helpers for HDF bug

gistfile1.txt

def sane_names(df):
    names = _gen_items()
    if isinstance(df, pd.Panel):
        return df.rename(items=names, minor_axis=names)
    return df.rename(columns=names)

def get_useful(df, strict=True):
    df = sane_names(df)
    # TODO: grab cols from convert_names.keys()
    # or both from a common function.
    cols = _gen_items().values()
    no_history = list(set(cols) - {'unemployed_history', 'nonemployed_history',
                                   'either_history', 'flow'})
    if isinstance(df, pd.Panel):
        try:
            res = df.loc[cols]
        except KeyError:
            if strict:
                raise KeyError
            else:
                res = df.loc[no_history]
    else:
        try:
            res = df[cols]
        except KeyError:
            if strict:
                raise KeyError
            else:
                res = df[no_history]
    return res

def fix_timestamp(wp_original):
    wp = wp_original.copy()
    wp = get_useful(wp)
    years = wp['year'].dropna()
    months = wp['month'].dropna()
    ts = wp['timestamp'].dropna()

    ncols = len(years.columns)
    idx = range(1, ncols + 1)

    stamp = df_unique(ts, index=idx)
    years = df_unique(years, index=idx)
    months = df_unique(months, index=idx)

    expected = pd.to_datetime(years.astype(int).astype(str) + '-' +
                              months.astype(int).astype(str) + '-01')

    needfix = expected[stamp != expected].index

    if needfix is not None:
        for month in needfix:
            wp_original['timestamp'][month] = expected[month]

        return wp_original