Changeset 2062

Timestamp:

2001-02-23T17:13:35+13:00 (23 years ago)

Author:

paynter

Message:

Added a function from the receptionist that decodes phrase CGI arguments
from URL-safe format (e.g. %43%41%54 means CAT) so we can do searches on
UTF8 text.

File:

• trunk/gsdl/src/phind/host/phindcgi.cpp (modified) (3 diffs)

trunk/gsdl/src/phind/host/phindcgi.cpp

@@ -69 +69 @@
             unsigned long &first_d, unsigned long &last_d,
             bool &XMLmode);
+
+void decode_cgi_arg (text_t &argstr);
 
 void print_expansions(char *cgi_script, char *collection, bool XMLmode, UCArray body,
@@ -900 +902 @@
       // p: the phrase text
       else if (key[0] == 'p') {
+    decode_cgi_arg(value);
     toUCArray(value, phrasetext);
       }
@@ -916 +919 @@
     strcpy(classifier, "1");
   }
+}
+
+
+// Convert %xx and + to their appropriate equivalents
+//
+// This function was copied from %GSDLHOME/src/recpt/cgiutils.cpp
+// because it was much easier to copy it than to link against it.
+
+static unsigned short hexdigit (unsigned short c) {
+  if (c >= '0' && c <= '9') return (c-'0');
+  if (c >= 'a' && c <= 'f') return (c-'a'+10);
+  if (c >= 'A' && c <= 'F') return (c-'A'+10);
+  return c;
+}
+
+void decode_cgi_arg (text_t &argstr) {
+  text_t::iterator in = argstr.begin();
+  text_t::iterator out = in;
+  text_t::iterator end = argstr.end();
+  
+  while (in != end) {
+    if (*in == '+') *out = ' ';
+    
+    else if (*in == '%') {
+      unsigned short c = '%';
+      in++;
+      if (in != end) {
+    c = hexdigit (*in);
+    in++;
+      }
+      if (in != end && c < 16) { // sanity check on the previous character
+    c = c*16 + hexdigit (*in);
+      }
+      
+      *out = c;
+    } else *out = *in;
+    
+    if (in != end) in++;
+    out++;
+  }
+  
+  // remove the excess characters
+  argstr.erase (out, end);
 }