--- parser3/src/classes/hash.C 2012/02/28 05:07:44 1.112 +++ parser3/src/classes/hash.C 2016/10/26 15:44:49 1.136 @@ -1,16 +1,15 @@ /** @file Parser: @b hash parser class. - Copyright (c) 2001-2009 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="$Id: hash.C,v 1.112 2012/02/28 05:07:44 moko Exp $"; - #include "classes.h" #include "pa_vmethod_frame.h" #include "pa_request.h" +#include "pa_charsets.h" #include "pa_vhash.h" #include "pa_vvoid.h" #include "pa_sql_connection.h" @@ -18,6 +17,8 @@ static const char * const IDENT="$Id: ha #include "pa_vbool.h" #include "pa_vmethod_frame.h" +volatile const char * IDENT_HASH_C="$Id: hash.C,v 1.136 2016/10/26 15:44:49 moko Exp $"; + // class class MHash: public Methoded { @@ -30,7 +31,7 @@ public: // global variable -DECLARE_CLASS_VAR(hash, new MHash, 0); +DECLARE_CLASS_VAR(hash, new MHash); // methods @@ -107,9 +108,9 @@ public: return false; } - bool add_row_cell(SQL_Error& error, const char *ptr, size_t ) { + bool add_row_cell(SQL_Error& error, const char *str, size_t ) { try { - String& cell=*new String(ptr, String::L_TAINTED /* no length as 0x00 can be inside */); + const String& cell=str?*new String(str, String::L_TAINTED /* no length as 0x00 can be inside */):String::Empty; bool duplicate=false; if(one_bool_column) { @@ -192,24 +193,30 @@ VBool Hash_sql_event_handlers::only_one_ static void _create_or_add(Request& r, MethodParams& params) { if(params.count()) { Value& vsrc=params.as_no_junction(0, PARAM_MUST_BE_HASH); - if(HashStringValue* src=vsrc.get_hash()) { - VHash& self=GET_SELF(r, VHash); - HashStringValue* self_hash=&(self.hash()); - if(src==self_hash) // same: doing nothing + VHashBase& self=GET_SELF(r, VHashBase); + HashStringValue* self_hash=&(self.hash()); + HashStringValue* src_hash; + + if(VHashBase* src=static_cast(vsrc.as(VHASH_TYPE))) { + src_hash=&(src->hash()); + + if(src_hash==self_hash) // same: doing nothing return; - src->for_each(copy_all_overwrite_to, self_hash); - if(VHash* vhash_src=static_cast(vsrc.as(VHASH_TYPE))) - if(Value* vdefault=vhash_src->get_default()) - if(vdefault->is_defined()) - self.set_default(vdefault); + if(Value* vdefault=src->get_default()) + self.set_default(vdefault); + } else { + src_hash=vsrc.get_hash(); } + + if(src_hash) + src_hash->for_each(copy_all_overwrite_to, self_hash); } } static void _sub(Request& r, MethodParams& params) { - if(HashStringValue* src=params.as_no_junction(0, PARAM_MUST_BE_HASH).get_hash()) { - HashStringValue* self=&(GET_SELF(r, VHash).hash()); + if(HashStringValue* src=params.as_hash(0, "param")) { + HashStringValue* self=&(GET_SELF(r, VHashBase).hash()); if(src==self) { // same: clearing self->clear(); return; @@ -218,17 +225,14 @@ static void _sub(Request& r, MethodParam } } -static void copy_all_dontoverwrite_to( - HashStringValue::key_type key, - HashStringValue::value_type value, - HashStringValue* dest) { +static void copy_all_dontoverwrite_to(HashStringValue::key_type key, HashStringValue::value_type value, HashStringValue* dest) { dest->put_dont_replace(key, value); } static void _union(Request& r, MethodParams& params) { // dest = copy of self - Value& result=*new VHash(GET_SELF(r, VHash).hash()); + Value& result=*new VHash(GET_SELF(r, VHashBase).hash()); // dest += b - if(HashStringValue* src=params.as_no_junction(0, PARAM_MUST_BE_HASH).get_hash()) + if(HashStringValue* src=params.as_hash(0, "param")) src->for_each(copy_all_dontoverwrite_to, result.get_hash()); // return result @@ -241,37 +245,37 @@ struct Copy_intersection_to_info { HashStringValue* dest; }; #endif -static void copy_intersection_to( - HashStringValue::key_type key, - HashStringValue::value_type value, - Copy_intersection_to_info *info) { +static void