personal-data/git_time_tracker/source/csv_file.py

from collections.abc import Iterator
from logging import getLogger
from pathlib import Path
from typing import Any

from personal_data.activity import ActivitySample, Label
from personal_data.csv_import import determine_possible_keys, load_csv_file, start_end

print(__name__)
logger = getLogger(__name__)


def iterate_samples_from_dicts(rows: list[dict[str, Any]]) -> Iterator[ActivitySample]:
    assert len(rows) > 0
    max_title_parts = 2

    if True:
        event_data = rows[len(rows) // 2]  # Hopefully select a useful representative.
        possible_keys = determine_possible_keys(event_data)
        logger.info('Found possible keys: %s', possible_keys)
        del event_data

    assert len(possible_keys.time_start) + len(possible_keys.time_end) >= 1
    assert len(possible_keys.image) >= 0

    for event_data in rows:
        """
        title = ': '.join(event_data[k] for k in possible_name_keys[:max_title_parts])
        description = '\n\n'.join(
            event_data[k] for k in possible_name_keys[max_title_parts:]
        )
        image = event_data[possible_keys.image[0]] if possible_keys.image else None
        """

        (start_at, end_at) = start_end(event_data, possible_keys)
        labels = [Label(k, event_data[k]) for k in possible_keys.misc]

        # Create event
        yield ActivitySample(
            labels=tuple(labels),
            start_at=start_at,
            end_at=end_at,
        )

        del event_data


def iterate_samples_from_csv_file(file_path: Path) -> Iterator[ActivitySample]:
    dicts = load_csv_file(file_path)
    samples = list(iterate_samples_from_dicts(dicts))
    assert len(samples) > 0, 'Did not found any samples'
    yield from samples