--- parser3/src/classes/string.C 2015/05/16 22:24:38 1.217 +++ parser3/src/classes/string.C 2019/09/06 10:17:07 1.239 @@ -1,7 +1,7 @@ /** @file Parser: @b string parser class. - Copyright (c) 2001-2012 Art. Lebedev Studio (http://www.artlebedev.com) + Copyright (c) 2001-2017 Art. Lebedev Studio (http://www.artlebedev.com) Author: Alexandr Petrosian (http://paf.design.ru) */ @@ -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 1.239 2019/09/06 10:17:07 moko Exp $"; // class @@ -31,7 +31,7 @@ public: // global variable -DECLARE_CLASS_VAR(string, new MString, 0); +DECLARE_CLASS_VAR(string, new MString); // void class, inherited from string and thus should be inited afterwards @@ -42,7 +42,7 @@ public: // void global variable should be after string global variable -DECLARE_CLASS_VAR(void, new MVoid, 0); +DECLARE_CLASS_VAR(void, new MVoid); // defines for statics @@ -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); @@ -61,7 +64,7 @@ static const String match_var_name(MATCH static void _length(Request& r, MethodParams&) { double result=GET_SELF(r, VString).string().length(r.charsets.source()); - r.write_no_lang(*new VDouble(result)); + r.write(*new VDouble(result)); } static void _int(Request& r, MethodParams& params) { @@ -84,30 +87,27 @@ static void _int(Request& r, MethodParam } } - r.write_no_lang(*new VInt(converted)); + r.write(*new VInt(converted)); } static void _double(Request& r, MethodParams& params) { const String& self_string=GET_SELF(r, VString).string(); - double converted; if(self_string.is_empty()) { if(params.count()>0) - converted=params.as_double(0, "default must be double", r); // (default) + r.write(*new VDouble(params.as_double(0, "default must be double", r))); // (default) else throw Exception(PARSER_RUNTIME, 0, "unable to convert empty string without default specified"); } else { try { - converted=self_string.as_double(); + r.write(*new VDouble(self_string.as_double())); } catch(...) { // convert problem if(params.count()>0) - converted=params.as_double(0, "default must be double", r); // (default) + r.write(*new VDouble(params.as_double(0, "default must be double", r))); // (default) else rethrow; // we have a problem when no default } } - - r.write_no_lang(*new VDouble(converted)); } static void _bool(Request& r, MethodParams& params) { @@ -137,7 +137,7 @@ static void _bool(Request& r, MethodPara } } - r.write_no_lang(VBool::get(converted)); + r.write(VBool::get(converted)); } /*not static*/void _string_format(Request& r, MethodParams& params) { @@ -148,13 +148,13 @@ static void _bool(Request& r, MethodPara const char* buf=format(r.get_self().as_double(), fmt.trim().cstrm()); - r.write_no_lang(String(buf)); + r.write(String(buf)); } static void _left(Request& r, MethodParams& params) { ssize_t sn=params.as_int(0, "n must be int", r); const String& string=GET_SELF(r, VString).string(); - r.write_assign_lang(sn<0 ? string : string.mid(r.charsets.source(), 0, (size_t)sn)); + r.write(sn<0 ? string : string.mid(r.charsets.source(), 0, (size_t)sn)); } static void _right(Request& r, MethodParams& params) { @@ -163,7 +163,7 @@ static void _right(Request& r, MethodPar 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=0", sbegin); + throw Exception(PARSER_RUNTIME, 0, "p(%d) must be >=0", sbegin); size_t begin=(size_t)sbegin; size_t end; @@ -182,16 +180,14 @@ static void _mid(Request& r, MethodParam if(params.count()>1) { ssize_t sn=params.as_int(1, "n must be int", r); if(sn<0) - throw Exception(PARSER_RUNTIME, - 0, - "n(%d) must be >=0", sn); + throw Exception(PARSER_RUNTIME, 0, "n(%d) must be >=0", sn); end=begin+(size_t)sn; } else { length=string.length(r.charsets.source()); end=length; } - r.write_assign_lang(string.mid(r.charsets.source(), begin, end, length)); + r.write(string.mid(r.charsets.source(), begin, end, length)); } static void _pos(Request& r, MethodParams& params) { @@ -202,21 +198,15 @@ static void _pos(Request& r, MethodParam if(params.count()>1){ offset=params.as_int(1, "n must be int", r); if(offset<0) - throw Exception(PARSER_RUNTIME, - 0, - "n(%d) must be >=0", offset); + throw Exception(PARSER_RUNTIME, 0, "n(%d) must be >=0", offset); } - r.write_no_lang(*new VInt((int)string.pos(r.charsets.source(), substr.as_string(), (size_t)offset))); + r.write(*new VInt((int)string.pos(r.charsets.source(), substr.as_string(), (size_t)offset))); } -static void split_list(MethodParams& params, int paramIndex, - const String& string, - ArrayString& result) { +static void split_list(MethodParams& params, int paramIndex, const String& string, ArrayString& result) { Value& delim_value=params.as_no_junction(paramIndex, "delimiter must not be code"); - - size_t pos_after=0; - string.split(result, pos_after, delim_value.as_string()); + string.split(result, 0, delim_value.as_string()); } #define SPLIT_LEFT 0x0001 @@ -244,9 +234,7 @@ static int split_options(const String* o if(options->pos(o->keyL)!=STRING_NOT_FOUND || (o->keyU && options->pos(o->keyU)!=STRING_NOT_FOUND)) { if(result & o->checkBit) - throw Exception(PARSER_RUNTIME, - options, - "conflicting split options"); + throw Exception(PARSER_RUNTIME, options, "conflicting split options"); result |= o->setBit; } } @@ -292,8 +280,7 @@ static Table& split_horizontal(ArrayStri return table; } -static void split_with_options(Request& r, MethodParams& params, - int bits) { +static void split_with_options(Request& r, MethodParams& params, int bits) { const String& string=GET_SELF(r, VString).string(); size_t params_count=params.count(); @@ -304,7 +291,6 @@ static void split_with_options(Request& const String* options=0; if(params_count>1) options=¶ms.as_string(1, OPTIONS_MUST_NOT_BE_CODE); - bits=split_options(options); } @@ -315,16 +301,14 @@ static void split_with_options(Request& if(params_count>2){ column_name=¶ms.as_string(2, COLUMN_NAME_MUST_BE_STRING); if (horizontal && !column_name->is_empty()) - throw Exception(PARSER_RUNTIME, - column_name, - "column name can't be specified with horisontal split"); + throw Exception(PARSER_RUNTIME, column_name, "column name can't be specified with horisontal split"); } if(!column_name || column_name->is_empty()) column_name=new String("piece"); Table& table=horizontal?split_horizontal(pieces, right):split_vertical(pieces, right, column_name); - r.write_no_lang(*new VTable(&table)); + r.write(*new VTable(&table)); } static void _split(Request& r, MethodParams& params) { split_with_options(r, params, 0 /* maybe-determine from param #2 */); @@ -351,10 +335,7 @@ struct Replace_action_info { }; #endif /// @todo they can do $global[$result] there, getting pointer to later-invalid local var, kill this -static void replace_action(Table& table, ArrayString* row, - int prestart, int prefinish, - int poststart, int postfinish, - void *info) { +static void replace_action(Table& table, ArrayString* row, int prestart, int prefinish, int poststart, int postfinish, void *info) { Replace_action_info& ai=*static_cast(info); if(row) { // begin&middle // piece from last match['prestart'] to beginning of this match['prefinish'] @@ -387,32 +368,24 @@ static void _match(Request& r, MethodPar if(Value* value=regexp.as(VREGEX_TYPE)){ if(options && options->is_defined()) - throw Exception(PARSER_RUNTIME, - 0, - "you can not specify regex-object and options together" - ); + throw Exception(PARSER_RUNTIME, 0, "you can not specify regex-object and options together"); vregex=static_cast(value); } else { - vregex=new VRegex(r.charsets.source(), - ®exp.as_string(), - (options)?