8
0
mirror of https://github.com/FirebirdSQL/firebird.git synced 2025-01-27 06:43:04 +01:00
firebird-mirror/src/jrd/par.cpp
2014-03-26 19:10:08 +00:00

1732 lines
45 KiB
C++

/*
* PROGRAM: JRD Access Method
* MODULE: par.cpp
* DESCRIPTION: BLR Parser
*
* The contents of this file are subject to the Interbase Public
* License Version 1.0 (the "License"); you may not use this file
* except in compliance with the License. You may obtain a copy
* of the License at http://www.Inprise.com/IPL.html
*
* Software distributed under the License is distributed on an
* "AS IS" basis, WITHOUT WARRANTY OF ANY KIND, either express
* or implied. See the License for the specific language governing
* rights and limitations under the License.
*
* The Original Code was created by Inprise Corporation
* and its predecessors. Portions created by Inprise Corporation are
* Copyright (C) Inprise Corporation.
*
* All Rights Reserved.
* Contributor(s): ______________________________________.
*
* 27-May-2001 Claudio Valderrama: par_plan() no longer uppercases
* an index's name before doing a lookup of such index.
* 2001.07.28: Added parse code for blr_skip to support LIMIT.
* 2002.09.28 Dmitry Yemanov: Reworked internal_info stuff, enhanced
* exception handling in SPs/triggers,
* implemented ROWS_AFFECTED system variable
* 2002.10.21 Nickolay Samofatov: Added support for explicit pessimistic locks
* 2002.10.28 Sean Leyne - Code cleanup, removed obsolete "MPEXL" port
* 2002.10.29 Mike Nordell - Fixed breakage.
* 2002.10.29 Nickolay Samofatov: Added support for savepoints
* 2002.10.29 Sean Leyne - Removed obsolete "Netware" port
* 2003.10.05 Dmitry Yemanov: Added support for explicit cursors in PSQL
* 2004.01.16 Vlad Horsun: Added support for default parameters
* Adriano dos Santos Fernandes
*/
#include "firebird.h"
#include <stdio.h>
#include <string.h>
#include "../jrd/jrd.h"
#include "../jrd/ibase.h"
#include "../jrd/ini.h"
#include "../jrd/val.h"
#include "../jrd/align.h"
#include "../jrd/exe.h"
#include "../jrd/extds/ExtDS.h"
#include "../jrd/lls.h"
#include "../jrd/scl.h"
#include "../jrd/req.h"
#include "../jrd/blb.h"
#include "../jrd/intl.h"
#include "../jrd/met.h"
#include "../jrd/cmp_proto.h"
#include "../jrd/cvt_proto.h"
#include "../jrd/err_proto.h"
#include "../jrd/fun_proto.h"
#include "../yvalve/gds_proto.h"
#include "../jrd/met_proto.h"
#include "../jrd/par_proto.h"
#include "../common/utils_proto.h"
#include "../jrd/RecordSourceNodes.h"
#include "../jrd/SysFunction.h"
#include "../common/classes/BlrReader.h"
#include "../jrd/Function.h"
#include "../jrd/Attachment.h"
#include "../dsql/BoolNodes.h"
#include "../dsql/ExprNodes.h"
#include "../dsql/StmtNodes.h"
using namespace Jrd;
using namespace Firebird;
#include "gen/codetext.h"
static NodeParseFunc blr_parsers[256] = {NULL};
static CompilerScratch* par_start(thread_db* tdbb, jrd_rel* relation, CompilerScratch* view_csb,
CompilerScratch** csb_ptr, const bool trigger, USHORT flags);
static void par_error(BlrReader& blrReader, const Arg::StatusVector& v, bool isSyntaxError = true);
static PlanNode* par_plan(thread_db*, CompilerScratch*);
static void getBlrVersion(CompilerScratch* csb);
namespace
{
class FetchNode
{
public:
// Parse a FETCH statement, and map it into FOR x IN relation WITH x.DBKEY EQ value ...
static DmlNode* parse(thread_db* tdbb, MemoryPool& pool, CompilerScratch* csb, UCHAR /*blrOp*/)
{
ForNode* forNode = FB_NEW(pool) ForNode(pool);
// Fake RseNode.
RseNode* rse = forNode->rse = FB_NEW(*tdbb->getDefaultPool()) RseNode(
*tdbb->getDefaultPool());
DmlNode* relationNode = PAR_parse_node(tdbb, csb);
if (relationNode->kind != DmlNode::KIND_REC_SOURCE)
PAR_syntax_error(csb, "TABLE");
RelationSourceNode* relationSource = static_cast<RelationSourceNode*>(relationNode);
if (relationSource->type != RelationSourceNode::TYPE)
PAR_syntax_error(csb, "TABLE");
rse->rse_relations.add(relationSource);
// Fake boolean.
ComparativeBoolNode* booleanNode = FB_NEW(csb->csb_pool) ComparativeBoolNode(
csb->csb_pool, blr_eql);
rse->rse_boolean = booleanNode;
booleanNode->arg2 = PAR_parse_value(tdbb, csb);
RecordKeyNode* dbKeyNode = FB_NEW(csb->csb_pool) RecordKeyNode(csb->csb_pool, blr_dbkey);
dbKeyNode->recStream = relationSource->getStream();
booleanNode->arg1 = dbKeyNode;
// Pick up statement.
forNode->statement = PAR_parse_stmt(tdbb, csb);
return forNode;
}
};
static RegisterNode<FetchNode> regFetch(blr_fetch);
} // namespace
// Common start for PAR_blr and PAR_preparsed_node. Returns the possibly created csb.
static CompilerScratch* par_start(thread_db* tdbb, jrd_rel* relation, CompilerScratch* view_csb,
CompilerScratch** csb_ptr, const bool trigger, USHORT flags)
{
SET_TDBB(tdbb);
CompilerScratch* csb;
if (!(csb_ptr && (csb = *csb_ptr)))
{
size_t count = 5;
if (view_csb)
count += view_csb->csb_rpt.getCapacity();
csb = CompilerScratch::newCsb(*tdbb->getDefaultPool(), count);
csb->csb_g_flags |= flags;
}
// If there is a request ptr, this is a trigger. Set up contexts 0 and 1 for
// the target relation
if (trigger)
{
StreamType stream = csb->nextStream();
CompilerScratch::csb_repeat* t1 = CMP_csb_element(csb, 0);
t1->csb_flags |= csb_used | csb_active | csb_trigger;
t1->csb_relation = relation;
t1->csb_stream = stream;
stream = csb->nextStream();
t1 = CMP_csb_element(csb, 1);
t1->csb_flags |= csb_used | csb_active | csb_trigger;
t1->csb_relation = relation;
t1->csb_stream = stream;
}
else if (relation)
{
CompilerScratch::csb_repeat* t1 = CMP_csb_element(csb, 0);
t1->csb_stream = csb->nextStream();
t1->csb_relation = relation;
t1->csb_flags = csb_used | csb_active;
}
if (view_csb)
{
CompilerScratch::rpt_itr ptr = view_csb->csb_rpt.begin();
// AB: csb_n_stream replaced by view_csb->csb_rpt.getCount(), because there could
// be more then just csb_n_stream-numbers that hold data.