copy_intersection_to(HashStringValue::key_type key, HashStringValue::value_type value, Copy_intersection_to_info *info) { if(info->b->get(key)) info->dest->put_dont_replace(key, value); } static void _intersection(Request& r, MethodParams& params) { Value& result=*new VHash; // dest += b - if(HashStringValue* b=params.as_no_junction(0, PARAM_MUST_BE_HASH).get_hash()) { + if(HashStringValue* b=params.as_hash(0, "param")) { Copy_intersection_to_info info={b, result.get_hash()}; - GET_SELF(r, VHash).hash().for_each(copy_intersection_to, &info); + GET_SELF(r, VHashBase).hash().for_each(copy_intersection_to, &info); } // return result r.write_no_lang(result); } -static bool intersects( - HashStringValue::key_type key, - HashStringValue::value_type /*value*/, - HashStringValue* b) { +static bool intersects( HashStringValue::key_type key, HashStringValue::value_type /*value*/, HashStringValue* b) { return b->get(key)!=0; } static void _intersects(Request& r, MethodParams& params) { bool result=false; - if(HashStringValue* b=params.as_no_junction(0, PARAM_MUST_BE_HASH).get_hash()) - result=GET_SELF(r, VHash).hash().first_that(intersects, b)!=0; + if(HashStringValue* b=params.as_hash(0, "param")) { + HashStringValue* self=&(GET_SELF(r, VHashBase).hash()); + if(b==self) { + r.write_no_lang(VBool::get(true)); + return; + } + result=self->first_that(intersects, b)!=0; + } // return result r.write_no_lang(VBool::get(result)); @@ -297,7 +301,7 @@ static void _sql(Request& r, MethodParam bool distinct=false; Table2hash_value_type value_type=C_HASH; if(params.count()>1) - if(HashStringValue* options=params.as_hash(1)) { + if(HashStringValue* options=params.as_hash(1, "sql options")) { int valid_options=0; if(Value* vbind=options->get(sql_bind_name)) { valid_options++; @@ -305,19 +309,19 @@ static void _sql(Request& r, MethodParam } if(Value* vlimit=options->get(sql_limit_name)) { valid_options++; - limit=(ulong)r.process_to_value(*vlimit).as_double(); + limit=(ulong)r.process(*vlimit).as_double(); } if(Value* voffset=options->get(sql_offset_name)) { valid_options++; - offset=(ulong)r.process_to_value(*voffset).as_double(); + offset=(ulong)r.process(*voffset).as_double(); } if(Value* vdistinct=options->get(sql_distinct_name)) { valid_options++; - distinct=r.process_to_value(*vdistinct).as_bool(); + distinct=r.process(*vdistinct).as_bool(); } if(Value* vvalue_type=options->get(sql_value_type_name)) { valid_options++; - value_type=get_value_type(r.process_to_value(*vvalue_type)); + value_type=get_value_type(r.process(*vvalue_type)); } if(valid_options!=options->count()) throw Exception(PARSER_RUNTIME, 0, CALLED_WITH_INVALID_OPTION); @@ -328,11 +332,10 @@ static void _sql(Request& r, MethodParam if(bind) placeholders_count=marshal_binds(*bind, placeholders); - Temp_lang temp_lang(r, String::L_SQL); const String& statement_string=r.process_to_string(statement); - const char* statement_cstr=statement_string.untaint_cstr(r.flang, r.connection()); + const char* statement_cstr=statement_string.untaint_cstr(String::L_SQL, r.connection()); - HashStringValue& hash=GET_SELF(r, VHash).hash(); + HashStringValue& hash=GET_SELF(r, VHashBase).hash(); hash.clear(); Hash_sql_event_handlers handlers( statement_string, statement_cstr, @@ -351,10 +354,7 @@ static void _sql(Request& r, MethodParam unmarshal_bind_updates(*bind, placeholders_count, placeholders); } -static void keys_collector( - HashStringValue::key_type key, - HashStringValue::value_type, - Table *table) { +static void keys_collector(HashStringValue::key_type key, HashStringValue::value_type, Table *table) { Table::element_type row(new ArrayString(1)); *row+=new String(key, String::L_TAINTED); *table+=row; @@ -370,98 +370,207 @@ static void _keys(Request& r, MethodPara *columns+=keys_column_name; Table* table=new Table(columns); - GET_SELF(r, VHash).hash().for_each(keys_collector, table); + GET_SELF(r, VHashBase).hash().for_each(keys_collector, table); r.write_no_lang(*new VTable(table)); } static void _count(Request& r, MethodParams&) { - r.write_no_lang(*new VInt(GET_SELF(r, VHash).hash().count())); + r.write_no_lang(*new VInt(GET_SELF(r, VHashBase).hash().count())); } static void _delete(Request& r, MethodParams& params) { - - GET_SELF(r, VHash).hash().remove(params.as_string(0, "key must be string")); + if(params.count()>0) + GET_SELF(r, VHashBase).hash().remove(params.as_string(0, "key must be string")); + else + GET_SELF(r, VHashBase).hash().clear(); } static void _contains(Request& r, MethodParams& params) { - bool result=GET_SELF(r, VHash).hash().contains(params.as_string(0, "key must be string")); + VHashBase& self=GET_SELF(r, VHashBase); + const String& key_name=params.as_string(0, "key must be string"); + bool result=SYMBOLS_EQ(key_name,_DEFAULT_SYMBOL) ? (self.get_default() != 0) : self.hash().contains(key_name); r.write_no_lang(VBool::get(result)); } -#ifndef DOXYGEN -struct Foreach_info { - Request *r; - const String* key_var_name; - const String* value_var_name; - Value* body_code; - Value* delim_maybe_code; +static void _foreach(Request& r, MethodParams& params) { + InCycle temp(r); - Value* var_context; - bool need_delim; -}; -#endif -static bool one_foreach_cycle( - HashStringValue::key_type akey, - HashStringValue::value_type avalue, - Foreach_info *info) { - Value& var_context=*info->var_context; - if(info->key_var_name){ - VString* vkey=new VString(*new String(akey, String::L_TAINTED)); - var_context.put_element(*info->key_var_name, vkey, false); - } - if(info->value_var_name) - var_context.put_element(*info->value_var_name, avalue, false); - - if(info->delim_maybe_code){ // delimiter set - StringOrValue sv_processed=info->r->process(*info->body_code); - Request::Skip lskip=info->r->get_skip(); info->r->set_skip(Request::SKIP_NOTHING); - - const String* s_processed=sv_processed.get_string(); - if(s_processed && !s_processed->is_empty()) { // we have body - if(info->need_delim) // need delim & iteration produced string? - info->r->write_pass_lang(info->r->process(*info->delim_maybe_code)); - else - info->need_delim=true; + const String* key_var_name=¶ms.as_string(0, "key-var name must be string"); + const String* value_var_name=¶ms.as_string(1, "value-var name must be string"); + Value* body_code=¶ms.as_junction(2, "body must be code"); + Value* delim_maybe_code=params.count()>3?¶ms[3]:0; + Value& caller=*r.get_method_frame()->caller(); + + if(key_var_name->is_empty()) key_var_name=0; + if(value_var_name->is_empty()) value_var_name=0; + + HashStringValue& hash=GET_SELF(r, VHashBase).hash(); + + if(delim_maybe_code){ // delimiter set + bool need_delim=false;; + for(HashStringValue::Iterator i(hash); i; i.next()){ + if(key_var_name){ + VString* vkey=new VString(*new String(i.key(), String::L_TAINTED)); + r.put_element(caller, *key_var_name, vkey); + } + + if(value_var_name) + r.put_element(caller, *value_var_name, i.value()); + + Value& sv_processed=r.process(*body_code); + Request::Skip lskip=r.get_skip(); r.set_skip(Request::SKIP_NOTHING); + + const String* s_processed=sv_processed.get_string(); + if(s_processed && !s_processed->is_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; } - info->r->write_pass_lang(sv_processed); - return lskip==Request::SKIP_BREAK; } else { - info->r->process_write(*info->body_code); - Request::Skip lskip=info->r->get_skip(); info->r->set_skip(Request::SKIP_NOTHING); - return lskip==Request::SKIP_BREAK; + for(HashStringValue::Iterator