(&options->as_string()):0); + vregex=new VRegex(r.charsets.source(), ®exp.as_string(), (options) ? (&options->as_string()) : 0); vregex->study(); vrcleaner.vregex=vregex; } - Temp_lang temp_lang(r, String::L_PASS_APPENDED); const String& src=GET_SELF(r, VString).string(); int matches_count=0; if(params_count<3) { // search - Table* table=src.match(vregex, - search_action, 0, - matches_count); + Table* table=src.match(vregex, search_action, 0, matches_count); if(table){ - r.write_no_lang(*new VTable(table)); + r.write(*new VTable(table)); } else { - r.write_no_lang(*new VInt(matches_count)); + r.write(*new VInt(matches_count)); } } else { // replace @@ -428,11 +401,9 @@ static void _match(Request& r, MethodPar if(replacement->is_defined()) replacement_code=replacement; } else if(!replacement->is_void()) - throw Exception(PARSER_RUNTIME, - 0, - "replacement option should be junction or string"); + throw Exception(PARSER_RUNTIME, 0, "replacement option should be junction or string"); - Value* default_code=(params_count==4)?¶ms.as_junction(3, "default value must be code"):0; + Value* default_code=(params_count==4) ? ¶ms.as_junction(3, "default value must be code") : 0; String result; VTable* vtable=new VTable; @@ -454,15 +425,14 @@ static void _match(Request& r, MethodPar if(!matches_count && default_code) r.process_write(*default_code); else - r.write_assign_lang(result); + r.write(result); } } -static void change_case(Request& r, MethodParams&, - String::Change_case_kind kind) { +static void change_case(Request& r, MethodParams&, String::Change_case_kind kind) { const String& src=GET_SELF(r, VString).string(); - r.write_assign_lang(src.change_case(r.charsets.source(), kind)); + r.write(src.change_case(r.charsets.source(), kind)); } static void _upper(Request& r, MethodParams& params) { change_case(r, params, String::CC_UPPER); @@ -488,9 +458,7 @@ public: bool add_column(SQL_Error& error, const char* /*str*/, size_t /*length*/) { if(got_column) { - error=SQL_Error(PARSER_RUNTIME, - //statement_string, - "result must contain exactly one column"); + error=SQL_Error(PARSER_RUNTIME, /*statement_string,*/ "result must contain exactly one column"); return true; } got_column=true; @@ -500,9 +468,7 @@ public: bool add_row(SQL_Error& /*error*/) { /* ignore */ return false; } bool add_row_cell(SQL_Error& error, const char* str, size_t) { if(got_cell) { - error=SQL_Error(PARSER_RUNTIME, - //statement_string, - "result must not contain more then one row"); + error=SQL_Error(PARSER_RUNTIME, /*statement_string,*/ "result must not contain more then one row"); return true; } @@ -511,17 +477,13 @@ public: result=new String(str, String::L_TAINTED /* no length as 0x00 can be inside */ ); return false; } catch(...) { - error=SQL_Error("exception occured in String_sql_event_handlers::add_row_cell"); + error=SQL_Error("exception occurred in String_sql_event_handlers::add_row_cell"); return true; } } }; #endif -extern String sql_bind_name; -extern String sql_limit_name; -extern String sql_offset_name; -extern String sql_default_name; -extern String sql_distinct_name; + extern int marshal_binds(HashStringValue& hash, SQL_Driver::Placeholder*& placeholders); extern void unmarshal_bind_updates(HashStringValue& hash, int