1 #include <stdio.h>

     2 #include <unicode/utf8.h>

     3 #include "dictre.h"

     4 

     5 static int url_get(FILE *f)

     6 {

     7     int val;

     8 

     9     val = fgetc(f);

    10     if (val == '%')

    11     {

    12         unsigned char num[3];

    13         num[2] = '\0';

    14         num[0] = fgetc(f);

    15         num[1] = fgetc(f);

    16         val = strtol(num, 0, 16);

    17     } else if (val == '&')

    18         return END_OF_URL;

    19     else if (val == '+')

    20         return ' ';

    21     /*printf("[%i]", val);*/

    22     return val;

    23 }

    24 

    25 static int char2num(unsigned char c)

    26 {

    27     if (c >= '0' && c <= '9')

    28         return c - '0';

    29     if (c >= 'A' && c <= 'Z')

    30         return c + 10 - 'A';

    31     if (c >= 'a' && c <= 'z')

    32         return c + 10 - 'a';

    33     return 0;

    34 }

    35 

    36 int http_getc(FILE *f)

    37 {

    38     int c;

    39     static unsigned char tmp[6]; /* for a UTF-8 string */

    40     static int itmp = -1;

    41     static int tmplen;

    42 

    43     if (itmp == -1)

    44     {

    45         c = url_get(f);

    46         if (c == '&')

    47         {

    48             c = url_get(f);

    49             if (c == '#')

    50             {

    51                 char iserror = 0;

    52                 int entval;

    53                 /*Get number*/

    54                 entval = 0;

    55                 while((c = url_get(f)) != ';')

    56                 {

    57                     /* Digits in base 10 */

    58                     entval = char2num(c) + entval * 10;

    59                 }

    60                 /*printf("{%i}", entval);*/

    61                 /*Get utf-8 version*/

    62                 tmplen = 0;

    63                 U8_APPEND(tmp, tmplen, 6, entval, iserror);

    64                 if (iserror)

    65                     return -3;

    66                 /* We need not to program itmp for the next run

    67                  * if we have only one character to send */

    68                 if (tmplen != 1)

    69                     itmp = 1;

    70                 return tmp[0];

    71             }

    72             else

    73                 return HTTP_DECODE_ERROR; /* ERROR! */

    74         }

    75     } else /* We already have a character to keep on sending */

    76     {

    77         int tosend;

    78         tosend = itmp;

    79         ++itmp;

    80         if (itmp == tmplen)

    81             itmp = -1;

    82         return tmp[tosend];

    83     }

    84     return c;

    85 }