--- parser3/src/classes/table.C 2008/05/27 19:01:46 1.243 +++ parser3/src/classes/table.C 2016/07/28 21:48:57 1.314 @@ -1,47 +1,41 @@ /** @file Parser: @b table parser class. - Copyright (c) 2001-2005 ArtLebedev Group (http://www.artlebedev.com) + Copyright (c) 2001-2015 Art. Lebedev Studio (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 $"; - -#ifndef NO_STRINGSTREAM -#include -using namespace std; -#endif - #include "classes.h" #include "pa_vmethod_frame.h" #include "pa_common.h" #include "pa_request.h" +#include "pa_charsets.h" #include "pa_vtable.h" #include "pa_vint.h" #include "pa_sql_connection.h" #include "pa_vbool.h" +#include "pa_array.h" + +#if (!defined(NO_STRINGSTREAM) && !defined(FREEBSD4)) +#include +#define USE_STRINGSTREAM +#endif + +volatile const char * IDENT_TABLE_C="$Id: table.C,v 1.314 2016/07/28 21:48:57 moko Exp $"; // class class MTable: public Methoded { public: // VStateless_class - Value* create_new_value(Pool&, HashStringValue&) { return new VTable(); } - + Value* create_new_value(Pool&) { return new VTable(); } public: MTable(); - -public: // Methoded - bool used_directly() { return true; } }; // global variable -DECLARE_CLASS_VAR(table, new MTable, 0); - -// externs - -extern String cycle_data_name; +DECLARE_CLASS_VAR(table, new MTable); #define TABLE_REVERSE_NAME "reverse" @@ -58,25 +52,12 @@ String table_reverse_name(TABLE_REVERSE_ // methods static Table::Action_options get_action_options(Request& r, MethodParams& params, - const Table& source) { + size_t options_index, const Table& source) { Table::Action_options result; - if(!params.count()) + if(params.count() <= options_index) return result; - Value& maybe_options=params.last(); -/* can not do it: - want to enable ^table::create[$source; -# $.option[] - ] - but there is ^table.locate[name;value] - - ...if(voptions.is_defined() && !voptions.is_string())) - if(maybe_options.is_string()) { // allow empty options - result.defined=true; - return result; - } -*/ - HashStringValue* options=maybe_options.get_hash(); + HashStringValue* options=params.as_hash(options_index); if(!options) return result; @@ -110,27 +91,17 @@ 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; } -static void check_option_param(bool options_defined, - MethodParams& params, size_t next_param_index, - const char *msg) { - if(next_param_index+(options_defined?1:0) != params.count()) - throw Exception(PARSER_RUNTIME, - 0, - "%s", msg); -} struct TableSeparators { char column; const String* scolumn; 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 ) { @@ -146,11 +117,15 @@ struct TableSeparators { } if(Value* vencloser=options.get(PA_COLUMN_ENCLOSER_NAME)) { sencloser=&vencloser->as_string(); + if(sencloser->is_empty()){ + encloser=0; + } else { if(sencloser->length()!=1) throw Exception(PARSER_RUNTIME, sencloser, "encloser must be one character long"); encloser=sencloser->first_char(); + } result++; } return result; @@ -160,9 +135,9 @@ 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); - check_option_param(o.defined, params, 1, - "too many parameters"); + Table::Action_options o=get_action_options(r, params, 1, *source); + if(params.count()>2) + throw Exception(PARSER_RUNTIME, 0, "too many parameters"); GET_SELF(r, VTable).set_table(*new Table(*source, o)); return; } @@ -188,7 +163,7 @@ static void _create(Request& r, MethodPa size_t options_param_index=data_param_index+1; if( options_param_indexsplit(*columns, col_pos_after, *separators.scolumn, String::L_AS_IS); + if(head[0]->is_empty()) + *columns += new String(); + else + head[0]->split(*columns, col_pos_after, *separators.scolumn, String::L_AS_IS); } } @@ -230,9 +208,9 @@ 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 +231,7 @@ 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; @@ -261,55 +239,59 @@ 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; write=read=string; char c; - while((c=*read++)) { + // we are enclosed, searching for second encloser + while(c=*read++) { if(c==encloser) { - char n=*read; - if(n==encloser) // double-encloser stands for encloser - read++; - else if(n==delim1 || n==delim2) { - result.delim=n; + if(*read==encloser) // double-encloser stands for encloser read++; - break; - } + else + break; // note: skipping encloser } - *write++=c; } + // we are no longer enclosed, searching for delimiter, skipping extra enclosers + while(c=*read++) { + if(c==delim1 || c==delim2) { + result.delim=c; + break; + } else if(c!=encloser) + *write++=c; + } *write=0; // terminate *string_ref=c? read: 0; result.piece=string; 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 +308,22 @@ 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; @@ -337,19 +335,21 @@ static void _load(Request& r, MethodPara HashStringValue *options=0; TableSeparators separators; if(options_param_indexcount()) // 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 +389,169 @@ static void _load(Request& r, MethodPara GET_SELF(r, VTable).set_table(table); } -#ifdef NO_STRINGSTREAM +#ifdef USE_STRINGSTREAM +#include "gc_allocator.h" -void maybe_enclose( String& to, const String& from, char encloser, const String* sencloser ) { - if(encloser) { - to<<*sencloser; +typedef std::basic_stringstream, gc_allocator > pa_stringstream; +typedef std::basic_string, gc_allocator > pa_string; + +static void enclose( pa_stringstream& to, const String* from, char encloser ) { + if(from){ + to<pos( encloser, pos_after ))!=STRING_NOT_FOUND; pos_after=pos_before) { pos_before++; // including first encloser (and skipping it for next pos) - to<mid(pos_after, pos_before).cstr(); + to<length(); if(pos_aftermid(pos_after, from_length).cstr(); + to< i(*table.columns()); i.has_next(); ) { + enclose( result, i.next(), separators.encloser ); + if(i.has_next()) + result< i(*table.columns()); i.has_next(); ) { + result<cstr(); + if(i.has_next()) + result<count():0) + for(int column=0; column i(table); + if(separators.encloser){ + while(i.has_next()) { + for(Array_iterator c(*i.next()); c.has_next(); ) { + enclose( result, c.next(), separators.encloser ); + if(c.has_next()) + result< c(*i.next()); c.has_next(); ) { + result<cstr(); + if(c.has_next()) + result<pos( encloser, pos_after ))!=STRING_NOT_FOUND; pos_after=pos_before) { pos_before++; // including first encloser (and skipping it for next pos) - to<mid(pos_after, pos_before); + to<<*sencloser; // doubling encloser } // last piece - size_t from_length=from.length(); + size_t from_length=from->length(); if(pos_aftermid(pos_after, from_length); + to<<*sencloser; + } else { + to<<*sencloser<<*sencloser; + } +} - to< i(*table.columns()); i.has_next(); ) { + enclose( result, i.next(), separators.encloser, separators.sencloser ); + if(i.has_next()) + result<<*separators.scolumn; + } + } else { + for(Array_iterator i(*table.columns()); i.has_next(); ) { + result<<*i.next(); + if(i.has_next()) + result<<*separators.scolumn; + } + } + } else { // nameless table [we were asked to output column names] + if(int lsize=table.count()?table[0]->count():0) + for(int column=0; column i(table); + if(separators.encloser){ + while(i.has_next()) { + for(Array_iterator c(*i.next()); c.has_next(); ) { + enclose( result, c.next(), separators.encloser, separators.sencloser ); + if(c.has_next()) + result<<*separators.scolumn; + } + result.append_know_length("\n", 1, String::L_CLEAN); + } + } else { + while(i.has_next()) { + for(Array_iterator c(*i.next()); c.has_next(); ) { + result<<*c.next(); + if(c.has_next()) + result<<*separators.scolumn; + } + result.append_know_length("\n", 