// Certainly csb_stream (see PAR_context where the context is retrieved)
const CompilerScratch::rpt_const_itr end = view_csb->csb_rpt.end();
for (StreamType stream = 0; ptr != end; ++ptr, ++stream)
{
CompilerScratch::csb_repeat* t2 = CMP_csb_element(csb, stream);
t2->csb_relation = ptr->csb_relation;
t2->csb_procedure = ptr->csb_procedure;
t2->csb_stream = ptr->csb_stream;
t2->csb_flags = ptr->csb_flags & csb_used;
}
csb->csb_n_stream = view_csb->csb_n_stream;
}
return csb;
}
// Parse blr, returning a compiler scratch block with the results.
// Caller must do pool handling.
DmlNode* PAR_blr(thread_db* tdbb, jrd_rel* relation, const UCHAR* blr, ULONG blr_length,
CompilerScratch* view_csb, CompilerScratch** csb_ptr, JrdStatement** statementPtr,
const bool trigger, USHORT flags)
{
#ifdef CMP_DEBUG
cmp_trace("BLR code given for JRD parsing:");
// CVC: Couldn't find isc_trace_printer, so changed it to gds__trace_printer.
fb_print_blr(blr, blr_length, gds__trace_printer, 0, 0);
#endif
CompilerScratch* csb = par_start(tdbb, relation, view_csb, csb_ptr, trigger, flags);
csb->csb_blr_reader = BlrReader(blr, blr_length);
getBlrVersion(csb);
DmlNode* node = PAR_parse_node(tdbb, csb);
csb->csb_node = node;
if (csb->csb_blr_reader.getByte() != (UCHAR) blr_eoc)
PAR_syntax_error(csb, "end_of_command");
if (statementPtr)
*statementPtr = JrdStatement::makeStatement(tdbb, csb, false);
if (csb_ptr)
*csb_ptr = csb;
else
delete csb;
return node;
}
// Finish parse of memory nodes, returning a compiler scratch block with the results.
// Caller must do pool handling.
void PAR_preparsed_node(thread_db* tdbb, jrd_rel* relation, DmlNode* node,
CompilerScratch* view_csb, CompilerScratch** csb_ptr, JrdStatement** statementPtr,
const bool trigger, USHORT flags)
{
CompilerScratch* csb = par_start(tdbb, relation, view_csb, csb_ptr, trigger, flags);
csb->blrVersion = 5; // blr_version5
csb->csb_node = node;
if (statementPtr)
*statementPtr = JrdStatement::makeStatement(tdbb, csb, false);
if (csb_ptr)
*csb_ptr = csb;
else
delete csb;
}
// PAR_blr equivalent for validation expressions.
// Validation expressions are boolean expressions, but may be prefixed with a blr_stmt_expr.
BoolExprNode* PAR_validation_blr(thread_db* tdbb, jrd_rel* relation, const UCHAR* blr, ULONG blr_length,
CompilerScratch* view_csb, CompilerScratch** csb_ptr, USHORT flags)
{
SET_TDBB(tdbb);
#ifdef CMP_DEBUG
cmp_trace("BLR code given for JRD parsing:");
// CVC: Couldn't find isc_trace_printer, so changed it to gds__trace_printer.
fb_print_blr(blr, blr_length, gds__trace_printer, 0, 0);
#endif
CompilerScratch* csb;
if (!(csb_ptr && (csb = *csb_ptr)))
{
size_t count = 5;
if (view_csb)
count += view_csb->csb_rpt.getCapacity();
csb = CompilerScratch::newCsb(*tdbb->getDefaultPool(), count);
csb->csb_g_flags |= flags;
}
// If there is a request ptr, this is a trigger. Set up contexts 0 and 1 for
// the target relation
if (relation)
{
CompilerScratch::csb_repeat* t1 = CMP_csb_element(csb, 0);
t1->csb_stream = csb->nextStream();
t1->csb_relation = relation;
t1->csb_flags = csb_used | csb_active;
}
csb->csb_blr_reader = BlrReader(blr, blr_length);
if (view_csb)
{
CompilerScratch::rpt_itr ptr = view_csb->csb_rpt.begin();
// AB: csb_n_stream replaced by view_csb->csb_rpt.getCount(), because there could
// be more then just csb_n_stream-numbers that hold data.
// Certainly csb_stream (see PAR_context where the context is retrieved)
const CompilerScratch::rpt_const_itr end = view_csb->csb_rpt.end();
for (StreamType stream = 0; ptr != end; ++ptr, ++stream)
{
CompilerScratch::csb_repeat* t2 = CMP_csb_element(csb, stream);
t2->csb_relation = ptr->csb_relation;
t2->csb_procedure = ptr->csb_procedure;
t2->csb_stream = ptr->csb_stream;
t2->csb_flags = ptr->csb_flags & csb_used;
}
csb->csb_n_stream = view_csb->csb_n_stream;
}
getBlrVersion(csb);
if (csb->csb_blr_reader.peekByte() == blr_stmt_expr)
{
// Parse it and ignore. This legacy and broken verb is transformed while parsing expressions.
csb->csb_blr_reader.getByte();
PAR_parse_stmt(tdbb, csb);
}
BoolExprNode* expr = PAR_parse_boolean(tdbb, csb);
if (csb->csb_blr_reader.getByte() != (UCHAR) blr_eoc)
PAR_syntax_error(csb, "end_of_command");
if (csb_ptr)
*csb_ptr = csb;
else
delete csb;
return expr;
}
// Parse a BLR datatype. Return the alignment requirements of the datatype.
