Source code for pycif.plugins.datastreams.meteos.dummy_csv.fetch

import os
import pandas as pd
import datetime
from .....utils import path
from .make import make




def fetch(ref_dir, ref_file, input_interval, target_dir, tracer=None, **kwargs):
    """Locate dummy meteo CSV files, generating them if missing, and link them.

    Iterates over period-start dates covering ``input_interval`` at
    ``tracer.file_freq``, and for each period computes the hourly (or
    ``tracer.resolution``) sub-periods it should cover. If the
    corresponding source file (``ref_dir/ref_file`` formatted for that
    period) exists, it is symlinked into ``target_dir``; otherwise it is
    generated on the fly via :func:`make.make`.

    Args:
        ref_dir: directory holding (or expected to hold) the dummy meteo
            CSV files.
        ref_file: strftime-style file name pattern (relative to
            ``ref_dir``), formatted once per period.
        input_interval: 2-element sequence ``(datei, datef)`` giving the
            requested date range.
        target_dir: directory where files are symlinked or generated.
        tracer: the meteo datastream Plugin, used for ``tracer.file_freq``
            and ``tracer.resolution``, and passed through to ``make`` when
            a file needs to be generated.

    Returns:
        tuple: ``(list_files, list_dates)``, dicts keyed by each period's
        start date. ``list_files`` maps each key to the (repeated) target
        file path, one entry per sub-period; ``list_dates`` maps each key
        to the corresponding list of ``[start, start + resolution]`` date
        pairs.
    """
    list_period_dates = \
        pd.date_range(input_interval[0], input_interval[1], freq=tracer.file_freq)
    list_dates = {}
    list_files = {}
    for dd in list_period_dates:
        file = dd.strftime(f"{ref_dir}/{ref_file}")
        file_hours = pd.date_range(
            dd, dd + pd.to_timedelta(tracer.file_freq),
            freq=tracer.resolution,
            inclusive="left"
        )
        list_dates[dd] = [[hh, hh + pd.to_timedelta(tracer.resolution)]
                          for hh in file_hours]
        target_file = f"{target_dir}/{os.path.basename(file)}"
        list_files[dd] = (len(file_hours) * [target_file])
        if os.path.isfile(file):
            # Fetching
            path.link(file, target_file)

        else:
            make(tracer, target_file, file_hours)

    return list_files, list_dates