--- parser3/src/classes/table.C 2008/05/27 19:01:46 1.243 +++ parser3/src/classes/table.C 2011/05/15 13:20:07 1.280 @@ -1,15 +1,14 @@ /** @file Parser: @b table parser class. - Copyright (c) 2001-2005 ArtLebedev Group (http://www.artlebedev.com) + Copyright (c) 2001-2009 ArtLebedev Group (http://www.artlebedev.com) Author: Alexandr Petrosian (http://paf.design.ru) */ -static const char * const IDENT_TABLE_C="$Date: 2008/05/27 19:01:46 $"; +static const char * const IDENT_TABLE_C="$Date: 2011/05/15 13:20:07 $"; -#ifndef NO_STRINGSTREAM +#if (!defined(NO_STRINGSTREAM) && !defined(FREEBSD4)) #include -using namespace std; #endif #include "classes.h" @@ -17,32 +16,26 @@ using namespace std; #include "pa_common.h" #include "pa_request.h" +#include "pa_charsets.h" #include "pa_vtable.h" #include "pa_vint.h" #include "pa_sql_connection.h" #include "pa_vbool.h" +#include "pa_array.h" // class class MTable: public Methoded { public: // VStateless_class - Value* create_new_value(Pool&, HashStringValue&) { return new VTable(); } - + Value* create_new_value(Pool&) { return new VTable(); } public: MTable(); - -public: // Methoded - bool used_directly() { return true; } }; // global variable DECLARE_CLASS_VAR(table, new MTable, 0); -// externs - -extern String cycle_data_name; - #define TABLE_REVERSE_NAME "reverse" // globals @@ -58,12 +51,12 @@ String table_reverse_name(TABLE_REVERSE_ // methods static Table::Action_options get_action_options(Request& r, MethodParams& params, - const Table& source) { + size_t options_index, const Table& source) { Table::Action_options result; - if(!params.count()) + if(params.count() <= options_index) return result; - Value& maybe_options=params.last(); + Value& maybe_options=params[options_index]; /* can not do it: want to enable ^table::create[$source; # $.option[] @@ -110,9 +103,7 @@ static Table::Action_options get_action_ } if(valid_options!=options->count()) - throw Exception(PARSER_RUNTIME, - 0, - "called with invalid option"); + throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION); return result; } @@ -130,7 +121,7 @@ struct TableSeparators { char encloser; const String* sencloser; TableSeparators(): - column('\t'), scolumn(new String("\t", false)), + column('\t'), scolumn(new String("\t")), encloser(0), sencloser(0) {} int load( HashStringValue& options ) { @@ -160,7 +151,7 @@ struct TableSeparators { static void _create(Request& r, MethodParams& params) { // clone/copy part? if(Table *source=params[0].get_table()) { - Table::Action_options o=get_action_options(r, params, *source); + Table::Action_options o=get_action_options(r, params, 1, *source); check_option_param(o.defined, params, 1, "too many parameters"); GET_SELF(r, VTable).set_table(*new Table(*source, o)); @@ -230,9 +221,9 @@ static void _create(Request& r, MethodPa Array_iterator i(rows); while(i.has_next()) { Table::element_type row(new ArrayString); - const String& string=*i.next(); + const String& string=*i.next(); // remove comment lines - if(!string.length()) + if(string.is_empty()) continue; size_t col_pos_after=0; @@ -253,7 +244,7 @@ struct lsplit_result { inline lsplit_result lsplit(char* string, char delim1, char delim2) { lsplit_result result; - if(string) { + if(string) { char delims[]={delim1, delim2, 0}; if(char* v=strpbrk(string, delims)) { result.delim=*v; @@ -261,27 +252,27 @@ inline lsplit_result lsplit(char* string result.piece=v+1; return result; } - } + } result.piece=0; result.delim=0; - return result; + return result; } inline lsplit_result lsplit(char* *string_ref, char delim1, char delim2) { - lsplit_result result; + lsplit_result result; result.piece=*string_ref; lsplit_result next=lsplit(*string_ref, delim1, delim2); result.delim=next.delim; *string_ref=next.piece; - return result; + return result; } static lsplit_result lsplit(char** string_ref, char delim1, char delim2, char encloser) { lsplit_result result; - if(char* string=*string_ref) { + if(char* string=*string_ref) { if(encloser && *string==encloser) { - string++; + string++; char *read; char *write; @@ -307,9 +298,9 @@ static lsplit_result lsplit(char** strin return result; } else return lsplit(string_ref, delim1, delim2); - } + } result.piece=0; - return result; + return result; } static void skip_empty_and_comment_lines( char** data_ref ) { @@ -326,6 +317,22 @@ static void skip_empty_and_comment_lines } } +static void skip_empty_lines( char** data_ref ) { + if(char *data=*data_ref) { + while( char c=*data ) { + if( c== '\n' ) { + /*nowhere=*/getrow(&data); // remove empty lines + if(!