From 1392882f35730480d7c89d6ac9193e9e48ae0f79 Mon Sep 17 00:00:00 2001 From: nagachika Date: Thu, 11 Apr 2013 19:13:29 +0000 Subject: merge revision(s) 39822,39856: [Backport #8141] * string.c (rb_str_conv_enc_opts): convert with one converter, instead of re-creating converters for each buffer expansion. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/branches/ruby_2_0_0@40256 b2dd03c8-39d4-4d8f-98ff-823fe69b080e --- string.c | 45 ++++++++++++++++++++++++++++++++------------- 1 file changed, 32 insertions(+), 13 deletions(-) (limited to 'string.c') diff --git a/string.c b/string.c index 6eebd9e096..b3893c9a95 100644 --- a/string.c +++ b/string.c @@ -490,12 +490,15 @@ RUBY_ALIAS_FUNCTION(rb_tainted_str_new2(const char *ptr), rb_tainted_str_new_cst VALUE rb_str_conv_enc_opts(VALUE str, rb_encoding *from, rb_encoding *to, int ecflags, VALUE ecopts) { + extern VALUE rb_cEncodingConverter; rb_econv_t *ec; rb_econv_result_t ret; - long len; + long len, olen; + VALUE econv_wrapper; VALUE newstr; - const unsigned char *sp; - unsigned char *dp; + const unsigned char *start, *sp; + unsigned char *dest, *dp; + size_t converted_output = 0; if (!to) return str; if (!from) from = rb_enc_get(str); @@ -511,23 +514,39 @@ rb_str_conv_enc_opts(VALUE str, rb_encoding *from, rb_encoding *to, int ecflags, len = RSTRING_LEN(str); newstr = rb_str_new(0, len); + olen = len; - retry: + econv_wrapper = rb_obj_alloc(rb_cEncodingConverter); + RBASIC(econv_wrapper)->klass = 0; ec = rb_econv_open_opts(from->name, to->name, ecflags, ecopts); if (!ec) return str; + DATA_PTR(econv_wrapper) = ec; sp = (unsigned char*)RSTRING_PTR(str); - dp = (unsigned char*)RSTRING_PTR(newstr); - ret = rb_econv_convert(ec, &sp, (unsigned char*)RSTRING_END(str), - &dp, (unsigned char*)RSTRING_END(newstr), 0); + start = sp; + while ((dest = (unsigned char*)RSTRING_PTR(newstr)), + (dp = dest + converted_output), + (ret = rb_econv_convert(ec, &sp, start + len, &dp, dest + olen, 0)), + ret == econv_destination_buffer_full) { + /* destination buffer short */ + size_t converted_input = sp - start; + size_t rest = len - converted_input; + converted_output = dp - dest; + rb_str_set_len(newstr, converted_output); + if (converted_input && converted_output && + rest < (LONG_MAX / converted_output)) { + rest = (rest * converted_output) / converted_input; + } + else { + rest = olen; + } + olen += rest < 2 ? 2 : rest; + rb_str_resize(newstr, olen); + } + DATA_PTR(econv_wrapper) = 0; rb_econv_close(ec); + rb_gc_force_recycle(econv_wrapper); switch (ret) { - case econv_destination_buffer_full: - /* destination buffer short */ - len = len < 2 ? 2 : len * 2; - rb_str_resize(newstr, len); - goto retry; - case econv_finished: len = dp - (unsigned char*)RSTRING_PTR(newstr); rb_str_set_len(newstr, len); -- cgit v1.2.3