From 0614c27d16467056f6cfa6a90ea2e619579b160f Mon Sep 17 00:00:00 2001
From: ColinGamez <235560865+ColinGamez@users.noreply.github.com>
Date: Fri, 24 Jul 2026 04:11:46 +0900
Subject: [PATCH] Normalize broadcast symbols for the Wii font

---
 tools/collect_hbnj_region.py | 22 ++++++++++++++++++++++
 tools/validate_hbnj_guide.py |  8 ++++++++
 2 files changed, 30 insertions(+)

diff --git a/tools/collect_hbnj_region.py b/tools/collect_hbnj_region.py
index 54626ff..8415edc 100644
--- a/tools/collect_hbnj_region.py
+++ b/tools/collect_hbnj_region.py
@@ -6,6 +6,7 @@
 import json
 import re
 import tempfile
+import unicodedata
 from collections import Counter
 from datetime import datetime
 from pathlib import Path
@@ -81,6 +82,21 @@ def compact_text(value: str) -> str:
     return re.sub(r"\s+", " ", html.unescape(value)).strip()
 
 
+def wii_safe_text(value: str) -> str:
+    """Replace supplementary broadcast symbols unsupported by the Wii font."""
+    result: list[str] = []
+    for character in value:
+        if ord(character) <= 0xFFFF:
+            result.append(character)
+            continue
+        normalized = unicodedata.normalize("NFKC", character)
+        if normalized != character and all(ord(item) <= 0xFFFF for item in normalized):
+            result.append(f"[{normalized}]")
+        # Other supplementary emoji have no reliable equivalent in HBNJ's
+        # Wii-era font and are omitted instead of rendering as surrogate ??.
+    return re.sub(r"\s+", " ", "".join(result)).strip()
+
+
 def parse_timestamp(value: str) -> datetime:
     if not re.fullmatch(r"\d{12}", value):
         raise ValueError(f"invalid Bangumi timestamp: {value!r}")
@@ -179,10 +195,16 @@ def parse_region(
                 title = compact_text(title_match.group("title"))
             if not title:
                 raise ValueError(f"missing program title on line {line_number}")
+            title = wii_safe_text(title)
+            if not title:
+                # Bangumi occasionally publishes placeholder events whose
+                # title contains only full-width spaces.
+                title = "番組情報なし"
             detail_match = DETAIL_PATTERN.search(anchor_body)
             description = (
                 compact_text(detail_match.group("detail")) if detail_match else ""
             )
+            description = wii_safe_text(description)
             genre_match = GENRE_PATTERN.search(item.group("body"))
             source_genre = genre_match.group("genre").lower() if genre_match else ""
             genre_id = BANGUMI_GENRE_IDS.get(source_genre, 0)
diff --git a/tools/validate_hbnj_guide.py b/tools/validate_hbnj_guide.py
index e01da47..6fc751b 100644
--- a/tools/validate_hbnj_guide.py
+++ b/tools/validate_hbnj_guide.py
@@ -78,6 +78,14 @@ def main() -> int:
         description = program.get("description", "")
         if not isinstance(description, str):
             fail(f"program {program['id']} has a non-string description")
+        for field_name, value in (("title", program["title"]), ("description", description)):
+            supplementary = [character for character in value if ord(character) > 0xFFFF]
+            if supplementary:
+                codes = ", ".join(f"U+{ord(character):05X}" for character in supplementary[:3])
+                fail(
+                    f"program {program['id']} {field_name} contains "
+                    f"Wii-unsupported supplementary characters: {codes}"
+                )
         descriptions += bool(description)
         programs_by_channel[channel_id].append((start, end))