Changeset 3151

Timestamp:

2002-06-17T15:27:27+12:00 (22 years ago)

Author:

jrm21

Message:

MSIE >= 6.0 encodes high bytes in URL as %uHHHH (unicode) rather than %HH%HH (utf8), apparently regardless of what character encoding is in use. We now convert that as well, although we assume we always convert it to utf8. This should really use whatever encoding the user interface is using...

File:

• trunk/gsdl/src/recpt/cgiutils.cpp (modified) (3 diffs)

trunk/gsdl/src/recpt/cgiutils.cpp

@@ -25 +25 @@
 
 #include "cgiutils.h"
-
+#include "gsdlunicode.h"
 
 static unsigned short hexdigit (unsigned short c) {
@@ -57 +57 @@
 
 // convert %xx and + to their appropriate equivalents
+// IE 6.0 and later use "%u" followed by 4 hex digits...
 void decode_cgi_arg (text_t &argstr) {
   text_t::iterator in = argstr.begin();
@@ -67 +68 @@
     else if (*in == '%') {
       unsigned short c = '%';
-      in++;
-      if (in != end) {
-    c = hexdigit (*in);
-    in++;
+      ++in;
+      if (in != end) { // this is an encoding...
+    if (*in == 'u') { // convert %uHHHH to unicode then current encoding
+      // this assumes a short int is at least 16 bits...
+      ++in;  
+      if (in != end)
+        c=hexdigit(*in++) << 12;
+      if (in != end)
+        c+=hexdigit(*in++) << 8;
+      if (in != end)
+        c+=hexdigit(*in++) << 4;
+      if (in != end)
+        c+=hexdigit(*in);
+      /* BAD!! The following assumes the interface is using utf-8. But
+         at this point we don't know what encoding we are using, unless
+         we can parse it out of the string we are currently decoding... */
+      text_t uni=" ";
+      uni[0]=c;
+      text_t utf8=to_utf8(uni);
+      int last_byte=utf8.size()-1;
+      for (int i=0;i<last_byte;i++)
+        *out++ = utf8[i];
+      c=utf8[last_byte];
+    } else {  // convert %HH to hex value
+      c = hexdigit (*in);
+      ++in;
+      if (in != end && c < 16) { // sanity check on the previous character
+        c = c*16 + hexdigit (*in);
+      }
+    }
       }
-      if (in != end && c < 16) { // sanity check on the previous character
-    c = c*16 + hexdigit (*in);
-      }
-      
       *out = c;
     } else *out = *in;