From 15c6ab5602ea580010e439de272f95c9aae6b577 Mon Sep 17 00:00:00 2001 From: Lamont Granquist Date: Tue, 5 Jan 2016 12:34:04 -0800 Subject: fix validate_utf8: false encoding coercion the string we get back from ffi is tagged as ascii encoded by default, so we must force encode it first. then we don't want to convert from binary to utf-8 since that will wind up mangling all the utf-8 characters, we want to convert from utf-8 to utf-8 while replacing invalid characters. --- lib/ffi_yajl/encoder.rb | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'lib') diff --git a/lib/ffi_yajl/encoder.rb b/lib/ffi_yajl/encoder.rb index f4dd8c8..4c3c570 100644 --- a/lib/ffi_yajl/encoder.rb +++ b/lib/ffi_yajl/encoder.rb @@ -41,7 +41,7 @@ module FFI_Yajl # call either the ext or ffi hook str = do_yajl_encode(obj, yajl_gen_opts, opts) # we can skip cleaning the whole string for utf-8 issues if we have yajl validate as we go - str.encode!("utf-8", "binary", undef: :replace) unless yajl_gen_opts[:yajl_gen_validate_utf8] + str.force_encoding("UTF-8").encode!("UTF-8", undef: :replace, invalid: :replace) unless yajl_gen_opts[:yajl_gen_validate_utf8] str end @@ -56,7 +56,7 @@ module FFI_Yajl def self.raise_error_for_status(status, token = nil) # scrub token to valid utf-8 since we may be issuing an exception on an invalid utf-8 token - token = token.to_s.encode("utf-8", "binary", undef: :replace) + token = token.to_s.force_encoding("UTF-8").encode("utf-8", undef: :replace, invalid: :replace) case status when 1 # yajl_gen_keys_must_be_strings raise FFI_Yajl::EncodeError, "YAJL internal error: attempted use of non-string object as key" -- cgit v1.2.1