--- parser3/src/classes/string.C 2001/10/09 07:06:00 1.78 +++ parser3/src/classes/string.C 2018/01/19 16:52:50 1.238 @@ -1,13 +1,13 @@ /** @file Parser: @b string parser class. - Copyright (c) 2001 ArtLebedev Group (http://www.artlebedev.com) - Author: Alexander Petrosyan (http://design.ru/paf) - - $Id: string.C,v 1.78 2001/10/09 07:06:00 parser Exp $ + Copyright (c) 2001-2017 Art. Lebedev Studio (http://www.artlebedev.com) + Author: Alexandr Petrosian (http://paf.design.ru) */ #include "classes.h" +#include "pa_vmethod_frame.h" + #include "pa_request.h" #include "pa_vdouble.h" #include "pa_vint.h" @@ -16,434 +16,748 @@ #include "pa_string.h" #include "pa_sql_connection.h" #include "pa_dictionary.h" +#include "pa_vmethod_frame.h" +#include "pa_vregex.h" +#include "pa_charsets.h" -// defines - -#define STRING_CLASS_NAME "string" +volatile const char * IDENT_STRING_C="$Id: string.C,v 1.238 2018/01/19 16:52:50 moko Exp $"; // class -class MString : public Methoded { +class MString: public Methoded { public: - MString(Pool& pool); -public: // Methoded - bool used_directly() { return true; } + MString(); }; +// global variable + +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 + +#define MATCH_VAR_NAME "match" +#define TRIM_START_OPTION "left" +#define TRIM_END_OPTION "right" +#define TRIM_BOTH_OPTION "both" + +#define MODE_APPEND "append" + +#define UNESCAPE_MODE_JS "js" +#define UNESCAPE_MODE_URI "uri" + +// statics + +static const String match_var_name(MATCH_VAR_NAME); + // methods -static void _length(Request& r, const String& method_name, MethodParams *) { - Pool& pool=r.pool(); - Value& result=*new(pool) VDouble(pool, r.self->get_string()->size()); - result.set_name(method_name); - r.write_no_lang(result); +static void _length(Request& r, MethodParams&) { + double result=GET_SELF(r, VString).string().length(r.charsets.source()); + r.write(*new VDouble(result)); } -static void _int(Request& r, const String& method_name, MethodParams *params) { - Pool& pool=r.pool(); - bool convert_problem=false; Exception rethrow_me; +static void _int(Request& r, MethodParams& params) { + const String& self_string=GET_SELF(r, VString).string(); int converted; - PTRY { - converted=r.self->as_int(); + + if(self_string.is_empty()) { + if(params.count()>0) + converted=params.as_int(0, "default must be int", r); // (default) + else + throw Exception(PARSER_RUNTIME, 0, "unable to convert empty string without default specified"); + } else { + try { + converted=self_string.as_int(); + } catch(...) { // convert problem + if(params.count()>0) + converted=params.as_int(0, "default must be int", r); // (default) + else + rethrow; // we have a problem when no default + } } - PCATCH(e) { // convert problem - if(convert_problem=params->size()==0) { // we have a problem when do not have default - rethrow_me=e; - converted=0; - } else - converted=params->as_int(0, "default must be int", r); // (default) + + r.write(*new VInt(converted)); +} + +static void _double(Request& r, MethodParams& params) { + const String& self_string=GET_SELF(r, VString).string(); + + if(self_string.is_empty()) { + if(params.count()>0) + 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 { + 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 + } } - PEND_CATCH - if(convert_problem) - PTHROW(rethrow_me.type(), rethrow_me.code(), - rethrow_me.problem_source(), - rethrow_me.comment()); - - Value& result=*new(pool) VInt(pool, converted); - result.set_name(method_name); - r.write_no_lang(result); -} - -static void _double(Request& r, const String& method_name, MethodParams *params) { - Pool& pool=r.pool(); - bool convert_problem=false; Exception rethrow_me; - double converted; - PTRY { - converted=r.self->as_double(); - } - PCATCH(e) { // convert problem - if(convert_problem=params->size()==0) { // we have a problem when do not have default - rethrow_me=e; - converted=0; - } else - converted=params->as_double(0, "default must be double", r); // (default) +} + +static void _bool(Request& r, MethodParams& params) { + const String& self_string=GET_SELF(r, VString).string(); + bool converted; + 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 + 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 + } } - PEND_CATCH - if(convert_problem) - PTHROW(rethrow_me.type(), rethrow_me.code(), - rethrow_me.problem_source(), - rethrow_me.comment()); - Value& result=*new(pool) VDouble(pool, converted); - result.set_name(method_name); - r.write_no_lang(result); + r.write(VBool::get(converted)); } -/*not static*/void _string_format(Request& r, const String& method_name, MethodParams *params) { - Pool& pool=r.pool(); +/*not static*/void _string_format(Request& r, MethodParams& params) { - Value& fmt=params->as_junction(0, "fmt must be code"); + Value& fmt_maybe_code=params[0]; + // for some time due to stupid {} in original design + const String& fmt=r.process_to_string(fmt_maybe_code); - Temp_lang temp_lang(r, String::UL_PASS_APPENDED); - char *buf=format(pool, r.self->as_double(), r.process(fmt).as_string().cstr()); + const char* buf=format(r.get_self().as_double(), fmt.trim().cstrm()); - String result(pool); - result.APPEND_CLEAN(buf, 0, - method_name.origin().file, - method_name.origin().line); - r.write_no_lang(result); + r.write(String(buf)); } -static void _left(Request& r, const String&, MethodParams *params) { - Pool& pool=r.pool(); +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(sn<0 ? string : string.mid(r.charsets.source(), 0, (size_t)sn)); +} - size_t n=(size_t)params->as_int(0, "n must be int", r); - - const String& string=*static_cast(r.self)->get_string(); - r.write_assign_lang(*new(pool) VString(string.mid(0, n))); +static void _right(Request& r, MethodParams& params) { + ssize_t sn=params.as_int(0, "n must be int", r); + if(sn>0){ + size_t n=(size_t)sn; + const String& string=GET_SELF(r, VString).string(); + size_t length=string.length(r.charsets.source()); + r.write(nas_int(0, "n must be int", r); - - const String& string=*static_cast(r.self)->get_string(); - r.write_assign_lang(*new(pool) VString(string.mid(string.size()-n, string.size()))); -} + ssize_t sbegin=params.as_int(0, "p must be int", r); + if(sbegin<0) + throw Exception(PARSER_RUNTIME, 0, "p(%d) must be >=0", sbegin); + size_t begin=(size_t)sbegin; -static void _mid(Request& r, const String&, MethodParams *params) { - Pool& pool=r.pool(); + 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; + } else { + length=string.length(r.charsets.source()); + end=length; + } - size_t p=(size_t)params->as_int(0, "p must be int", r); - size_t n=(size_t)params->as_int(1, "n must be int", r); - - const String& string=*static_cast(r.self)->get_string(); - r.write_assign_lang(*new(pool) VString(string.mid(p, p+n))); + r.write(string.mid(r.charsets.source(), begin, end, length)); } -static void _pos(Request& r, const String& method_name, MethodParams *params) { - Pool& pool=r.pool(); - - Value& substr=params->as_no_junction(0, "substr must not be code"); +static void _pos(Request& r, MethodParams& params) { + Value& substr=params.as_no_junction(0, "substr must not be code"); - const String& string=*static_cast(r.self)->get_string(); - r.write_assign_lang(*new(pool) VInt(pool, string.pos(substr.as_string()))); -} + const String& string=GET_SELF(r, VString).string(); + ssize_t 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); + } -static void split_list(Request& r, const String& method_name, MethodParams *params, - const String& string, - Array& result) { - Value& delim_value=params->as_no_junction(0, "delimiter must not be code"); + 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) { + Value& delim_value=params.as_no_junction(paramIndex, "delimiter must not be code"); string.split(result, 0, delim_value.as_string()); } -static void _lsplit(Request& r, const String& method_name, MethodParams *params) { - Pool& pool=r.pool(); - const String& string=*static_cast(r.self)->get_string(); - - Array pieces(pool); - split_list(r, method_name, params, string, pieces); +#define SPLIT_LEFT 0x0001 +#define SPLIT_RIGHT 0x0010 +#define SPLIT_HORIZONTAL 0x0100 +#define SPLIT_VERTICAL 0x1000 + +static int split_options(const String* options) { + struct Split_option { + const char* keyL; + const char* keyU; + int setBit; + int checkBit; + } split_option[]={ + {"l", "L", SPLIT_LEFT, SPLIT_RIGHT}, // 0xVHRL + {"r", "R", SPLIT_RIGHT, SPLIT_LEFT}, + {"h", "H", SPLIT_HORIZONTAL, SPLIT_VERTICAL}, + {"v", "V", SPLIT_VERTICAL, SPLIT_HORIZONTAL}, + {0, 0, 0, 0} + }; + + int result=0; + if(options) { + for(Split_option *o=split_option; o->keyL; 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"); + result |= o->setBit; + } + } + + return result; +} + +static Table& split_vertical(ArrayString& pieces, bool right, const String* column_name) { + Table::columns_type columns(new ArrayString); + *columns+=column_name; + + Table& table=*new Table(columns, pieces.count()); + if(right) { // right + for(int i=pieces.count(); --i>=0; ) { + Table::element_type row(new ArrayString); + *row+=pieces[i]; + table+=row; + } + } else { // left + Array_iterator i(pieces); + while(i.has_next()) { + Table::element_type row(new ArrayString); + *row+=i.next(); + table+=row; + } + } - Array& columns=*new(pool) Array(pool); - columns+=new(pool) String(pool, "piece"); + return table; +} - Table& table=*new(pool) Table(pool, &string, - &columns, pieces.size()); - Array_iter i(pieces); - while(i.has_next()) { - Array& row=*new(pool) Array(pool); - row+=i.next(); - table+=&row; +static Table& split_horizontal(ArrayString& pieces, bool right) { + Table& table=*new Table(Table::columns_type(0) /* nameless */); + Table::element_type row(new ArrayString(pieces.count())); + if(right) { // right + for(int i=pieces.count(); --i>=0; ) + *row+=pieces[i]; + } else { // left + for(Array_iterator i(pieces); i.has_next(); ) + *row+=i.next(); } - r.write_no_lang(*new(pool) VTable(pool, &table)); -} + table+=row; -static void _rsplit(Request& r, const String& method_name, MethodParams *params) { - Pool& pool=r.pool(); - const String& string=*static_cast(r.self)->get_string(); + return table; +} - Array pieces(pool); - split_list(r, method_name, params, string, pieces); +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(); - Array& columns=*new(pool) Array(pool); - columns+=new(pool) String(pool, "piece"); + ArrayString pieces; + split_list(params, 0, string, pieces); - Table& table=*new(pool) Table(pool, &string, - &columns, pieces.size()); - for(int i=pieces.size(); --i>=0; ) { - Array& row=*new(pool) Array(pool); - row+=pieces.get(i); - table+=&row; + if(!bits) { + const String* options=0; + if(params_count>1) + options=¶ms.as_string(1, OPTIONS_MUST_NOT_BE_CODE); + bits=split_options(options); } - r.write_no_lang(*new(pool) VTable(pool, &table)); + bool right=(bits & SPLIT_RIGHT) != 0; + bool horizontal=(bits & SPLIT_HORIZONTAL) !=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); + + r.write(*new VTable(&table)); +} +static void _split(Request& r, MethodParams& params) { + split_with_options(r, params, 0 /* maybe-determine from param #2 */); +} +static void _lsplit(Request& r, MethodParams& params) { + split_with_options(r, params, SPLIT_LEFT); +} +static void _rsplit(Request& r, MethodParams& params) { + split_with_options(r, params, SPLIT_RIGHT); } -static void search_action(Table& table, Array *row, int, int, void *) { +static void search_action(Table& table, Table::element_type row, int, int, int, int, void *) { if(row) table+=row; } #ifndef DOXYGEN struct Replace_action_info { - Request *request; const String *origin; - const String *src; String *dest; - Value *replacement_code; - const String *post_match; + Request* request; + const String* src; + String* dest; + VTable* vtable; + Value* replacement_code; }; #endif -static void replace_action(Table& table, Array *row, int start, int finish, - void *info) { +/// @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) { Replace_action_info& ai=*static_cast(info); if(row) { // begin&middle - // piece from last match['start'] to beginning of this match['finish'] - if(start!=finish) - *ai.dest << ai.src->mid(start, finish);//ai.dest->APPEND_CONST("-"); + // piece from last match['prestart'] to beginning of this match['prefinish'] + if(prestart!