Changeset e8161ec

Timestamp:

2012-12-01T12:13:44Z (12 years ago)

Author:

Wilmer van der Gaast <wilmer@…>

Branches:

master

Children:

3ca001b

Parents:

85cd12d

Message:

Importing json.c UTF-8 encoding fix from upstream. Not importing the
strto(d|l) fix because I can just use GLib instead of reinventing the wheel.
This should fix corruption issues with tweets in Cyrillic. I've double-
checked this code against ​http://en.wikipedia.org/wiki/UTF-8 now and it looks
correct to me.

File:

• lib/json.c (modified) (3 diffs)

lib/json.c

@@ -43 +43 @@
 #endif
 
+#include <glib.h>
 #include <stdlib.h>
 #include <stdio.h>
@@ -287 +288 @@
                            string_length += 2;
                         else
-                        {  string [string_length ++] = 0xC0 | ((uc_b2 & 0xC0) >> 6) | ((uc_b1 & 0x3) << 3);
+                        {  string [string_length ++] = 0xC0 | ((uc_b2 & 0xC0) >> 6) | ((uc_b1 & 0x7) << 2);
                            string [string_length ++] = 0x80 | (uc_b2 & 0x3F);
                         }
@@ -483 +484 @@
 
                            if (top->type == json_double)
-                              top->u.dbl = strtod (i, (json_char **) &i);
+                              top->u.dbl = g_ascii_strtod (i, (json_char **) &i);
                            else
-                              top->u.integer = strtoll (i, (json_char **) &i, 10);
+                              top->u.integer = g_ascii_strtoll (i, (json_char **) &i, 10);
 
                            flags |= flag_next | flag_reproc;