import typing

import pandas
from dataclasses import dataclass
from math import floor
from datetime import datetime


def timestamp_range_seconds(start: pandas.Timestamp, end: pandas.Timestamp) -> typing.Iterator[pandas.Timestamp]:
    assert end >= start
    start = int(floor(start.timestamp()))
    end = int(floor(end.timestamp()))
    for second in range(start, end):
        yield pandas.Timestamp(second, unit='s')

def _add_accumulated_score(df: pandas.DataFrame):
    acc_col = pandas.Series([0.0]).repeat(len(df)).reset_index(drop=True)

    acc = 0.0
    for i, row in enumerate(df.itertuples()):
        acc += row.score
        acc_col[i] = acc

    df['accumulated_score'] = acc_col

def load_score_log(path: str) -> pandas.DataFrame:
    return pandas.read_csv(path, sep=',',
                           dtype={'score': int, 'sourcename': str, 'name': str, 'mapx': int, 'mapy': int},
                           parse_dates=['when'], date_format='%d/%m/%Y %H:%M')

def get_score_per(sourcename: str) -> float | None:
    if sourcename == 'Capture':
        return 1.0
    elif sourcename == 'Output Boost':
        return 0.1
    else:
        return None

def extend_score_log(scores: pandas.DataFrame):
    scores.sort_values('when', inplace=True)
    _add_accumulated_score(scores)

    def _calc_duration(row) -> int:
        score_per = get_score_per(row['sourcename'])
        if score_per is None:
            return 1
        return int(floor(row['score'] / score_per))

    def _calc_event_start(row) -> pandas.Timestamp:
        return pandas.Timestamp(row['when'].timestamp() - row['seconds'], unit='s')

    scores['seconds'] = scores.apply(_calc_duration, axis=1)
    scores['when_start'] = scores.apply(_calc_event_start, axis=1)


def generate_station_stats(score_log: pandas.DataFrame) -> pandas.DataFrame:
    station_count = len(score_log['name'].unique())

    # every station in the score log should have a first visit, so create summary based on that
    summary = score_log[score_log['sourcename'] == 'First Visit'][['name', 'mapx', 'mapy', 'when']]
    summary.rename(columns={'when': 'first_visit'}, inplace=True)

    assert len(summary) == station_count

    common_join_args = {'on': 'name', 'how': 'left', 'validate': '1:1'}

    # add total score
    summary = pandas.merge(summary, score_log[['name', 'score']].groupby('name').sum(), **common_join_args)
    summary.rename(columns={'score': 'total_score'}, inplace=True)
    assert len(summary) == station_count

    boosts = score_log[score_log['sourcename'] == 'Output Boost'][['name', 'score']].groupby('name')

    # add total boosts
    total_boosts = boosts.sum()
    total_boosts['totalboostduration'] = total_boosts['score'].apply(lambda x: 10 * x)
    total_boosts.rename(columns={'score': 'totalboostscore'}, inplace=True)
    summary = pandas.merge(summary, total_boosts, **common_join_args)

    # add max boosts
    max_boosts = boosts.max()
    max_boosts['maxboostduration'] = max_boosts['score'].apply(lambda x: 10 * x)
    max_boosts.rename(columns={'score': 'maxboostscore'}, inplace=True)
    summary = pandas.merge(summary, max_boosts, **common_join_args)

    visits = score_log[(score_log['sourcename'] == 'Visit') | (score_log['sourcename'] == 'First Visit')][
        ['name', 'score']].groupby('name')

    # add total visits (count)
    summary = pandas.merge(summary, visits.count(), **common_join_args)
    summary.rename(columns={'score': 'totalvisits'}, inplace=True)

    captures = score_log[score_log['sourcename'] == 'Capture'][['name', 'score']].groupby('name')

    # add captures (count)
    summary = pandas.merge(summary, captures.count(), **common_join_args)
    summary.rename(columns={'score': 'captures'}, inplace=True)

    # add max held duration (max capture score)
    summary = pandas.merge(summary, captures.max(), **common_join_args)
    summary.rename(columns={'score': 'maxheldduration'}, inplace=True)

    # add total held duration (sum capture score)
    summary = pandas.merge(summary, captures.sum(), **common_join_args)
    summary.rename(columns={'score': 'totalheldduration'}, inplace=True)

    assert len(summary) == station_count
    return summary


def generate_score_per_second(score_log: pandas.DataFrame) -> pandas.DataFrame:
    @dataclass
    class ScoreSecond:
        name: str
        sourcename: str
        when: datetime
        score: float
        once: bool
        event_start: bool
        mapx: int
        mapy: int

    def gen_scoreseconds() -> typing.Iterator[ScoreSecond]:
        for row in score_log.itertuples():
            # TODO: the code below should work with 0s now
            if row.seconds < 2:  # one-off
                yield ScoreSecond(name=row.name, sourcename=row.sourcename, mapx=row.mapx, mapy=row.mapy, when=row.when,
                                  score=row.score, once=True, event_start=True)
                continue

            once = row.seconds == 1
            score_per = get_score_per(row.sourcename)

            event_start = True
            for when in timestamp_range_seconds(row.when_start, row.when):
                yield ScoreSecond(when=when, once=once, event_start=event_start,
                                  name=row.name, sourcename=row.sourcename, mapx=row.mapx, mapy=row.mapy,
                                  score=score_per, )
                event_start = False

    scoreseconds = pandas.DataFrame(gen_scoreseconds())
    scoreseconds.sort_values(by=['when'], inplace=True)
    scoreseconds.reset_index(drop=True, inplace=True)

    _add_accumulated_score(scoreseconds)
    return scoreseconds

def get_known_player_locations(score_log):
    locations = score_log[score_log['mapx'] != 0][['name', 'when_start', 'mapx', 'mapy']].copy()
    locations.rename(columns={'when_start': 'when'}, inplace=True)
    locations.sort_values(by=['when'], inplace=True)
    locations.reset_index(drop=True, inplace=True)
    return locations

def interpolate_player_locations(locations: pandas.DataFrame) -> pandas.DataFrame:
    from dataclasses import dataclass

    IGNORED_GAP_SECONDS = 60 * 60 * 1

    @dataclass
    class LocationSecond:
        when: pandas.Timestamp
        mapx: int
        mapy: int

    def interpolate_locations():
        skipped = False
        for pair in locations.rolling(window=2, closed='right'):
            if not skipped:
                skipped = True
                continue

            left = pair.iloc[0]
            right = pair.iloc[1]

            start = left['when']
            end = right['when']

            seconds = end.timestamp() - start.timestamp()
            if seconds > IGNORED_GAP_SECONDS:
                end = start
                seconds = 1

            x = left['mapx']
            y = left['mapy']

            if seconds < 0.1:
                x_increment = 0
                y_increment = 0
            else:
                x_increment = (right['mapx'] - left['mapx']) / seconds
                y_increment = (right['mapy'] - left['mapy']) / seconds

            for elapsed, timestamp in enumerate(timestamp_range_seconds(start, end)):
                yield LocationSecond(when=timestamp, mapx=x, mapy=y)
                x += x_increment
                y += y_increment

    return pandas.DataFrame(interpolate_locations())