From 3c7d6f3528432a433cd368605e4dc6f0fecaa898 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Bert=20M=C3=BCnnich?= <ber.t@posteo.de>
Date: Thu, 7 Dec 2017 14:57:02 +0100
Subject: Replace utf8codepoint with Chris Wellons' utf8_decode

Code under a different license should be kept in a separate file. This
implemention is a single header file with ~65 lines, so it better fits this
requirement.
---
 utf8.h | 68 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 68 insertions(+)
 create mode 100644 utf8.h

(limited to 'utf8.h')

diff --git a/utf8.h b/utf8.h
new file mode 100644
index 0000000..8c6a7a0
--- /dev/null
+++ b/utf8.h
@@ -0,0 +1,68 @@
+/* Branchless UTF-8 decoder
+ *
+ * This is free and unencumbered software released into the public domain.
+ */
+#ifndef UTF8_H
+#define UTF8_H
+
+#include <stdint.h>
+
+/* Decode the next character, C, from BUF, reporting errors in E.
+ *
+ * Since this is a branchless decoder, four bytes will be read from the
+ * buffer regardless of the actual length of the next character. This
+ * means the buffer _must_ have at least three bytes of zero padding
+ * following the end of the data stream.
+ *
+ * Errors are reported in E, which will be non-zero if the parsed
+ * character was somehow invalid: invalid byte sequence, non-canonical
+ * encoding, or a surrogate half.
+ *
+ * The function returns a pointer to the next character. When an error
+ * occurs, this pointer will be a guess that depends on the particular
+ * error, but it will always advance at least one byte.
+ */
+static void *
+utf8_decode(void *buf, uint32_t *c, int *e)
+{
+    static const char lengths[] = {
+        1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
+        0, 0, 0, 0, 0, 0, 0, 0, 2, 2, 2, 2, 3, 3, 4, 0
+    };
+    static const int masks[]  = {0x00, 0x7f, 0x1f, 0x0f, 0x07};
+    static const uint32_t mins[] = {4194304, 0, 128, 2048, 65536};
+    static const int shiftc[] = {0, 18, 12, 6, 0};
+    static const int shifte[] = {0, 6, 4, 2, 0};
+
+    unsigned char *s = buf;
+    int len = lengths[s[0] >> 3];
+
+    /* Compute the pointer to the next character early so that the next
+     * iteration can start working on the next character. Neither Clang
+     * nor GCC figure out this reordering on their own.
+     */
+    unsigned char *next = s + len + !len;
+
+    /* Assume a four-byte character and load four bytes. Unused bits are
+     * shifted out.
+     */
+    *c  = (uint32_t)(s[0] & masks[len]) << 18;
+    *c |= (uint32_t)(s[1] & 0x3f) << 12;
+    *c |= (uint32_t)(s[2] & 0x3f) <<  6;
+    *c |= (uint32_t)(s[3] & 0x3f) <<  0;
+    *c >>= shiftc[len];
+
+    /* Accumulate the various error conditions. */
+    *e  = (*c < mins[len]) << 6; // non-canonical encoding
+    *e |= ((*c >> 11) == 0x1b) << 7;  // surrogate half?
+    *e |= (*c > 0x10FFFF) << 8;  // out of range?
+    *e |= (s[1] & 0xc0) >> 2;
+    *e |= (s[2] & 0xc0) >> 4;
+    *e |= (s[3]       ) >> 6;
+    *e ^= 0x2a; // top two bits of each tail byte correct?
+    *e >>= shifte[len];
+
+    return next;
+}
+
+#endif
-- 
cgit v1.2.3