1, String::L_CLEAN); + } + } } +#endif // don't use stringstream -#endif static void _save(Request& r, MethodParams& params) { const String& first_arg=params.as_string(0, FIRST_ARG_MUST_NOT_BE_CODE); @@ -451,24 +569,19 @@ static void _save(Request& r, MethodPara --param_index; const String& file_name=params.as_string(param_index++, FILE_NAME_MUST_NOT_BE_CODE); + String file_spec=r.absolute(file_name); + + if(do_append && file_exist(file_spec)) + output_column_names=false; TableSeparators separators; - if(param_indexcount()) - throw Exception(PARSER_RUNTIME, - 0, - "invalid option passed"); - } else { - throw Exception(PARSER_RUNTIME, - 0, - "additional params must be hash (did you spell mode parameter correctly?)"); - } + if(param_indexcount()) + throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION); } - } + if(param_index i(*table.columns()); i.has_next(); ) { - maybe_enclose( sdata, *i.next(), separators.encloser, separators.sencloser ); - if(i.has_next()) - sdata<<*separators.scolumn; - } - } else { // nameless table [we were asked to output column names] - if(int lsize=table.count()?table[0]->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 ); - if(c.has_next()) - sdata<<*separators.scolumn; - } - sdata.append_know_length("\n", 1, String::L_CLEAN); - } + table_to_csv(sdata, table, separators, output_column_names); // write - { const char* data_cstr=sdata.cstr(); - file_write(r.absolute(file_name), - data_cstr, sdata.length(), true, do_append); + file_write(r.charsets, file_spec, data_cstr, sdata.length(), true /* as text */, do_append); if(*data_cstr) // not empty (when empty it's not heap memory) pa_free((void*)data_cstr); // not needed anymore - } - -#else - - ostringstream ost(stringstream::out); +#endif +} - // process header - if(output_column_names) { - if(table.columns()) { // named table - for(Array_iterator i(*table.columns()); i.has_next(); ) { - maybe_enclose( ost, *i.next(), separators.encloser ); - if(i.has_next()){ - ost<count():0) - for(int column=0; column0 && params[0].is_string()) { + if(params.as_string(0, FIRST_ARG_MUST_NOT_BE_CODE)=="nameless") { + output_column_names=false; + param_index++; + } else { + throw Exception(PARSER_RUNTIME, + 0, + "bad mode (must be nameless)"); } - ost<<'\n'; } - // process data lines - Array_iterator i(table); - while(i.has_next()) { - for(Array_iterator c(*i.next()); c.has_next(); ) { - maybe_enclose( ost, *c.next(), separators.encloser ); - if(c.has_next()) - ost<count()) + throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION); } - ost<<'\n'; - } - // write - { - string data=ost.str(); - const char* data_cstr = data.c_str(); + Table& table=GET_SELF(r, VTable).table(); - file_write(r.absolute(file_name), data_cstr, data.length(), true /* as text */, do_append); - } +#ifdef USE_STRINGSTREAM + pa_stringstream ost(std::stringstream::out); + table_to_csv(ost, table, separators, output_column_names); + + r.write_no_lang(*new VString(*new String(pa_strdup(ost.str().c_str()), String::L_CLEAN))); +#else + String sdata; + + table_to_csv(sdata, table, separators, output_column_names); + + r.write_no_lang(*new VString(*new String(sdata.cstr(), String::L_CLEAN))); #endif } -static void _count(Request& r, MethodParams&) { - int result=GET_SELF(r, VTable).table().count(); +static void _count(Request& r, MethodParams& params) { + Table& table=GET_SELF(r, VTable).table(); + size_t result=0; + if(params.count()) { + const String& param=params.as_string(0, PARAMETER_MUST_BE_STRING); + if(param == "columns") + result = table.columns() ? table.columns()->count() : table.max_cells(); + 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 +697,47 @@ 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; - } + size_t