--- parser3/src/classes/string.C 2001/03/29 17:23:18 1.18 +++ parser3/src/classes/string.C 2009/07/07 05:47:43 1.185 @@ -1,151 +1,797 @@ -/* - Parser - Copyright (c) 2001 ArtLebedev Group (http://www.artlebedev.com) - Author: Alexander Petrosyan (http://design.ru/paf) +/** @file + Parser: @b string parser class. - $Id: string.C,v 1.18 2001/03/29 17:23:18 paf Exp $ + Copyright (c) 2001-2009 ArtLebedev Group (http://www.artlebedev.com) + Author: Alexandr Petrosian (http://paf.design.ru) */ +static const char * const IDENT_STRING_C="$Date: 2009/07/07 05:47:43 $"; + +#include "classes.h" +#include "pa_vmethod_frame.h" + #include "pa_request.h" -#include "_string.h" #include "pa_vdouble.h" #include "pa_vint.h" #include "pa_vtable.h" +#include "pa_vbool.h" +#include "pa_string.h" +#include "pa_sql_connection.h" +#include "pa_dictionary.h" +#include "pa_vmethod_frame.h" +#include "pa_vregex.h" + +// class + +class MString: public Methoded { +public: + MString(); +public: // Methoded + bool used_directly() { return true; } +}; + +// global variable + +DECLARE_CLASS_VAR(string, new MString, 0); + +// defines for statics -// global var +#define MATCH_VAR_NAME "match" +#define TRIM_START_OPTION "left" +#define TRIM_END_OPTION "right" +#define TRIM_BOTH_OPTION "both" -VStateless_class *string_class; +// statics + +static const String match_var_name(MATCH_VAR_NAME); // methods -static void _length(Request& r, const String&, Array *) { - Pool& pool=r.pool(); - Value& value=*new(pool) VDouble(pool, r.self->get_string()->size()); - r.write_no_lang(value); +static void _length(Request& r, MethodParams&) { + double result=GET_SELF(r, VString).string().length(r.charsets.source()); + r.write_no_lang(*new VDouble(result)); } -static void _int(Request& r, const String&, Array *) { - Pool& pool=r.pool(); - Value& value=*new(pool) VInt(pool, (int)r.self->as_double()); - r.write_no_lang(value); +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(params.count()>0) + converted=params.as_int(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, const String&, Array *) { - Pool& pool=r.pool(); - Value& value=*new(pool) VDouble(pool, r.self->as_double()); - r.write_no_lang(value); +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(params.count()>0) + converted=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 + 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)); } -/*not static*/void _string_format(Request& r, const String& method_name, Array *params) { - Pool& pool=r.pool(); +/*not static*/void _string_format(Request& r, MethodParams& params) { - Value& fmt=*static_cast(params->get(0)); - // forcing [this param type] - r.fail_if_junction_(true, fmt, method_name, "fmt must not be junction"); + 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); - char *buf=format(pool, r.self->as_double(), fmt.as_string().cstr()); - - r.write_no_lang(String(pool, buf)); + const char* buf=format(r.get_self().as_double(), fmt.trim().cstrm()); + + r.write_no_lang(String(buf)); } -static void _left(Request& r, const String&, Array *params) { - Pool& pool=r.pool(); +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; - size_t n=(size_t)r.process(*static_cast(params->get(0))).as_double(); - - const String& string=*static_cast(r.self)->get_string(); - r.write_assign_lang(*new(pool) VString(string.piece(0, n))); + const String& string=GET_SELF(r, VString).string(); + r.write_assign_lang(string.mid(r.charsets.source(), 0, n)); +} + +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(n(params->get(0))).as_double(); + 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; + + 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; + } + + r.write_assign_lang(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=*static_cast(r.self)->get_string(); - r.write_assign_lang(*new(pool) VString(string.piece(string.size()-n, string.size()))); + 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); + } + + r.write_no_lang(*new VInt((int)string.pos(r.charsets.source(), substr.as_string(), (size_t)offset))); } -static void _mid(Request& r, const String&, Array *params) { - Pool& pool=r.pool(); +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 p=(size_t)r.process(*static_cast(params->get(0))).as_double(); - size_t n=(size_t)r.process(*static_cast(params->get(1))).as_double(); - - const String& string=*static_cast(r.self)->get_string(); - r.write_assign_lang(*new(pool) VString(string.piece(p, p+n))); + size_t pos_after=0; + string.split(result, pos_after, delim_value.as_string()); } -static void _pos(Request& r, const String& method_name, Array *params) { - Pool& pool=r.pool(); +#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; + } + } - Value& substr=*static_cast(params->get(0)); - // forcing [this param type] - r.fail_if_junction_(true, substr, method_name, "substr must not be junction"); - - const String& string=*static_cast(r.self)->get_string(); - r.write_assign_lang(*new(pool) VInt(pool, string.pos(substr.as_string()))); + 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; + } + } + + return table; +} + +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(); + } + table+=row; + + return