--- parser3/src/classes/string.C 2010/04/05 00:57:01 1.192 +++ parser3/src/classes/string.C 2026/01/06 16:36:39 1.265 @@ -1,14 +1,13 @@ /** @file Parser: @b string parser class. - Copyright (c) 2001-2009 ArtLebedev Group (http://www.artlebedev.com) - Author: Alexandr Petrosian (http://paf.design.ru) + Copyright (c) 2001-2024 Art. Lebedev Studio (http://www.artlebedev.com) + Authors: Konstantin Morshnev , Alexandr Petrosian */ -static const char * const IDENT_STRING_C="$Date: 2010/04/05 00:57:01 $"; - #include "classes.h" #include "pa_vmethod_frame.h" +#include "pa_base64.h" #include "pa_request.h" #include "pa_vdouble.h" @@ -16,24 +15,36 @@ static const char * const IDENT_STRING_C #include "pa_vtable.h" #include "pa_vbool.h" #include "pa_string.h" +#include "pa_varray.h" #include "pa_sql_connection.h" #include "pa_dictionary.h" #include "pa_vmethod_frame.h" #include "pa_vregex.h" #include "pa_charsets.h" +volatile const char * IDENT_STRING_C="$Id: string.C,v 1.265 2026/01/06 16:36:39 moko Exp $"; + // class class MString: public Methoded { public: MString(); -public: // Methoded - bool used_directly() { return true; } }; // 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 + +class MVoid: public Methoded { +public: + MVoid(); +}; + +// void global variable should be after string global variable + +DECLARE_CLASS_VAR(void, new MVoid); // defines for statics @@ -44,6 +55,9 @@ DECLARE_CLASS_VAR(string, new MString, 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); @@ -52,75 +66,77 @@ 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) { const String& self_string=GET_SELF(r, VString).string(); - int converted; - try { - if(self_string.is_empty()) - throw Exception(PARSER_RUNTIME, - 0, - "parameter is empty string, error converting"); - converted=self_string.as_int(); - } catch(...) { // convert problem + + if(self_string.is_empty()) { if(params.count()>0) - converted=params.as_int(0, "default must be int", r); // (default) + r.write(*new VInt(params.as_wint(0, "default must be int", r))); // (default) else - rethrow; // we have a problem when no default + throw Exception(PARSER_RUNTIME, 0, "unable to convert empty string without default specified"); + } else { + try { + r.write(*new VInt(self_string.as_wint())); + } catch(...) { // convert problem + if(params.count()>0) + r.write(*new VInt(params.as_wint(0, "default must be int", r))); // (default) + else + rethrow; // we have a problem when no default + } } - r.write_no_lang(*new VInt(converted)); } static void _double(Request& r, MethodParams& params) { const String& self_string=GET_SELF(r, VString).string(); - double converted; - try { - if(self_string.is_empty()) - throw Exception(PARSER_RUNTIME, - 0, - "parameter is empty string, error converting"); - converted=self_string.as_double(); - } catch(...) { // convert problem + + 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 - rethrow; // we have a problem when no default + throw Exception(PARSER_RUNTIME, 0, "unable to convert empty string without default specified"); + } else { + try { + r.write(*new VDouble(self_string.as_double())); + } catch(...) { // convert problem + if(params.count()>0) + 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) { const String& self_string=GET_SELF(r, VString).string(); bool converted; - try { - if(self_string.is_empty()) - throw Exception(PARSER_RUNTIME, - 0, - "parameter is empty string, error converting"); - - try { - converted=self_string.as_bool(); - } catch(...) { - const String& lower_string=self_string.change_case(r.charsets.source(), String::CC_LOWER); - if(lower_string == "true"){ - converted=true; - } else if (lower_string == "false"){ - converted=false; - } else { - rethrow; - } - } - } catch(...) { // convert problem + const char *str=self_string.cstr(); + + if(self_string.is_empty()) { if(params.count()>0) converted=params.as_bool(0, "default