USHORT PAR_datatype(BlrReader& blrReader, dsc* desc)
{
desc->clear();
const USHORT dtype = blrReader.getByte();
USHORT textType;
switch (dtype)
{
case blr_text:
desc->makeText(blrReader.getWord(), ttype_dynamic);
desc->dsc_flags |= DSC_no_subtype;
break;
case blr_cstring:
desc->dsc_dtype = dtype_cstring;
desc->dsc_flags |= DSC_no_subtype;
desc->dsc_length = blrReader.getWord();
desc->setTextType(ttype_dynamic);
break;
case blr_varying:
desc->makeVarying(blrReader.getWord(), ttype_dynamic);
desc->dsc_flags |= DSC_no_subtype;
break;
case blr_text2:
textType = blrReader.getWord();
desc->makeText(blrReader.getWord(), textType);
break;
case blr_cstring2:
desc->dsc_dtype = dtype_cstring;
desc->setTextType(blrReader.getWord());
desc->dsc_length = blrReader.getWord();
break;
case blr_varying2:
textType = blrReader.getWord();
desc->makeVarying(blrReader.getWord(), textType);
break;
case blr_short:
desc->dsc_dtype = dtype_short;
desc->dsc_length = sizeof(SSHORT);
desc->dsc_scale = (int) blrReader.getByte();
break;
case blr_long:
desc->dsc_dtype = dtype_long;
desc->dsc_length = sizeof(SLONG);
desc->dsc_scale = (int) blrReader.getByte();
break;
case blr_int64:
desc->dsc_dtype = dtype_int64;
desc->dsc_length = sizeof(SINT64);
desc->dsc_scale = (int) blrReader.getByte();
break;
case blr_quad:
desc->dsc_dtype = dtype_quad;
desc->dsc_length = sizeof(ISC_QUAD);
desc->dsc_scale = (int) blrReader.getByte();
break;
case blr_float:
desc->dsc_dtype = dtype_real;
desc->dsc_length = sizeof(float);
break;
case blr_timestamp:
desc->dsc_dtype = dtype_timestamp;
desc->dsc_length = sizeof(ISC_QUAD);
break;
case blr_sql_date:
desc->dsc_dtype = dtype_sql_date;
desc->dsc_length = type_lengths[dtype_sql_date];
break;
case blr_sql_time:
desc->dsc_dtype = dtype_sql_time;
desc->dsc_length = type_lengths[dtype_sql_time];
break;
case blr_double:
case blr_d_float:
desc->dsc_dtype = dtype_double;
desc->dsc_length = sizeof(double);
break;
case blr_blob2:
desc->dsc_dtype = dtype_blob;
desc->dsc_length = sizeof(ISC_QUAD);
desc->dsc_sub_type = blrReader.getWord();
textType = blrReader.getWord();
desc->dsc_scale = textType & 0xFF; // BLOB character set
desc->dsc_flags = textType & 0xFF00; // BLOB collation
break;
case blr_bool:
desc->makeBoolean();
break;
default:
par_error(blrReader, Arg::Gds(isc_datnotsup));
}
return type_alignments[desc->dsc_dtype];
}
// Parse a BLR descriptor. Return the alignment requirements of the datatype.
USHORT PAR_desc(thread_db* tdbb, CompilerScratch* csb, dsc* desc, ItemInfo* itemInfo)
{
if (itemInfo)
{
itemInfo->nullable = true;
itemInfo->explicitCollation = false;
itemInfo->fullDomain = false;
}
desc->clear();
const USHORT dtype = csb->csb_blr_reader.getByte();
switch (dtype)
{
case blr_not_nullable:
PAR_desc(tdbb, csb, desc, itemInfo);
if (itemInfo)
itemInfo->nullable = false;
break;
case blr_domain_name:
case blr_domain_name2:
{
const bool fullDomain = (csb->csb_blr_reader.getByte() == blr_domain_full);
MetaName* name = FB_NEW(csb->csb_pool) MetaName(csb->csb_pool);
PAR_name(csb, *name);
MetaNamePair namePair(*name, "");
FieldInfo fieldInfo;
bool exist = csb->csb_map_field_info.get(namePair, fieldInfo);
MET_get_domain(tdbb, csb->csb_pool, *name, desc, (exist ? NULL : &fieldInfo));
if (!exist)
csb->csb_map_field_info.put(namePair, fieldInfo);
if (itemInfo)
{
itemInfo->field = namePair;
if (fullDomain)
{
itemInfo->nullable = fieldInfo.nullable;
itemInfo->fullDomain = true;
}
else
itemInfo->nullable = true;
}
if (dtype == blr_domain_name2)
{
const USHORT ttype = csb->csb_blr_reader.getWord();
switch (desc->dsc_dtype)
{
case dtype_cstring:
case dtype_text:
case dtype_varying:
desc->setTextType(ttype);
break;
case dtype_blob:
desc->dsc_scale = ttype & 0xFF; // BLOB character set
desc->dsc_flags = ttype & 0xFF00; // BLOB collation
break;
default:
PAR_error(csb, Arg::Gds(isc_collation_requires_text));
break;
}
}
CompilerScratch::Dependency dependency(obj_field);
dependency.name = name;
csb->csb_dependencies.push(dependency);
break;
}
case blr_column_name:
case blr_column_name2:
{
const bool fullDomain = (csb->csb_blr_reader.getByte() == blr_domain_full);
MetaName* relationName = FB_NEW(csb->csb_pool) MetaName(csb->csb_pool);
PAR_name(csb, *relationName);
MetaName* fieldName = FB_NEW(csb->csb_pool) MetaName(csb->csb_pool);
PAR_name(csb, *fieldName);
MetaNamePair namePair(*relationName, *fieldName);
FieldInfo fieldInfo;
bool exist = csb->csb_map_field_info.get(namePair, fieldInfo);
MET_get_relation_field(tdbb, csb->csb_pool, *relationName, *fieldName, desc,
(exist ? NULL : &fieldInfo));
if (!exist)
csb->csb_map_field_info.put(namePair, fieldInfo);
if (itemInfo)
{
itemInfo->field = namePair;
if (fullDomain)
{
itemInfo->nullable = fieldInfo.nullable;
itemInfo->fullDomain = true;
}
else
itemInfo->nullable = true;
}
if (dtype == blr_column_name2)
{
const USHORT ttype = csb->csb_blr_reader.getWord();
switch (desc->dsc_dtype)
{
case dtype_cstring:
case dtype_text:
case dtype_varying:
desc->setTextType(ttype);
break;
case dtype_blob:
desc->dsc_scale = ttype & 0xFF; // BLOB character set
desc->dsc_flags = ttype & 0xFF00; // BLOB collation
break;
default:
PAR_error(csb, Arg::Gds(isc_collation_requires_text));
break;
}
}
CompilerScratch::Dependency dependency(obj_relation);
dependency.relation = MET_lookup_relation(tdbb, *relationName);
dependency.subName = fieldName;
csb->csb_dependencies.push(dependency);
break;
}
default:
csb->csb_blr_reader.seekBackward(1);
PAR_datatype(csb->csb_blr_reader, desc);
break;
}
if (desc->getTextType() != CS_NONE)
{
CompilerScratch::Dependency dependency(obj_collation);
dependency.number = INTL_TEXT_TYPE(*desc);
csb->csb_dependencies.push(dependency);
}
if (itemInfo)
{
if (dtype == blr_cstring2 || dtype == blr_text2 || dtype == blr_varying2 ||
dtype == blr_blob2 || dtype == blr_domain_name2)
{
itemInfo->explicitCollation = true;
}
}
return type_alignments[desc->dsc_dtype];
}
ValueExprNode* PAR_gen_field(thread_db* tdbb, StreamType stream, USHORT id, bool byId)
{
/**************************************
*
* P A R _ g e n _ f i e l d
*
**************************************
*
* Functional description
* Generate a field block.