(*data_ref=data)) + break; + continue; + } + break; + } + } +} + +typedef void (*Skip_lines_action)(char** data_ref); + static void _load(Request& r, MethodParams& params) { const String& first_param=params.as_string(0, FILE_NAME_MUST_BE_STRING); int filename_param_index=0; @@ -344,12 +351,14 @@ static void _load(Request& r, MethodPara } // loading text - char *data=file_read_text(r.charsets, + char *data=file_load_text(r, r.absolute(params.as_string(filename_param_index, FILE_NAME_MUST_BE_STRING)), true, options ); + Skip_lines_action skip_lines_action = (separators.column=='#' || separators.encloser=='#') ? skip_empty_lines : skip_empty_and_comment_lines; + // parse columns Table::columns_type columns; if(nameless) { @@ -357,9 +366,9 @@ static void _load(Request& r, MethodPara } else { columns=Table::columns_type(new ArrayString); - skip_empty_and_comment_lines(&data); + skip_lines_action(&data); while( lsplit_result sr=lsplit(&data, separators.column, '\n', separators.encloser) ) { - *columns+=new String(sr.piece, 0, true); + *columns+=new String(sr.piece, String::L_TAINTED); if(sr.delim=='\n') break; } @@ -370,15 +379,15 @@ static void _load(Request& r, MethodPara // parse cells Table::element_type row(new ArrayString(columns_count)); - skip_empty_and_comment_lines(&data); + skip_lines_action(&data); while( lsplit_result sr=lsplit(&data, separators.column, '\n', separators.encloser) ) { if(!*sr.piece && !sr.delim && !row->count()) // append last empty column [if without \n] break; - *row+=new String(sr.piece, 0, true); + *row+=new String(sr.piece, String::L_TAINTED); if(sr.delim=='\n') { table+=row; row=new ArrayString(columns_count); - skip_empty_and_comment_lines(&data); + skip_lines_action(&data); } } // last line [if without \n] @@ -389,51 +398,56 @@ static void _load(Request& r, MethodPara GET_SELF(r, VTable).set_table(table); } -#ifdef NO_STRINGSTREAM +#if (!defined(NO_STRINGSTREAM) && !defined(FREEBSD4)) -void maybe_enclose( String& to, const String& from, char encloser, const String* sencloser ) { +#include "gc_allocator.h" + +typedef std::basic_stringstream, gc_allocator > pa_stringstream; +typedef std::basic_string, gc_allocator > pa_string; + +void maybe_enclose( pa_stringstream& to, const String& from, char encloser ) { if(encloser) { - to<<*sencloser; + to<count()) - throw Exception(PARSER_RUNTIME, - 0, - "invalid option passed"); + throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION); } else { throw Exception(PARSER_RUNTIME, 0, @@ -476,106 +492,119 @@ static void _save(Request& r, MethodPara Table& table=GET_SELF(r, VTable).table(); -#ifdef NO_STRINGSTREAM +#if (!defined(NO_STRINGSTREAM) && !defined(FREEBSD4)) - String sdata; + pa_stringstream ost(std::stringstream::out); + + // process header if(output_column_names) { if(table.columns()) { // named table for(Array_iterator i(*table.columns()); i.has_next(); ) { - maybe_enclose( sdata, *i.next(), separators.encloser, separators.sencloser ); - if(i.has_next()) - sdata<<*separators.scolumn; + maybe_enclose( ost, *i.next(), separators.encloser ); + if(i.has_next()){ + ost<count():0) for(int column=0; column i(table); while(i.has_next()) { for(Array_iterator c(*i.next()); c.has_next(); ) { - maybe_enclose( sdata, *c.next(), separators.encloser, separators.sencloser ); + maybe_enclose( ost, *c.next(), separators.encloser ); if(c.has_next()) - sdata<<*separators.scolumn; + ost< i(*table.columns()); i.has_next(); ) { - maybe_enclose( ost, *i.next(), separators.encloser ); - if(i.has_next()){ - ost<count():0) for(int column=0; column i(table); while(i.has_next()) { for(Array_iterator c(*i.next()); c.has_next(); ) { - maybe_enclose( ost, *c.next(), separators.encloser ); + maybe_enclose( sdata, *c.next(), separators.encloser, separators.sencloser ); if(c.has_next()) - ost<count() : 0; + else if(param == "cells") + result = table.count() ? table[table.current()]->count() : 0; + else if(param == "rows") // synonim for ^table.count[] + result = table.count(); + else + throw Exception(PARSER_RUNTIME, ¶m, "parameter must be 'columns', 'cells' and 'rows' only"); + } else + result = table.count(); + r.write_no_lang(*new VInt(result)); } @@ -607,35 +636,47 @@ static void _offset(Request& r, MethodPa } static void _menu(Request& r, MethodParams& params) { - Temp_hash_value - cycle_data_setter(r.classes_conf, cycle_data_name, /*any not null flag*/&r); + InCycle temp(r); Value& body_code=params.as_junction(0, "body must be code"); Value* delim_maybe_code=params.count()>1?