must be bool", r); // (default) else - rethrow; // we have a problem when no default + throw Exception(PARSER_RUNTIME, 0, "unable to convert empty string without default specified"); + } else if( (str[0]=='T' || str[0]=='t') && (str[1]=='R' || str[1]=='r') && (str[2]=='U' || str[2]=='u') && + (str[3]=='E' || str[3]=='e') && str[4]==0 ) { // "true" + converted=true; + } else if( (str[0]=='F' || str[0]=='f') && (str[1]=='A' || str[1]=='a') && (str[2]=='L' || str[2]=='l') && + (str[3]=='S' || str[3]=='s') && (str[4]=='E' || str[4]=='e') && str[5]==0 ) { // "false" + converted=false; + } else { + try { + converted=self_string.as_bool(); + } catch(...) { // convert problem + if(params.count()>0) + converted=params.as_bool(0, "default must be bool", r); // (default) + else + rethrow; // we have a problem when no default + } } - r.write_no_lang(VBool::get(converted)); + r.write(VBool::get(converted)); } /*not static*/void _string_format(Request& r, MethodParams& params) { @@ -129,92 +145,97 @@ static void _bool(Request& r, MethodPara // for some time due to stupid {} in original design const String& fmt=r.process_to_string(fmt_maybe_code); - const char* buf=format(r.get_self().as_double(), fmt.trim().cstrm()); + const char* buf=format_double(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); - if(sn<0) - throw Exception(PARSER_RUNTIME, - 0, - "n(%d) must be >=0", sn); - size_t n=(size_t)sn; - + int sn=params.as_int(0, "n must be int", r); const String& string=GET_SELF(r, VString).string(); - r.write_assign_lang(string.mid(r.charsets.source(), 0, n)); + r.write(sn<0 ? string : string.mid(r.charsets.source(), 0, 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=sn; + const String& string=GET_SELF(r, VString).string(); + size_t length=string.length(r.charsets.source()); + r.write(n=0", sbegin); - size_t begin=(size_t)sbegin; + int begin=params.as_int(0, "p must be int", r); + if(begin<0) + throw Exception(PARSER_RUNTIME, 0, "p(%d) must be >=0", begin); size_t end; size_t length=0; + 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); - end=begin+(size_t)sn; + int n=params.as_int(1, "n must be int", r); + if(n<0) + throw Exception(PARSER_RUNTIME, 0, "n(%d) must be >=0", n); + end=begin+n; } 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) { Value& substr=params.as_no_junction(0, "substr must not be code"); const String& string=GET_SELF(r, VString).string(); - ssize_t offset=0; + int offset=0; 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(), offset))); +} + +struct Split_action_info { + const String& src; + ArrayString &result; + Split_action_info(const String& s, ArrayString& r) : src(s), result(r) {} +}; + +static void split_action(Table& , ArrayString* row, int prestart, int prefinish, int poststart, int postfinish, void *info) { + Split_action_info& ai=*static_cast(info); + if(row) { // begin&middle + // piece from last match['prestart'] to beginning of this match['prefinish'] + ai.result += &ai.src.mid(prestart, prefinish); + } else // end + if(poststart != postfinish) + ai.result += &ai.src.mid(poststart, postfinish); } -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"); +static void split_list(Value& delim_value, const String& string, ArrayString& result) { + if(VRegex *vregex=dynamic_cast(&delim_value)){ + vregex->study(); + + int matches_count=0; + Split_action_info ai(string, result); - size_t pos_after=0; - string.split(result, pos_after, delim_value.as_string()); + string.match(vregex, split_action, &ai, matches_count); + } else + string.split(result, 0, delim_value.as_string()); } #define SPLIT_LEFT 0x0001 #define SPLIT_RIGHT 0x0010 #define SPLIT_HORIZONTAL 0x0100 #define SPLIT_VERTICAL 0x1000 +#define SPLIT_ARRAY 0x10000 static int split_options(const String* options) { struct Split_option { @@ -223,10 +244,11 @@ static