*
**************************************/
SET_TDBB(tdbb);
return FB_NEW(*tdbb->getDefaultPool()) FieldNode(*tdbb->getDefaultPool(), stream, id, byId);
}
ValueExprNode* PAR_make_field(thread_db* tdbb, CompilerScratch* csb, USHORT context,
const MetaName& base_field)
{
/**************************************
*
* P A R _ m a k e _ f i e l d
*
**************************************
*
* Functional description
* Make up a field node in the permanent pool. This is used
* by MET_scan_relation to handle view fields.
*
**************************************/
SET_TDBB(tdbb);
if (context >= csb->csb_rpt.getCount() || !(csb->csb_rpt[context].csb_flags & csb_used))
return NULL;
const StreamType stream = csb->csb_rpt[context].csb_stream;
jrd_rel* const relation = csb->csb_rpt[stream].csb_relation;
jrd_prc* const procedure = csb->csb_rpt[stream].csb_procedure;
const SSHORT id =
relation ? MET_lookup_field(tdbb, relation, base_field) :
procedure ? PAR_find_proc_field(procedure, base_field) : -1;
if (id < 0)
return NULL;
if (csb->csb_g_flags & csb_get_dependencies)
PAR_dependency(tdbb, csb, stream, id, base_field);
return PAR_gen_field(tdbb, stream, id);
}
// Make a list node out of a stack.
CompoundStmtNode* PAR_make_list(thread_db* tdbb, StmtNodeStack& stack)
{
SET_TDBB(tdbb);
// Count the number of nodes.
USHORT count = stack.getCount();
CompoundStmtNode* node = FB_NEW(*tdbb->getDefaultPool()) CompoundStmtNode(*tdbb->getDefaultPool());
NestConst<StmtNode>* ptr = node->statements.getBuffer(count) + count;
while (stack.hasData())
*--ptr = stack.pop();
return node;
}
CompilerScratch* PAR_parse(thread_db* tdbb, const UCHAR* blr, ULONG blr_length,
bool internal_flag, ULONG dbginfo_length, const UCHAR* dbginfo)
{
/**************************************
*
* P A R _ p a r s e
*
**************************************
*
* Functional description
* Parse blr, returning a compiler scratch block with the results.
*
**************************************/
SET_TDBB(tdbb);
CompilerScratch* csb = CompilerScratch::newCsb(*tdbb->getDefaultPool(), 5);
csb->csb_blr_reader = BlrReader(blr, blr_length);
if (internal_flag)
csb->csb_g_flags |= csb_internal;
getBlrVersion(csb);
if (dbginfo_length > 0)
DBG_parse_debug_info(dbginfo_length, dbginfo, *csb->csb_dbg_info);
DmlNode* node = PAR_parse_node(tdbb, csb);
csb->csb_node = node;
if (csb->csb_blr_reader.getByte() != (UCHAR) blr_eoc)
PAR_syntax_error(csb, "end_of_command");
return csb;
}
SLONG PAR_symbol_to_gdscode(const Firebird::string& name)
{
/**************************************
*
* P A R _ s y m b o l _ t o _ g d s c o d e
*
**************************************
*
* Functional description
* Symbolic ASCII names are used in blr for posting and handling
* exceptions. They are also used to identify error codes
* within system triggers in a database.
*
* Returns the gds error status code for the given symbolic
* name, or 0 if not found.
*
* Symbolic names may be null or space terminated.
*
**************************************/
for (int i = 0; codes[i].code_number; ++i)
{
if (name == codes[i].code_string) {
return codes[i].code_number;
}
}
return 0;
}
// Registers a parse function (DmlNode creator) for a BLR code.
void PAR_register(UCHAR blr, NodeParseFunc parseFunc)
{
fb_assert(!blr_parsers[blr] || blr_parsers[blr] == parseFunc);
blr_parsers[blr] = parseFunc;
}
// We've got a blr error other than a syntax error. Handle it.
static void par_error(BlrReader& blrReader, const Arg::StatusVector& v, bool isSyntaxError)
{
fb_assert(v.value()[0] == isc_arg_gds);
// Don't bother to pass tdbb for error handling
thread_db* tdbb = JRD_get_thread_data();
if (isSyntaxError)
{
blrReader.seekBackward(1);
Arg::Gds p(isc_invalid_blr);
p << Arg::Num(blrReader.getOffset());
p.append(v);
p.copyTo(tdbb->tdbb_status_vector);
}
else
v.copyTo(tdbb->tdbb_status_vector);
ERR_make_permanent(tdbb->tdbb_status_vector);
// Give up whatever we were doing and return to the user.
ERR_punt();
}
// We've got a blr error other than a syntax error. Handle it.
void PAR_error(CompilerScratch* csb, const Arg::StatusVector& v, bool isSyntaxError)
{
par_error(csb->csb_blr_reader, v, isSyntaxError);
}
// Look for named field in procedure output fields.
SSHORT PAR_find_proc_field(const jrd_prc* procedure, const Firebird::MetaName& name)
{
const Array<NestConst<Parameter> >& list = procedure->getOutputFields();
Array<NestConst<Parameter> >::const_iterator ptr = list.begin();
for (const Array<NestConst<Parameter> >::const_iterator end = list.end(); ptr < end; ++ptr)
{
const Parameter* param = *ptr;
if (name == param->prm_name)
return param->prm_number;
}
return -1;
}
// Parse a counted argument list, given the count.
