diff options
| author | hiro <hiro@ee746299-78ed-0310-b773-934348b2243d> | 2005-09-07 07:09:33 +0000 |
|---|---|---|
| committer | hiro <hiro@ee746299-78ed-0310-b773-934348b2243d> | 2005-09-07 07:09:33 +0000 |
| commit | e8a89709045c2dcb85098ae9ac515cccf6c9bb2d (patch) | |
| tree | 83e6d2fbe8fd938fa2e1a33e075f7c8ae14041ff /libsylph | |
| parent | f59c735b2a727a1359c485e9568e8c278a05c1ca (diff) | |
support UTF-8 in guessing Japanese encoding.
git-svn-id: svn://sylpheed.sraoss.jp/sylpheed/trunk@553 ee746299-78ed-0310-b773-934348b2243d
Diffstat (limited to 'libsylph')
| -rw-r--r-- | libsylph/codeconv.c | 37 |
1 files changed, 33 insertions, 4 deletions
diff --git a/libsylph/codeconv.c b/libsylph/codeconv.c index c88b588c..b3be5f5c 100644 --- a/libsylph/codeconv.c +++ b/libsylph/codeconv.c @@ -61,6 +61,7 @@ typedef enum (((c) & 0xff) >= 0xa1 && ((c) & 0xff) <= 0xdf) #define iseucaux(c) \ (((c) & 0xff) == 0x8f) + #define issjiskanji1(c) \ ((((c) & 0xff) >= 0x81 && ((c) & 0xff) <= 0x9f) || \ (((c) & 0xff) >= 0xe0 && ((c) & 0xff) <= 0xfc)) @@ -70,6 +71,17 @@ typedef enum #define issjishwkana(c) \ (((c) & 0xff) >= 0xa1 && ((c) & 0xff) <= 0xdf) +/* U+0080 - U+07FF */ +#define isutf8_2_1(c) \ + (((c) & 0xe0) == 0xc0) +#define isutf8_2_2(c) \ + (((c) & 0xc0) == 0x80) +/* U+0800 - U+FFFF */ +#define isutf8_3_1(c) \ + (((c) & 0xf0) == 0xe0) +#define isutf8_3_2(c) \ + (((c) & 0xc0) == 0x80) + #define K_IN() \ if (state != JIS_KANJI) { \ *out++ = ESC; \ @@ -799,19 +811,36 @@ CharSet conv_guess_ja_encoding(const gchar *str) guessed = C_EUC_JP; p += 2; } else if (issjiskanji1(*p) && issjiskanji2(*(p + 1))) { - if (iseuchwkana1(*p) && iseuchwkana2(*(p + 1))) - guessed = C_SHIFT_JIS; - else - return C_SHIFT_JIS; + guessed = C_SHIFT_JIS; p += 2; } else if (issjishwkana(*p)) { guessed = C_SHIFT_JIS; p++; } else { + if (guessed == C_US_ASCII) + guessed = C_AUTO; p++; } } + if (guessed != C_US_ASCII) { + p = (const guchar *)str; + + while (*p != '\0') { + if (isascii(*p)) { + p++; + } else if (isutf8_3_1(*p) && + isutf8_3_2(*(p + 1)) && + isutf8_3_2(*(p + 2))) { + p += 3; + } else { + return guessed; + } + } + + return C_UTF_8; + } + return guessed; } |