=prefinish) + *ai.dest << ai.src->mid(prestart, prefinish);//ai.dest->APPEND_CONST("-"); // store found parts in one-record VTable - if(table.size()) // middle + if(table.count()) // middle table.put(0, row); else // begin table+=row; - { // execute 'replacement_code' in 'table' context - VTable& vtable=*new(table.pool()) VTable(table.pool(), &table); - vtable.set_name(*ai.origin); - Junction *junction=ai.replacement_code->get_junction(); - Value *saved_match_var_value=junction->root->get_element(*match_var_name); - junction->root->put_element(*match_var_name, &vtable); - Value& replaced=ai.request->process(*ai.replacement_code, ai.origin, false); - junction->root->put_element(*match_var_name, saved_match_var_value); - - /* - ai.dest->APPEND_CONST("("); - *ai.dest << *(String *)row->get(1/*match* /); - ai.dest->APPEND_CONST(")"); - */ - *ai.dest << replaced.as_string(); + { // execute 'replacement_code' in 'table' context + if(ai.replacement_code){ + ai.vtable->set_table(table); + *ai.dest << ai.request->process_to_string(*ai.replacement_code); + } } - ai.post_match=(String *)row->get(2/*post_match*/); } else // end - *ai.dest << *ai.post_match; + *ai.dest << ai.src->mid(poststart, postfinish); } -/// @todo use pcre:study somehow -static void _match(Request& r, const String& method_name, MethodParams *params) { - Pool& pool=r.pool(); - const String& src=*static_cast(r.self)->get_string(); - - Value& regexp=params->as_no_junction(0, "regexp must not be code"); - - const String *options= - params->size()>1? - ¶ms->as_no_junction(1, "options must not be code").as_string():0; - - Value *result; - Temp_lang temp_lang(r, String::UL_PASS_APPENDED); - Table *table; - if(params->size()<3) { // search - bool was_global; - bool matched=src.match(r.pcre_tables(), - &method_name, - regexp.as_string(), options, - &table, - search_action, 0, - &was_global); - // matched - // not (just matched[3=pre/match/post], no substrings) or Global search - if(table->columns()->size()>3 || was_global) - result=new(pool) VTable(pool, table); // table of pre/match/post+substrings - else - result=new(pool) VBool(pool, matched); +static void _match(Request& r, MethodParams& params) { + size_t params_count=params.count(); + + 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; + VRegexCleaner vrcleaner; + + 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"); + vregex=static_cast(value); + } else { + vregex=new VRegex(r.charsets.source(), ®exp.as_string(), (options) ? (&options->as_string()) : 0); + vregex->study(); + vrcleaner.vregex=vregex; + } + + 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); + + if(table){ + r.write(*new VTable(table)); + } else { + r.write(*new VInt(matches_count)); + } + } else { // replace - Value& replacement_code=params->as_junction(2, "replacement code must be code"); - String& dest=*new(pool) String(pool); - Replace_action_info replace_action_info={ - &r, &method_name, - &src, &dest, - &replacement_code, - &src + Value* replacement_code=0; + bool is_junction=false; + + Value* replacement=¶ms[2]; + if(replacement->get_junction()){ + replacement_code=replacement; + is_junction=true; + } else if(replacement->is_string()){ + if(replacement->is_defined()) + replacement_code=replacement; + } else if(!replacement->is_void()) + 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; + + String result; + VTable* vtable=new VTable; + Replace_action_info info={ + &r, + &src, + &result, + vtable, + replacement_code }; - src.match(r.pcre_tables(), - &method_name, - r.process(regexp).as_string(), options, - &table, - replace_action, &replace_action_info); - result=new(pool) VString(dest); - } - result->set_name(method_name); - r.write_assign_lang(*result); -} - -static void change_case(Request& r, const String& method_name, MethodParams *params, - String::Change_case_kind kind) { - Pool& pool=r.pool(); - const String& src=*static_cast(r.self)->get_string(); - r.write_assign_lang(*new(pool) VString(src.change_case(pool, r.pcre_tables(), - kind))); + if(is_junction){ + 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); + } + + if(!matches_count && default_code) + r.process_write(*default_code); + else + r.write(result); + } } -static void _upper(Request& r, const String& method_name, MethodParams *params) { - change_case(r, method_name, params, String::CC_UPPER); + +static void change_case(Request& r, MethodParams&, String::Change_case_kind kind) { + const String& src=GET_SELF(r, VString).string(); + + r.write(src.change_case(r.charsets.source(), kind)); +} +static