placeholder_count, SQL_Driver::Placeholder* placeholders); @@ -541,11 +503,11 @@ const String* sql_result_string(Request& } if(Value* vlimit=options->get(sql_limit_name)) { valid_options++; - limit=(ulong)r.process_to_value(*vlimit).as_double(); + limit=(ulong)r.process(*vlimit).as_double(); } if(Value* voffset=options->get(sql_offset_name)) { valid_options++; - offset=(ulong)r.process_to_value(*voffset).as_double(); + offset=(ulong)r.process(*voffset).as_double(); } if((default_code=options->get(sql_default_name))) { valid_options++; @@ -559,9 +521,8 @@ const String* sql_result_string(Request& if(bind) placeholders_count=marshal_binds(*bind, placeholders); - Temp_lang temp_lang(r, String::L_SQL); const String& statement_string=r.process_to_string(statement); - const char* statement_cstr=statement_string.untaint_cstr(r.flang, r.connection()); + const char* statement_cstr=statement_string.untaint_cstr(String::L_SQL, r.connection()); String_sql_event_handlers handlers(statement_string, statement_cstr); @@ -589,12 +550,10 @@ static void _sql(Request& r, MethodParam if(default_code) { string=&r.process_to_string(*default_code); } else - throw Exception(PARSER_RUNTIME, - 0, - "produced no result, but no default option specified"); + throw Exception(PARSER_RUNTIME, 0, "produced no result, but no default option specified"); } - r.write_assign_lang(*string); + r.write(*string); } static void _replace(Request& r, MethodParams& params) { @@ -604,14 +563,11 @@ static void _replace(Request& r, MethodP // ^string.replace[table] Table* table=params.as_table(0, "param"); Dictionary dict(*table); - r.write_assign_lang(src.replace(dict)); + r.write(src.replace(dict)); } else { // ^string.replace[from-string;to-string] - Dictionary dict( - params.as_string(0, "from must be string"), - params.as_string(1, "to must be string") - ); - r.write_assign_lang(src.replace(dict)); + Dictionary dict(params.as_string(0, "from must be string"), params.as_string(1, "to must be string")); + r.write(src.replace(dict)); } } @@ -626,7 +582,7 @@ static void _save(Request& r, MethodPara // ^file.save[filespec;$.charset[] $.append(true)] int valid_options=0; if(Value* vcharset_name=options->get(PA_CHARSET_NAME)){ - asked_charset=&::charsets.get(vcharset_name->as_string().change_case(r.charsets.source(), String::CC_UPPER)); + asked_charset=&pa_charsets.get(vcharset_name->as_string()); valid_options++; } if(Value* vappend=options->get(MODE_APPEND)){ @@ -660,7 +616,7 @@ static void _save(Request& r, MethodPara static void _normalize(Request& r, MethodParams&) { const String& src=GET_SELF(r, VString).string(); - r.write_assign_lang(src); + r.write(src); } static void _trim(Request& r, MethodParams& params) { @@ -678,10 +634,10 @@ static void _trim(Request& r, MethodPara kind=String::TRIM_START; else if(skind==TRIM_END_OPTION || skind=="end") kind=String::TRIM_END; + else if(params_count==1) + chars=skind.cstr(); else - throw Exception(PARSER_RUNTIME, - &skind, - "'kind' must be one of " TRIM_START_OPTION ", " TRIM_BOTH_OPTION ", " TRIM_END_OPTION); + throw Exception(PARSER_RUNTIME, &skind, "'kind' must be one of " TRIM_START_OPTION ", " TRIM_BOTH_OPTION ", " TRIM_END_OPTION); } if(params_count>1) { @@ -691,13 +647,13 @@ static void _trim(Request& r, MethodPara } } - r.write_assign_lang(src.trim(kind, chars, &r.charsets.source())); + r.write(src.trim(kind, chars, &r.charsets.source())); } static void _base64(Request& r, MethodParams& params) { - if(params.count()) { + if(&r.get_self() == string_class) { // decode: ^string:base64[encoded[;$.