ValueListNode* PAR_args(thread_db* tdbb, CompilerScratch* csb, UCHAR count,
USHORT allocCount)
{
SET_TDBB(tdbb);
fb_assert(allocCount >= count);
MemoryPool& pool = *tdbb->getDefaultPool();
ValueListNode* node = FB_NEW(pool) ValueListNode(pool, allocCount);
NestConst<ValueExprNode>* ptr = node->items.begin();
if (count)
{
do
{
*ptr++ = PAR_parse_value(tdbb, csb);
} while (--count);
}
return node;
}
// Parse a counted argument list.
ValueListNode* PAR_args(thread_db* tdbb, CompilerScratch* csb)
{
SET_TDBB(tdbb);
UCHAR count = csb->csb_blr_reader.getByte();
return PAR_args(tdbb, csb, count, count);
}
StreamType PAR_context(CompilerScratch* csb, SSHORT* context_ptr)
{
/**************************************
*
* P A R _ c o n t e x t
*
**************************************
*
* Functional description
* Introduce a new context into the system. This involves
* assigning a stream and possibly extending the compile
* scratch block.
*
**************************************/
// CVC: Bottleneck
const SSHORT context = (unsigned int) csb->csb_blr_reader.getByte();
if (context_ptr)
*context_ptr = context;
CompilerScratch::csb_repeat* tail = CMP_csb_element(csb, context);
if (tail->csb_flags & csb_used)
{
if (csb->csb_g_flags & csb_reuse_context) {
return tail->csb_stream;
}
PAR_error(csb, Arg::Gds(isc_ctxinuse));
}
const StreamType stream = csb->nextStream(false);
if (stream >= MAX_STREAMS)
{
PAR_error(csb, Arg::Gds(isc_too_many_contexts));
}
tail->csb_flags |= csb_used;
tail->csb_stream = stream;
CMP_csb_element(csb, stream);
return stream;
}
void PAR_dependency(thread_db* tdbb, CompilerScratch* csb, StreamType stream, SSHORT id,
const MetaName& field_name)
{
/**************************************
*
* P A R _ d e p e n d e n c y
*
**************************************
*
* Functional description
* Register a field, relation, procedure or exception reference
* as a dependency.
*
**************************************/
SET_TDBB(tdbb);
CompilerScratch::Dependency dependency(0);
if (csb->csb_rpt[stream].csb_relation)
{
dependency.relation = csb->csb_rpt[stream].csb_relation;
// How do I determine reliably this is a view?
// At this time, rel_view_rse is still null.
//if (is_view)
// dependency.objType = obj_view;
//else
dependency.objType = obj_relation;
}
else if (csb->csb_rpt[stream].csb_procedure)
{
if (csb->csb_rpt[stream].csb_procedure->isSubRoutine())
return;
dependency.procedure = csb->csb_rpt[stream].csb_procedure;
dependency.objType = obj_procedure;
}
if (field_name.length() > 0)
dependency.subName = FB_NEW(*tdbb->getDefaultPool()) MetaName(*tdbb->getDefaultPool(), field_name);
else if (id >= 0)
dependency.subNumber = id;
csb->csb_dependencies.push(dependency);
}
USHORT PAR_name(CompilerScratch* csb, Firebird::MetaName& name)
{
/**************************************
*
* P A R _ n a m e
*
**************************************
*
* Functional description
* Parse a counted string, returning count.
*
**************************************/
size_t l = csb->csb_blr_reader.getByte();
// Check for overly long identifiers at BLR parse stage to prevent unwanted
// surprises in deeper layers of the engine.
if (l > MAX_SQL_IDENTIFIER_LEN)
{
SqlIdentifier st;
char* s = st;
l = MAX_SQL_IDENTIFIER_LEN;
while (l--)
*s++ = csb->csb_blr_reader.getByte();
*s = 0;
ERR_post(Arg::Gds(isc_identifier_too_long) << Arg::Str(st));
}
char* s = name.getBuffer(l);
while (l--)
*s++ = csb->csb_blr_reader.getByte();
return name.length();
}
size_t PAR_name(CompilerScratch* csb, Firebird::string& name)
{
/**************************************
*
* P A R _ n a m e
*
**************************************
*
* Functional description
* Parse a counted string of virtually unlimited size
* (up to 64K, actually <= 255), returning count.
*
**************************************/
size_t l = csb->csb_blr_reader.getByte();
char* s = name.getBuffer(l);
while (l--)
*s++ = csb->csb_blr_reader.getByte();
return name.length();
}
static PlanNode* par_plan(thread_db* tdbb, CompilerScratch* csb)
{
/**************************************
*
* p a r _ p l a n
*
**************************************
*
* Functional description
* Parse an access plan expression.
* At this stage we are just generating the
* parse tree and checking contexts
* and indices.
*
**************************************/
SET_TDBB(tdbb);
USHORT node_type = (USHORT) csb->csb_blr_reader.getByte();
// a join type indicates a cross of two or more streams
if (node_type == blr_join || node_type == blr_merge)
{
// CVC: bottleneck
int count = (USHORT) csb->csb_blr_reader.getByte();
PlanNode* plan = FB_NEW(csb->csb_pool) PlanNode(csb->csb_pool, PlanNode::TYPE_JOIN);
while (count-- > 0)
plan->subNodes.add(par_plan(tdbb, csb));
return plan;
}
// we have hit a stream; parse the context number and access type
if (node_type == blr_retrieve)
{
PlanNode* plan = FB_NEW(csb->csb_pool) PlanNode(csb->csb_pool, PlanNode::TYPE_RETRIEVE);
// parse the relation name and context--the relation
// itself is redundant except in the case of a view,
// in which case the base relation (and alias) must be specified
USHORT n = (unsigned int) csb->csb_blr_reader.getByte();
if (n != blr_relation && n != blr_relation2 && n != blr_rid && n != blr_rid2)
PAR_syntax_error(csb, "TABLE");
// don't make RelationSourceNode::parse() parse the context, because
// this would add a new context; while this is a reference to
// an existing context
plan->relationNode = RelationSourceNode::parse(tdbb, csb, n, false);
jrd_rel* relation = plan->relationNode->relation;
// CVC: bottleneck
n = csb->csb_blr_reader.getByte();
if (n >= csb->csb_rpt.getCount() || !(csb->csb_rpt[n].csb_flags & csb_used))
PAR_error(csb, Arg::Gds(isc_ctxnotdef));
const StreamType stream = csb->csb_rpt[n].csb_stream;
plan->relationNode->setStream(stream);
plan->relationNode->context = n;
// Access plan types (sequential is default)
node_type = (USHORT) csb->csb_blr_reader.getByte();
MetaName name;
switch (node_type)
{
case blr_navigational:
{
plan->accessType = FB_NEW(csb->csb_pool) PlanNode::AccessType(csb->csb_pool,
PlanNode::AccessType::TYPE_NAVIGATIONAL);
// pick up the index name and look up the appropriate ids
PAR_name(csb, name);
/* CVC: We can't do this. Index names are identifiers.
