Fix PUA characters

Related: #6 (not in effect yet, planned for Qieyun.js v0.15)

build.py

@@ -25,6 +25,15 @@ def process_音韻地位(row: list[str]) -> str:
     return 母 + 呼 + 等類 + 韻 + 聲
 
 
+def fix_pua(s: str) -> str:
+    fixed = s.replace('\uee42', '𧞬').replace('\uece0', '勳')
+    for ch in fixed:
+        assert not (
+            0xE000 <= ord(ch) <= 0xF8FF
+        ), f'PUA character U+{ord(ch):04x} in {repr(s)}'
+    return fixed
+
+
 def main():
     小韻_data: dict[str, list[str]] = {}
     with open('src/rime-table-bfa9b50.tsv') as fin:
@@ -147,8 +156,8 @@ def main():
                 last_原小韻號 = 原小韻號
                 小韻內字序 = 0
             小韻內字序 += 1
-            row[1] = 小韻內字序
-            print(*row, sep=',', file=fout)
+            row[1] = str(小韻內字序)
+            print(fix_pua(','.join(row)), file=fout)
 
 
 if __name__ == '__main__':

韻書/廣韻.csv

@@ -1220,8 +1220,8 @@
 130,1,支,生開三支平,所宜,釃,,下酒所宜切又山爾切七,
 130,2,支,生開三支平,所宜,簁,,下物竹器又所綺切,
 130,3,支,生開三支平,所宜,欐,,梁棟別名又禮麗二音,
-130,4,支,生開三支平,所宜,襹,,襹毛羽衣皃,
-130,5,支,生開三支平,所宜,褷,,上同,襹毛羽衣皃
+130,4,支,生開三支平,所宜,襹,,𧞬襹毛羽衣皃,
+130,5,支,生開三支平,所宜,褷,,上同,𧞬襹毛羽衣皃
 130,6,支,生開三支平,所宜,𧕯,,蚰蜒別名,
 130,7,支,生開三支平,所宜,籭,,𥂖也又山佳切,
 131,1,支,生合三支平,山垂,䪎,,鞍鞘一曰垂皃山垂切二,
@@ -16031,7 +16031,7 @@
 2368,9,霽,疑開四齊去,五計,甈,,破罌,
 2368,10,霽,疑開四齊去,五計,堄,,埤堄女牆也見博雅,
 2368,11,霽,疑開四齊去,五計,霓,,虹又音倪,
-2369,1,霽,見開四齊去,古詣,計,,籌計說文會也筭也又姓後漢有計子古詣切十二,
+2369,1,霽,見開四齊去,古詣,計,,籌計說文會也筭也又姓後漢有計子勳古詣切十二,
 2369,2,霽,見開四齊去,古詣,係,,連係,
 2369,3,霽,見開四齊去,古詣,繼,,紹繼俗作継,
 2369,4,霽,見開四齊去,古詣,繫,,縛繫又口奚胡計二切,