Merge pull request #721 from ac1982/fix/subtitle-cue-escaping

fix(subtitle): escape literal transcript text in WebVTT

Author: calesthio

tests/tools/test_subtitle_escaping.py

@@ -0,0 +1,58 @@
+"""Transcript text stays literal while generated karaoke markup remains active."""
+import json
+import shutil
+import subprocess
+
+import pytest
+
+from tools.subtitle.subtitle_gen import SubtitleGen
+
+
+@pytest.mark.parametrize("style", ["none", "word_by_word", "karaoke"])
+def test_transcript_markup_is_escaped(tmp_path, style):
+    fmt = "vtt"
+    output = tmp_path / f"captions.{fmt}"
+    result = SubtitleGen().execute({
+        "segments": [{"words": [
+            {"word": "<b>tag</b>", "start": 0, "end": 1},
+            {"word": "&amp;", "start": 1, "end": 2},
+        ]}],
+        "format": fmt,
+        "highlight_style": style,
+        "output_path": str(output),
+    })
+    assert result.success, result.error
+    content = output.read_text()
+    assert "&lt;b&gt;tag&lt;/b&gt;" in content
+    assert "&amp;amp;" in content
+    if style == "karaoke":
+        assert "<b>&lt;b&gt;tag&lt;/b&gt;</b>" in content
+    else:
+        assert "<b>" not in content
+
+
+def test_caption_json_keeps_raw_text(tmp_path):
+    output = tmp_path / "captions.json"
+    result = SubtitleGen().execute({
+        "segments": [{"text": "<b>tag</b> &amp;", "start": 0, "end": 1}],
+        "format": "json", "output_path": str(output),
+    })
+    assert result.success, result.error
+    assert json.loads(output.read_text())["cues"][0]["text"] == "<b>tag</b> &amp;"
+
+
+@pytest.mark.skipif(shutil.which("ffmpeg") is None, reason="ffmpeg not installed")
+@pytest.mark.parametrize("style", ["none", "word_by_word", "karaoke"])
+def test_srt_remains_compatible_with_ffmpeg(tmp_path, style):
+    source = tmp_path / "captions.srt"
+    output = tmp_path / "captions.ass"
+    result = SubtitleGen().execute({
+        "segments": [{"words": [{"word": "AT&T", "start": 0, "end": 1}]}],
+        "format": "srt", "highlight_style": style, "output_path": str(source),
+    })
+    assert result.success, result.error
+    subprocess.run(["ffmpeg", "-v", "error", "-y", "-i", str(source), str(output)],
+                   check=True, capture_output=True, timeout=10)
+    rendered = output.read_text()
+    assert "AT&T" in rendered
+    assert "&amp;" not in rendered

tools/subtitle/subtitle_gen.py

@@ -8,6 +8,7 @@
 from __future__ import annotations
 
 import json
+from html import escape
 import time
 from pathlib import Path
 from typing import Any
@@ -279,14 +280,14 @@ def _render_vtt(self, cues: list[dict], highlight_style: str = "none") -> str:
                     lines.append(
                         f"{self._ts_vtt(word_info['start'])} --> {self._ts_vtt(word_info['end'])}"
                     )
-                    lines.append(word_info["word"])
+                    lines.append(escape(word_info["word"], quote=False))
                     lines.append("")
         elif highlight_style == "karaoke":
             for cue in cues:
                 words = cue.get("words", [])
                 if not words:
                     lines.append(f"{self._ts_vtt(cue['start'])} --> {self._ts_vtt(cue['end'])}")
-                    lines.append(cue["text"])
+                    lines.append(escape(cue["text"], quote=False))
                     lines.append("")
                     continue
                 for wi, word_info in enumerate(words):
@@ -296,15 +297,15 @@ def _render_vtt(self, cues: list[dict], highlight_style: str = "none") -> str:
                     parts = []
                     for wj, w in enumerate(words):
                         if wj == wi:
-                            parts.append(f"<b>{w['word']}</b>")
+                            parts.append(f"<b>{escape(w['word'], quote=False)}</b>")
                         else:
-                            parts.append(w["word"])
+                            parts.append(escape(w["word"], quote=False))
                     lines.append(" ".join(parts))
                     lines.append("")
         else:
             for cue in cues:
                 lines.append(f"{self._ts_vtt(cue['start'])} --> {self._ts_vtt(cue['end'])}")
-                lines.append(cue["text"])
+                lines.append(escape(cue["text"], quote=False))
                 lines.append("")
         return "\n".join(lines)