22 lines
732 B
Python
22 lines
732 B
Python
from pathlib import Path
|
|
import pandas as pd
|
|
from subtitles.parser import parse_srt_to_df # your existing parser
|
|
|
|
|
|
def load_subtitles_from_dir(folder: str | Path, ext: str = ".srt") -> pd.DataFrame:
|
|
"""
|
|
Читает каждый файл из заданной папки и возвращает обьединенный pd.DataFrame
|
|
"""
|
|
folder = Path(folder)
|
|
all_frames: list[pd.DataFrame] = []
|
|
|
|
for file in folder.glob(f"*{ext}"):
|
|
df_file = parse_srt_to_df(file)
|
|
df_file["source_file"] = file.name
|
|
all_frames.append(df_file)
|
|
|
|
if not all_frames:
|
|
raise FileNotFoundError(f"No {ext} files found in {folder.resolve()}")
|
|
|
|
return pd.concat(all_frames, ignore_index=True)
|