diff options
| -rw-r--r-- | ChangeLog | 10 | ||||
| -rw-r--r-- | enc/trans/escape.trans | 50 | ||||
| -rw-r--r-- | include/ruby/encoding.h | 4 | ||||
| -rw-r--r-- | test/ruby/test_econv.rb | 32 | ||||
| -rw-r--r-- | transcode.c | 44 |
5 files changed, 75 insertions, 65 deletions
@@ -1,3 +1,13 @@ +Sat Sep 6 23:03:41 2008 Tanaka Akira <akr@fsij.org> + + * include/ruby/encoding.h (ECONV_XML_TEXT_ENCODER): renamed from + ECONV_HTML_TEXT_ENCODER. + (ECONV_XML_ATTR_ENCODER): renamed from ECONV_HTML_ATTR_ENCODER. + + * enc/trans/escape.trans: follow the renaming. + + * transcode.c: ditto. + Sat Sep 6 21:35:52 2008 Yukihiro Matsumoto <matz@ruby-lang.org> * ext/socket/mkconstants.rb: add some IPV6 constans. [ruby-dev:36175] diff --git a/enc/trans/escape.trans b/enc/trans/escape.trans index b2aefbb19a..0641c6e251 100644 --- a/enc/trans/escape.trans +++ b/enc/trans/escape.trans @@ -1,7 +1,7 @@ #include "transcode_data.h" static int -fun_so_escape_html_chref(void *statep, const unsigned char *s, size_t l, unsigned char *o) +fun_so_escape_xml_chref(void *statep, const unsigned char *s, size_t l, unsigned char *o) { switch (*s) { case '&': @@ -45,16 +45,16 @@ fun_so_escape_html_chref(void *statep, const unsigned char *s, size_t l, unsigne map_amp["26"] = :func_so transcode_generate_node(ActionMap.parse(map_amp), "escape_amp_as_chref") - map_html_text = {} - map_html_text["{00-25,27-3B,3D,3F-FF}"] = :nomap - map_html_text["26"] = :func_so - map_html_text["3C"] = :func_so - map_html_text["3E"] = :func_so - transcode_generate_node(ActionMap.parse(map_html_text), "escape_html_text") + map_xml_text = {} + map_xml_text["{00-25,27-3B,3D,3F-FF}"] = :nomap + map_xml_text["26"] = :func_so + map_xml_text["3C"] = :func_so + map_xml_text["3E"] = :func_so + transcode_generate_node(ActionMap.parse(map_xml_text), "escape_xml_text") - map_html_attr = {} - map_html_attr["{00-FF}"] = :func_so - transcode_generate_node(ActionMap.parse(map_html_attr), "escape_html_attr") + map_xml_attr = {} + map_xml_attr["{00-FF}"] = :func_so + transcode_generate_node(ActionMap.parse(map_xml_attr), "escape_xml_attr") %> <%= transcode_generated_code %> @@ -68,26 +68,26 @@ rb_escape_amp_as_chref = { 5, /* max_output */ stateless_converter, /* stateful_type */ 0, NULL, NULL, - NULL, NULL, NULL, &fun_so_escape_html_chref + NULL, NULL, NULL, &fun_so_escape_xml_chref }; static const rb_transcoder -rb_escape_html_text = { - "", "html-text-escaped", escape_html_text, +rb_escape_xml_text = { + "", "xml-text-escaped", escape_xml_text, TRANSCODE_TABLE_INFO, 1, /* input_unit_length */ 1, /* max_input */ 5, /* max_output */ stateless_converter, /* stateful_type */ 0, NULL, NULL, - NULL, NULL, NULL, &fun_so_escape_html_chref + NULL, NULL, NULL, &fun_so_escape_xml_chref }; #define END 0 #define NORMAL 1 static int -escape_html_attr_init(void *statep) +escape_xml_attr_init(void *statep) { unsigned char *sp = statep; *sp = END; @@ -95,7 +95,7 @@ escape_html_attr_init(void *statep) } static int -fun_so_escape_html_attr(void *statep, const unsigned char *s, size_t l, unsigned char *o) +fun_so_escape_xml_attr(void *statep, const unsigned char *s, size_t l, unsigned char *o) { unsigned char *sp = statep; int n = 0; @@ -108,7 +108,7 @@ fun_so_escape_html_attr(void *statep, const unsigned char *s, size_t l, unsigned case '<': case '>': case '"': - n += fun_so_escape_html_chref(statep, s, l, o+n); + n += fun_so_escape_xml_chref(statep, s, l, o+n); break; default: @@ -119,7 +119,7 @@ fun_so_escape_html_attr(void *statep, const unsigned char *s, size_t l, unsigned } static int -escape_html_attr_finish(void *statep, unsigned char *o) +escape_xml_attr_finish(void *statep, unsigned char *o) { unsigned char *sp = statep; int n = 0; @@ -135,23 +135,23 @@ escape_html_attr_finish(void *statep, unsigned char *o) } static const rb_transcoder -rb_escape_html_attr = { - "", "html-attr-escaped", escape_html_attr, +rb_escape_xml_attr = { + "", "xml-attr-escaped", escape_xml_attr, TRANSCODE_TABLE_INFO, 1, /* input_unit_length */ 1, /* max_input */ 7, /* max_output */ stateful_encoder, /* stateful_type */ - 1, escape_html_attr_init, escape_html_attr_init, - NULL, NULL, NULL, fun_so_escape_html_attr, - escape_html_attr_finish + 1, escape_xml_attr_init, escape_xml_attr_init, + NULL, NULL, NULL, fun_so_escape_xml_attr, + escape_xml_attr_finish }; void Init_escape(void) { rb_register_transcoder(&rb_escape_amp_as_chref); - rb_register_transcoder(&rb_escape_html_text); - rb_register_transcoder(&rb_escape_html_attr); + rb_register_transcoder(&rb_escape_xml_text); + rb_register_transcoder(&rb_escape_xml_attr); } diff --git a/include/ruby/encoding.h b/include/ruby/encoding.h index 3959ae478e..7a616a1934 100644 --- a/include/ruby/encoding.h +++ b/include/ruby/encoding.h @@ -263,8 +263,8 @@ void rb_econv_binmode(rb_econv_t *ec); /* usable only if source encoding is ascii compatible */ #define ECONV_CRLF_NEWLINE_ENCODER 0x0200 #define ECONV_CR_NEWLINE_ENCODER 0x0400 -#define ECONV_HTML_TEXT_ENCODER 0x0800 -#define ECONV_HTML_ATTR_ENCODER 0x1000 +#define ECONV_XML_TEXT_ENCODER 0x0800 +#define ECONV_XML_ATTR_ENCODER 0x1000 /* end of flags for rb_econv_open */ diff --git a/test/ruby/test_econv.rb b/test/ruby/test_econv.rb index 2fcc6ab786..5c9fc143b1 100644 --- a/test/ruby/test_econv.rb +++ b/test/ruby/test_econv.rb @@ -34,7 +34,7 @@ class TestEncodingConverter < Test::Unit::TestCase assert_nil(Encoding::Converter.stateless_encoding("UTF-8")) assert_nil(Encoding::Converter.stateless_encoding("UTF-16BE")) assert_nil(Encoding::Converter.stateless_encoding(Encoding::UTF_8)) - assert_nil(Encoding::Converter.stateless_encoding("html-attr-escaped")) + assert_nil(Encoding::Converter.stateless_encoding("xml-attr-escaped")) end def test_stateless_encoding_iso2022jp @@ -728,54 +728,54 @@ class TestEncodingConverter < Test::Unit::TestCase assert_equal("&", ec.convert("&")) end - def test_html_escape_text + def test_xml_escape_text ec = Encoding::Converter.new("", "amp-escaped") assert_equal('&<>"', ec.convert("&<>\"")) assert_equal('', ec.finish) - ec = Encoding::Converter.new("", "html-text-escaped") + ec = Encoding::Converter.new("", "xml-text-escaped") assert_equal('&<>"', ec.convert("&<>\"")) assert_equal('', ec.finish) end - def test_html_escape_attr - ec = Encoding::Converter.new("", "html-attr-escaped") + def test_xml_escape_attr + ec = Encoding::Converter.new("", "xml-attr-escaped") assert_equal('""', ec.finish) - ec = Encoding::Converter.new("", "html-attr-escaped") + ec = Encoding::Converter.new("", "xml-attr-escaped") assert_equal('', ec.convert("")) assert_equal('""', ec.finish) - ec = Encoding::Converter.new("", "html-attr-escaped") + ec = Encoding::Converter.new("", "xml-attr-escaped") assert_equal('""', ec.convert('"')) assert_equal('"', ec.finish) - ec = Encoding::Converter.new("", "html-attr-escaped") + ec = Encoding::Converter.new("", "xml-attr-escaped") assert_equal('"&<>"', ec.convert("&<>\"")) assert_equal('"', ec.finish) end - def test_html_escape_with_charref - ec = Encoding::Converter.new("utf-8", "euc-jp", Encoding::Converter::HTML_TEXT_ENCODER|Encoding::Converter::UNDEF_HEX_CHARREF) + def test_xml_escape_with_charref + ec = Encoding::Converter.new("utf-8", "euc-jp", Encoding::Converter::XML_TEXT_ENCODER|Encoding::Converter::UNDEF_HEX_CHARREF) assert_equal('<♥>&"♡"', ec.convert("<\u2665>&\"\u2661\"")) assert_equal('', ec.finish) - ec = Encoding::Converter.new("utf-8", "euc-jp", Encoding::Converter::HTML_ATTR_ENCODER|Encoding::Converter::UNDEF_HEX_CHARREF) + ec = Encoding::Converter.new("utf-8", "euc-jp", Encoding::Converter::XML_ATTR_ENCODER|Encoding::Converter::UNDEF_HEX_CHARREF) assert_equal('"<♥>&"♡"', ec.convert("<\u2665>&\"\u2661\"")) assert_equal('"', ec.finish) - ec = Encoding::Converter.new("utf-8", "iso-2022-jp", Encoding::Converter::HTML_TEXT_ENCODER) + ec = Encoding::Converter.new("utf-8", "iso-2022-jp", Encoding::Converter::XML_TEXT_ENCODER) assert_equal("&\e$B$&\e(B&".force_encoding("iso-2022-jp"), ec.convert("&\u3046&")) assert_equal('', ec.finish) end - def test_html_hasharg + def test_xml_hasharg assert_equal("&\e$B$&\e(B♥&\"'".force_encoding("iso-2022-jp"), - "&\u3046\u2665&\"'".encode("iso-2022-jp", html: :text)) + "&\u3046\u2665&\"'".encode("iso-2022-jp", xml: :text)) assert_equal("\"&\e$B$&\e(B♡&"'\"".force_encoding("iso-2022-jp"), - "&\u3046\u2661&\"'".encode("iso-2022-jp", html: :attr)) + "&\u3046\u2661&\"'".encode("iso-2022-jp", xml: :attr)) assert_equal("&\u3046\u2661&\"'".force_encoding("utf-8"), - "&\u3046\u2661&\"'".encode("utf-8", html: :text)) + "&\u3046\u2661&\"'".encode("utf-8", xml: :text)) end end diff --git a/transcode.c b/transcode.c index 505811a3bf..acfe688ca4 100644 --- a/transcode.c +++ b/transcode.c @@ -21,7 +21,7 @@ VALUE rb_eNoConverter; VALUE rb_cEncodingConverter; static VALUE sym_invalid, sym_undef, sym_ignore, sym_replace; -static VALUE sym_html, sym_text, sym_attr; +static VALUE sym_xml, sym_text, sym_attr; static VALUE sym_universal_newline_decoder; static VALUE sym_crlf_newline_encoder; static VALUE sym_cr_newline_encoder; @@ -895,8 +895,8 @@ rb_econv_open(const char *sname, const char *dname, int ecflags) (ecflags & ECONV_UNIVERSAL_NEWLINE_DECODER)) return NULL; - if ((ecflags & ECONV_HTML_TEXT_ENCODER) && - (ecflags & ECONV_HTML_ATTR_ENCODER)) + if ((ecflags & ECONV_XML_TEXT_ENCODER) && + (ecflags & ECONV_XML_ATTR_ENCODER)) return NULL; num_encoders = 0; @@ -906,11 +906,11 @@ rb_econv_open(const char *sname, const char *dname, int ecflags) if (ecflags & ECONV_CR_NEWLINE_ENCODER) if (!(encoders[num_encoders++] = get_transcoder_entry("", "cr_newline"))) return NULL; - if (ecflags & ECONV_HTML_TEXT_ENCODER) - if (!(encoders[num_encoders++] = get_transcoder_entry("", "html-text-escaped"))) + if (ecflags & ECONV_XML_TEXT_ENCODER) + if (!(encoders[num_encoders++] = get_transcoder_entry("", "xml-text-escaped"))) return NULL; - if (ecflags & ECONV_HTML_ATTR_ENCODER) - if (!(encoders[num_encoders++] = get_transcoder_entry("", "html-attr-escaped"))) + if (ecflags & ECONV_XML_ATTR_ENCODER) + if (!