From ecebfc0c718c81b7f0d6d6552e81c71fe9bf2053 Mon Sep 17 00:00:00 2001
From: jaime-m-p <>
Date: Fri, 26 Jul 2024 00:16:24 +0200
Subject: [PATCH] Update unicode data: sorted whitespaces

---
 scripts/gen-unicode-data.py | 2 +-
 src/unicode-data.cpp        | 3 +--
 src/unicode-data.h          | 3 +--
 src/unicode.cpp             | 2 +-
 4 files changed, 4 insertions(+), 6 deletions(-)

diff --git a/scripts/gen-unicode-data.py b/scripts/gen-unicode-data.py
index 542a9edba..d774fcabe 100644
--- a/scripts/gen-unicode-data.py
+++ b/scripts/gen-unicode-data.py
@@ -170,7 +170,7 @@ for rle in codepoint_categs_runs:
     out("0x%04X," % rle)
 out("};\n")
 
-out("const std::unordered_set<uint32_t> unicode_set_whitespace = {")
+out("const std::vector<uint32_t> unicode_vec_whitespace = {")
 for codepoint in table_whitespace:
     out("0x%06X," % codepoint)
 out("};\n")
diff --git a/src/unicode-data.cpp b/src/unicode-data.cpp
index 4a0c0547c..2591723ce 100644
--- a/src/unicode-data.cpp
+++ b/src/unicode-data.cpp
@@ -5,7 +5,6 @@
 #include <cstdint>
 #include <vector>
 #include <unordered_map>
-#include <unordered_set>
 
 const std::vector<uint16_t> unicode_rle_codepoints_categs = {  // run length encoding, 5 bits categ + 11 bits length
 0x03E1,
@@ -4527,7 +4526,7 @@ const std::vector<uint16_t> unicode_rle_codepoints_categs = {  // run length enc
 0x0020,
 };
 
-const std::unordered_set<uint32_t> unicode_set_whitespace = {
+const std::vector<uint32_t> unicode_vec_whitespace = {
 0x000009,
 0x00000A,
 0x00000B,
diff --git a/src/unicode-data.h b/src/unicode-data.h
index cd6a6451a..682f79c37 100644
--- a/src/unicode-data.h
+++ b/src/unicode-data.h
@@ -3,7 +3,6 @@
 #include <cstdint>
 #include <vector>
 #include <unordered_map>
-#include <unordered_set>
 
 struct range_nfd {
     uint32_t first;
@@ -14,7 +13,7 @@ struct range_nfd {
 static const uint32_t MAX_CODEPOINTS = 0x110000;
 
 extern const std::vector<uint16_t> unicode_rle_codepoints_categs;
-extern const std::unordered_set<uint32_t> unicode_set_whitespace;
+extern const std::vector<uint32_t> unicode_vec_whitespace;
 extern const std::unordered_map<uint32_t, uint32_t> unicode_map_lowercase;
 extern const std::unordered_map<uint32_t, uint32_t> unicode_map_uppercase;
 extern const std::vector<range_nfd> unicode_ranges_nfd;
diff --git a/src/unicode.cpp b/src/unicode.cpp
index 4c3335974..dd413c809 100644
--- a/src/unicode.cpp
+++ b/src/unicode.cpp
@@ -591,7 +591,7 @@ codepoint_categ unicode_cpt_category(const uint32_t cp) {
         }
         assert (cpt == MAX_CODEPOINTS);
 
-        for (auto cpt : unicode_set_whitespace) {
+        for (auto cpt : unicode_vec_whitespace) {
             cpt_categs[cpt].set_flag(codepoint_categ::WHITESPACE);
         }