Fail closed when manim.unsafe_unicode is not a YAML list

milestones/README.md

@@ -5,11 +5,13 @@ repositories that install `docgen` and maintain their own demo bundle. The
 library no longer ships an in-repo dogfood; consumers are the integration test
 of record.
 
-**Active:** **[validation-list-keys.md](validation-list-keys.md)** —
-`validation.ocr.error_patterns`, `av_sync.visual_types`, and narration-lint
-deny pattern lists must be YAML lists, not scalars.
+**Active:** **[manim-unsafe-unicode.md](manim-unsafe-unicode.md)** —
+`manim.unsafe_unicode` must be a YAML list at config load, not a scalar.
 
 **Shipped:**
+- **[validation-list-keys.md](validation-list-keys.md)** —
+  `validation.ocr.error_patterns`, `av_sync.visual_types`, and narration-lint
+  deny pattern lists must be YAML lists (#95).
 - **[wizard-exclude-lists.md](wizard-exclude-lists.md)** —
   `wizard.exclude_patterns` / `scan_extensions` must be string lists (#94).
 - **[timestamps-merge-stems.md](timestamps-merge-stems.md)** — CLI `timestamps`

milestones/manim-unsafe-unicode.md

@@ -0,0 +1,33 @@
+# Milestone: `manim.unsafe_unicode` must be a list at config load
+
+**Status:** Active  
+**PR:** pending  
+**Depends on:** `milestones/validation-list-keys.md` (PR #95),
+`milestones/config-mapping-keys.md` (PR #83)
+
+## Problem
+
+`manim.unsafe_unicode` as a YAML string already raised `ConfigError` when the
+property was **read** (validate / scene-spec-generate). `Config.from_yaml` did
+not touch it, so `yaml-generate`, `tts`, and other commands could load a bad
+file. A string would also have been iterated as characters inside unicode lint
+if anything called the property after a bypass.
+
+## Goal
+
+Evaluate `manim_unsafe_unicode` during `Config.from_yaml` (same as
+`manim.scenes`). Use `string_list_block` so items must be YAML strings.
+
+## Done when
+
+- [x] `manim.unsafe_unicode` if present must be a YAML list of strings.
+- [x] Missing / null still uses the built-in Pango fallback character list.
+- [x] Empty list still disables the unicode lint (`[]`).
+- [x] Test for a string value at `Config.from_yaml`.
+- [x] `ruff check src/ tests/`
+- [x] `pytest tests/`
+- [x] `docgen benchmark` (no clock change)
+
+## Out of scope
+
+- The default character list itself is unchanged.

milestones/validation-list-keys.md

@@ -1,7 +1,7 @@
 # Milestone: validation list keys must not be scalars
 
-**Status:** Active  
-**PR:** pending  
+**Status:** Shipped  
+**PR:** #95  
 **Depends on:** `milestones/wizard-exclude-lists.md` (PR #94),
 `milestones/config-mapping-keys.md` (PR #83)
 

src/docgen/config.py

@@ -206,6 +206,7 @@ def __post_init__(self) -> None:
         # List-valued keys: a string must not be iterated as characters.
         _ = self.segments_all
         _ = self.manim_scenes
+        _ = self.manim_unsafe_unicode
         concat = self._block("concat")
         src = self._source_label()
         for name, segs in concat.items():
@@ -538,15 +539,13 @@ def manim_unsafe_unicode(self) -> list[str]:
                    "\u2260", "\u2264", "\u2265", "\u2014", "\u2013",
                    "\u2018", "\u2019", "\u201c", "\u201d", "\u2022",
                    "\u2026"]
-        val = self._block("manim").get("unsafe_unicode", default)
-        if val is None:
-            return list(default)
-        if not isinstance(val, list):
-            raise ConfigError(
-                f"{self._source_label()}: manim.unsafe_unicode must be a YAML list, "
-                f"not {type(val).__name__}"
-            )
-        return [str(x) for x in val]
+        return string_list_block(
+            self._block("manim"),
+            "unsafe_unicode",
+            fallback=default,
+            label="manim.unsafe_unicode",
+            source=self._source_label(),
+        )
 
     # -- Compose ----------------------------------------------------------------
 

tests/test_config.py

@@ -371,3 +371,10 @@ def test_from_yaml_string_post_tts_deny_patterns_raises(tmp_path: Path) -> None:
         ConfigError, match="validation.narration_lint.post_tts_deny_patterns must be a YAML list"
     ):
         Config.from_yaml(p)
+
+
+def test_from_yaml_string_manim_unsafe_unicode_raises(tmp_path: Path) -> None:
+    p = tmp_path / "docgen.yaml"
+    p.write_text("manim:\n  unsafe_unicode: \"\\u2192\"\n", encoding="utf-8")
+    with pytest.raises(ConfigError, match="manim.unsafe_unicode must be a YAML list"):
+        Config.from_yaml(p)