max(
                [track.AR[v]['DBConfidence'] for v in ('v1', 'v2')
                 if track.AR[v]['DBConfidence'] is not None], default=None
            )
            if max_conf:
                if max_conf < track.AR['DBMaxConfidence']:
                    conf = '(confidence %3d of %3d)' % (
                        max_conf, track.AR['DBMaxConfidence']
                    )
        # htoa tracks (i == 0) do not have an ARCRC
        if track.number == 0:
            print('track  0: unknown          (not tracked)')
            continue
        if not (track.AR['v1']['CRC'] or track.AR['v2']['CRC']):
            logger.error('no track AR CRC on non-HTOA track %d', track.number)
            print('track %2d: unknown          (error)' % track.number)
        else:
            print('track %2d: %-16s %-23s v1 [%s], v2 [%s], DB [%s]' % (
                track.number, status, conf,
                track.AR['v1']['CRC'], track.AR['v2']['CRC'], db
            ))