From d429f732e1c7a95e205830f56ab1d80098e568f3 Mon Sep 17 00:00:00 2001
From: Kovid Goyal <kovid@kovidgoyal.net>
Date: Tue, 25 Mar 2025 13:45:56 +0530
Subject: [PATCH] DRYer

---
 gen/wcwidth.py                    | 14 +++++++++++--
 kitty/char-props-data.h           | 32 +++++++++++++++++++++++++++++
 tools/wcswidth/char-props-data.go | 34 +++++++++++++++++++++++++++++++
 3 files changed, 78 insertions(+), 2 deletions(-)

diff --git a/gen/wcwidth.py b/gen/wcwidth.py
index 2cd550b61..2899a1306 100755
--- a/gen/wcwidth.py
+++ b/gen/wcwidth.py
@@ -12,6 +12,7 @@ from contextlib import contextmanager
 from functools import lru_cache, partial
 from html.entities import html5
 from itertools import groupby
+from math import ceil, log
 from operator import itemgetter
 from typing import (
     Callable,
@@ -598,15 +599,21 @@ def gen_multistage_table(
 
 width_shift = 4
 
+
+def bitsize(maxval: int) -> int:  # number of bits needed to store maxval
+    return ceil(log(maxval, 2))
+
+
 class CharProps(NamedTuple):
 
     width: int = 3
-    grapheme_break: str = '4'
-    indic_conjunct_break: str = '2'
+    grapheme_break: str = ''  # set at runtime
+    indic_conjunct_break: str = ''  # set at runtime
     is_extended_pictographic: bool = True
     is_emoji: bool = True
     is_emoji_presentation_base: bool = True
 
+    # derived properties for fast lookup
     is_invalid: bool = True
     is_non_rendered: bool = True
     is_symbol: bool = True
@@ -693,6 +700,8 @@ def generate_enum(p: Callable[..., None], gp: Callable[..., None], name: str, *i
 
 
 def gen_char_props() -> None:
+    CharProps._field_defaults['grapheme_break'] = str(bitsize(len(grapheme_segmentation_maps) + 2))
+    CharProps._field_defaults['indic_conjunct_break'] = str(bitsize(len(incb_map) + 1))
     invalid = class_maps['Cc'] | class_maps['Cs']
     non_printing = invalid | class_maps['Cf']
     width_map: dict[int, int] = {}
@@ -732,6 +741,7 @@ def gen_char_props() -> None:
         gp('package wcswidth')
         generate_enum(c, gp, 'GraphemeBreakProperty', 'AtStart', 'None', *grapheme_segmentation_maps, prefix='GBP_')
         generate_enum(c, gp, 'IndicConjunctBreak', 'None', *incb_map, prefix='ICB_')
+        generate_enum(c, gp, 'UnicodeCategory', *class_maps, prefix='UC_')
         bf = make_bitfield('tools/wcswidth', 'CharProps', *CharProps().go_fields, add_package=False)[1]
         gp(bf)
         gp(f'''
diff --git a/kitty/char-props-data.h b/kitty/char-props-data.h
index 6866f0905..1f5be9fdd 100644
--- a/kitty/char-props-data.h
+++ b/kitty/char-props-data.h
@@ -28,6 +28,38 @@ typedef enum IndicConjunctBreak {
 	ICB_Extend,
 } IndicConjunctBreak;
 
+typedef enum UnicodeCategory {
+	UC_Cc,
+	UC_Zs,
+	UC_Po,
+	UC_Sc,
+	UC_Ps,
+	UC_Pe,
+	UC_Sm,
+	UC_Pd,
+	UC_Nd,
+	UC_Lu,
+	UC_Sk,
+	UC_Pc,
+	UC_Ll,
+	UC_So,
+	UC_Lo,
+	UC_Pi,
+	UC_Cf,
+	UC_No,
+	UC_Pf,
+	UC_Lt,
+	UC_Lm,
+	UC_Mn,
+	UC_Me,
+	UC_Mc,
+	UC_Nl,
+	UC_Zl,
+	UC_Zp,
+	UC_Cs,
+	UC_Co,
+} UnicodeCategory;
+
 static const char_type CharProps_mask = 255u;
 static const char_type CharProps_shift = 8u;
 static const uint8_t CharProps_t1[4352] = {
diff --git a/tools/wcswidth/char-props-data.go b/tools/wcswidth/char-props-data.go
index 332644aa5..b2121b92e 100644
--- a/tools/wcswidth/char-props-data.go
+++ b/tools/wcswidth/char-props-data.go
@@ -29,6 +29,40 @@ const (
 	ICB_Extend
 )
 
+type UnicodeCategory uint8
+
+const (
+	UC_Cc UnicodeCategory = iota
+	UC_Zs
+	UC_Po
+	UC_Sc
+	UC_Ps
+	UC_Pe
+	UC_Sm
+	UC_Pd
+	UC_Nd
+	UC_Lu
+	UC_Sk
+	UC_Pc
+	UC_Ll
+	UC_So
+	UC_Lo
+	UC_Pi
+	UC_Cf
+	UC_No
+	UC_Pf
+	UC_Lt
+	UC_Lm
+	UC_Mn
+	UC_Me
+	UC_Mc
+	UC_Nl
+	UC_Zl
+	UC_Zp
+	UC_Cs
+	UC_Co
+)
+
 // Total number of bits used: 16
 type CharProps uint16