viric@19: #include <stdio.h>
viric@19: #include <unicode/utf8.h>
viric@19: #include "dictre.h"
viric@19: 
viric@19: static int url_get(FILE *f)
viric@19: {
viric@19:     int val;
viric@19: 
viric@19:     val = fgetc(f);
viric@19:     if (val == '%')
viric@19:     {
viric@19:         unsigned char num[3];
viric@19:         num[2] = '\0';
viric@19:         num[0] = fgetc(f);
viric@19:         num[1] = fgetc(f);
viric@19:         val = strtol(num, 0, 16);
viric@19:     } else if (val == '&')
viric@19:         return END_OF_URL;
viric@19:     else if (val == '+')
viric@19:         return ' ';
viric@19:     /*printf("[%i]", val);*/
viric@19:     return val;
viric@19: }
viric@19: 
viric@19: static int char2num(unsigned char c)
viric@19: {
viric@19:     if (c >= '0' && c <= '9')
viric@19:         return c - '0';
viric@19:     if (c >= 'A' && c <= 'Z')
viric@19:         return c + 10 - 'A';
viric@19:     if (c >= 'a' && c <= 'z')
viric@19:         return c + 10 - 'a';
viric@19:     return 0;
viric@19: }
viric@19: 
viric@19: int http_getc(FILE *f)
viric@19: {
viric@19:     int c;
viric@19:     static unsigned char tmp[6]; /* for a UTF-8 string */
viric@19:     static int itmp = -1;
viric@19:     static int tmplen;
viric@19: 
viric@19:     if (itmp == -1)
viric@19:     {
viric@19:         c = url_get(f);
viric@19:         if (c == '&')
viric@19:         {
viric@19:             c = url_get(f);
viric@19:             if (c == '#')
viric@19:             {
viric@19:                 char iserror = 0;
viric@19:                 int entval;
viric@19:                 /*Get number*/
viric@19:                 entval = 0;
viric@19:                 while((c = url_get(f)) != ';')
viric@19:                 {
viric@19:                     /* Digits in base 10 */
viric@19:                     entval = char2num(c) + entval * 10;
viric@19:                 }
viric@19:                 /*printf("{%i}", entval);*/
viric@19:                 /*Get utf-8 version*/
viric@19:                 tmplen = 0;
viric@19:                 U8_APPEND(tmp, tmplen, 6, entval, iserror);
viric@19:                 if (iserror)
viric@19:                     return -3;
viric@19:                 /* We need not to program itmp for the next run
viric@19:                  * if we have only one character to send */
viric@19:                 if (tmplen != 1)
viric@19:                     itmp = 1;
viric@19:                 return tmp[0];
viric@19:             }
viric@19:             else
viric@19:                 return HTTP_DECODE_ERROR; /* ERROR! */
viric@19:         }
viric@19:     } else /* We already have a character to keep on sending */
viric@19:     {
viric@19:         int tosend;
viric@19:         tosend = itmp;
viric@19:         ++itmp;
viric@19:         if (itmp == tmplen)
viric@19:             itmp = -1;
viric@19:         return tmp[tosend];
viric@19:     }
viric@19:     return c;
viric@19: }