saved_current=table.current(); + size_t size=table.count(); - r.write_pass_lang(sv_processed); + if(delim_maybe_code) { // delimiter set + bool need_delim=false; + for(size_t 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; + } - if(lskip==Request::SKIP_BREAK) - break; + r.write_pass_lang(sv_processed); + + if(lskip==Request::SKIP_BREAK) + break; + } + } else { + for(size_t 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: { @@ -678,9 +781,7 @@ static void table_row_to_hash(Table::ele for(Array_iterator i(*info->value_fields); i.has_next(); ) { size_t value_field=i.next(); if(value_fieldcount()) - hash.put( - *info->table->columns()->get(value_field), - new VString(*row->get(value_field))); + hash.put(*info->table->columns()->get(value_field), new VString(*row->get(value_field))); } exist=info->hash->put_dont_replace(*key, vhash); @@ -701,14 +802,14 @@ static void table_row_to_hash(Table::ele table=new Table(*info->table, table_options/*no rows, just structure*/); info->hash->put(*key, new VTable(table)); } - *table+=row; + Table::element_type row_copy(new ArrayString(row->count())); + row_copy->append(*row); + *table+=row_copy; break; } } if(exist && info->distinct==D_ILLEGAL) - throw Exception(PARSER_RUNTIME, - key, - "duplicate key"); + throw Exception(PARSER_RUNTIME, key, "duplicate key"); } Table2hash_value_type get_value_type(Value& vvalue_type){ @@ -721,14 +822,10 @@ Table2hash_value_type get_value_type(Val } else if (svalue_type == "hash") { return C_HASH; } else { - throw Exception(PARSER_RUNTIME, - &svalue_type, - "must be 'hash', 'table' or 'string'"); + throw Exception(PARSER_RUNTIME, &svalue_type, "must be 'hash', 'table' or 'string'"); } } else { - throw Exception(PARSER_RUNTIME, - 0, - "'type' must be hash"); + throw Exception(PARSER_RUNTIME, 0, "'type' must be string"); } } @@ -741,7 +838,7 @@ static void _hash(Request& r, MethodPara Table2hash_value_type value_type=C_HASH; int param_index=params.count()-1; if(param_index>0) { - if(HashStringValue* options=params.as_no_junction(param_index, PARAM_MUST_NOT_BE_CODE).get_hash()){ // options where specified + if(HashStringValue* options=params.as_no_junction(param_index, PARAM_MUST_NOT_BE_CODE).get_hash()){ // can't use .as_has because the 2nd param could be table so .as_hash throws an error --param_index; int valid_options=0; if(Value* vdistinct_code=options->get(sql_distinct_name)) { // $.distinct ? @@ -753,9 +850,7 @@ static void _hash(Request& r, MethodPara value_type=C_TABLE; distinct=D_FIRST; } else { - throw Exception(PARSER_RUNTIME, - &sdistinct, - "must be 'tables' or true/false"); + throw Exception(PARSER_RUNTIME, &sdistinct, "must be 'tables' or true/false"); } } else { distinct=vdistinct_value.as_bool()?D_FIRST:D_ILLEGAL; @@ -763,41 +858,30 @@ static void _hash(Request& r, MethodPara } if(Value* vvalue_type_code=options->get(sql_value_type_name)) { // $.type ? if(value_type==C_TABLE) // $.distinct[tables] already was specified - throw Exception(PARSER_RUNTIME, - 0, - "you can't specify $.distinct[tables] and $.type[] together"); - + throw Exception(PARSER_RUNTIME, 0, "you can't specify $.distinct[tables] and $.type[] together"); valid_options++; value_type=get_value_type(r.process_to_value(*vvalue_type_code)); } 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 if(value_type==C_STRING) // $.type[string] - throw Exception(PARSER_RUNTIME, - 0, - "you must specify one value field with option $.type[string]"); + throw Exception(PARSER_RUNTIME, 0, "you must specify one value field with option $.type[string]"); for(size_t i=0; icount(); i++) // by all columns, including key value_fields+=i; } else { // list of columns was specified if(value_type==C_TABLE) - throw Exception(PARSER_RUNTIME, - 0, - "you can't