--- parser3/src/classes/string.C 2015/05/16 22:24:38 1.217 +++ parser3/src/classes/string.C 2015/10/08 23:14:56 1.221 @@ -20,7 +20,7 @@ #include "pa_vregex.h" #include "pa_charsets.h" -volatile const char * IDENT_STRING_C="$Id: string.C,v 1.217 2015/05/16 22:24:38 moko Exp $"; +volatile const char * IDENT_STRING_C="$Id: string.C,v 1.221 2015/10/08 23:14:56 moko Exp $"; // class @@ -53,6 +53,9 @@ DECLARE_CLASS_VAR(void, new MVoid, 0); #define MODE_APPEND "append" +#define UNESCAPE_MODE_JS "js" +#define UNESCAPE_MODE_URI "uri" + // statics static const String match_var_name(MATCH_VAR_NAME); @@ -678,6 +681,8 @@ static void _trim(Request& r, MethodPara kind=String::TRIM_START; else if(skind==TRIM_END_OPTION || skind=="end") kind=String::TRIM_END; + else if(params_count==1) + chars=skind.cstr(); else throw Exception(PARSER_RUNTIME, &skind, @@ -695,9 +700,9 @@ static void _trim(Request& r, MethodPara } static void _base64(Request& r, MethodParams& params) { - if(params.count()) { + if(&r.get_self() == string_class) { // decode: ^string:base64[encoded[;$.strict(true|false)]] - const char* cstr=params.as_string(0, PARAMETER_MUST_BE_STRING).cstr(); + const char* cstr=params.count() ? params.as_string(0, PARAMETER_MUST_BE_STRING).cstr() : ""; char* decoded=0; size_t length=0; @@ -734,9 +739,9 @@ static void _base64(Request& r, MethodPa } static void _idna(Request& r, MethodParams& params) { - if(params.count()) { + if(&r.get_self() == string_class) { // decode: ^string:idna[encoded] - const char* cstr=params.as_string(0, PARAMETER_MUST_BE_STRING).cstr(); + const char* cstr=params.count() ? params.as_string(0, PARAMETER_MUST_BE_STRING).cstr() : ""; r.write_assign_lang(*new String(pa_idna_decode(cstr, r.charsets.source()), String::L_TAINTED)); } else { // encode: ^str.idna[] @@ -746,17 +751,50 @@ static void _idna(Request& r, MethodPara } } -static void _escape(Request& r, MethodParams&){ +static void _js_escape(Request& r, MethodParams&){ const String& src=GET_SELF(r, VString).string(); r.write_assign_lang(src.escape(r.charsets.source())); } -static void _unescape(Request& r, MethodParams& params){ +static void _js_unescape(Request& r, MethodParams& params){ const String& src=params.as_string(0, PARAMETER_MUST_BE_STRING); if(const char* result=unescape_chars(src.cstr(), src.length(), &r.charsets.source(), true)) r.write_assign_lang(*new String(result, String::L_TAINTED)); } +static void _unescape(Request& r, MethodParams& params){ + const String& mode=params.as_string(0, MODE_MUST_NOT_BE_CODE); + const String& src=params.as_string(1, PARAMETER_MUST_BE_STRING); + + Charset* from_charset=&r.charsets.client(); + + if(params.count() > 2) + if(HashStringValue* options=params.as_hash(2)) { + int valid_options=0; + if(Value* vcharset_name=options->get(PA_CHARSET_NAME)){ + from_charset=&::charsets.get(vcharset_name->as_string().change_case(r.charsets.source(), String::CC_UPPER)); + valid_options++; + } + if(valid_options!=options->count()) + throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION); + } + + bool mode_js; + if(mode==UNESCAPE_MODE_JS){ + mode_js=true; + } else if(mode==UNESCAPE_MODE_URI){ + mode_js=false; + } else { + throw Exception(PARSER_RUNTIME, &mode, "is invalid mode, must be either '"UNESCAPE_MODE_JS"' or '"UNESCAPE_MODE_URI"'"); + } + + const char* unescaped=unescape_chars(src.cstr(), src.length(), from_charset, mode_js); + if(*unescaped){ + const String* result=new String(Charset::transcode(unescaped, *from_charset, r.charsets.source()), String::L_TAINTED); + r.write_assign_lang(*result); + } +} + // constructor MString::MString(): Methoded("string") { @@ -835,8 +873,11 @@ MString::MString(): Methoded("string") { add_native_method("idna", Method::CT_ANY, _idna, 0, 1); // ^string.js-escape[] - add_native_method("js-escape", Method::CT_ANY, _escape, 0, 0); + add_native_method("js-escape", Method::CT_DYNAMIC, _js_escape, 0, 0); // ^string:js-unescape[escaped%uXXXXstring] - add_native_method("js-unescape", Method::CT_STATIC, _unescape, 1, 1); -} + add_native_method("js-unescape", Method::CT_STATIC, _js_unescape, 1, 1); + + // ^string:unescape[js|uri;escaped;$.charset[...]] + add_native_method("unescape", Method::CT_STATIC, _unescape, 2, 3); +}