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; @@ -706,7 +662,7 @@ static void _base64(Request& r, MethodPa if(HashStringValue* options=params.as_hash(1)) { int valid_options=0; if(Value* vstrict=options->get(BASE64_STRICT_OPTION_NAME)) { - strict=r.process_to_value(*vstrict).as_bool(); + strict=r.process(*vstrict).as_bool(); valid_options++; } if(valid_options!=options->count()) @@ -716,45 +672,83 @@ static void _base64(Request& r, MethodPa pa_base64_decode(cstr, strlen(cstr), decoded, length, strict); if(decoded && length){ if(memchr((const char*)decoded, 0, length)) - throw Exception(PARSER_RUNTIME, - 0, - "Invalid \\x00 character found while decode to string. Decode it to file instead."); + throw Exception(PARSER_RUNTIME, 0, "Invalid \\x00 character found while decode to string. Decode it to file instead."); fix_line_breaks(decoded, length); if(length) - r.write_assign_lang(*new String(decoded, String::L_TAINTED)); + r.write(*new String(decoded, String::L_TAINTED)); } } else { // encode: ^str.base64[] VString& self=GET_SELF(r, VString); const char* cstr=self.string().cstr(); const char* encoded=pa_base64_encode(cstr, strlen(cstr)); - r.write_assign_lang(*new String(encoded, String::L_TAINTED/*once ?param=base64(something) was needed*/)); + r.write(*new String(encoded, String::L_TAINTED/*once ?param=base64(something) was needed*/)); } } 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(); - r.write_assign_lang(*new String(pa_idna_decode(cstr, r.charsets.source()), String::L_TAINTED)); + const char* cstr=params.count() ? params.as_string(0, PARAMETER_MUST_BE_STRING).cstr() : ""; + r.write(*new String(pa_idna_decode(cstr, r.charsets.source()), String::L_TAINTED)); } else { // encode: ^str.idna[] VString& self=GET_SELF(r, VString); const char* cstr=self.string().cstr(); - r.write_assign_lang(*new String(pa_idna_encode(cstr, r.charsets.source()), String::L_TAINTED)); + r.write(*new String(pa_idna_encode(cstr, r.charsets.source()), String::L_TAINTED)); } } -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())); + r.write(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)); + r.write(*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=&pa_charsets.get(vcharset_name->as_string()); + 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(*result); + } +} + +static void _contains(Request& r, MethodParams& params) { + // empty or whitespace string is hash compatible + GET_SELF(r, VString).get_element(params.as_string(0, "key must be string")); + // ignoring result as it allways null + r.write(VBool::get(false)); } // constructor @@ -792,12 +786,12 @@ MString::MString(): Methoded("string") { // ^string.split[delim][options] // ^string.split[delim][options][column name] add_native_method("split", Method::CT_DYNAMIC, _split, 1, 3); - // old names for backward compatibility - // ^string.lsplit[delim] - add_native_method("lsplit", Method::CT_DYNAMIC, _lsplit, 1, 1); - // ^string.rsplit[delim] - add_native_method("rsplit", Method::CT_DYNAMIC, _rsplit, 1, 1); - + // old names for backward compatibility + // ^string.lsplit[delim] + add_native_method("lsplit", Method::CT_DYNAMIC, _lsplit, 1, 1); + // ^string.rsplit[delim] + add_native_method("rsplit", Method::CT_DYNAMIC, _rsplit, 1, 1); + // ^string.match[regexp][options] // ^string.match[regexp][options]{replacement-code} // ^string.match[regexp][options]{replacement-code}{code-if-nothing-is-found} @@ -835,8 +829,14 @@ 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); + + // ^string.contains[key] for hash compatibility + add_native_method("contains", Method::CT_DYNAMIC, _contains, 1, 1); +}