specify value field(s) with option $.distinct[tables] or $.type[tables]"); + throw Exception(PARSER_RUNTIME, 0, "you can't specify value field(s) with option $.distinct[tables] or $.type[tables]"); Value& value_fields_param=params.as_no_junction(param_index, "value field(s) must not be code"); if(value_fields_param.is_string()) { // one column as string was specified @@ -808,15 +892,11 @@ static void _hash(Request& r, MethodPara value_fields +=self_table.column_name2index(value_field_name, true); } } else - throw Exception(PARSER_RUNTIME, - 0, - "value field(s) must be string or table"); + throw Exception(PARSER_RUNTIME, 0, "value field(s) must be string or table"); } if(value_type==C_STRING && value_fields.count()!=1) - throw Exception(PARSER_RUNTIME, - 0, - "you can specify only one value field with option $.type[string]"); + throw Exception(PARSER_RUNTIME, 0, "you can specify only one value field with option $.type[string]"); { Value* key_param=¶ms[0]; @@ -831,8 +911,7 @@ static void _hash(Request& r, MethodPara /*row=*/0, value_type }; - info.key_field=(info.key_code?-1 - :self_table.column_name2index(key_param->as_string(), true)); + info.key_field=(info.key_code ? -1 : self_table.column_name2index(key_param->as_string(), true)); int saved_current=self_table.current(); self_table.for_each(table_row_to_hash, &info); @@ -890,18 +969,25 @@ static void _sort(Request& r, MethodPara old_table.set_current(i); // calculate key value seq[i].row=old_table[i]; - Value& value=r.process_to_value(key_maker).as_expr_result(true/*return string as-is*/); + Value& value=r.process_to_value(key_maker); if(i==0) // determining key values type by first one key_values_are_strings=value.is_string(); if(key_values_are_strings) seq[i].value.c_str=value.as_string().cstr(); else - seq[i].value.d=value.as_double(); + seq[i].value.d=value.as_expr_result().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"); - Value& expression_code=params.as_junction(0, "must be expression"); +static bool _locate_expression(Table& table, Request& r, MethodParams& params) { + Value& expression_code=params.as_junction(0, "must be expression"); + Table::Action_options o=get_action_options(r, params, 1, table); + if(params.count()>2) + throw Exception(PARSER_RUNTIME, 0, "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); + Table::Action_options o=get_action_options(r, params, 2, table); 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)); + bool result=params[0].get_junction() || (params.count() == 1) ? + _locate_expression(table, r, params) : + _locate_name_value(table, r, params); + r.write_no_lang(VBool::get(result)); } @@ -955,7 +1040,7 @@ 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.max_cells()) for(size_t column=0; column3?¶ms[3]:0; + + Table& table=GET_SELF(r, VTable).table(); + size_t saved_current=table.current(); + size_t size=table.count(); + + const String* rownum_var_name=rownum_name.is_empty()? 0 : &rownum_name; + const String* value_var_name=value_name.is_empty()? 0 : &value_name; + + Value* var_context=r.get_method_frame()->caller(); + + if(delim_maybe_code) { // delimiter set + bool need_delim=false; + for(size_t 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); + + if(lskip==Request::SKIP_BREAK) + break; + } + } else { + for(size_t row=0; rowput_element(String(aname, String::L_CLEAN), avalue); // new not required +} + +inline Table::element_type row_from_string(Request& r, Value ¶m){ + if(!param.is_string() && !param.get_junction()) + throw Exception(PARSER_RUNTIME, 0, "row must be string, code or hash"); + Temp_lang temp_lang(r, String::L_PASS_APPENDED); - const String& string=r.process_to_string(params.as_junction(0, "body must be code")); + const String& string=r.process_to_string(param); // parse