| 1 | /* |
| 2 | ** Copyright 2003-2009 Double Precision, Inc. See COPYING for |
| 3 | ** distribution information. |
| 4 | */ |
| 5 | |
| 6 | /* |
| 7 | ** $Id: encodeautodetect.c,v 1.2 2009/11/08 18:14:47 mrsam Exp $ |
| 8 | */ |
| 9 | #include "encode.h" |
| 10 | #include <string.h> |
| 11 | #include <stdlib.h> |
| 12 | #include "../unicode/unicode.h" |
| 13 | |
| 14 | static const char *libmail_encode_autodetect(const char *charset, |
| 15 | int (*func)(void *), void *arg) |
| 16 | { |
| 17 | const char *encoding="7bit"; |
| 18 | int l=0; |
| 19 | int longline=0; |
| 20 | int c; |
| 21 | const struct unicode_info *ci = unicode_find(charset); |
| 22 | |
| 23 | while ((c = (*func)(arg)) != EOF) |
| 24 | { |
| 25 | unsigned char ch= (unsigned char)c; |
| 26 | |
| 27 | if (ch >= 0x80) |
| 28 | { |
| 29 | |
| 30 | if (!charset || !*charset) |
| 31 | encoding="8bit"; |
| 32 | else if (ci && ci->flags & UNICODE_BODY_QUOPRI) |
| 33 | encoding="quoted-printable"; |
| 34 | else if (!ci || ci->flags & UNICODE_BODY_BASE64) |
| 35 | encoding="base64"; |
| 36 | else |
| 37 | encoding="8bit"; |
| 38 | } |
| 39 | |
| 40 | if (ch < 0x20 && |
| 41 | ch != '\t' && ch != '\r' && ch != '\n') |
| 42 | { |
| 43 | if (!charset || !*charset) |
| 44 | ; |
| 45 | else if (ci && ci->flags & UNICODE_BODY_QUOPRI) |
| 46 | encoding="quoted-printable"; |
| 47 | else if (!ci || ci->flags & UNICODE_BODY_BASE64) |
| 48 | encoding="base64"; |
| 49 | } |
| 50 | |
| 51 | if (ch == 0) |
| 52 | return "base64"; |
| 53 | |
| 54 | if (ch == '\n') l=0; |
| 55 | else if (++l > 990) |
| 56 | { |
| 57 | longline=1; |
| 58 | if (ci && ci->flags & UNICODE_BODY_QUOPRI) |
| 59 | encoding="quoted-printable"; |
| 60 | } |
| 61 | |
| 62 | } |
| 63 | |
| 64 | if (longline) |
| 65 | { |
| 66 | if (ci && ci->flags & UNICODE_BODY_QUOPRI) |
| 67 | encoding="quoted-printable"; |
| 68 | else |
| 69 | encoding="base64"; |
| 70 | } |
| 71 | return encoding; |
| 72 | } |
| 73 | |
| 74 | struct file_info { |
| 75 | FILE *fp; |
| 76 | off_t pos; |
| 77 | off_t end; |
| 78 | }; |
| 79 | |
| 80 | static int read_file(void *arg) |
| 81 | { |
| 82 | int c; |
| 83 | struct file_info *fi = (struct file_info *)arg; |
| 84 | if (fi->end >= 0 && fi->pos > fi->end) |
| 85 | return EOF; |
| 86 | c = getc(fi->fp); |
| 87 | fi->pos++; |
| 88 | return c; |
| 89 | } |
| 90 | |
| 91 | static int read_string(void * arg) |
| 92 | { |
| 93 | int c; |
| 94 | unsigned char **strp = (unsigned char **)arg; |
| 95 | if (**strp == 0) |
| 96 | return EOF; |
| 97 | c = (int)**strp; |
| 98 | (*strp)++; |
| 99 | return c; |
| 100 | } |
| 101 | |
| 102 | const char *libmail_encode_autodetect_fp(FILE *fp, int okQp) |
| 103 | { |
| 104 | if (okQp) |
| 105 | return libmail_encode_autodetect_fppos(fp, "ISO-8859-1", 0, -1); |
| 106 | else |
| 107 | return libmail_encode_autodetect_fppos(fp, NULL, 0, -1); |
| 108 | } |
| 109 | |
| 110 | const char *libmail_encode_autodetect_fppos(FILE *fp, const char *charset, |
| 111 | off_t start_pos, off_t end_pos) |
| 112 | { |
| 113 | struct file_info fi; |
| 114 | off_t orig_pos = ftell(fp); |
| 115 | off_t pos = orig_pos; |
| 116 | const char *rc; |
| 117 | |
| 118 | if (start_pos >= 0) |
| 119 | { |
| 120 | if (fseek(fp, start_pos, SEEK_SET) == (off_t)-1) |
| 121 | return NULL; |
| 122 | else |
| 123 | pos = start_pos; |
| 124 | } |
| 125 | |
| 126 | fi.fp = fp; |
| 127 | fi.pos = pos; |
| 128 | fi.end = end_pos; |
| 129 | rc = libmail_encode_autodetect(charset, &read_file, &fi); |
| 130 | |
| 131 | if (fseek(fp, orig_pos, SEEK_SET) == (off_t)-1) |
| 132 | return NULL; |
| 133 | return rc; |
| 134 | } |
| 135 | |
| 136 | const char *libmail_encode_autodetect_str(const char *str, const char *charset) |
| 137 | { |
| 138 | return libmail_encode_autodetect(charset, &read_string, &str); |
| 139 | } |