int split_options(const String* o int setBit; int checkBit; } split_option[]={ - {"l", "L", SPLIT_LEFT, SPLIT_RIGHT}, // 0xVHRL + {"l", "L", SPLIT_LEFT, SPLIT_RIGHT}, // 0xAVHRL {"r", "R", SPLIT_RIGHT, SPLIT_LEFT}, - {"h", "H", SPLIT_HORIZONTAL, SPLIT_VERTICAL}, - {"v", "V", SPLIT_VERTICAL, SPLIT_HORIZONTAL}, + {"h", "H", SPLIT_HORIZONTAL, SPLIT_VERTICAL | SPLIT_ARRAY}, + {"v", "V", SPLIT_VERTICAL, SPLIT_HORIZONTAL | SPLIT_ARRAY}, + {"a", "A", SPLIT_ARRAY, SPLIT_VERTICAL | SPLIT_HORIZONTAL}, {0, 0, 0, 0} }; @@ -236,9 +258,7 @@ 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; } } @@ -258,8 +278,7 @@ static Table& split_vertical(ArrayString table+=row; } } else { // left - Array_iterator i(pieces); - while(i.has_next()) { + for(ArrayString::Iterator i(pieces); i; ) { Table::element_type row(new ArrayString); *row+=i.next(); table+=row; @@ -276,7 +295,7 @@ static Table& split_horizontal(ArrayStri for(int i=pieces.count(); --i>=0; ) *row+=pieces[i]; } else { // left - for(Array_iterator i(pieces); i.has_next(); ) + for(ArrayString::Iterator i(pieces); i; ) *row+=i.next(); } table+=row; @@ -284,39 +303,54 @@ static Table& split_horizontal(ArrayStri return table; } -static void split_with_options(Request& r, MethodParams& params, - int bits) { +static VArray& split_array(ArrayString& pieces, bool right) { + VArray& result=*new VArray(pieces.count()); + ArrayValue &array=result.array(); + if(right) { // right + for(int i=pieces.count(); --i>=0; ) + array+=new VString(*pieces[i]); + } else { // left + for(ArrayString::Iterator i(pieces); i; ) + array+=new VString(*i.next()); + } + return result; +} + +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(); ArrayString pieces; - split_list(params, 0, string, pieces); + split_list(params.as_no_junction(0, "delimiter must not be code"), string, pieces); if(!bits) { const String* options=0; if(params_count>1) options=¶ms.as_string(1, OPTIONS_MUST_NOT_BE_CODE); - bits=split_options(options); } bool right=(bits & SPLIT_RIGHT) != 0; bool horizontal=(bits & SPLIT_HORIZONTAL) !=0; + bool array=(bits & SPLIT_ARRAY) !=0; const String* column_name=0; 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"); - } - 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); + if (horizontal && !column_name->is_empty()) + throw Exception(PARSER_RUNTIME, column_name, "column name cannot be specified when splitting into a table horizontal row"); + if (array && !column_name->is_empty()) + throw Exception(PARSER_RUNTIME, column_name, "column name cannot be specified when splitting into an array"); + } - r.write_no_lang(*new VTable(&table)); + if(array){ + r.write(split_array(pieces, right)); + } else { + 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(*new VTable(&table)); + } } static void _split(Request& r, MethodParams& params) { split_with_options(r, params, 0 /* maybe-determine from param #2 */); @@ -343,10 +377,7 @@ 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'] @@ -374,37 +405,28 @@ static void _match(Request& r, MethodPar Value& regexp=params.as_no_junction(0, "regexp must not be code"); Value* options=(params_count>1)?¶ms.as_no_junction(1, OPTIONS_MUST_NOT_BE_CODE):0; - VRegex* vregex; + VRegex* vregex=dynamic_cast(®exp); VRegexCleaner vrcleaner; - if(Value* value=regexp.as(VREGEX_TYPE)){ + if(vregex){ if(options && options->is_defined()) - throw Exception(PARSER_RUNTIME, - 0, - "you can not specify regex-object and options together" - ); - vregex=static_cast(value); + throw Exception(PARSER_RUNTIME, 0, "you cannot specify regex-object and options together"); } 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 @@ -420,11 +442,9 @@ 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; @@ -437,7 +457,7 @@ static void _match(Request& r, MethodPar }; if(is_junction){ - Temp_value_element