cells Table::element_type row=new ArrayString; size_t pos_after=0; string.split(*row, pos_after, "\t", String::L_AS_IS); - GET_SELF(r, VTable).table()+=row; + return row; +} + +static void _append(Request& r, MethodParams& params) { + VTable vtable=GET_SELF(r, VTable); + Table& table=vtable.table(); + + HashStringValue* hash=params[0].get_hash(); + if(hash){ + table+=new ArrayString(); + size_t saved_current=table.current(); + table.set_current(table.count()-1); + hash->for_each(update_cell, &vtable); + table.set_current(saved_current); + } else { + table+=row_from_string(r, params[0]); + } +} + +static void _insert(Request& r, MethodParams& params) { + VTable vtable=GET_SELF(r, VTable); + Table& table=vtable.table(); + HashStringValue* hash=params[0].get_hash(); + if(hash){ + table.insert(table.current(), new ArrayString()); + hash->for_each(update_cell, &vtable); + } else { + table.insert(table.current(), row_from_string(r, params[0])); + } +} + +static void _delete(Request& r, MethodParams&) { + Table& table=GET_SELF(r, VTable).table(); + table.remove_current(); } static void join_named_row(Table& src, Table* dest) { @@ -995,16 +1183,9 @@ static void join_nameless_row(Table& src *dest+=src[src.current()]; } static void _join(Request& r, MethodParams& params) { - Table* maybe_src=params.as_no_junction(0, "table ref must not be code").get_table(); - if(!maybe_src) - throw Exception(PARSER_RUNTIME, - 0, - "source is not a table"); - Table& src=*maybe_src; + Table& src=*params.as_table(0, "source"); - Table::Action_options o=get_action_options(r, params, src); - check_option_param(o.defined, params, 1, - "invalid extra parameter"); + Table::Action_options o=get_action_options(r, params, 1, src); Table& dest=GET_SELF(r, VTable).table(); if(&src == &dest) @@ -1030,9 +1211,9 @@ public: columns(*new ArrayString), row(0), table(0) { } - bool add_column(SQL_Error& error, const char *str, size_t length) { + bool add_column(SQL_Error& error, const char *str, size_t ) { try { - columns+=new String(str, length, true); + columns+=new String(str, String::L_TAINTED /* no length as 0x00 can be inside */); return false; } catch(...) { error=SQL_Error("exception occured in Table_sql_event_handlers::add_column"); @@ -1058,12 +1239,9 @@ public: return true; } } - 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 ) { try { - String& cell=*new String; - if(length) - cell.append_know_length(str, length, String::L_TAINTED); - *row+=&cell; + *row+=str?new String(str, String::L_TAINTED /* no length as 0x00 can be inside */):&String::Empty; return false; } catch(...) { error=SQL_Error("exception occured in Table_sql_event_handlers::add_row_cell"); @@ -1080,7 +1258,7 @@ static void marshal_bind( { SQL_Driver::Placeholder& ph=**pptr; ph.name=aname.cstr(); - ph.value=avalue->as_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; @@ -1090,7 +1268,7 @@ static void marshal_bind( // not static, used elsewhere int marshal_binds(HashStringValue& hash, SQL_Driver::Placeholder*& placeholders) { int hash_count=hash.count(); - placeholders=new(UseGC) SQL_Driver::Placeholder[hash_count]; + placeholders=new SQL_Driver::Placeholder[hash_count]; SQL_Driver::Placeholder* ptr=placeholders; hash.for_each(marshal_bind, &ptr); return hash_count; @@ -1103,12 +1281,9 @@ 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); } } @@ -1117,34 +1292,26 @@ 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"); - if(voptions.is_defined() && !voptions.is_string()) - if(HashStringValue* 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(valid_options!=options->count()) - throw Exception(PARSER_RUNTIME, - 0, - "called with invalid option"); - } else - throw Exception(PARSER_RUNTIME, - 0, - "options must be hash"); - } + 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_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(valid_options!