void _upper(Request& r, MethodParams& params) { + change_case(r, params, String::CC_UPPER); } -static void _lower(Request& r, const String& method_name, MethodParams *params) { - change_case(r, method_name, params, String::CC_LOWER); +static void _lower(Request& r, MethodParams& params) { + change_case(r, params, String::CC_LOWER); } #ifndef DOXYGEN -class String_sql_event_handlers : public SQL_Driver_query_event_handlers { +class String_sql_event_handlers: public SQL_Driver_query_event_handlers { + const String& statement_string; const char* statement_cstr; + bool got_column; public: - String_sql_event_handlers(Pool& apool, - const String& astatement_string, const char *astatement_cstr) : - pool(apool), - statement_string(astatement_string), - statement_cstr(astatement_cstr), - got_column(false), got_cell(false) { - result=new(pool) String(pool); - } - - void add_column(void *ptr, size_t size) { - if(got_column) - PTHROW(0, 0, - &statement_string, - "result must contain exactly one column"); + bool got_cell; + const String* result; +public: + String_sql_event_handlers( + const String& astatement_string, const char* astatement_cstr): + statement_string(astatement_string), statement_cstr(astatement_cstr), + got_column(false), + got_cell(false), + 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"); + return true; + } got_column=true; + return false; } - void before_rows() { /* ignore */ } - void add_row() { /* ignore */ } - void add_row_cell(void *ptr, size_t size) { - if(got_cell) - PTHROW(0, 0, - &statement_string, - "result must not contain more then one row"); - got_cell=true; - - result->APPEND_TAINTED((const char *)ptr, size, statement_cstr, 0); - } + 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) { + if(got_cell) { + error=SQL_Error(PARSER_RUNTIME, /*statement_string,*/ "result must not contain more then one row"); + return true; + } -private: - Pool& pool; - const String& statement_string; const char *statement_cstr; - bool got_column; -public: - bool got_cell; - String *result; + try { + got_cell=true; + result=new String(str, String::L_TAINTED /* no length as 0x00 can be inside */ ); + return false; + } catch(...) { + error=SQL_Error("exception occurred in String_sql_event_handlers::add_row_cell"); + return true; + } + } }; #endif -const String* sql_result_string(Request& r, const String& method_name, MethodParams *params, - Hash *&options) { - Pool& pool=r.pool(); - - if(!r.connection) - PTHROW(0, 0, - &method_name, - "without connect"); +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); - Value& statement=params->as_junction(0, "statement must be code"); +const String* sql_result_string(Request& r, MethodParams& params, Value*& default_code) { + Value& statement=params.as_junction(0, "statement must be code"); - ulong limit=0; + HashStringValue* bind=0; + ulong limit=SQL_NO_LIMIT; ulong offset=0; - if(params->size()>1) { - Value& voptions=params->as_no_junction(1, "options must be hash, not code"); - if(voptions.is_defined()) - if(options=voptions.get_hash()) { - if(Value *vlimit=(Value *)options->get(*sql_limit_name)) - limit=(ulong)r.process(*vlimit).as_double(); - if(Value *voffset=(Value *)options->get(*sql_offset_name)) - offset=(ulong)r.process(*voffset).as_double(); - } else - PTHROW(0, 0, - &method_name, - "options must be hash"); - } else - options=0; - - Temp_lang temp_lang(r, String::UL_SQL); - const String& statement_string=r.process(statement).as_string(); - const char *statement_cstr= - statement_string.cstr(String::UL_UNSPECIFIED, r.connection); - String_sql_event_handlers handlers(pool, statement_string, statement_cstr); - bool need_rethrow=false; Exception rethrow_me; - PTRY { - r.connection->query( - statement_cstr, offset, limit, - handlers); - } - PCATCH(e) { // query problem - rethrow_me=e; need_rethrow=true; - } - PEND_CATCH - if(need_rethrow) - PTHROW(rethrow_me.type(), rethrow_me.code(), - &statement_string, // setting more specific source [were url] - rethrow_me.comment()); + default_code=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); + + const String& statement_string=r.process_to_string(statement); + const