From b3fd7b26be63c76711dcd70df79453fa0175cd9d Mon Sep 17 00:00:00 2001
From: Jean Boussier <jean.boussier@gmail.com>
Date: Sat, 1 Nov 2025 12:15:04 +0100
Subject: [PATCH] parser.c: Use SWAR to skip consecutive spaces
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Closes: https://github.com/ruby/json/pull/881

If we encounter a newline, it is likely that the document is pretty printed,
hence that the newline is followed by multiple spaces.

In such case we can use SWAR to count up to eight consecutive spaces at once.

```
== Parsing activitypub.json (58160 bytes)
ruby 3.4.6 (2025-09-16 revision dbd83256b1) +YJIT +PRISM [arm64-darwin24]
Warming up --------------------------------------
               after     1.118k i/100ms
Calculating -------------------------------------
               after     11.223k (± 0.7%) i/s   (89.10 μs/i) -     57.018k in   5.080522s

Comparison:
              before:    10834.4 i/s
               after:    11223.4 i/s - 1.04x  faster

== Parsing twitter.json (567916 bytes)
ruby 3.4.6 (2025-09-16 revision dbd83256b1) +YJIT +PRISM [arm64-darwin24]
Warming up --------------------------------------
               after   118.000 i/100ms
Calculating -------------------------------------
               after      1.188k (± 1.0%) i/s  (841.62 μs/i) -      6.018k in   5.065355s

Comparison:
              before:     1094.8 i/s
               after:     1188.2 i/s - 1.09x  faster

== Parsing citm_catalog.json (1727030 bytes)
ruby 3.4.6 (2025-09-16 revision dbd83256b1) +YJIT +PRISM [arm64-darwin24]
Warming up --------------------------------------
               after    58.000 i/100ms
Calculating -------------------------------------
               after    570.506 (± 3.7%) i/s    (1.75 ms/i) -      2.900k in   5.091529s

Comparison:
              before:      419.6 i/s
               after:      570.5 i/s - 1.36x  faster

== Parsing float parsing (2251051 bytes)
ruby 3.4.6 (2025-09-16 revision dbd83256b1) +YJIT +PRISM [arm64-darwin24]
Warming up --------------------------------------
               after    22.000 i/100ms
Calculating -------------------------------------
               after    212.010 (± 1.9%) i/s    (4.72 ms/i) -      1.078k in   5.086885s

Comparison:
              before:      189.4 i/s
               after:      212.0 i/s - 1.12x  faster
```

Co-Authored-By: Scott Myron <samyron@gmail.com>
---
 ext/json/ext/parser/parser.c | 46 +++++++++++++++++++++++++-----------
 ext/json/ext/simd/simd.h     |  3 +--
 2 files changed, 33 insertions(+), 16 deletions(-)

diff --git a/ext/json/ext/parser/parser.c b/ext/json/ext/parser/parser.c
index e591ca2c5..01234b5a8 100644
--- a/ext/json/ext/parser/parser.c
+++ b/ext/json/ext/parser/parser.c
@@ -557,14 +557,6 @@ static uint32_t unescape_unicode(JSON_ParserState *state, const unsigned char *p
 
 static const rb_data_type_t JSON_ParserConfig_type;
 
-static const bool whitespace[256] = {
-    [' '] = 1,
-    ['\t'] = 1,
-    ['\n'] = 1,
-    ['\r'] = 1,
-    ['/'] = 1,
-};
-
 static void
 json_eat_comments(JSON_ParserState *state)
 {
@@ -607,12 +599,38 @@ json_eat_comments(JSON_ParserState *state)
 static inline void
 json_eat_whitespace(JSON_ParserState *state)
 {
-    unsigned char cursor;
-    while (RB_UNLIKELY(whitespace[cursor = (unsigned char)peek(state)])) {
-        if (RB_UNLIKELY(cursor == '/')) {
-            json_eat_comments(state);
-        } else {
-            state->cursor++;
+    while (true) {
+        switch (peek(state)) {
+            case ' ':
+                state->cursor++;
+                break;
+            case '\n':
+                state->cursor++;
+
+                // Heuristic: if we see a newline, there is likely consecutive spaces after it.
+#if defined(__BYTE_ORDER__) && __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
+                while (rest(state) > 8) {
+                    uint64_t chunk;
+                    memcpy(&chunk, state->cursor, sizeof(uint64_t));
+                    size_t consecutive_spaces = trailing_zeros64(chunk ^ 0x2020202020202020) / CHAR_BIT;
+
+                    state->cursor += consecutive_spaces;
+                    if (consecutive_spaces != 8) {
+                        break;
+                    }
+                }
+#endif
+                break;
+            case '\t':
+            case '\r':
+                state->cursor++;
+                break;
+            case '/':
+                json_eat_comments(state);
+                break;
+
+            default:
+                return;
         }
     }
 }
diff --git a/ext/json/ext/simd/simd.h b/ext/json/ext/simd/simd.h
index 3abbdb020..2aa6c3d04 100644
--- a/ext/json/ext/simd/simd.h
+++ b/ext/json/ext/simd/simd.h
@@ -4,8 +4,6 @@ typedef enum {
     SIMD_SSE2
 } SIMD_Implementation;
 
-#ifdef JSON_ENABLE_SIMD
-
 #ifdef __clang__
 # if __has_builtin(__builtin_ctzll)
 #   define HAVE_BUILTIN_CTZLL 1
@@ -54,6 +52,7 @@ static inline int trailing_zeros(int input)
 #define FORCE_INLINE
 #endif
 
+#ifdef JSON_ENABLE_SIMD
 
 #define SIMD_MINIMUM_THRESHOLD 6