temp( *replacement_code->get_junction()->method_frame, match_var_name, vtable); + Temp_value_element temp(r, *replacement_code->get_junction()->method_frame, match_var_name, vtable); src.match(vregex, replace_action, &info, matches_count); } else { src.match(vregex, replace_action, &info, matches_count); @@ -446,15 +466,14 @@ 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); @@ -465,24 +484,19 @@ static void _lower(Request& r, MethodPar #ifndef DOXYGEN class String_sql_event_handlers: public SQL_Driver_query_event_handlers { - const String& statement_string; const char* statement_cstr; bool got_column; public: bool got_cell; - String& result; + const String* result; public: - String_sql_event_handlers( - const String& astatement_string, const char* astatement_cstr): - statement_string(astatement_string), statement_cstr(astatement_cstr), + String_sql_event_handlers(): got_column(false), got_cell(false), - result(*new String) {} + result(&String::Empty) {} 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("result must contain exactly one column"); return true; } got_column=true; @@ -490,89 +504,66 @@ public: } bool before_rows(SQL_Error& /*error*/ ) { /* ignore */ return false; } bool add_row(SQL_Error& /*error*/) { /* ignore */ return false; } - bool add_row_cell(SQL_Error& error, const char* str, size_t length) { + 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("result must contain no more than one row"); return true; } - try { got_cell=true; - result.append_know_length(str, length, String::L_TAINTED); + 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); -const String* sql_result_string(Request& r, MethodParams& params, - HashStringValue*& options, Value*& default_code) { +const String* sql_result_string(Request& r, MethodParams& params, Value*& default_code) { Value& statement=params.as_junction(0, "statement must be code"); HashStringValue* bind=0; ulong limit=SQL_NO_LIMIT; ulong offset=0; default_code=0; - if(params.count()>1) { - Value& voptions=params.as_no_junction(1, "options must be hash, not code"); - if(voptions.is_defined() && !voptions.is_string()) - if((options=voptions.get_hash())) { - int valid_options=0; - if(Value* vbind=options->get(sql_bind_name)) { - valid_options++; - bind=vbind->get_hash(); - } - if(Value* vlimit=options->get(sql_limit_name)) { - valid_options++; - limit=(ulong)r.process_to_value(*vlimit).as_double(); - } - if(Value* voffset=options->get(sql_offset_name)) { - valid_options++; - offset=(ulong)r.process_to_value(*voffset).as_double(); - } - if((default_code=options->get(sql_default_name))) { - valid_options++; - } - if(valid_options!=options->count()) - throw Exception(PARSER_RUNTIME, - 0, - "called with invalid option"); - } else - throw Exception(PARSER_RUNTIME, - 0, - "options must be hash"); - } else - options=0; + if(params.count()>1) + if(HashStringValue* options=params.as_hash(1, "sql options")) { + int valid_options=0; + if(Value* vbind=options->get(sql_bind_name)) { + valid_options++; + bind=vbind->get_hash(); + } + if(Value* vlimit=options->get(sql_limit_name)) { + valid_options++; + limit=(ulong)r.process(*vlimit).as_double(); + } + if(Value* voffset=options->get(sql_offset_name)) { + valid_options++; + offset=(ulong)r.process(*voffset).as_double(); + } + if((default_code=options->get(sql_default_name))) { + valid_options++; + } + if(valid_options!