/* * PROGRAM: JRD Access Method * MODULE: cvt2.cpp * DESCRIPTION: Data mover and converter and comparator, etc. * Routines used ONLY within engine. * * 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): ______________________________________. * * 2001.6.18 Claudio Valderrama: Implement comparison on blobs and blobs against * other datatypes by request from Ann Harrison. */ #include "firebird.h" #include #include "../jrd/common.h" #include "../jrd/ibase.h" #include "../jrd/jrd.h" #include "../jrd/val.h" #include "../jrd/quad.h" #include "gen/iberror.h" #include "../jrd/intl.h" #include "../jrd/gdsassert.h" #include "../jrd/cvt_proto.h" #include "../jrd/cvt2_proto.h" #include "../common/cvt.h" #include "../jrd/err_proto.h" #include "../jrd/intl_proto.h" #include "../jrd/intl_classes.h" #include "../jrd/gds_proto.h" // CVC: I needed them here. #include "../jrd/jrd.h" #include "../jrd/blb_proto.h" #include "../jrd/tra.h" #include "../jrd/req.h" #include "../jrd/constants.h" #include "../common/utils_proto.h" #include "../common/classes/VaryStr.h" using namespace Jrd; using namespace Firebird; /* The original order of dsc_type values corresponded to the priority of conversion (that is, always convert the lesser to the greater type.) Introduction of dtype_int64 breaks that assumption: its position on the scale should be between dtype_long and dtype_real, but the types are integers, and dtype_quad occupies the only available place. Renumbering all the higher-numbered types would be a major ODS change and a fundamental discomfort This table permits us to put the entries in the right order for comparison purpose, even though isc_int64 had to get number 19, which is otherwise too high. This table is used in CVT2_compare, is indexed by dsc_dtype, and returns the relative priority of types for use when different types are compared. */ static const BYTE compare_priority[] = { dtype_unknown, // dtype_unknown through dtype_varying dtype_text, // have their natural values stored dtype_cstring, // in the table. dtype_varying, 0, 0, // dtypes and 4, 5 are unused. dtype_packed, // packed through long also have dtype_byte, // their natural values in the table dtype_short, dtype_long, dtype_quad + 1, // quad through array all move up dtype_real + 1, // by one to make room for int64 dtype_double + 1, // at its proper place in the table. dtype_d_float + 1, dtype_sql_date + 1, dtype_sql_time + 1, dtype_timestamp + 1, dtype_blob + 1, dtype_array + 1, dtype_long + 1, // int64 goes right after long dtype_dbkey // compares with nothing except itself }; SSHORT CVT2_compare(const dsc* arg1, const dsc* arg2) { /************************************** * * C V T 2 _ c o m p a r e * ************************************** * * Functional description * Compare two descriptors. Return (-1, 0, 1) if ab. * **************************************/ thread_db* tdbb = NULL; // AB: Maybe we need a other error-message, but at least throw // a message when 1 or both input paramters are empty. if (!arg1 || !arg2) { BUGCHECK(189); // msg 189 comparison not supported for specified data types. } // Handle the simple (matched) ones first if (arg1->dsc_dtype == arg2->dsc_dtype && arg1->dsc_scale == arg2->dsc_scale) { const UCHAR* p1 = arg1->dsc_address; const UCHAR* p2 = arg2->dsc_address; switch (arg1->dsc_dtype) { case dtype_short: if (*(SSHORT *) p1 == *(SSHORT *) p2) return 0; if (*(SSHORT *) p1 > *(SSHORT *) p2) return 1; return -1; case dtype_sql_time: if (*(ULONG *) p1 == *(ULONG *) p2) return 0; if (*(ULONG *) p1 > *(ULONG *) p2) return 1; return -1; case dtype_long: case dtype_sql_date: if (*(SLONG *) p1 == *(SLONG *) p2) return 0; if (*(SLONG *) p1 > *(SLONG *) p2) return 1; return -1; case dtype_quad: return QUAD_COMPARE(*(SQUAD *) p1, *(SQUAD *) p2); case dtype_int64: if (*(SINT64 *) p1 == *(SINT64 *) p2) return 0; if (*(SINT64 *) p1 > *(SINT64 *) p2) return 1; return -1; case dtype_dbkey: { // keep old ttype_binary compare rules USHORT l = MIN(arg1->dsc_length, arg2->dsc_length); SSHORT rc = memcmp(p1, p2, l); if (rc) { return rc; } return (arg1->dsc_length > l) ? 1 : (arg2->dsc_length > l) ? -1 : 0; } case dtype_timestamp: if (((SLONG *) p1)[0] > ((SLONG *) p2)[0]) return 1; if (((SLONG *) p1)[0] < ((SLONG *) p2)[0]) return -1; if (((ULONG *) p1)[1] > ((ULONG *) p2)[1]) return 1; if (((ULONG *) p1)[1] < ((ULONG *) p2)[1]) return -1; return 0; case dtype_real: if (*(float *) p1 == *(float *) p2) return 0; if (*(float *) p1 > *(float *) p2) return 1; return -1; case DEFAULT_DOUBLE: if (*(double *) p1 == *(double *) p2) return 0; if (*(double *) p1 > *(double *) p2) return 1; return -1; case dtype_text: case dtype_varying: case dtype_cstring: case dtype_array: case dtype_blob: // Special processing below break; default: // the two arguments have identical dtype and scale, but the // dtype is not one of your defined types! fb_assert(FALSE); break; } // switch on dtype } // if dtypes and scales are equal // Handle mixed string comparisons if (arg1->dsc_dtype <= dtype_varying && arg2->dsc_dtype <= dtype_varying) { /* * For the sake of optimization, we call INTL_compare * only when we cannot just do byte-by-byte compare. * We can do a local compare here, if * (a) one of the arguments is charset ttype_binary * OR (b) both of the arguments are char set ttype_none * OR (c) both of the arguments are char set ttype_ascii * If any argument is ttype_dynamic, we must see the * charset of the attachment. */ SET_TDBB(tdbb); CHARSET_ID charset1 = INTL_TTYPE(arg1); if (charset1 == ttype_dynamic) charset1 = INTL_charset(tdbb, charset1); CHARSET_ID charset2 = INTL_TTYPE(arg2); if (charset2 == ttype_dynamic) charset2 = INTL_charset(tdbb, charset2); if ((IS_INTL_DATA(arg1) || IS_INTL_DATA(arg2)) && (charset1 != ttype_binary) && (charset2 != ttype_binary) && ((charset1 != ttype_ascii) || (charset2 != ttype_ascii)) && ((charset1 != ttype_none) || (charset2 != ttype_none))) { return INTL_compare(tdbb, arg1, arg2, ERR_post); } UCHAR* p1 = NULL; UCHAR* p2 = NULL; USHORT t1, t2; // unused later USHORT length = CVT_get_string_ptr(arg1, &t1, &p1, NULL, 0); USHORT length2 = CVT_get_string_ptr(arg2, &t2, &p2, NULL, 0); int fill = length - length2; const UCHAR pad = charset1 == ttype_binary || charset2 == ttype_binary ? '\0' : ' '; if (length >= length2) { if (length2) { do { if (*p1++ != *p2++) return (p1[-1] > p2[-1]) ? 1 : -1; } while (--length2); } if (fill > 0) { do { if (*p1++ != pad) return (p1[-1] > pad) ? 1 : -1; } while (--fill); } return 0; } if (length) { do { if (*p1++ != *p2++) return (p1[-1] > p2[-1]) ? 1 : -1; } while (--length); } do { if (*p2++ != pad) return (pad > p2[-1]) ? 