¶ms[1]:0; Table& table=GET_SELF(r, VTable).table(); - bool need_delim=false; int saved_current=table.current(); int size=table.count(); - for(int row=0; rowlength()) { // delimiter set and we have body - if(need_delim) // need delim & iteration produced string? - r.write_pass_lang(r.process(*delim_maybe_code)); - else - need_delim=true; - } + if(delim_maybe_code) { // delimiter set + bool need_delim=false; + for(int row=0; rowis_empty()) { // we have body + if(need_delim) // need delim & iteration produced string? + r.write_pass_lang(r.process(*delim_maybe_code)); + else + need_delim=true; + } - r.write_pass_lang(sv_processed); + r.write_pass_lang(sv_processed); - if(lskip==Request::SKIP_BREAK) - break; + if(lskip==Request::SKIP_BREAK) + break; + } + } else { + for(int row=0; rowvalue_type) { case C_STRING: { - exist=info->hash->put_dont_replace(*key, new VString(*row->get(info->value_fields->get(0)))); + size_t index=info->value_fields->get(0); + exist=info->hash->put_dont_replace(*key, (index < row->count()) ? new VString(*row->get(index)) : new VString()); break; } case C_HASH: { @@ -772,16 +814,12 @@ static void _hash(Request& r, MethodPara } if(valid_options!=options->count()) - throw Exception(PARSER_RUNTIME, - 0, - "called with invalid option"); + throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION); } } if(param_index==2) // options was specified but not as hash - throw Exception(PARSER_RUNTIME, - 0, - "options must be hash"); + throw Exception(PARSER_RUNTIME, 0, OPTIONS_MUST_BE_HASH); Array value_fields; if(param_index==0){ // list of columns wasn't specified @@ -899,6 +937,14 @@ static void _sort(Request& r, MethodPara else seq[i].value.d=value.as_double(); } + + // @todo: handle this elsewhere + if(r.charsets.source().NAME()=="KOI8-R" && key_values_are_strings) { + for(i=0; ir->process_to_value(*info->expression_code).as_bool(); } -static bool _locate_expression(Table& table, Table::Action_options o, - Request& r, MethodParams& params) { - check_option_param(o.defined, params, 1, - "locate by expression only has parameters: expression and, maybe, options"); + +static bool _locate_expression(Table& table, Request& r, MethodParams& params) { Value& expression_code=params.as_junction(0, "must be expression"); + const size_t options_index=1; + Table::Action_options o=get_action_options(r, params, options_index, table); + check_option_param(o.defined, params, options_index, "locate by expression only has parameters: expression and, maybe, options"); Expression_is_true_info info={&r, &expression_code}; return table.table_first_that(expression_is_true, &info, o); } -static bool _locate_name_value(Table& table, Table::Action_options o, - Request&, MethodParams& params) { - check_option_param(o.defined, params, 2, - "locate by locate by name has parameters: name, value and, maybe, options"); + +static bool _locate_name_value(Table& table, Request& r, MethodParams& params) { const String& name=params.as_string(0, "column name must be string"); const String& value=params.as_string(1, VALUE_MUST_BE_STRING); + const size_t options_index=2; + Table::Action_options o=get_action_options(r, params, options_index, table); + check_option_param(o.defined, params, options_index, "locate by name has parameters: name, value and, maybe, options"); + return table.locate(name, value, o); } + static void _locate(Request& r, MethodParams& params) { Table& table=GET_SELF(r, VTable).table(); - Table::Action_options o=get_action_options(r, params, table); - bool result=params[0].get_junction()? - _locate_expression(table, o, r, params) : - _locate_name_value(table, o, r, params); - r.write_no_lang(*new VBool(result)); + _locate_expression(table, r, params) : + _locate_name_value(table, r, params); + r.write_no_lang(VBool::get(result)); } @@ -955,7 +1004,7 @@ static void _flip(Request& r, MethodPara Table& old_table=GET_SELF(r, VTable).table(); Table& new_table=*new Table(0); if(size_t old_count=old_table.count()) - if(size_t old_cols=old_table[0]->count()) + if(size_t