From a415045ae6d33b23316b77e96f638c9723d6d80c Mon Sep 17 00:00:00 2001
From: Wilson Lin <code@wilsonl.in>
Date: Mon, 24 Aug 2020 21:48:58 +1000
Subject: [PATCH] Fix hex numeric entity parsing

---
 gen/codepoints.ts  | 10 +++++-----
 src/proc/entity.rs |  6 +++---
 src/tests/mod.rs   |  4 ++++
 3 files changed, 12 insertions(+), 8 deletions(-)

diff --git a/gen/codepoints.ts b/gen/codepoints.ts
index 945bef7..ce81438 100644
--- a/gen/codepoints.ts
+++ b/gen/codepoints.ts
@@ -14,9 +14,9 @@ const WHITESPACE = [0x09, 0x0a, 0x0c, 0x0d, 0x20];
 const C0_CONTROL = rangeInclusive(0, 0x1f);
 const CONTROL = [...C0_CONTROL, ...rangeInclusive(0x7f, 0x9f)];
 const DIGIT = rangeInclusive(c('0'), c('9'));
-const UPPER_HEX_DIGIT = [...DIGIT, ...rangeInclusive(c('A'), c('F'))];
-const LOWER_HEX_DIGIT = [...DIGIT, ...rangeInclusive(c('a'), c('f'))];
-const HEX_DIGIT = [...UPPER_HEX_DIGIT, ...LOWER_HEX_DIGIT];
+const UPPER_HEX_ALPHA = [...rangeInclusive(c('A'), c('F'))];
+const LOWER_HEX_ALPHA = [...rangeInclusive(c('a'), c('f'))];
+const HEX_DIGIT = [...DIGIT, ...UPPER_HEX_ALPHA, ...LOWER_HEX_ALPHA];
 const UPPER_ALPHA = rangeInclusive(c('A'), c('Z'));
 const LOWER_ALPHA = rangeInclusive(c('a'), c('z'));
 const ALPHA = [...UPPER_ALPHA, ...LOWER_ALPHA];
@@ -61,8 +61,8 @@ impl std::ops::Index<u8> for Lookup {
 ` + Object.entries({
   WHITESPACE,
   DIGIT,
-  UPPER_HEX_DIGIT,
-  LOWER_HEX_DIGIT,
+  UPPER_HEX_ALPHA,
+  LOWER_HEX_ALPHA,
   HEX_DIGIT,
 
   ATTR_NAME_CHAR,
diff --git a/src/proc/entity.rs b/src/proc/entity.rs
index d527c22..0be7bb7 100644
--- a/src/proc/entity.rs
+++ b/src/proc/entity.rs
@@ -17,7 +17,7 @@ use crate::gen::entities::{ENTITY, EntityType};
 use crate::pattern::TrieNodeMatch;
 use std::char::from_u32;
 use crate::proc::Processor;
-use crate::gen::codepoints::{DIGIT, HEX_DIGIT, LOWER_HEX_DIGIT, UPPER_HEX_DIGIT, Lookup};
+use crate::gen::codepoints::{DIGIT, HEX_DIGIT, LOWER_HEX_ALPHA, UPPER_HEX_ALPHA, Lookup};
 
 enum Parsed {
     // This includes numeric entities that were invalid and decoded to 0xFFFD.
@@ -93,8 +93,8 @@ fn parse_entity(code: &mut [u8], read_pos: usize, write_pos: usize) -> Parsed {
                 HEX_DIGIT,
                 |value, c| value.wrapping_mul(16).wrapping_add(match c {
                     c if DIGIT[c] => (c - b'0') as u32,
-                    c if LOWER_HEX_DIGIT[c] => (c - b'a') as u32,
-                    c if UPPER_HEX_DIGIT[c] => (c - b'A') as u32,
+                    c if LOWER_HEX_ALPHA[c] => 10 + (c - b'a') as u32,
+                    c if UPPER_HEX_ALPHA[c] => 10 + (c - b'A') as u32,
                     _ => unreachable!(),
                 }),
                 6,
diff --git a/src/tests/mod.rs b/src/tests/mod.rs
index c9efd95..7cb4669 100644
--- a/src/tests/mod.rs
+++ b/src/tests/mod.rs
@@ -260,6 +260,10 @@ fn test_attr_value_backtick() {
 
 #[test]
 fn test_hexadecimal_entity_decoding() {
+    eval(b"&#x2E", b".");
+    eval(b"&#x2F", b"/");
+    eval(b"&#x2f", b"/");
+    eval(b"&#x00", b"\0");
     eval(b"&#x30", b"0");
     eval(b"&#x0030", b"0");
     eval(b"&#x000000000000000000000000000000000000000000030", b"0");