=options->count()) + throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION); + } SQL_Driver::Placeholder* placeholders=0; uint placeholders_count=0; 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); + String_sql_event_handlers handlers; - r.connection()->query( - statement_cstr, - placeholders_count, placeholders, - offset, limit, - handlers, - statement_string); + r.connection()->query(statement_cstr, placeholders_count, placeholders, offset, limit, handlers, statement_string); if(bind) unmarshal_bind_updates(*bind, placeholders_count, placeholders); @@ -580,37 +571,39 @@ const String* sql_result_string(Request& if(!handlers.got_cell) return 0; // no lines, caller should return second param[default value] - return &handlers.result; + return handlers.result; } static void _sql(Request& r, MethodParams& params) { - HashStringValue* options; Value* default_code; - const String* string=sql_result_string(r, params, options, default_code); + const String* string=sql_result_string(r, params, default_code); if(!string) { 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) { const String& src=GET_SELF(r, VString).string(); - Table* table=params.as_no_junction(0, PARAM_MUST_NOT_BE_CODE).get_table(); - if(!table) - throw Exception(PARSER_RUNTIME, - 0, - "parameter must be table"); + if(params.count()==1) { + // ^string.replace[table] + if(Table* table=params.as_table(0, "param")){ + Dictionary dict(*table); + r.write(src.replace(dict)); + } else + r.write(src); + } 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(src.replace(dict)); + } - Dictionary dict(*table); - r.write_assign_lang(src.replace(dict)); } static void _save(Request& r, MethodParams& params) { @@ -618,11 +611,12 @@ static void _save(Request& r, MethodPara Charset* asked_charset=0; size_t file_name_index=0; - if(params.count()>1) + if(params.count()>1) { if(HashStringValue* options=params.as_no_junction(1, "second parameter should be string or hash").get_hash()){ + // ^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)){ @@ -630,10 +624,9 @@ static void _save(Request& r, MethodPara valid_options++; } if(valid_options != options->count()) - throw Exception(PARSER_RUNTIME, - 0, - INVALID_OPTION_PASSED); + throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION); } else { + // ^file.save[append;filespec] const String& mode=params.as_string(0, "mode must be string"); if(mode==MODE_APPEND){ do_append=true; @@ -643,20 +636,21 @@ static void _save(Request& r, MethodPara &mode, "unknown mode, must be 'append'"); } + } - const String& file_name=params.as_string(file_name_index, FILE_NAME_MUST_BE_STRING); + const String& file_name=params.as_file_name(file_name_index); const String& src=GET_SELF(r, VString).string(); - String::Body sbody=src.cstr_to_string_body_untaint(String::L_AS_IS, r.connection(false/*no error if none*/)); + String::Body sbody=src.cstr_to_string_body_untaint(String::L_AS_IS, r.connection(false), &r.charsets); // write - file_write(r.charsets, r.absolute(file_name), sbody.cstr(), sbody.length(), true, do_append, asked_charset); + file_write(r.charsets, r.full_disk_path(file_name), sbody.cstr(), sbody.length(), true, do_append, asked_charset); } 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) { @@ -667,17 +661,18 @@ static void _trim(Request& r, MethodPara const char* chars=0; if(params_count>0) { const String& skind=params.as_string(0, "'where' must be string"); - if(!skind.is_empty()) + if(!skind.is_empty()) { if(skind==TRIM_BOTH_OPTION) kind=String::TRIM_BOTH; else if(skind==TRIM_START_OPTION || skind=="start") 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) { const String& schars=params.as_string(1, "'chars' must be string"); @@ -686,54 +681,139 @@ 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())); +} + +Base64Options base64_encode_options(Request& r, HashStringValue* options) { + Base64Options result; + if(options) { + for(HashStringValue::Iterator i(*options); i; i.next()) { + String::Body key=i.key(); + Value* value=i.value(); + if(key == "pad") { + result.pad=r.process(*value).as_bool(); + } else if(key == "wrap") { + result.wrap=r.process(*value).as_bool(); + } else if(key == "url-safe") { + if(r.process(*value).as_bool()) + result.set_url_safe_abc(); + } else + throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION); + } + } + return result; } -static void _append(Request& r, MethodParams& params) { - // c=a+b - VString& va=GET_SELF(r, VString); - const String& a=va.string(); - const String& b=params.as_string(0, PARAMETER_MUST_BE_STRING); - String& c=*new String(a); - c.append(b, String::L_PASS_APPENDED); - va.set_string(c); +Base64Options base64_decode_options(Request& r, HashStringValue* options) { + Base64Options result; + if(options) { + for(HashStringValue::Iterator i(*options); i; i.next() ) { + String::Body key=i.key(); + Value* value=i.value(); + if(key == "pad") { + result.pad=r.process(*value).as_bool(); + } else if(key == "strict") { + result.strict=r.process(*value).as_bool(); + } else if(key == "url-safe") { + if(r.process(*value).as_bool()) + result.set_url_safe_abc(); + } else + throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION); + } + } + return result; } static void _base64(Request& r, MethodParams& params) { - if(params.count()) { - // decode: ^string:base64[encoded] - const char* cstr=params.as_string(0, PARAMETER_MUST_BE_STRING).cstr(); + if(&r.get_self() == string_class) { + // decode: ^string:base64[encoded[;options]] + const char* cstr=params.count() ? params.as_string(0, PARAMETER_MUST_BE_STRING).cstr() : ""; + Base64Options options = base64_decode_options(r, params.count() > 1 ? params.as_hash(1) : NULL); + char* decoded=0; - size_t length=0; - pa_base64_decode(cstr, strlen(cstr), decoded, length); + size_t length=pa_base64_decode(cstr, strlen(cstr), decoded, options); + 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."); + if(memchr(decoded, 0, length)) + throw Exception(PARSER_RUNTIME, 0, "Invalid \\x00 character found while decoding to string. Decode 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[] + // encode: ^str.base64[options] + VString& self=GET_SELF(r, VString); + const char* cstr=self.string().cstr(); + Base64Options options = base64_encode_options(r, params.count() ? params.as_hash(0) : NULL); + const char* encoded=pa_base64_encode(cstr, strlen(cstr), options); + r.write(*new String(encoded, String::L_TAINTED /*once ?param=base64(something) was needed*/ )); + } +} + +static void _idna(Request& r, MethodParams& params) { + if(&r.get_self() == string_class) { + // decode: ^string:idna[encoded] + 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(); - 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(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/* don't unescape '+' char */)) - r.write_assign_lang(*new String(result)); + if(const char* result=unescape_chars(src.cstr(), src.length(), &r.charsets.source(), true)) + r.write(*new String(result, String::L_TAINTED)); +} + +static void _unescape(Request& r, MethodParams& params){ + const String& mode=params.as_string(0, MODE_MUST_BE_STRING); + 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 an 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 @@ -771,12 +851,12 @@ 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} @@ -792,7 +872,7 @@ MString::MString(): Methoded("string") { add_native_method("sql", Method::CT_STATIC, _sql, 1, 2); // ^string.replace[table] - add_native_method("replace", Method::CT_DYNAMIC, _replace, 1, 1); + add_native_method("replace", Method::CT_DYNAMIC, _replace, 1, 2); // ^string.save[append][file] // ^string.save[file] @@ -805,16 +885,23 @@ MString::MString(): Methoded("string") { // ^string.trim[[start|both|end][;chars]] add_native_method("trim", Method::CT_DYNAMIC, _trim, 0, 2); - // ^string.append[string] - add_native_method("append", Method::CT_DYNAMIC, _append, 1, 1); - // ^string.base64[] << encode - // ^string:base64[encoded string] << decode - add_native_method("base64", Method::CT_ANY, _base64, 0, 1); + // ^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); + + // ^string.contains[key] for hash compatibility + add_native_method("contains", Method::CT_DYNAMIC, _contains, 1, 1); +}