From 9803364adb77cff60e42ffe99c4683e9a0c9245f Mon Sep 17 00:00:00 2001
From: Andrew Noyes <andrew@weaselab.dev>
Date: Mon, 23 Jun 2025 21:52:17 -0400
Subject: [PATCH] Use table for hex values

---
 src/parser3.h | 36 ++++++++++++------------------------
 src/tables.h  | 13 +++++++++++++
 2 files changed, 25 insertions(+), 24 deletions(-)

diff --git a/src/parser3.h b/src/parser3.h
index d2e0881..fc3ce9e 100644
--- a/src/parser3.h
+++ b/src/parser3.h
@@ -514,16 +514,12 @@ inline PRESERVE_NONE WeaselJsonStatus n_string_following_escape(Parser3 *self,
 
 inline PRESERVE_NONE WeaselJsonStatus t_hex(Parser3 *self, char *buf,
                                             char *bufEnd) {
-  self->utf8Codepoint <<= 4;
-  if (('0' <= *buf && *buf <= '9')) {
-    self->utf8Codepoint |= *buf - '0';
-  } else if ('a' <= *buf && *buf <= 'f') {
-    self->utf8Codepoint |= 10 + *buf - 'a';
-  } else if ('A' <= *buf && *buf <= 'F') {
-    self->utf8Codepoint |= 10 + *buf - 'A';
-  } else [[unlikely]] {
+  auto hexVal = tables.hex[uint8_t(*buf)];
+  if (hexVal < 0) [[unlikely]] {
     return WeaselJson_REJECT;
   }
+  self->utf8Codepoint <<= 4;
+  self->utf8Codepoint |= hexVal;
   ++buf;
   self->pop();
   MUSTTAIL return Parser3::keepGoing(self, buf, bufEnd);
@@ -531,16 +527,12 @@ inline PRESERVE_NONE WeaselJsonStatus t_hex(Parser3 *self, char *buf,
 
 inline PRESERVE_NONE WeaselJsonStatus t_hex2(Parser3 *self, char *buf,
                                              char *bufEnd) {
-  self->utf8Codepoint <<= 4;
-  if (('0' <= *buf && *buf <= '9')) {
-    self->utf8Codepoint |= *buf - '0';
-  } else if ('a' <= *buf && *buf <= 'f') {
-    self->utf8Codepoint |= 10 + *buf - 'a';
-  } else if ('A' <= *buf && *buf <= 'F') {
-    self->utf8Codepoint |= 10 + *buf - 'A';
-  } else [[unlikely]] {
+  auto hexVal = tables.hex[uint8_t(*buf)];
+  if (hexVal < 0) [[unlikely]] {
     return WeaselJson_REJECT;
   }
+  self->utf8Codepoint <<= 4;
+  self->utf8Codepoint |= hexVal;
   ++buf;
 
   // Write codepoint in utf-8 if there's room in the user provided buffer. If
@@ -599,16 +591,12 @@ inline PRESERVE_NONE WeaselJsonStatus t_hex2(Parser3 *self, char *buf,
 
 inline PRESERVE_NONE WeaselJsonStatus t_hex3(Parser3 *self, char *buf,
                                              char *bufEnd) {
-  self->utf8Codepoint <<= 4;
-  if (('0' <= *buf && *buf <= '9')) {
-    self->utf8Codepoint |= *buf - '0';
-  } else if ('a' <= *buf && *buf <= 'f') {
-    self->utf8Codepoint |= 10 + *buf - 'a';
-  } else if ('A' <= *buf && *buf <= 'F') {
-    self->utf8Codepoint |= 10 + *buf - 'A';
-  } else [[unlikely]] {
+  auto hexVal = tables.hex[uint8_t(*buf)];
+  if (hexVal < 0) [[unlikely]] {
     return WeaselJson_REJECT;
   }
+  self->utf8Codepoint <<= 4;
+  self->utf8Codepoint |= hexVal;
   ++buf;
 
   if (!(0xdc00 <= self->utf8Codepoint && self->utf8Codepoint <= 0xdfff))
diff --git a/src/tables.h b/src/tables.h
index 68b3e16..3349585 100644
--- a/src/tables.h
+++ b/src/tables.h
@@ -17,9 +17,22 @@ constexpr inline struct Tables {
     unescape['f'] = '\f';
     unescape['\\'] = '\\';
     unescape['/'] = '/';
+    for (int i = 0; i < 256; ++i) {
+      hex[i] = -1;
+    }
+    for (int i = '0'; i <= '9'; ++i) {
+      hex[i] = i - '0';
+    }
+    for (int i = 'a'; i <= 'f'; ++i) {
+      hex[i] = 10 + i - 'a';
+    }
+    for (int i = 'A'; i <= 'F'; ++i) {
+      hex[i] = 10 + i - 'A';
+    }
   }
   bool whitespace[256]{};
   char unescape[256]{};
+  int8_t hex[256]{};
 } tables;
 
 // See https://gist.github.com/pervognsen/218ea17743e1442e59bb60d29b1aa725 for