for (p = name; *p; *p++)
*p = UPPER (*p);
*/
SLONG relation_id;
SSHORT idx_status;
const SLONG index_id = MET_lookup_index_name(tdbb, name, &relation_id, &idx_status);
if (idx_status == MET_object_unknown || idx_status == MET_object_inactive)
{
if (tdbb->getAttachment()->att_flags & ATT_gbak_attachment)
{
PAR_warning(Arg::Warning(isc_indexname) << Arg::Str(name) <<
Arg::Str(relation->rel_name));
}
else
{
PAR_error(csb, Arg::Gds(isc_indexname) << Arg::Str(name) <<
Arg::Str(relation->rel_name));
}
}
// save both the relation id and the index id, since
// the relation could be a base relation of a view;
// save the index name also, for convenience
PlanNode::AccessItem& item = plan->accessType->items.add();
item.relationId = relation_id;
item.indexId = index_id;
item.indexName = name;
if (csb->csb_g_flags & csb_get_dependencies)
{
CompilerScratch::Dependency dependency(obj_index);
dependency.name = &item.indexName;
csb->csb_dependencies.push(dependency);
}
if (csb->csb_blr_reader.peekByte() != blr_indices)
break;
// dimitr: FALL INTO, if the plan item is ORDER ... INDEX (...)
}
case blr_indices:
{
if (plan->accessType)
csb->csb_blr_reader.getByte(); // skip blr_indices
else
{
plan->accessType = FB_NEW(csb->csb_pool) PlanNode::AccessType(csb->csb_pool,
PlanNode::AccessType::TYPE_INDICES);
}
int count = (USHORT) csb->csb_blr_reader.getByte();
// pick up the index names and look up the appropriate ids
while (count-- > 0)
{
PAR_name(csb, name);
/* Nickolay Samofatov: We can't do this. Index names are identifiers.
for (p = name; *p; *p++)
*p = UPPER(*p);
*/
SLONG relation_id;
SSHORT idx_status;
const SLONG index_id = MET_lookup_index_name(tdbb, name, &relation_id, &idx_status);
if (idx_status == MET_object_unknown || idx_status == MET_object_inactive)
{
if (tdbb->getAttachment()->att_flags & ATT_gbak_attachment)
{
PAR_warning(Arg::Warning(isc_indexname) << Arg::Str(name) <<
Arg::Str(relation->rel_name));
}
else
{
PAR_error(csb, Arg::Gds(isc_indexname) << Arg::Str(name) <<
Arg::Str(relation->rel_name));
}
}
// save both the relation id and the index id, since
// the relation could be a base relation of a view;
// save the index name also, for convenience
PlanNode::AccessItem& item = plan->accessType->items.add();
item.relationId = relation_id;
item.indexId = index_id;
item.indexName = name;
if (csb->csb_g_flags & csb_get_dependencies)
{
CompilerScratch::Dependency dependency(obj_index);
dependency.name = &item.indexName;
csb->csb_dependencies.push(dependency);
}
}
}
break;
case blr_sequential:
break;
default:
PAR_syntax_error(csb, "access type");
}
return plan;
}
PAR_syntax_error(csb, "plan item");
return NULL; // Added to remove compiler warning
}
// Parse some procedure parameters.
void PAR_procedure_parms(thread_db* tdbb, CompilerScratch* csb, jrd_prc* procedure,
MessageNode** message_ptr, ValueListNode** sourceList, ValueListNode** targetList, bool input_flag)
{
SET_TDBB(tdbb);
SLONG count = csb->csb_blr_reader.getWord();
const SLONG inputCount = procedure->getInputFields().getCount();
// Check to see if the parameter count matches
if (input_flag ?
(count < (inputCount - procedure->getDefaultCount()) || (count > inputCount) ) :
(count != SLONG(procedure->getOutputFields().getCount())))
{
PAR_error(csb, Arg::Gds(input_flag ? isc_prcmismat : isc_prc_out_param_mismatch) <<
Arg::Str(procedure->getName().toString()));
}
if (count || input_flag && procedure->getDefaultCount())
{
MemoryPool& pool = *tdbb->getDefaultPool();
// We have a few parameters. Get on with creating the message block
USHORT n = ++csb->csb_msg_number;
if (n < 2)
csb->csb_msg_number = n = 2;
CompilerScratch::csb_repeat* tail = CMP_csb_element(csb, n);
MessageNode* message = tail->csb_message = *message_ptr = FB_NEW(pool) MessageNode(pool);
message->messageNumber = n;
const Format* format = input_flag ? procedure->getInputFormat() : procedure->getOutputFormat();
/* dimitr: procedure (with its parameter formats) is allocated out of
its own pool (prc_request->req_pool) and can be freed during
the cache cleanup (MET_clear_cache). Since the current
tdbb default pool is different from the procedure's one,
it's dangerous to copy a pointer from one request to another.
As an experiment, I've decided to copy format by value
instead of copying the reference. Since Format structure
doesn't contain any pointers, it should be safe to use a
default assignment operator which does a simple byte copy.
This change fixes one serious bug in the current codebase.
I think that this situation can (and probably should) be
handled by the metadata cache (via incrementing prc_use_count)
to avoid unexpected cache cleanups, but that area is out of my
knowledge. So this fix should be considered a temporary solution.
message->format = format;
*/
Format* fmt_copy = Format::newFormat(pool, format->fmt_count);
*fmt_copy = *format;
message->format = fmt_copy;
// --- end of fix ---
n = format->fmt_count / 2;
ValueListNode* sourceValues = *sourceList = FB_NEW(pool) ValueListNode(pool, n);
ValueListNode* targetValues = *targetList = FB_NEW(pool) ValueListNode(pool, n);
NestConst<ValueExprNode>* sourcePtr =
input_flag ? sourceValues->items.begin() : targetValues->items.begin();
NestConst<ValueExprNode>* targetPtr =
input_flag ? targetValues->items.begin() : sourceValues->items.begin();
for (USHORT i = 0; n; count--, n--)
{
// default value for parameter
if (count <= 0 && input_flag)
{
Parameter* parameter = procedure->getInputFields()[inputCount - n];
*sourcePtr++ = CMP_clone_node(tdbb, csb, parameter->prm_default_value);
}
else
*sourcePtr++ = PAR_parse_value(tdbb, csb);
ParameterNode* paramNode = FB_NEW(csb->csb_pool) ParameterNode(csb->csb_pool);
paramNode->message = message;
paramNode->argNumber = i++;
ParameterNode* paramFlagNode = FB_NEW(csb->csb_pool) ParameterNode(csb->csb_pool);
paramFlagNode->message = message;
paramFlagNode->argNumber = i++;
paramNode->argFlag = paramFlagNode;
*targetPtr++ = paramNode;
}
}
else if (input_flag ? inputCount : procedure->getOutputFields().getCount())
{
PAR_error(csb, Arg::Gds(input_flag ? isc_prcmismat : isc_prc_out_param_mismatch) <<
Arg::Str(procedure->getName().toString()));
}
}
// Parse a RecordSourceNode.
