rb_str_buf_append: fastpath to str_buf_cat

If the LHS is ASCII compatible and the RHS is 7BIT
we can directly concat without being concerned about
anything else.

Benchmark:
```
compare-ruby: ruby 3.2.0dev (2022-07-12T15:01:11Z master 71aec68566) [arm64-darwin21]
built-ruby: ruby 3.2.0dev (2022-07-13T10:13:53Z faster-buffer-conc.. a04c10476d) [arm64-darwin21]
warming up...

|                      |compare-ruby|built-ruby|
|:---------------------|-----------:|---------:|
|binary_append_utf8    |    385.315k|  573.663k|
|                      |           -|     1.49x|
|binary_append_binary  |    446.579k|  574.898k|
|                      |           -|     1.29x|
|utf8_append_utf8      |    430.936k|  573.394k|
|                      |           -|     1.33x|
```

Note that in the benchmark, the RHS always have a precomputed
coderange. So the benchmark never enter the slowpath of having to
scan the RHS. However it's extremly likely that we'll end
up scanning it anyway in rb_enc_cr_str_buf_cat

1 files changed, 19 insertions, 3 deletions

diff --git a/string.c b/string.c
index 4cca9b1eac..d6c79c7988 100644
--- a/string.c
+++ b/string.c
@@ -3303,12 +3303,28 @@ rb_str_buf_cat_ascii(VALUE str, const char *ptr)
     }
 }
 
+static inline bool
+str_enc_fastpath(VALUE str)
+{
+    // The overwhelming majority of strings are in one of these 3 encodings.
+    switch (ENCODING_GET_INLINED(str)) {
+      case ENCINDEX_ASCII_8BIT:
+      case ENCINDEX_UTF_8:
+      case ENCINDEX_US_ASCII:
+        return true;
+      default:
+        return false;
+    }
+}
+
 VALUE
 rb_str_buf_append(VALUE str, VALUE str2)
 {
-    int str2_cr;
-
-    str2_cr = ENC_CODERANGE(str2);
+    int str2_cr = rb_enc_str_coderange(str2);
+    if (str2_cr == ENC_CODERANGE_7BIT && str_enc_fastpath(str)) {
+        str_buf_cat(str, RSTRING_PTR(str2), RSTRING_LEN(str2));
+        return str;
+    }
 
     rb_enc_cr_str_buf_cat(str, RSTRING_PTR(str2), RSTRING_LEN(str2),
         ENCODING_GET(str2), str2_cr, &str2_cr);