char* statement_cstr=statement_string.untaint_cstr(String::L_SQL, r.connection()); + + String_sql_event_handlers handlers(statement_string, statement_cstr); + + r.connection()->query( + statement_cstr, + placeholders_count, placeholders, + offset, limit, + handlers, + statement_string); + if(bind) + unmarshal_bind_updates(*bind, placeholders_count, placeholders); + if(!handlers.got_cell) return 0; // no lines, caller should return second param[default value] return handlers.result; } -static void _sql(Request& r, const String& method_name, MethodParams *params) { - Pool& pool=r.pool(); +static void _sql(Request& r, MethodParams& params) { - Hash *options; - const String *string=sql_result_string(r, method_name, params, options); + Value* default_code; + const String* string=sql_result_string(r, params, default_code); if(!string) { - if(options) { - if(Value *vdefault=(Value *)options->get(*sql_default_name)) { - if(!vdefault->get_junction()) - PTHROW(0, 0, - &method_name, - "default option must be code"); - string=r.process(*vdefault).get_string(); - if(!string) - string=empty_string; - } else - PTHROW(0, 0, - &method_name, - "produced no result, but no default option specified"); + if(default_code) { + string=&r.process_to_string(*default_code); } else - PTHROW(0, 0, - &method_name, - "produced no result, but no options (no default) specified"); - } - VString& result=*new(pool) VString(*string); - result.set_name(method_name); - r.write_assign_lang(result); -} - -static void _replace(Request& r, const String& method_name, MethodParams *params) { - Pool& pool=r.pool(); - const String& src=*static_cast(r.self)->get_string(); - - Table *table=params->as_no_junction(0, "parameter must not be code").get_table(); - if(!table) - PTHROW(0, 0, - &method_name, - "parameter must be table"); + throw Exception(PARSER_RUNTIME, 0, "produced no result, but no default option specified"); + } - Dictionary dict(*table); - r.write_assign_lang(*new(pool) VString(src.replace(pool, dict))); + r.write(*string); } -// constructor -MString::MString(Pool& apool) : Methoded(apool) { - set_name(*NEW String(pool(), STRING_CLASS_NAME)); +static void _replace(Request& r, MethodParams& params) { + const String& src=GET_SELF(r, VString).string(); + + if(params.count()==1) { + // ^string.replace[table] + Table* table=params.as_table(0, "param"); + Dictionary dict(*table); + 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(src.replace(dict)); + } + +} +static void _save(Request& r, MethodParams& params) { + bool do_append=false; + Charset* asked_charset=0; + + size_t file_name_index=0; + 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=&pa_charsets.get(vcharset_name->as_string()); + valid_options++; + } + if(Value* vappend=options->get(MODE_APPEND)){ + do_append=vappend->as_bool(); + valid_options++; + } + if(valid_options != options->count()) + 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; + file_name_index++; + } else + throw Exception(PARSER_RUNTIME, + &mode, + "unknown mode, must be 'append'"); + } + } + const String& file_name=params.as_string(file_name_index, FILE_NAME_MUST_BE_STRING); + const String& src=GET_SELF(r, VString).string(); + + 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); +} + +static void _normalize(Request& r, MethodParams&) { + const String& src=GET_SELF(r, VString).string(); + + r.write(src); +} + +static void _trim(Request& r, MethodParams& params) { + const String& src=GET_SELF(r, VString).string(); + + String::Trim_kind kind=String::TRIM_BOTH; + size_t params_count=params.count(); + 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==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); + } + + if(params_count>1) { + const String& schars=params.as_string(1, "'chars' must be string"); + if(!schars.is_empty()) + chars=schars.cstr(); + } + } + + r.write(src.trim(kind, chars, &r.charsets.source())); +} + +static void _base64(Request& r, MethodParams& params) { + if(&r.get_self() == string_class) { + // decode: ^string:base64[encoded[;$.strict(true|false)]] + const char* cstr=params.count() ? params.as_string(0, PARAMETER_MUST_BE_STRING).cstr() : ""; + char* decoded=0; + size_t length=0; + + bool strict=false; + if(params.count() > 1) + if(HashStringValue* options=params.as_hash(1)) { + int valid_options=0; + if(Value* vstrict=options->get(BASE64_STRICT_OPTION_NAME)) { + strict=r.process(*vstrict).as_bool(); + valid_options++; + } + if(valid_options!