scripts/export_transcript.py - master - Max_Cherep/super-transcription

scripts/export_transcript.py
116 строк3 KB
maxim.cherepanov
Add VL-first hybrid screenshots, EN markers, and lightbox.
31 июл 2026, 19:22
31 июл 2026, 19:229df8b25
"""Экспорт сегментов транскрипта в timed TXT / SRT."""

from __future__ import annotations

from dataclasses import dataclass
from pathlib import Path


@dataclass(frozen=True)
class Segment:
    start: float
    end: float
    text: str
    speaker: int | None = None


def speaker_prefix(speaker: int | None) -> str:
    """Префикс `[Speaker N]` или пустая строка."""
    if speaker is None:
        return ""
    return f"[Speaker {int(speaker)}]"


def format_segment_text(seg: Segment) -> str:
    """Текст сегмента с опциональным префиксом спикера."""
    text = (seg.text or "").strip()
    if not text:
        return ""
    prefix = speaker_prefix(seg.speaker)
    if not prefix:
        return text
    return f"{prefix} {text}"


def _clamp_sec(sec: float) -> float:
    return 0.0 if sec < 0 else sec


def _fmt_ts_txt(sec: float) -> str:
    sec = _clamp_sec(sec)
    h = int(sec // 3600)
    m = int((sec % 3600) // 60)
    s = sec % 60
    return f"{h:02d}:{m:02d}:{s:06.3f}"


def _fmt_ts_srt(sec: float) -> str:
    """SRT: HH:MM:SS,mmm (запятая)."""
    sec = _clamp_sec(sec)
    total_ms = int(round(sec * 1000.0))
    h, rem = divmod(total_ms, 3_600_000)
    m, rem = divmod(rem, 60_000)
    s, ms = divmod(rem, 1000)
    return f"{h:02d}:{m:02d}:{s:02d},{ms:03d}"


def format_timed_txt(segments: list[Segment], *, with_timestamps: bool = True) -> str:
    lines: list[str] = []
    for seg in segments:
        text = format_segment_text(seg)
        if not text:
            continue
        if with_timestamps:
            lines.append(f"[{_fmt_ts_txt(seg.start)} → {_fmt_ts_txt(seg.end)}] {text}")
        else:
            lines.append(text)
    return "\n".join(lines) + ("\n" if lines else "")


def format_srt(segments: list[Segment]) -> str:
    blocks: list[str] = []
    idx = 0
    for seg in segments:
        text = format_segment_text(seg)
        if not text:
            continue
        idx += 1
        start = _fmt_ts_srt(seg.start)
        end = _fmt_ts_srt(max(seg.end, seg.start))
        blocks.append(f"{idx}\n{start} --> {end}\n{text}\n")
    return "\n".join(blocks)


def write_timed_txt(
    path: Path,
    segments: list[Segment],
    *,
    with_timestamps: bool = True,
) -> None:
    path.parent.mkdir(parents=True, exist_ok=True)
    path.write_text(
        format_timed_txt(segments, with_timestamps=with_timestamps),
        encoding="utf-8",
    )


def write_srt(path: Path, segments: list[Segment]) -> None:
    path.parent.mkdir(parents=True, exist_ok=True)
    path.write_text(format_srt(segments), encoding="utf-8")


def write_transcript(
    path: Path,
    segments: list[Segment],
    *,
    output_format: str = "txt",
    with_timestamps: bool = True,
) -> None:
    fmt = (output_format or "txt").lower().lstrip(".")
    if fmt == "srt":
        write_srt(path, segments)
        return
    if fmt == "txt":
        write_timed_txt(path, segments, with_timestamps=with_timestamps)
        return
    raise ValueError(f"unsupported output_format: {output_format}")