1from __future__ import annotations
2 
3import json
4from pathlib import Path
5 
6from video_timeline_editor.domain.model import CaptionTrack, WordToken
7 
8 
9def resolve_path(value: str | Path, base_dir: Path) -> Path:
10    path = Path(value)
11    return path if path.is_absolute() else (base_dir / path).resolve()
12 
13 
14def load_transcript_words(track: CaptionTrack, timeline_dir: Path) -> tuple[WordToken, ...]:
15    transcript_path = resolve_path(track.transcript, timeline_dir)
16    if not transcript_path.exists():
17        raise FileNotFoundError(f"Missing transcript {transcript_path}")
18    payload = json.loads(transcript_path.read_text())
19    words: list[WordToken] = []
20    replacements = dict(track.replacements)
21    skipped = set(track.skip_tokens)
22    offset = float(track.source_offset)
23    for item in payload.get("words", []):
24        if item.get("type") != "word":
25            continue
26        text = str(item.get("text", "")).strip()
27        if not text or text in skipped:
28            continue
29        text = replacements.get(text, text)
30        words.append(
31            WordToken(
32                text=text,
33                start=float(item["start"]) + offset,
34                end=float(item["end"]) + offset,
35            )
36        )
37    return tuple(words)
38