1 : -1; } while (++fill); return 0; } // Handle heterogeneous compares if (compare_priority[arg1->dsc_dtype] < compare_priority[arg2->dsc_dtype]) return -CVT2_compare(arg2, arg1); // At this point, the type of arg1 is guaranteed to be "greater than" arg2, // in the sense that it is the preferred type for comparing the two. switch (arg1->dsc_dtype) { case dtype_timestamp: { DSC desc; MOVE_CLEAR(&desc, sizeof(desc)); desc.dsc_dtype = dtype_timestamp; SLONG datetime[2]; desc.dsc_length = sizeof(datetime); desc.dsc_address = (UCHAR*) datetime; CVT_move(arg2, &desc); return CVT2_compare(arg1, &desc); } case dtype_sql_time: { DSC desc; MOVE_CLEAR(&desc, sizeof(desc)); desc.dsc_dtype = dtype_sql_time; SLONG atime; desc.dsc_length = sizeof(atime); desc.dsc_address = (UCHAR*) &atime; CVT_move(arg2, &desc); return CVT2_compare(arg1, &desc); } case dtype_sql_date: { DSC desc; MOVE_CLEAR(&desc, sizeof(desc)); desc.dsc_dtype = dtype_sql_date; SLONG date; desc.dsc_length = sizeof(date); desc.dsc_address = (UCHAR*) &date; CVT_move(arg2, &desc); return CVT2_compare(arg1, &desc); } case dtype_short: { SSHORT scale; if (arg2->dsc_dtype > dtype_varying) scale = MIN(arg1->dsc_scale, arg2->dsc_scale); else scale = arg1->dsc_scale; const SLONG temp1 = CVT_get_long(arg1, scale, ERR_post); const SLONG temp2 = CVT_get_long(arg2, scale, ERR_post); if (temp1 == temp2) return 0; if (temp1 > temp2) return 1; return -1; } case dtype_long: // Since longs may overflow when scaled, use int64 instead case dtype_int64: { SSHORT scale; if (arg2->dsc_dtype > dtype_varying) scale = MIN(arg1->dsc_scale, arg2->dsc_scale); else scale = arg1->dsc_scale; const SINT64 temp1 = CVT_get_int64(arg1, scale, ERR_post); const SINT64 temp2 = CVT_get_int64(arg2, scale, ERR_post); if (temp1 == temp2) return 0; if (temp1 > temp2) return 1; return -1; } case dtype_quad: { SSHORT scale; if (arg2->dsc_dtype > dtype_varying) scale = MIN(arg1->dsc_scale, arg2->dsc_scale); else scale = arg1->dsc_scale; const SQUAD temp1 = CVT_get_quad(arg1, scale, ERR_post); const SQUAD temp2 = CVT_get_quad(arg2, scale, ERR_post); return QUAD_COMPARE(temp1, temp2); } case dtype_real: { const float temp1 = (float) CVT_get_double(arg1, ERR_post); const float temp2 = (float) CVT_get_double(arg2, ERR_post); if (temp1 == temp2) return 0; if (temp1 > temp2) return 1; return -1; } case dtype_double: { const double temp1 = CVT_get_double(arg1, ERR_post); const double temp2 = CVT_get_double(arg2, ERR_post); if (temp1 == temp2) return 0; if (temp1 > temp2) return 1; return -1; } case dtype_blob: return CVT2_blob_compare(arg1, arg2); case dtype_array: ERR_post(Arg::Gds(isc_wish_list) << Arg::Gds(isc_blobnotsup) << "compare"); break; case dtype_dbkey: if (arg2->dsc_dtype <= dtype_any_text) { UCHAR* p = NULL; USHORT t; // unused later USHORT length = CVT_get_string_ptr(arg2, &t, &p, NULL, 0); USHORT l = MIN(arg1->dsc_length, length); SSHORT rc = memcmp(arg1->dsc_address, p, l); if (rc) { return rc; } return (arg1->dsc_length > l) ? 1 : (length > l) ? -1 : 0; } ERR_post(Arg::Gds(isc_wish_list) << Arg::Gds(isc_random) << "DB_KEY compare"); break; default: BUGCHECK(189); // msg 189 comparison not supported for specified data types break; } return 0; } SSHORT CVT2_blob_compare(const dsc* arg1, const dsc* arg2) { /************************************** * * C V T 2 _ b l o b _ c o m p a r e * ************************************** * * Functional description * Compare two blobs. Return (-1, 0, 1) if ab. * Alternatively, it will try to compare a blob against a string; * in this case, the string should be the second argument. * CVC: Ann Harrison asked for this function to make comparisons more * complete in the engine. * **************************************/ SLONG l1, l2; USHORT ttype2; SSHORT ret_val = 0; thread_db* tdbb = NULL; SET_TDBB(tdbb); // DEV_BLKCHK (node, type_nod); if (arg1->dsc_dtype != dtype_blob) ERR_post(Arg::Gds(isc_wish_list) << Arg::Gds(isc_datnotsup)); USHORT ttype1; if (arg1->dsc_sub_type == isc_blob_text) ttype1 = arg1->dsc_blob_ttype(); // Load blob character set and collation else ttype1 = ttype_binary; TextType* obj1 = INTL_texttype_lookup(tdbb, ttype1); ttype1 = obj1->getType(); // Is arg2 a blob? if (arg2->dsc_dtype == dtype_blob) { // Same blob id address? if (arg1->dsc_address == arg2->dsc_address) return 0; // Second test for blob id, checking relation and slot. const bid* bid1 = (bid*) arg1->dsc_address; const bid* bid2 = (bid*) arg2->dsc_address; if (*bid1 == *bid2) { return 0; } if (arg2->dsc_sub_type == isc_blob_text) ttype2 = arg2->dsc_blob_ttype(); // Load blob character set and collation else ttype2 = ttype_binary; TextType* obj2 = INTL_texttype_lookup(tdbb, ttype2); ttype2 = obj2->getType(); if (ttype1 == ttype_binary || ttype2 == ttype_binary) ttype1 = ttype2 = ttype_binary; else if (ttype1 == ttype_none || ttype2 == ttype_none) ttype1 = ttype2 = ttype_none; obj1 = INTL_texttype_lookup(tdbb, ttype1); obj2 = INTL_texttype_lookup(tdbb, ttype2); CharSet* charSet1 = obj1->getCharSet(); CharSet* charSet2 = obj2->getCharSet(); Firebird::HalfStaticArray buffer1; Firebird::HalfStaticArray buffer2; fb_assert(BUFFER_LARGE % 4 == 0); // 4 is our maximum character length UCHAR bpb[] = {isc_bpb_version1, isc_bpb_source_type, 1, isc_blob_text, isc_bpb_source_interp, 1, 0, isc_bpb_target_type, 1, isc_blob_text, isc_bpb_target_interp, 1, 0}; USHORT bpbLength = 0; if (arg1->dsc_sub_type == isc_blob_text && arg2->dsc_sub_type == isc_blob_text) { bpb[6] = arg2->dsc_scale; // source charset bpb[12] = arg1->dsc_scale; // destination charset bpbLength = sizeof(bpb); } blb* blob1 = BLB_open(tdbb, tdbb->getRequest()->req_transaction, (bid*) arg1->dsc_address); blb* blob2 = BLB_open2(tdbb, tdbb->getRequest()->req_transaction, (bid*) arg2->dsc_address, bpbLength, bpb); if (charSet1->isMultiByte()) { buffer1.getBuffer(blob1->blb_length); buffer2.getBuffer(blob2->blb_length / charSet2->minBytesPerChar() * charSet1->maxBytesPerChar()); } while (ret_val == 0 && !(blob1->blb_flags & BLB_eof) && !(blob2->blb_flags & BLB_eof)) { l1 = BLB_get_data(tdbb, blob1, buffer1.begin(), buffer1.getCapacity(), false); l2 = BLB_get_data(tdbb, blob2, buffer2.begin(), buffer2.getCapacity(), false); ret_val = obj1->compare(l1, buffer1.begin(), l2, buffer2.begin()); } if (ret_val == 0) { if ((blob1->blb_flags & BLB_eof) == BLB_eof) l1 = 0; if ((blob2->blb_flags & BLB_eof) == BLB_eof) l2 = 0; while (ret_val == 0 && !((blob1->blb_flags & BLB_eof) == BLB_eof && (blob2->blb_flags & BLB_eof) == BLB_eof)) { if (!(blob1->blb_flags & BLB_eof)) l1 = BLB_get_data(tdbb, blob1, buffer1.begin(), buffer1.getCapacity(), false); if (!(blob2->blb_flags & BLB_eof)) l2 = BLB_get_data(tdbb, blob2, buffer2.begin(), buffer2.getCapacity(), false); ret_val = obj1->compare(l1, buffer1.begin(), l2, buffer2.begin()); } } BLB_close(tdbb, blob1); BLB_close(tdbb, blob2); } else if (arg2->dsc_dtype == dtype_array) { // We do not accept arrays for now. Maybe InternalArrayDesc in the future. ERR_post(Arg::Gds(isc_wish_list) << Arg::Gds(isc_datnotsup)); } else { // The second parameter should be a string. if (arg2->dsc_dtype <= dtype_varying) { if ((ttype2 = arg2->dsc_ttype()) != ttype_binary) ttype2 = ttype1; } else ttype2 = ttype1; if (ttype1 == ttype_binary || ttype2 == ttype_binary) ttype1 = ttype2 = ttype_binary; else if (ttype1 == ttype_none || ttype2 == ttype_none) ttype1 = ttype2 = ttype_none; obj1 = INTL_texttype_lookup(tdbb, ttype1); CharSet* charSet1 = obj1->getCharSet(); Firebird::HalfStaticArray buffer1; UCHAR* p; MoveBuffer temp_str; l2 = CVT2_make_string2(arg2, ttype1, &p, temp_str); blb* blob1 = BLB_open(tdbb, tdbb->getRequest()->req_transaction, (bid*) arg1->dsc_address); if (charSet1->isMultiByte()) buffer1.getBuffer(blob1->blb_length); else buffer1.getBuffer(l2); l1 = BLB_get_data(tdbb, blob1, buffer1.begin(), buffer1.getCapacity(), false); ret_val = obj1->compare(l1, buffer1.begin(), l2, p); while (ret_val == 0 && (blob1->blb_flags & BLB_eof) != BLB_eof) { l1 = BLB_get_data(tdbb, blob1, buffer1.begin(), buffer1.getCapacity(), false); ret_val = obj1->compare(l1, buffer1.begin(), 0, p); } BLB_close(tdbb, blob1); } return ret_val; } void CVT2_get_name(const dsc* desc, TEXT* string) { /************************************** * * C V T 2 _ g e t _ n a m e * ************************************** * * Functional description * Get a name (max length 31, NULL terminated) from a descriptor. * **************************************/ VaryStr temp; // 31 bytes + 1 NULL const char* p; const USHORT length = CVT_make_string(desc, ttype_metadata, &p, &temp, sizeof(temp), ERR_post); memcpy(string, p, length); string[length] = 0; fb_utils::exact_name(string); } USHORT CVT2_make_string2(const dsc* desc, USHORT to_interp, UCHAR** address, Jrd::MoveBuffer& temp) { /************************************** * * C V T 2 _ m a k e _ s t r i n g 2 * ************************************** * * Functional description * * Convert the data from the desc to a string in the specified interp. * The pointer to this string is returned in address. * **************************************/ UCHAR* from_buf; USHORT from_len; USHORT from_interp; fb_assert(desc != NULL); fb_assert(address != NULL); switch (desc->dsc_dtype) { case dtype_text: from_buf = desc->dsc_address; from_len = desc->dsc_length; from_interp = INTL_TTYPE(desc); break; case dtype_cstring: from_buf = desc->dsc_address; from_len = MIN(strlen((char *) desc->dsc_address), (unsigned) (desc->dsc_length - 1)); from_interp = INTL_TTYPE(desc); break; case dtype_varying: { vary* varying = (vary*) desc->dsc_address; from_buf = reinterpret_cast(varying->vary_string); from_len = MIN(varying->vary_length, (USHORT) (desc->dsc_length - sizeof(SSHORT))); from_interp = INTL_TTYPE(desc); } break; } if (desc->dsc_dtype <= dtype_any_text) { if (to_interp == from_interp) { *address = from_buf; return from_len; } thread_db* tdbb = JRD_get_thread_data(); const USHORT cs1 = INTL_charset(tdbb, to_interp); const USHORT cs2 = INTL_charset(tdbb, from_interp); if (cs1 == cs2) { *address = from_buf; return from_len; } USHORT length = INTL_convert_bytes(tdbb, cs1, NULL, 0, cs2, from_buf, from_len, ERR_post); UCHAR* tempptr = temp.getBuffer(length); length = INTL_convert_bytes(tdbb, cs1, tempptr, length, cs2, from_buf, from_len, ERR_post); *address = tempptr; temp.resize(length); return length; } // Not string data, then -- convert value to varying string. dsc temp_desc; MOVE_CLEAR(&temp_desc, sizeof(temp_desc)); temp_desc.dsc_length = temp.getCapacity(); temp_desc.dsc_address = temp.getBuffer(temp_desc.dsc_length); vary* vtmp = reinterpret_cast(temp_desc.dsc_address); temp_desc.dsc_dtype = dtype_varying; temp_desc.setTextType(to_interp); CVT_move(desc, &temp_desc); *address = reinterpret_cast(vtmp->vary_string); return vtmp->vary_length; }