=options->count()) + throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION); + } SQL_Driver::Placeholder* placeholders=0; uint placeholders_count=0; @@ -1153,14 +1320,10 @@ 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]; - //measure:before - gettimeofday(&mt[0],NULL); -#endif + r.connection()->query( statement_cstr, placeholders_count, placeholders, @@ -1168,17 +1331,6 @@ static void _sql(Request& r, MethodParam handlers, statement_string); -#ifdef RESOURCES_DEBUG - //measure:after connect - gettimeofday(&mt[1],NULL); - - double t[2]; - for(int i=0;i<2;i++) - t[i]=mt[i].tv_sec+mt[i].tv_usec/1000000.0; - - r.sql_request_time+=t[1]-t[0]; -#endif - if(bind) unmarshal_bind_updates(*bind, placeholders_count, placeholders); @@ -1217,20 +1369,61 @@ static void _select(Request& r, MethodPa Value& vcondition=params.as_expression(0, "condition must be number, bool or expression"); Table& source_table=GET_SELF(r, VTable).table(); + + int limit=source_table.count(); + int offset=0; + bool reverse=false; + + if(params.count()>1) + if(HashStringValue* options=params.as_hash(1)) { + 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); + } + Table& result_table=*new Table(source_table.columns()); - int saved_current=source_table.current(); - int size=source_table.count(); - for(int row=0; row0 && offset>=0 && (size_t)offset (size_t)offset) // ...condition is true, adding to the result + result_table+=source_table[row]; + if(row==0) break; + } + } else { + for(size_t row=0; row < size && result_table.count() < (size_t)limit; row++) { + source_table.set_current(row); - bool condition=r.process_to_value(vcondition, - false/*don't intercept string*/).as_bool(); + bool condition=r.process_to_value(vcondition, false/*don't intercept string*/).as_bool(); - if(condition) // ...condition is true= - result_table+=source_table[row]; // =green light to go to result + if(condition && ++appended > (size_t)offset) // ...condition is true, adding to the result + result_table+=source_table[row]; + } + } + source_table.set_current(saved_current); } - source_table.set_current(saved_current); r.write_no_lang(*new VTable(&result_table)); } @@ -1255,8 +1448,16 @@ MTable::MTable(): Methoded("table") { // add_native_method("save_old", Method::CT_DYNAMIC, _save_old, 1, 3); + // ^table.csv-string[] + // ^table.csv-string[nameless] + // ^table.csv-string[nameless;$.encloser["] $.separator[,]] + add_native_method("csv-string", Method::CT_DYNAMIC, _csv_string, 0, 2); + // ^table.count[] - add_native_method("count", Method::CT_DYNAMIC, _count, 0, 0); + // ^table.count[rows] + // ^table.count[columns] + // ^table.count[cells] + add_native_method("count", Method::CT_DYNAMIC, _count, 0, 1); // ^table.line[] add_native_method("line", Method::CT_DYNAMIC, _line, 0, 0); @@ -1284,9 +1485,19 @@ MTable::MTable(): Methoded("table") { // ^table.flip[] add_native_method("flip", Method::CT_DYNAMIC, _flip, 0, 0); - // ^table.append{r{tab}e{tab}c{tab}o{tab}r{tab}d} + // ^table.foreach[row-num;value]{code} + // ^table.foreach[row-num;value]{code}[delim] + add_native_method("foreach", Method::CT_DYNAMIC, _foreach, 3, 4); + + // ^table.append{row{tab}data} add_native_method("append", Method::CT_DYNAMIC, _append, 1, 1); + // ^table.insert{row{tab}data} before current row + add_native_method("insert", Method::CT_DYNAMIC, _insert, 1, 1); + + // ^table.delete[] current row + add_native_method("delete", Method::CT_DYNAMIC, _delete, 0, 0); + // ^table.join[table][$.limit(10) $.offset(1) $.offset[cur] ] add_native_method("join", Method::CT_DYNAMIC, _join, 1, 2); @@ -1299,5 +1510,5 @@ 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); }