RecordSourceNode* PAR_parseRecordSource(thread_db* tdbb, CompilerScratch* csb)
{
SET_TDBB(tdbb);
const SSHORT blrOp = csb->csb_blr_reader.getByte();
switch (blrOp)
{
case blr_pid:
case blr_pid2:
case blr_procedure:
case blr_procedure2:
case blr_procedure3:
case blr_procedure4:
case blr_subproc:
return ProcedureSourceNode::parse(tdbb, csb, blrOp);
case blr_rse:
case blr_rs_stream:
return PAR_rse(tdbb, csb, blrOp);
case blr_relation:
case blr_rid:
case blr_relation2:
case blr_rid2:
return RelationSourceNode::parse(tdbb, csb, blrOp, true);
case blr_union:
case blr_recurse:
return UnionSourceNode::parse(tdbb, csb, blrOp);
case blr_window:
return WindowSourceNode::parse(tdbb, csb);
case blr_aggregate:
return AggregateSourceNode::parse(tdbb, csb);
default:
PAR_syntax_error(csb, "record source");
}
return NULL; // silence warning
}
// Parse a record selection expression.
RseNode* PAR_rse(thread_db* tdbb, CompilerScratch* csb, SSHORT rse_op)
{
SET_TDBB(tdbb);
int count = (unsigned int) csb->csb_blr_reader.getByte();
RseNode* rse = FB_NEW(*tdbb->getDefaultPool()) RseNode(*tdbb->getDefaultPool());
while (--count >= 0)
rse->rse_relations.add(PAR_parseRecordSource(tdbb, csb));
while (true)
{
const UCHAR op = csb->csb_blr_reader.getByte();
switch (op)
{
case blr_boolean:
rse->rse_boolean = PAR_parse_boolean(tdbb, csb);
break;
case blr_first:
if (rse_op == blr_rs_stream)
PAR_syntax_error(csb, "RecordSelExpr stream clause");
rse->rse_first = PAR_parse_value(tdbb, csb);
rse->flags |= RseNode::FLAG_OPT_FIRST_ROWS;
break;
case blr_skip:
if (rse_op == blr_rs_stream)
PAR_syntax_error(csb, "RecordSelExpr stream clause");
rse->rse_skip = PAR_parse_value(tdbb, csb);
break;
case blr_sort:
if (rse_op == blr_rs_stream)
PAR_syntax_error(csb, "RecordSelExpr stream clause");
csb->csb_blr_reader.seekBackward(1);
rse->rse_sorted = PAR_sort(tdbb, csb, op, false);
break;
case blr_project:
if (rse_op == blr_rs_stream)
PAR_syntax_error(csb, "RecordSelExpr stream clause");
csb->csb_blr_reader.seekBackward(1);
rse->rse_projection = PAR_sort(tdbb, csb, op, false);
break;
case blr_join_type:
{
const USHORT jointype = (USHORT) csb->csb_blr_reader.getByte();
rse->rse_jointype = jointype;
if (jointype != blr_inner &&
jointype != blr_left &&
jointype != blr_right &&
jointype != blr_full)
{
PAR_syntax_error(csb, "join type clause");
}
break;
}
case blr_plan:
rse->rse_plan = par_plan(tdbb, csb);
break;
case blr_writelock:
rse->flags |= RseNode::FLAG_WRITELOCK;
break;
default:
if (op == (UCHAR) blr_end)
{
// An outer join is only allowed when the stream count is 2
// and a boolean expression has been supplied
if (!rse->rse_jointype || (rse->rse_relations.getCount() == 2 && rse->rse_boolean))
{
// Convert right outer joins to left joins to avoid
// RIGHT JOIN handling at lower engine levels
if (rse->rse_jointype == blr_right)
{
// Swap sub-streams
RecordSourceNode* temp = rse->rse_relations[0];
rse->rse_relations[0] = rse->rse_relations[1];
rse->rse_relations[1] = temp;
rse->rse_jointype = blr_left;
}
return rse;
}
}
PAR_syntax_error(csb, ((rse_op == blr_rs_stream) ?
"RecordSelExpr stream clause" :
"record selection expression clause"));
}
}
}
// Parse a record selection expression.
RseNode* PAR_rse(thread_db* tdbb, CompilerScratch* csb)
{
SET_TDBB(tdbb);
const SSHORT blrOp = csb->csb_blr_reader.getByte();
switch (blrOp)
{
case blr_rse:
case blr_rs_stream:
return PAR_rse(tdbb, csb, blrOp);
case blr_singular:
{
RseNode* rseNode = PAR_rse(tdbb, csb);
rseNode->flags |= RseNode::FLAG_SINGULAR;
return rseNode;
}
case blr_scrollable:
{
RseNode* rseNode = PAR_rse(tdbb, csb);
rseNode->flags |= RseNode::FLAG_SCROLLABLE;
return rseNode;
}
default:
PAR_syntax_error(csb, "RecordSelExpr");
}
return NULL; // warning
}
// Parse a sort clause (sans header byte). This is used for blr_sort, blr_project and blr_group_by.
SortNode* PAR_sort(thread_db* tdbb, CompilerScratch* csb, UCHAR expectedBlr,
bool nullForEmpty)
{
SET_TDBB(tdbb);
const UCHAR blrOp = csb->csb_blr_reader.getByte();
if (blrOp != expectedBlr)
{
char s[20];
sprintf(s, "blr code %d", expectedBlr);
PAR_syntax_error(csb, s);
}
USHORT count = csb->csb_blr_reader.getByte();
if (count == 0 && nullForEmpty)
return NULL;
SortNode* sort = PAR_sort_internal(tdbb, csb, blrOp, count);
if (blrOp != blr_sort)
sort->unique = true;
return sort;
}
// Parse the internals of a sort clause. This is used for blr_sort, blr_project, blr_group_by
// and blr_partition_by.