=options->count()) + throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION); + } + + 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."); + + fix_line_breaks(decoded, length); + if(length) + 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(*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(); + r.write(*new String(pa_idna_encode(cstr, r.charsets.source()), String::L_TAINTED)); + } +} + +static void _js_escape(Request& r, MethodParams&){ + const String& src=GET_SELF(r, VString).string(); + r.write(src.escape(r.charsets.source())); +} + +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(*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 + +MString::MString(): Methoded("string") { // ^string.length[] add_native_method("length", Method::CT_DYNAMIC, _length, 0, 0); @@ -453,20 +767,30 @@ MString::MString(Pool& apool) : Methoded // ^string.double[] // ^string.double(default) add_native_method("double", Method::CT_DYNAMIC, _double, 0, 1); + // ^void.bool[] + // ^void.bool(default) + add_native_method("bool", Method::CT_DYNAMIC, _bool, 0, 1); - // ^string.format{format} + // ^string.format[format] add_native_method("format", Method::CT_DYNAMIC, _string_format, 1, 1); // ^string.left(n) add_native_method("left", Method::CT_DYNAMIC, _left, 1, 1); // ^string.right(n) add_native_method("right", Method::CT_DYNAMIC, _right, 1, 1); + // ^string.mid(p) // ^string.mid(p;n) - add_native_method("mid", Method::CT_DYNAMIC, _mid, 2, 2); + add_native_method("mid", Method::CT_DYNAMIC, _mid, 1, 2); // ^string.pos[substr] - add_native_method("pos", Method::CT_DYNAMIC, _pos, 1, 1); + // ^string.pos[substr](n) + add_native_method("pos", Method::CT_DYNAMIC, _pos, 1, 2); + // ^string.split[delim] + // ^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] @@ -474,27 +798,49 @@ MString::MString(Pool& apool) : Methoded // ^string.match[regexp][options] // ^string.match[regexp][options]{replacement-code} - add_native_method("match", Method::CT_DYNAMIC, _match, 1, 3); + // ^string.match[regexp][options]{replacement-code}{code-if-nothing-is-found} + add_native_method("match", Method::CT_DYNAMIC, _match, 1, 4); - // ^string.toupper[] + // ^string.upper[] add_native_method("upper", Method::CT_DYNAMIC, _upper, 0, 0); - // ^string.tolower[] + // ^string.lower[] add_native_method("lower", Method::CT_DYNAMIC, _lower, 0, 0); - // ^sql[query] - // ^sql[query][$.limit(1) $.offset(2) $.default[n/a]] + // ^string:sql{query} + // ^string:sql{query}[options hash] add_native_method("sql", Method::CT_STATIC, _sql, 1, 2); // ^string.replace[table] - add_native_method("replace", Method::CT_DYNAMIC, _replace, 1, 1); -} + add_native_method("replace", Method::CT_DYNAMIC, _replace, 1, 2); -// global variable + // ^string.save[append][file] + // ^string.save[file] + // ^string.save[file][$.append(true) $.charset[...]] + add_native_method("save", Method::CT_DYNAMIC, _save, 1, 2); + + // ^string.normalize[] + add_native_method("normalize", Method::CT_DYNAMIC, _normalize, 0, 0); + + // ^string.trim[[start|both|end][;chars]] + add_native_method("trim", Method::CT_DYNAMIC, _trim, 0, 2); + + // ^string.base64[] << encode + // ^string:base64[encoded string] << decode + add_native_method("base64", Method::CT_ANY, _base64, 0, 2); + + // ^string.idna[] << encode + // ^string:idna[encoded string] << decode + add_native_method("idna", Method::CT_ANY, _idna, 0, 1); + + // ^string.js-escape[] + add_native_method("js-escape", Method::CT_DYNAMIC, _js_escape, 0, 0); -Methoded *string_class; + // ^string:js-unescape[escaped%uXXXXstring] + add_native_method("js-unescape", Method::CT_STATIC, _js_unescape, 1, 1); -// creator + // ^string:unescape[js|uri;escaped;$.charset[...]] + add_native_method("unescape", Method::CT_STATIC, _unescape, 2, 3); -Methoded *MString_create(Pool& pool) { - return string_class=new(pool) MString(pool); + // ^string.contains[key] for hash compatibility + add_native_method("contains", Method::CT_DYNAMIC, _contains, 1, 1); }