From c4f5c776e259c9c388594595b7821470fbffd68d Mon Sep 17 00:00:00 2001
From: xenofem <xenofem@xeno.science>
Date: Tue, 2 Apr 2024 14:24:42 -0400
Subject: [PATCH] kanatype-insensitive sorting using Unicode Collation
 Algorithm

---
 dlibrary/dlibrary.py | 4 +++-
 flake.nix            | 1 +
 pyproject.toml       | 1 +
 3 files changed, 5 insertions(+), 1 deletion(-)

diff --git a/dlibrary/dlibrary.py b/dlibrary/dlibrary.py
index 7bce8b9..67ba0ea 100755
--- a/dlibrary/dlibrary.py
+++ b/dlibrary/dlibrary.py
@@ -23,6 +23,7 @@ import fitz
 from libsixel import *
 from PIL import Image
 from jinja2 import Environment, PackageLoader, select_autoescape
+import pyuca
 import rarfile
 import requests
 
@@ -1151,10 +1152,11 @@ def generate(args):
 
         print(f'\x1b[2K\r{idx+1}/{len(works)} works processed...', end=('' if idx+1 < len(works) else '\n'))
 
+    uca = pyuca.Collator().sort_key
     def make_categorization(categorization, query, work_filter, work_style_cards=False):
         categorization_dir = site_dir / categorization
 
-        cats = sorted((cat for (cat,) in cur.execute(query)), key=str.casefold)
+        cats = sorted((cat for (cat,) in cur.execute(query)), key=uca)
         cat_samples = {}
         for cat in cats:
             cat_works = list(filter(work_filter(cat), works))
diff --git a/flake.nix b/flake.nix
index 6391e97..c4d3f8d 100644
--- a/flake.nix
+++ b/flake.nix
@@ -60,6 +60,7 @@
           setuptools
           libsixel
           (rarfile.override { useUnrar = true; })
+          pyuca
         ];
         src = ./.;
       };
diff --git a/pyproject.toml b/pyproject.toml
index 7897dec..99a07c1 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -13,6 +13,7 @@ dependencies = [
     "importlib_resources",
     "libsixel-python",
     "rarfile",
+    "pyuca",
 ]
 
 [project.scripts]