SortNode* PAR_sort_internal(thread_db* tdbb, CompilerScratch* csb, UCHAR blrOp,
USHORT count)
{
SET_TDBB(tdbb);
SortNode* sort = FB_NEW(*tdbb->getDefaultPool()) SortNode(
*tdbb->getDefaultPool());
NestConst<ValueExprNode>* ptr = sort->expressions.getBuffer(count);
bool* ptr2 = sort->descending.getBuffer(count);
int* ptr3 = sort->nullOrder.getBuffer(count);
while (count-- > 0)
{
if (blrOp == blr_sort)
{
UCHAR code = csb->csb_blr_reader.getByte();
switch (code)
{
case blr_nullsfirst:
*ptr3++ = rse_nulls_first;
code = csb->csb_blr_reader.getByte();
break;
case blr_nullslast:
*ptr3++ = rse_nulls_last;
code = csb->csb_blr_reader.getByte();
break;
default:
*ptr3++ = rse_nulls_default;
}
*ptr2++ = (code == blr_descending);
}
else
{
*ptr2++ = false; // ascending
*ptr3++ = rse_nulls_default;
}
*ptr++ = PAR_parse_value(tdbb, csb);
}
return sort;
}
// Parse a boolean node.
BoolExprNode* PAR_parse_boolean(thread_db* tdbb, CompilerScratch* csb)
{
DmlNode* node = PAR_parse_node(tdbb, csb);
if (node->kind != DmlNode::KIND_BOOLEAN)
PAR_syntax_error(csb, "boolean");
return static_cast<BoolExprNode*>(node);
}
// Parse a value node.
ValueExprNode* PAR_parse_value(thread_db* tdbb, CompilerScratch* csb)
{
DmlNode* node = PAR_parse_node(tdbb, csb);
if (node->kind != DmlNode::KIND_VALUE)
PAR_syntax_error(csb, "value");
return static_cast<ValueExprNode*>(node);
}
// Parse a statement node.
StmtNode* PAR_parse_stmt(thread_db* tdbb, CompilerScratch* csb)
{
DmlNode* node = PAR_parse_node(tdbb, csb);
if (node->kind != DmlNode::KIND_STATEMENT)
PAR_syntax_error(csb, "statement");
return static_cast<StmtNode*>(node);
}
// Parse a BLR node.
DmlNode* PAR_parse_node(thread_db* tdbb, CompilerScratch* csb)
{
SET_TDBB(tdbb);
const ULONG blr_offset = csb->csb_blr_reader.getOffset();
const SSHORT blr_operator = csb->csb_blr_reader.getByte();
if (blr_operator < 0 || blr_operator >= FB_NELEM(blr_parsers))
{
// NS: This error string is correct, please do not mangle it again and again.
// The whole error message is "BLR syntax error: expected %s at offset %d, encountered %d"
PAR_syntax_error(csb, "valid BLR code");
}
// Dispatch on operator type.
switch (blr_operator)
{
case blr_rse:
case blr_rs_stream:
case blr_singular:
case blr_scrollable:
csb->csb_blr_reader.seekBackward(1);
return PAR_rse(tdbb, csb);
case blr_pid:
case blr_pid2:
case blr_procedure:
case blr_procedure2:
case blr_procedure3:
case blr_procedure4:
case blr_subproc:
case blr_relation:
case blr_rid:
case blr_relation2:
case blr_rid2:
case blr_union:
case blr_recurse:
case blr_window:
case blr_aggregate:
csb->csb_blr_reader.seekBackward(1);
return PAR_parseRecordSource(tdbb, csb);
}
if (!blr_parsers[blr_operator])
PAR_syntax_error(csb, "valid BLR code");
DmlNode* node = blr_parsers[blr_operator](tdbb, *tdbb->getDefaultPool(), csb, blr_operator);
size_t pos = 0;
if (node->kind == DmlNode::KIND_STATEMENT && csb->csb_dbg_info->blrToSrc.find(blr_offset, pos))
{
MapBlrToSrcItem& i = csb->csb_dbg_info->blrToSrc[pos];
StmtNode* stmt = static_cast<StmtNode*>(node);
stmt->hasLineColumn = true;
stmt->line = i.mbs_src_line;
stmt->column = i.mbs_src_col;
}
return node;
}
void PAR_syntax_error(CompilerScratch* csb, const TEXT* string)
{
/**************************************
*
* P A R _ s y n t a x _ e r r o r
*
**************************************
*
* Functional description
* Post a syntax error message.
*
**************************************/
csb->csb_blr_reader.seekBackward(1);
PAR_error(csb, Arg::Gds(isc_syntaxerr) << Arg::Str(string) <<
Arg::Num(csb->csb_blr_reader.getOffset()) <<
Arg::Num(csb->csb_blr_reader.peekByte()));
}
void PAR_warning(const Arg::StatusVector& v)
{
/**************************************
*
* P A R _ w a r n i n g
*
**************************************
*
* Functional description
* This is for GBAK so that we can pass warning messages
* back to the client. DO NOT USE this function until we
* fully implement warning at the engine level.
*
* We will use the status vector like a warning vector. What
* we are going to do is leave the [1] position of the vector
* as 0 so that this will not be treated as an error, and we
* will place our warning message in the consecutive positions.
* It will be up to the caller to check these positions for
* the message.
*
**************************************/
fb_assert(v.value()[0] == isc_arg_warning);
thread_db* tdbb = JRD_get_thread_data();
// Make sure that the [1] position is 0 indicating that no error has occurred
Arg::Gds p(FB_SUCCESS);
// Now place your warning messages
p.append(v);
// Save into tdbb
p.copyTo(tdbb->tdbb_status_vector);
ERR_make_permanent(tdbb->tdbb_status_vector);
}
// Get the BLR version from the CSB stream and complain if it's unknown.
static void getBlrVersion(CompilerScratch* csb)
{
const SSHORT version = csb->csb_blr_reader.getByte();
switch (version)
{
case blr_version4:
csb->blrVersion = 4;
break;
case blr_version5:
csb->blrVersion = 5;
break;
//case blr_version6:
// csb->blrVersion = 6;
// break;
default:
PAR_error(csb, Arg::Gds(isc_metadata_corrupt) <<
Arg::Gds(isc_wroblrver2) << Arg::Num(blr_version4) << Arg::Num(blr_version5/*6*/) <<
Arg::Num(version));
}
}