import datetime
from collections.abc import Iterator

from ..data import HIDDEN_LABEL_PREFIX, HIDDEN_LABEL_TOTAL, WorkSample

DEFAULT_EST_TIME = datetime.timedelta(hours=1)
ZERO_DURATION = datetime.timedelta(seconds=0)
HOUR = datetime.timedelta(hours=1)
MINUTE = datetime.timedelta(minutes=1)


def fmt_year_ranges_internal(years: list[int]) -> Iterator[str]:
    years = sorted(years)
    for idx, year in enumerate(years):
        at_end = idx == len(years) - 1
        range_before = idx > 0 and years[idx - 1] == year - 1
        range_after = not at_end and years[idx + 1] == year + 1

        if not range_before or not range_after:
            yield str(year)

        if not at_end:
            if not range_before and range_after:
                yield '-'
            elif not range_after:
                yield ','


def fmt_year_ranges(years: list[int]) -> str:
    return ''.join(list(fmt_year_ranges_internal(years)))


def fmt_line(label_type: str, label: str, total_time: datetime.timedelta) -> str:
    hours = int(total_time / HOUR)
    minutes = int((total_time - hours * HOUR) / MINUTE)
    return f'  {label_type:10} {label:40}  {hours:-4d}h {minutes:-2d}m'


def generate_report(
    samples: list[WorkSample],
    sample_filter=frozenset(),
) -> Iterator[str]:
    LABEL_FILTER = {}

    # Time spent per label
    time_per_label: dict[str, datetime.timedelta] = {}
    years_per_label: dict[str, set[int]] = {}
    prev_time = datetime.datetime.fromtimestamp(0, datetime.UTC)
    for sample in samples:
        est_time: datetime.timedelta = DEFAULT_EST_TIME
        est_time = min(sample.registered_at - prev_time, est_time)

        if len(sample_filter) == 0:
            pass
        elif not set(sample.labels).intersection(sample_filter):
            continue

        for label in sample.labels:
            time_per_label.setdefault(label, ZERO_DURATION)
            time_per_label[label] += est_time
            years_per_label.setdefault(label, set()).add(sample.registered_at.year)

        prev_time = sample.registered_at
        del sample, est_time

    time_and_label = [(duration, label) for label, duration in time_per_label.items()]
    time_and_label.sort(reverse=True)

    #
    yield '-' * 66
    yield '\n'
    for total_time, label_and_type in time_and_label:
        if label_and_type.startswith(HIDDEN_LABEL_PREFIX):
            continue

        label_type, label = label_and_type.split(':', 1)

        if len(LABEL_FILTER) > 0 and label_type not in LABEL_FILTER:
            continue

        yield fmt_line(label_type, label, total_time)
        yield '  ('
        yield fmt_year_ranges(years_per_label.get(label_and_type, []))
        yield ')'
        yield '\n'
        del label, total_time

    yield '-' * 66
    yield '\n'

    yield fmt_line('', 'TOTAL', time_per_label.get(HIDDEN_LABEL_TOTAL, ZERO_DURATION))
    yield '\n'