old_cols=old_table.columns()?old_table.columns()->count():old_table[0]->count()) for(size_t column=0; columnas_string().cstr(String::L_UNSPECIFIED); + ph.value=avalue->as_string().untaint_cstr(String::L_AS_IS); ph.is_null=avalue->get_class()==void_class; ph.were_updated=false; @@ -1103,12 +1148,9 @@ void unmarshal_bind_updates(HashStringVa if(ph->were_updated) { Value* value; if(ph->is_null) - value=new VVoid(); + value=VVoid::get(); else - if(ph->value) - value=new VString(*new String(ph->value, 0, true/*tainted*/)); - else - value=new VString(*new String()); + value=new VString(*new String(ph->value, String::L_TAINTED)); hash.put(ph->name, value); } } @@ -1117,7 +1159,7 @@ static void _sql(Request& r, MethodParam Value& statement=params.as_junction(0, "statement must be code"); HashStringValue* bind=0; - ulong limit=0; + ulong limit=SQL_NO_LIMIT; ulong offset=0; if(params.count()>1) { Value& voptions=params.as_no_junction(1, "options must be hash, not code"); @@ -1137,13 +1179,9 @@ static void _sql(Request& r, MethodParam offset=(ulong)r.process_to_value(*voffset).as_double(); } if(valid_options!=options->count()) - throw Exception(PARSER_RUNTIME, - 0, - "called with invalid option"); + throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION); } else - throw Exception(PARSER_RUNTIME, - 0, - "options must be hash"); + throw Exception(PARSER_RUNTIME, 0, OPTIONS_MUST_BE_HASH); } SQL_Driver::Placeholder* placeholders=0; @@ -1153,8 +1191,8 @@ static void _sql(Request& r, MethodParam Temp_lang temp_lang(r, String::L_SQL); const String& statement_string=r.process_to_string(statement); - const char* statement_cstr= - statement_string.cstr(String::L_UNSPECIFIED, r.connection()); + const char* statement_cstr=statement_string.untaint_cstr(r.flang, r.connection()); + Table_sql_event_handlers handlers; #ifdef RESOURCES_DEBUG struct timeval mt[2]; @@ -1219,16 +1257,59 @@ static void _select(Request& r, MethodPa Table& source_table=GET_SELF(r, VTable).table(); Table& result_table=*new Table(source_table.columns()); + int limit=source_table.count(); + int offset = 0; + bool reverse = false; + + 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(HashStringValue* options=voptions.get_hash()) { + int valid_options=0; + if(Value* vlimit=options->get(sql_limit_name)) { + valid_options++; + limit=r.process_to_value(*vlimit).as_int(); + } + if(Value* voffset=options->get(sql_offset_name)) { + valid_options++; + offset=r.process_to_value(*voffset).as_int(); + } + if(Value* vreverse=options->get(table_reverse_name)) { + valid_options++; + reverse=r.process_to_value(*vreverse).as_bool(); + } + if(valid_options!=options->count()) + throw Exception(PARSER_RUNTIME, + 0, + "called with invalid option"); + } else + throw Exception(PARSER_RUNTIME, + 0, + "options must be hash"); + } + int saved_current=source_table.current(); int size=source_table.count(); - for(int row=0; row=0 && result_table.count() < limit; row--) { + source_table.set_current(row); - if(condition) // ...condition is true= - result_table+=source_table[row]; // =green light to go to result + bool condition=r.process_to_value(vcondition, false/*don't intercept string*/).as_bool(); + + if(condition && ++appended > offset) // ...condition is true, adding to the result + result_table+=source_table[row]; + } + } else { + for(int row=0; row < size && result_table.count() < limit; row++) { + source_table.set_current(row); + + bool condition=r.process_to_value(vcondition, false/*don't intercept string*/).as_bool(); + + if(condition && ++appended > offset) // ...condition is true, adding to the result + result_table+=source_table[row]; + } } source_table.set_current(saved_current); @@ -1256,7 +1337,10 @@ MTable::MTable(): Methoded("table") { // add_native_method("save_old", Method::CT_DYNAMIC, _save_old, 1, 3); // ^table.count[] - add_native_method("count", Method::CT_DYNAMIC, _count, 0, 0); + // ^table.count[rows] + // ^table.count[columns] + // ^table.count[cells] + add_native_method("count", Method::CT_DYNAMIC, _count, 0, 1); // ^table.line[] add_native_method("line", Method::CT_DYNAMIC, _line, 0, 0); @@ -1299,5 +1383,5 @@ MTable::MTable(): Methoded("table") { add_native_method("columns", Method::CT_DYNAMIC, _columns, 0, 1); // ^table.select(expression) = table - add_native_method("select", Method::CT_DYNAMIC, _select, 1, 1); + add_native_method("select", Method::CT_DYNAMIC, _select, 1, 2); }