diff options
-rw-r--r-- | ChangeLog | 7 | ||||
-rw-r--r-- | io.c | 13 | ||||
-rw-r--r-- | string.c | 55 | ||||
-rw-r--r-- | version.h | 4 |
4 files changed, 75 insertions, 4 deletions
@@ -1,3 +1,10 @@ +Wed Feb 27 22:55:42 2008 NARUSE, Yui <naruse@ruby-lang.org> + + * string.c (rb_str_coderange_scan_restartable): coderange scaning + for partial read. + + * io.c (read_all): set coderange when not convert encoding. + Wed Feb 27 03:55:58 2008 Nobuyoshi Nakada <nobu@ruby-lang.org> * ext/extmk.rb, enc/make_encmake.rb: load current mkmf.rb even if @@ -1358,6 +1358,9 @@ read_all(rb_io_t *fptr, long siz, VALUE str) { long bytes = 0; long n; + long pos = 0; + rb_encoding *enc = io_read_encoding(fptr); + int cr = 0; if (siz == 0) siz = BUFSIZ; if (NIL_P(str)) { @@ -1370,15 +1373,21 @@ read_all(rb_io_t *fptr, long siz, VALUE str) READ_CHECK(fptr); n = io_fread(str, bytes, fptr); if (n == 0 && bytes == 0) { - break; + break; } bytes += n; + if (cr != ENC_CODERANGE_BROKEN) + pos = rb_str_coderange_scan_restartable(RSTRING_PTR(str) + pos, RSTRING_PTR(str) + bytes, enc, &cr); if (bytes < siz) break; siz += BUFSIZ; rb_str_resize(str, siz); } if (bytes != siz) rb_str_resize(str, bytes); - return io_enc_str(str, fptr); + str = io_enc_str(str, fptr); + if (fptr->enc2) { + ENC_CODERANGE_SET(str, cr); + } + return str; } void @@ -201,6 +201,61 @@ coderange_scan(const char *p, long len, rb_encoding *enc) return ENC_CODERANGE_VALID; } +long +rb_str_coderange_scan_restartable(const char *s, const char *e, rb_encoding *enc, int *cr) +{ + long c; + const char *p; + int ret; + p = s; + + if (*cr == ENC_CODERANGE_BROKEN) + return e - s; + + if (rb_enc_to_index(enc) == 0) { + /* enc is ASCII-8BIT. ASCII-8BIT string never be broken. */ + p = search_nonascii(p, e); + *cr = (!p && *cr != ENC_CODERANGE_VALID) ? ENC_CODERANGE_7BIT : ENC_CODERANGE_VALID; + return e - s; + } + else if (rb_enc_asciicompat(enc)) { + p = search_nonascii(p, e); + if (!p) { + if (*cr != ENC_CODERANGE_VALID) *cr = ENC_CODERANGE_7BIT; + return e - s; + } + while (p < e) { + int ret = rb_enc_precise_mbclen(p, e, enc); + if (!MBCLEN_CHARFOUND_P(ret)) { + *cr = MBCLEN_INVALID_P(ret) ? ENC_CODERANGE_BROKEN: ret; + return e - s; + } + p += MBCLEN_CHARFOUND_LEN(ret); + if (p < e) { + p = search_nonascii(p, e); + if (!p) { + *cr = ENC_CODERANGE_VALID; + return e - s; + } + } + } + *cr = e < p ? ENC_CODERANGE_BROKEN: ENC_CODERANGE_VALID; + return p - s; + } + else { + while (p < e) { + int ret = rb_enc_precise_mbclen(p, e, enc); + if (!MBCLEN_CHARFOUND_P(ret)) { + *cr = MBCLEN_INVALID_P(ret) ? ENC_CODERANGE_BROKEN: ret; + return p - s; + } + p += MBCLEN_CHARFOUND_LEN(ret); + } + *cr = e < p ? ENC_CODERANGE_BROKEN: ENC_CODERANGE_VALID; + return p - s; + } +} + static void rb_enc_str_copy(VALUE dest, VALUE src) { @@ -1,7 +1,7 @@ #define RUBY_VERSION "1.9.0" -#define RUBY_RELEASE_DATE "2008-02-27" +#define RUBY_RELEASE_DATE "2008-02-28" #define RUBY_VERSION_CODE 190 -#define RUBY_RELEASE_CODE 20080227 +#define RUBY_RELEASE_CODE 20080228 #define RUBY_PATCHLEVEL 0 #define RUBY_VERSION_MAJOR 1 |