--- parser3/src/classes/string.C 2015/04/08 18:08:52 1.214 +++ parser3/src/classes/string.C 2015/10/07 22:25:40 1.220 @@ -20,7 +20,7 @@ #include "pa_vregex.h" #include "pa_charsets.h" -volatile const char * IDENT_STRING_C="$Id: string.C,v 1.214 2015/04/08 18:08:52 moko Exp $"; +volatile const char * IDENT_STRING_C="$Id: string.C,v 1.220 2015/10/07 22:25:40 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); @@ -153,27 +156,18 @@ static void _bool(Request& r, MethodPara static void _left(Request& r, MethodParams& params) { ssize_t sn=params.as_int(0, "n must be int", r); - if(sn<0) - throw Exception(PARSER_RUNTIME, - 0, - "n(%d) must be >=0", sn); - size_t n=(size_t)sn; - const String& string=GET_SELF(r, VString).string(); - r.write_assign_lang(string.mid(r.charsets.source(), 0, n)); + r.write_assign_lang(sn<0 ? string : string.mid(r.charsets.source(), 0, (size_t)sn)); } static void _right(Request& r, MethodParams& params) { - ssize_t sn=(size_t)params.as_int(0, "n must be int", r); - if(sn<0) - throw Exception(PARSER_RUNTIME, - 0, - "n(%d) must be >=0", sn); - size_t n=(size_t)sn; - - const String& string=GET_SELF(r, VString).string(); - size_t length=string.length(r.charsets.source()); - r.write_assign_lang(n0){ + size_t n=(size_t)sn; + const String& string=GET_SELF(r, VString).string(); + size_t length=string.length(r.charsets.source()); + r.write_assign_lang(n 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(String::Body(unescaped), *from_charset, r.charsets.source()), String::L_TAINTED); + r.write_assign_lang(*result); + } +} + // constructor MString::MString(): Methoded("string") { @@ -823,12 +865,19 @@ MString::MString(): Methoded("string") { add_native_method("trim", Method::CT_DYNAMIC, _trim, 0, 2); // ^string.base64[] << encode - // ^string:base64[encoded string] << decode + // ^string:base64[encoded string] << decode add_native_method("base64", Method::CT_ANY, _base64, 0, 2); + // ^string.idna[] << encode + // ^string:idna[encoded string] << decode + 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); +}