table; +} + +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); + + 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; + + 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_no_lang(*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, Table::element_type row, int, int, int, int, void *) { + if(row) + table+=row; +} + +#ifndef DOXYGEN +struct Replace_action_info { + Request* request; + const String* src; + String* dest; + VTable* vtable; + Value* replacement_code; +}; +#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) { + Replace_action_info& ai=*static_cast(info); + if(row) { // begin&middle + // 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.count()) // middle + table.put(0, row); + else // begin + table+=row; + + { // 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); + } + } + } else // end + *ai.dest << ai.src->mid(poststart, postfinish); } -static void _lsplit(Request& r, const String& method_name, Array *params) { - Pool& pool=r.pool(); +static void _match(Request& r, MethodParams& params) { + size_t params_count=params.count(); - Value& delim_value=*static_cast(params->get(0)); - // forcing [this param type] - r.fail_if_junction_(true, delim_value, method_name, "delimiter must not be junction"); - const String& delim=delim_value.as_string(); + 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, false)){ + 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; + } + + 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); + + if(table){ + r.write_no_lang(*new VTable(table)); + } else { + r.write_no_lang(*new VInt(matches_count)); + } + + } else { // replace + + 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"); + + String result; + VTable* vtable=new VTable; + Replace_action_info info={ + &r, + &src, + &result, + vtable, + replacement_code + }; + + Temp_value_element* temp_match_var=0; + + if(is_junction) + temp_match_var=new Temp_value_element( + *replacement_code->get_junction()->method_frame, + match_var_name, vtable); + + src.match(vregex, + replace_action, &info, + matches_count); + + if(temp_match_var) + delete temp_match_var; + + r.write_assign_lang(result); + } +} + +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)); +} +static void _upper(Request& r, MethodParams& params) { + change_case(r, params, String::CC_UPPER); +} +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 { + const String& statement_string; const char* statement_cstr; + bool got_column; +public: + bool got_cell; + 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(*new String) {} + + 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; + } + 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) { + if(got_cell) { + error=SQL_Error(PARSER_RUNTIME, + //statement_string, + "result must not contain more then one row"); + return true; + } + + try { + got_cell=true; + result.append_know_length(str, length, String::L_TAINTED); + return false; + } catch(...) { + error=SQL_Error("exception occured 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) { + 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; + + 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()); + + String_sql_event_handlers handlers(statement_string, statement_cstr); + + r.connection()->query( + statement_cstr, + placeholders_count, placeholders, + offset, limit, + handlers, + statement_string); - const String& string=*static_cast(r.self)->get_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; +} - Table& result=*new(pool) Table(pool, &string, 0); +static void _sql(Request& r, MethodParams& params) { - if(delim.size()) { - size_t pos_after=0; - int pos_before; - while((pos_before=string.pos(delim, pos_after))>=0) { // we have 'delim' in 'string'? - Array& row=*new(pool) Array(pool, 1); - result+=&(row+=&string.piece(pos_after, pos_before)); - pos_after=pos_before+delim.size(); + HashStringValue* options; + Value* default_code; + const String* string=sql_result_string(r, params, options, 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"); + } + + r.write_assign_lang(*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"); + + Dictionary dict(*table); + r.write_assign_lang(src.replace(dict)); +} + +static void _save(Request& r, MethodParams& params) { + size_t params_count=params.count(); + const String& file_name=params.as_string(params_count-1, FILE_NAME_MUST_BE_STRING); + + const String& src=GET_SELF(r, VString).string(); + + bool do_append=false; + if(params_count>1) { + const String& mode=params.as_string(0, "mode must be string"); + if(mode=="append") + do_append=true; + else + throw Exception(PARSER_RUNTIME, + &mode, + "unknown mode, must be 'append'"); + } + + // write + String::Body sbody=src.cstr_to_string_body_untaint(String::L_AS_IS, r.connection(false/*no error if none*/)); + file_write(r.absolute(file_name), sbody.cstr(), sbody.length(), true, do_append); +} + +static void _normalize(Request& r, MethodParams&) { + const String& src=GET_SELF(r, VString).string(); + + r.write_assign_lang(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 + 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(); } - // last piece - if(pos_after