(encoders[num_encoders++] = get_transcoder_entry("", "xml-attr-escaped"))) return NULL; num_decoders = 0; @@ -1791,8 +1791,8 @@ econv_description(const char *sname, const char *dname, int ecflags, VALUE mesg) if (ecflags & (ECONV_UNIVERSAL_NEWLINE_DECODER| ECONV_CRLF_NEWLINE_ENCODER| ECONV_CR_NEWLINE_ENCODER| - ECONV_HTML_TEXT_ENCODER| - ECONV_HTML_ATTR_ENCODER)) { + ECONV_XML_TEXT_ENCODER| + ECONV_XML_ATTR_ENCODER)) { const char *pre = ""; if (has_description) rb_str_cat2(mesg, " with "); @@ -1808,13 +1808,13 @@ econv_description(const char *sname, const char *dname, int ecflags, VALUE mesg) rb_str_cat2(mesg, pre); pre = ","; rb_str_cat2(mesg, "CR-newline"); } - if (ecflags & ECONV_HTML_TEXT_ENCODER) { + if (ecflags & ECONV_XML_TEXT_ENCODER) { rb_str_cat2(mesg, pre); pre = ","; - rb_str_cat2(mesg, "HTML-text"); + rb_str_cat2(mesg, "XML-text"); } - if (ecflags & ECONV_HTML_ATTR_ENCODER) { + if (ecflags & ECONV_XML_ATTR_ENCODER) { rb_str_cat2(mesg, pre); pre = ","; - rb_str_cat2(mesg, "HTML-attr"); + rb_str_cat2(mesg, "XML-attr"); } has_description = 1; } @@ -2166,17 +2166,17 @@ econv_opts(VALUE opt) rb_raise(rb_eArgError, "unknown value for undefined character option"); } - v = rb_hash_aref(opt, sym_html); + v = rb_hash_aref(opt, sym_xml); if (!NIL_P(v)) { v = rb_convert_type(v, T_SYMBOL, "Symbol", "to_sym"); if (v==sym_text) { - ecflags |= ECONV_HTML_TEXT_ENCODER|ECONV_UNDEF_HEX_CHARREF; + ecflags |= ECONV_XML_TEXT_ENCODER|ECONV_UNDEF_HEX_CHARREF; } else if (v==sym_attr) { - ecflags |= ECONV_HTML_ATTR_ENCODER|ECONV_UNDEF_HEX_CHARREF; + ecflags |= ECONV_XML_ATTR_ENCODER|ECONV_UNDEF_HEX_CHARREF; } else { - rb_raise(rb_eArgError, "unexpected value for html option: %s", rb_id2name(SYM2ID(v))); + rb_raise(rb_eArgError, "unexpected value for xml option: %s", rb_id2name(SYM2ID(v))); } } @@ -2328,8 +2328,8 @@ str_transcode0(int argc, VALUE *argv, VALUE *self, int ecflags, VALUE ecopts) if ((ecflags & (ECONV_UNIVERSAL_NEWLINE_DECODER| ECONV_CRLF_NEWLINE_ENCODER| ECONV_CR_NEWLINE_ENCODER| - ECONV_HTML_TEXT_ENCODER| - ECONV_HTML_ATTR_ENCODER)) == 0) { + ECONV_XML_TEXT_ENCODER| + ECONV_XML_ATTR_ENCODER)) == 0) { if (senc && senc == denc) { return -1; } @@ -3526,7 +3526,7 @@ Init_transcode(void) sym_undef = ID2SYM(rb_intern("undef")); sym_ignore = ID2SYM(rb_intern("ignore")); sym_replace = ID2SYM(rb_intern("replace")); - sym_html = ID2SYM(rb_intern("html")); + sym_xml = ID2SYM(rb_intern("xml")); sym_text = ID2SYM(rb_intern("text")); sym_attr = ID2SYM(rb_intern("attr")); @@ -3572,8 +3572,8 @@ Init_transcode(void) rb_define_const(rb_cEncodingConverter, "UNIVERSAL_NEWLINE_DECODER", INT2FIX(ECONV_UNIVERSAL_NEWLINE_DECODER)); rb_define_const(rb_cEncodingConverter, "CRLF_NEWLINE_ENCODER", INT2FIX(ECONV_CRLF_NEWLINE_ENCODER)); rb_define_const(rb_cEncodingConverter, "CR_NEWLINE_ENCODER", INT2FIX(ECONV_CR_NEWLINE_ENCODER)); - rb_define_const(rb_cEncodingConverter, "HTML_TEXT_ENCODER", INT2FIX(ECONV_HTML_TEXT_ENCODER)); - rb_define_const(rb_cEncodingConverter, "HTML_ATTR_ENCODER", INT2FIX(ECONV_HTML_ATTR_ENCODER)); + rb_define_const(rb_cEncodingConverter, "XML_TEXT_ENCODER", INT2FIX(ECONV_XML_TEXT_ENCODER)); + rb_define_const(rb_cEncodingConverter, "XML_ATTR_ENCODER", INT2FIX(ECONV_XML_ATTR_ENCODER)); rb_define_method(rb_eConversionUndefined, "source_encoding_name", ecerr_source_encoding_name, 0); rb_define_method(rb_eConversionUndefined, "destination_encoding_name", ecerr_destination_encoding_name, 0); |