i(hash); i; i.next()){ + if(key_var_name){ + VString* vkey=new VString(*new String(i.key(), String::L_TAINTED)); + r.put_element(caller, *key_var_name, vkey); + } + + if(value_var_name) + r.put_element(caller, *value_var_name, i.value()); + + r.process_write(*body_code); + Request::Skip lskip=r.get_skip(); r.set_skip(Request::SKIP_NOTHING); + + if(lskip==Request::SKIP_BREAK) + break; + } } } -static void _foreach(Request& r, MethodParams& params) { - InCycle temp(r); +enum AtResultType { + AtResultTypeValue = 0, + AtResultTypeKey = 1, + AtResultTypeHash = 2 +}; + +inline Value& SingleElementHash(String::Body akey, Value* avalue) { + Value& result=*new VHash; + result.put_element(*new String(akey, String::L_TAINTED), avalue); + return result; +} + +#ifndef DOXYGEN +struct Hash_seq_item { + HashStringValue::Pair *hash_pair; + union { + const char *c_str; + double d; + } value; +}; +#endif +static int sort_cmp_string(const void *a, const void *b) { + return strcmp( + static_cast(a)->value.c_str, + static_cast(b)->value.c_str + ); +} +static int sort_cmp_double(const void *a, const void *b) { + double va=static_cast(a)->value.d; + double vb=static_cast(b)->value.d; + if(vavb) + return +1; + else + return 0; +} +static void _sort(Request& r, MethodParams& params){ +#ifdef HASH_ORDER const String& key_var_name=params.as_string(0, "key-var name must be string"); const String& value_var_name=params.as_string(1, "value-var name must be string"); + Value& key_maker=params.as_junction(2, "key-maker must be code"); + bool reverse=params.count()>3/*..[desc|asc|]*/? + reverse=params.as_no_junction(3, "order must not be code").as_string()=="desc": + false; // default=asc + + const String* key_var=key_var_name.is_empty()? 0 : &key_var_name; + const String* value_var=value_var_name.is_empty()? 0 : &value_var_name; + VMethodFrame* context=r.get_method_frame()->caller(); + + HashStringValue& hash=GET_SELF(r, VHashBase).hash(); + int count=hash.count(); + + Hash_seq_item* seq=new(PointerFreeGC) Hash_seq_item[count]; + int pos=0; + bool key_values_are_strings=true; + + for(HashStringValue::Iterator i(hash); i; i.next(), pos++ ){ + if(key_var) + r.put_element(*context, *key_var, new VString(*new String(i.key(), String::L_TAINTED))); + if(value_var) + r.put_element(*context, *value_var, i.value()); + + Value& value=r.process(key_maker); + if(pos==0) // determining key values type by first one + key_values_are_strings=value.is_string(); + + seq[pos].hash_pair=i.pair(); + if(key_values_are_strings) + seq[pos].value.c_str=value.as_string().cstr(); + else + seq[pos].value.d=value.as_expr_result().as_double(); + } + + // @todo: handle this elsewhere + if(r.charsets.source().NAME()=="KOI8-R" && key_values_are_strings) + for(pos=0; pos=0; pos--) + hash.order_next(seq[pos].hash_pair); + else + for(pos=0; pos3?params.get(3):0, - /*var_context*/r.get_method_frame()->caller(), - false - }; - - VHash& self=GET_SELF(r, VHash); - HashStringValue& hash=self.hash(); - VHash_lock lock(self); - hash.first_that(one_foreach_cycle, &info); + delete[] seq; +#endif } static void _at(Request& r, MethodParams& params) { - HashStringValue& hash=GET_SELF(r, VHash).hash(); + HashStringValue& hash=GET_SELF(r, VHashBase).hash(); size_t count=hash.count(); int pos=0; - Value& vwhence=*params.get(0); - if(vwhence.is_string()){ + // misha@ + // I do not like that type is checked before whence. + // But I do not like the idea to move it after whence (where process can be called) even more. + AtResultType result_type=AtResultTypeValue; + if(params.count() > 1) { + const String& stype=params.as_string(1, "type must be string"); + if(stype == "key") + result_type=AtResultTypeKey; + else if(stype == "hash") + result_type=AtResultTypeHash; + else if(stype != "value") + throw Exception(PARSER_RUNTIME, &stype, "type must be 'key', 'value' or 'hash'"); + } + + Value& vwhence=params[0]; + if(vwhence.is_string()) { const String& swhence=*vwhence.get_string(); if(swhence == "last") pos=count-1; @@ -470,24 +579,69 @@ static void _at(Request& r, MethodParams &swhence, "whence must be 'first', 'last' or expression"); } else { - pos=r.process_to_value(vwhence).as_int(); + pos=r.process(vwhence).as_int(); if(pos < 0) pos+=count; } if(count && pos >= 0 && (size_t)pos < count){ - if(pos == 0) - r.write_assign_lang(*hash.first_value()); - else if((size_t)pos == count-1) - r.write_assign_lang(*hash.last_value()); - else - for(HashStringValue::Iterator i(hash); i; i.next(), pos-- ) - if(!pos){ - r.write_assign_lang(*i.value()); + switch(result_type) { + case AtResultTypeKey: + { +#ifdef HASH_ORDER + if(pos == 0) { + r.write_pass_lang(*new VString(*new String(hash.first_key(), String::L_TAINTED))); + } else if((size_t)pos == count-1) { + r.write_pass_lang(*new VString(*new String(hash.last_key(), String::L_TAINTED))); + } else +#endif + { + for(HashStringValue::Iterator i(hash); i; i.next(), pos-- ) + if(!pos){ + r.write_pass_lang(*new VString(*new String(i.key(), String::L_TAINTED))); + break; + } + } break; } + case AtResultTypeValue: + { +#ifdef HASH_ORDER + if(pos == 0) { + r.write_pass_lang(*hash.first_value()); + } else if((size_t)pos == count-1) { + r.write_pass_lang(*hash.last_value()); + } else +#endif + { + for(HashStringValue::Iterator i(hash); i; i.next(), pos-- ) + if(!pos){ + r.write_pass_lang(*i.value()); + break; + } + } + break; + } + case AtResultTypeHash: + { +#ifdef HASH_ORDER + if(pos == 0) { + r.write_no_lang(SingleElementHash(hash.first_key(), hash.first_value())); + } else if((size_t)pos == count-1) { + r.write_no_lang(SingleElementHash(hash.last_key(), hash.last_value())); + } else +#endif + { + for(HashStringValue::Iterator i(hash); i; i.next(), pos-- ) + if(!pos){ + r.write_no_lang(SingleElementHash(i.key(), i.value())); + break; + } + } + break; + } + } } - } // constructor @@ -508,7 +662,7 @@ MHash::MHash(): Methoded("hash") add_native_method("intersects", Method::CT_DYNAMIC, _intersects, 1, 1); // ^a.delete[key] - add_native_method("delete", Method::CT_DYNAMIC, _delete, 1, 1); + add_native_method("delete", Method::CT_DYNAMIC, _delete, 0, 1); // ^a.contains[key] add_native_method("contains", Method::CT_DYNAMIC, _contains, 1, 1); @@ -519,15 +673,27 @@ MHash::MHash(): Methoded("hash") add_native_method("sql", Method::CT_DYNAMIC, _sql, 1, 2); // ^hash._keys[[column name]] - add_native_method("_keys", Method::CT_DYNAMIC, _keys, 0, 1); + add_native_method("_keys", Method::CT_DYNAMIC, _keys, 0, 1); // ^hash._count[] - add_native_method("_count", Method::CT_DYNAMIC, _count, 0, 0); + add_native_method("_count", Method::CT_DYNAMIC, _count, 0, 0); // ^hash.foreach[key;value]{code}[delim] add_native_method("foreach", Method::CT_DYNAMIC, _foreach, 2+1, 2+1+1); - // ^hash._at[first|last] - // ^hash._at([-]offset) - add_native_method("_at", Method::CT_DYNAMIC, _at, 1, 1); + // ^hash.sort[key;value]{string-key-maker}[[asc|desc]] + // ^hash.sort[key;value](numeric-key-maker)[[asc|desc]] + add_native_method("sort", Method::CT_DYNAMIC, _sort, 2+1, 2+1+1); + + // ^hash._at[first|last[;'key'|'value'|'hash']] + // ^hash._at([-+]offset)[['key'|'value'|'hash']] + add_native_method("_at", Method::CT_DYNAMIC, _at, 1, 2); + +#ifdef FEATURE_GET_ELEMENT4CALL + // aliases without "_" + add_native_method("keys", Method::CT_DYNAMIC, _keys, 0, 1); + add_native_method("count", Method::CT_DYNAMIC, _count, 0, 0); + add_native_method("at", Method::CT_DYNAMIC, _at, 1, 2); +#endif + }