Module music_df.humdrum_export.df_utils.spell_df

Functions

def humdrum_spelling_from_row(row)

Expand source code

def humdrum_spelling_from_row(row):
    if isinstance(row.spelling, float) and isnan(row.spelling):
        return row.spelling
    assert isinstance(row.spelling, str)
    return shell_spelling_to_humdrum_spelling(row.spelling, int(row.pitch))

def spell_df(df: pandas.DataFrame,
speller: mspell.group_speller.GroupSpeller | None = None,
chunk_len=32)

Expand source code

def spell_df(df: pd.DataFrame, speller: t.Optional[GroupSpeller] = None, chunk_len=32):
    """
    Keyword args:
        chunk_len: "chunk" length of df to spell at once. If we spell the whole
            dataframe at once, it is liable to contain all 12 pitches which
            tends to lead to nonsensical spellings. Thus we take contiguous
            "chunks" from the dataframe and spell them one by one. This
            parameter controls the length of each chunk.
    """
    df = df.copy()
    if speller is None:
        speller = GroupSpeller(pitches=True, letter_format="kern")
    df["humdrum_spelling"] = ""
    if "spelling" in df.columns:
        df["humdrum_spelling"] = df.apply(humdrum_spelling_from_row, axis=1)
    else:
        for i in range(0, len(df), chunk_len):
            # To avoid chained indexing as follows we use `row_mask`
            # df.iloc[i : i + chunk_len].loc[df.type == "note", "humdrum_spelling"] = (
            row_mask = (
                (df.index >= i) & (df.index < (i + chunk_len)) & (df.type == "note")
            )
            df.loc[row_mask, "humdrum_spelling"] = speller(
                # df.iloc[i : i + chunk_len]
                # .loc[df.type == "note", "pitch"]
                df.loc[row_mask, "pitch"].astype(int)  # type:ignore
            )
    return df

Keyword args: chunk_len: "chunk" length of df to spell at once. If we spell the whole dataframe at once, it is liable to contain all 12 pitches which tends to lead to nonsensical spellings. Thus we take contiguous "chunks" from the dataframe and spell them one by one. This parameter controls the length of each chunk.