From da86c3076c0adf5d0ae8026b3a44942def657b6a Mon Sep 17 00:00:00 2001
From: Christopher Wellons <wellons@nullprogram.com>
Date: Sun, 8 Oct 2017 13:12:37 -0400
Subject: Add simple decoder to the benchmarks

---
 test/benchmark.c | 66 +++++++++++++++++++++++++++++++++++++++++++++++++++++---
 1 file changed, 63 insertions(+), 3 deletions(-)

diff --git a/test/benchmark.c b/test/benchmark.c
index 4eb51da..4e93408 100644
--- a/test/benchmark.c
+++ b/test/benchmark.c
@@ -71,9 +71,50 @@ buffer_fill(void *buf, size_t z)
     return p;
 }
 
+static unsigned char *
+utf8_simple(unsigned char *s, long *c)
+{
+    unsigned char *next;
+    if (s[0] < 0x80) {
+        *c = s[0];
+        next = s + 1;
+    } else if ((s[0] & 0xe0) == 0xc0) {
+        *c = ((long)(s[0] & 0x1f) <<  6) |
+             ((long)(s[1] & 0x3f) <<  0);
+        if ((s[1] & 0xc0) != 0x80)
+            *c = -1;
+        next = s + 2;
+    } else if ((s[0] & 0xf0) == 0xe0) {
+        *c = ((long)(s[0] & 0x0f) << 12) |
+             ((long)(s[1] & 0x3f) <<  6) |
+             ((long)(s[2] & 0x3f) <<  0);
+        if ((s[1] & 0xc0) != 0x80 ||
+            (s[2] & 0xc0) != 0x80)
+            *c = -1;
+        next = s + 3;
+    } else if ((s[0] & 0xf8) == 0xf0 && (s[0] <= 0xf4)) {
+        *c = ((long)(s[0] & 0x07) << 18) |
+             ((long)(s[1] & 0x3f) << 12) |
+             ((long)(s[2] & 0x3f) <<  6) |
+             ((long)(s[3] & 0x3f) <<  0);
+        if ((s[1] & 0xc0) != 0x80 ||
+            (s[2] & 0xc0) != 0x80 ||
+            (s[3] & 0xc0) != 0x80)
+            *c = -1;
+        next = s + 4;
+    } else {
+        *c = -1; // invalid
+        next = s + 1; // skip this byte
+    }
+    if (*c >= 0xd800 && *c <= 0xdfff)
+        *c = -1; // surrogate half
+    return next;
+}
+
 int
 main(void)
 {
+    double rate;
     long errors, n;
     size_t z = BUFLEN * 1024L * 1024;
     unsigned char *buffer = malloc(z);
@@ -97,8 +138,7 @@ main(void)
         if (p == end) // reached the end successfully?
             n++;
     } while (running);
-
-    double rate = n * (end - buffer) / (double)SECONDS / 1024 / 1024;
+    rate = n * (end - buffer) / (double)SECONDS / 1024 / 1024;
     printf("branchless: %f MB/s, %ld errors\n", rate, errors);
 
     /* Benchmark Bjoern Hoehrmann's decoder */
@@ -120,9 +160,29 @@ main(void)
         if (p == end) // reached the end successfully?
             n++;
     } while (running);
-
     rate = n * (end - buffer) / (double)SECONDS / 1024 / 1024;
     printf("Hoehrmann:  %f MB/s, %ld errors\n", rate, errors);
 
+    /* Benchmark simple decoder */
+    running = 1;
+    signal(SIGALRM, alarm_handler);
+    alarm(SECONDS);
+    errors = n = 0;
+    do {
+        unsigned char *p = buffer;
+        long c;
+        long count = 0;
+        while (p < end) {
+            p = utf8_simple(p, &c);
+            count++;
+            if (c < 0)
+                errors++;
+        }
+        if (p == end) // reached the end successfully?
+            n++;
+    } while (running);
+    rate = n * (end - buffer) / (double)SECONDS / 1024 / 1024;
+    printf("Simple:     %f MB/s, %ld errors\n", rate, errors);
+
     free(buffer);
 }
-- 
cgit v1.2.3