Merge lp:~vanvugt/mir/typo-implement-utf8 into lp:mir

[Daniel van Vugt]

Implement typo::Renderer::unicode_from_utf8

Copied from myself:
https://sourceforge.net/p/dapar/code/ci/master/tree/dapar.c

Unmerged revisions

3800. By Daniel van Vugt on 2016-11-03

Implement typo::Renderer::unicode_from_utf8

=== modified file 'playground/demo-shell/typo/typo_renderer.cpp'
--- playground/demo-shell/typo/typo_renderer.cpp	2016-10-11 12:51:40 +0000
+++ playground/demo-shell/typo/typo_renderer.cpp	2016-11-03 07:51:47 +0000
@@ -18,6 +18,7 @@
 
 #include "typo_renderer.h"
 #include <cstring>
+#include <cstdint>
 
 using namespace mir::typo;
 
@@ -49,11 +50,47 @@
 {
 }
 
+const unsigned long Renderer::invalid_unicode = 0x80000000UL;
+
+namespace {
+int utf8_char_len(uint8_t const* c)
+{
+    return (c[0] < 0x80) ? 1 :
+           ((c[0] & 0xE0) == 0xC0) ? 2 :
+           ((c[0] & 0xF0) == 0xE0) ? 3 :
+           ((c[0] & 0xF8) == 0xF0) ? 4 : 0;
+}
+}
+
 unsigned long Renderer::unicode_from_utf8(char const** utf8)
 {
-    int char_len = 1; // TODO: Add support for non-ASCII UTF-8
-    unsigned long unicode = **utf8;
-    if (unicode)
-        *utf8 += char_len;
-    return unicode;
+    uint8_t b0 = **(const uint8_t**)utf8;
+    int len = utf8_char_len(&b0);
+    unsigned long u;
+    if (len == 1)
+    {
+        u = b0;
+    }
+    else if (len > 1)
+    {
+        u = b0 & (0x7F >> len);
+        for (int i = 1; i < len; i++)
+        {
+            uint8_t bi = (*(const uint8_t**)utf8)[i];
+            if ((bi & 0xC0) != 0x80)
+            {
+                len = i + 1;
+                u = invalid_unicode;
+                break;
+            }
+            u = (u << 6) | ((unsigned long)bi & 0x3F);
+        }
+    }
+    else
+    {
+        u = invalid_unicode;
+        len = 1;
+    }
+    *utf8 += len;
+    return u;
 }

=== modified file 'playground/demo-shell/typo/typo_renderer.h'
--- playground/demo-shell/typo/typo_renderer.h	2016-10-11 12:51:40 +0000
+++ playground/demo-shell/typo/typo_renderer.h	2016-11-03 07:51:47 +0000
@@ -53,6 +53,7 @@
     virtual void render(char const* str, Image& img) = 0;
 
 protected:
+    static const unsigned long invalid_unicode;
     static unsigned long unicode_from_utf8(char const** utf8);
 };