src/io/utf8.c

   1 /*
   2 Copyright (C) 2001-2010, Parrot Foundation.
   3 $Id$
   4
   5 =head1 NAME
   6
   7 src/io/utf8.c - UTF8 I/O utility functions
   8
   9 =head1 DESCRIPTION
  10
  11 Convert output to utf8. Convert input to Parrot's internal string
  12 representation.
  13
  14 =head2 Utility functions
  15
  16 =over 4
  17
  18 =cut
  19
  20 */
  21
  22 #include "parrot/parrot.h"
  23 #include "io_private.h"
  24 #include "../string/unicode.h"
  25
  26 /* HEADERIZER HFILE: include/parrot/io.h */
  27
  28 /* HEADERIZER BEGIN: static */
  29 /* Don't modify between HEADERIZER BEGIN / HEADERIZER END.  Your changes will be lost. */
  30
  31 /* Don't modify between HEADERIZER BEGIN / HEADERIZER END.  Your changes will be lost. */
  32 /* HEADERIZER END: static */
  33
  34 /*
  35
  36 =item C<size_t Parrot_io_read_utf8(PARROT_INTERP, PMC *filehandle, STRING
  37 **buf)>
  38
  39 Read a string from a filehandle in UTF-8 format and convert it to a Parrot
  40 string.
  41
  42 =cut
  43
  44 */
  45
  46 size_t
  47 Parrot_io_read_utf8(PARROT_INTERP, ARGMOD(PMC *filehandle),
  48         ARGMOD(STRING **buf))
  49 {
  50     ASSERT_ARGS(Parrot_io_read_utf8)
  51     STRING *s, *s2;
  52     String_iter iter;
  53
  54     size_t len  = Parrot_io_read_buffer(interp, filehandle, buf);
  55     s           = *buf;
  56     s->encoding = Parrot_utf8_encoding_ptr;
  57
  58     /* count chars, verify utf8 */
  59     STRING_ITER_INIT(interp, &iter);
  60
  61     while (iter.bytepos < s->bufused) {
  62         if (iter.bytepos + 4 > s->bufused) {
  63             const utf8_t *u8ptr = (utf8_t *)((char *)s->strstart +
  64                     iter.bytepos);
  65             const UINTVAL c = *u8ptr;
  66
  67             if (UTF8_IS_START(c)) {
  68                 UINTVAL len2 = UTF8SKIP(u8ptr);
  69                 INTVAL  read;
  70
  71                 if (iter.bytepos + len2 <= s->bufused)
  72                     goto ok;
  73
  74                 /* need len - 1 more chars */
  75                 --len2;
  76                 s2 = Parrot_str_new_init(interp, NULL, len2,
  77                         Parrot_utf8_encoding_ptr, 0);
  78                 s2->bufused  = len2;
  79
  80                 read = Parrot_io_read_buffer(interp, filehandle, &s2);
  81                 UNUSED(read);
  82
  83                 s->strlen    = iter.charpos;
  84                 s            = Parrot_str_concat(interp, s, s2);
  85                 *buf         = s;
  86                 len         += len2 + 1;
  87
  88                 /* check last char */
  89             }
  90         }
  91 ok:
  92         Parrot_utf8_encoding_ptr->iter_get_and_advance(interp, *buf, &iter);
  93     }
  94     s->strlen = iter.charpos;
  95     return len;
  96 }
  97
  98 /*
  99
 100 =item C<size_t Parrot_io_write_utf8(PARROT_INTERP, PMC *filehandle, const STRING
 101 *s)>
 102
 103 Write a Parrot string to a filehandle in UTF-8 format.
 104
 105 =cut
 106
 107 */
 108
 109 size_t
 110 Parrot_io_write_utf8(PARROT_INTERP, ARGMOD(PMC *filehandle), ARGIN(const STRING *s))
 111 {
 112     ASSERT_ARGS(Parrot_io_write_utf8)
 113     STRING *dest;
 114
 115     if (s->encoding == Parrot_utf8_encoding_ptr)
 116         return Parrot_io_write_buffer(interp, filehandle, s);
 117
 118     dest = Parrot_utf8_encoding_ptr->to_encoding(interp, s);
 119     return Parrot_io_write_buffer(interp, filehandle, dest);
 120 }
 121
 122 /*
 123
 124 =back
 125
 126 =head1 SEE ALSO
 127
 128 F<src/io/io_passdown.c>,
 129 F<src/io/io.c>,
 130 F<src/io/io_layers.c>,
 131 F<src/io/io_private.h>.
 132
 133 =cut
 134
 135 */
 136
 137
 138 /*
 139  * Local variables:
 140  *   c-file-style: "parrot"
 141  * End:
 142  * vim: expandtab shiftwidth=4:
 143  */