| 1 | /* |
| 2 | Copyright (c) 2000, 2010, Oracle and/or its affiliates |
| 3 | |
| 4 | This program is free software; you can redistribute it and/or modify |
| 5 | it under the terms of the GNU General Public License as published by |
| 6 | the Free Software Foundation; version 2 of the License. |
| 7 | |
| 8 | This program is distributed in the hope that it will be useful, |
| 9 | but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 10 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 11 | GNU General Public License for more details. |
| 12 | |
| 13 | You should have received a copy of the GNU General Public License |
| 14 | along with this program; if not, write to the Free Software |
| 15 | Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA */ |
| 16 | |
| 17 | /* Functions to check if a row is unique */ |
| 18 | |
| 19 | #include "myisamdef.h" |
| 20 | #include <m_ctype.h> |
| 21 | |
| 22 | my_bool mi_check_unique(MI_INFO *info, MI_UNIQUEDEF *def, const uchar *record, |
| 23 | ha_checksum unique_hash, my_off_t disk_pos) |
| 24 | { |
| 25 | my_off_t lastpos=info->lastpos; |
| 26 | MI_KEYDEF *key= &info->s->keyinfo[def->key]; |
| 27 | uchar *key_buff=info->lastkey2; |
| 28 | DBUG_ENTER("mi_check_unique" ); |
| 29 | |
| 30 | /* We need to store the hash value as a key in the record, breaking const */ |
| 31 | mi_unique_store(((uchar*) record)+key->seg->start, unique_hash); |
| 32 | _mi_make_key(info,def->key,key_buff,record,0); |
| 33 | |
| 34 | /* The above changed info->lastkey2. Inform mi_rnext_same(). */ |
| 35 | info->update&= ~HA_STATE_RNEXT_SAME; |
| 36 | |
| 37 | if (_mi_search(info,info->s->keyinfo+def->key,key_buff,MI_UNIQUE_HASH_LENGTH, |
| 38 | SEARCH_FIND,info->s->state.key_root[def->key])) |
| 39 | { |
| 40 | info->page_changed=1; /* Can't optimize read next */ |
| 41 | info->lastpos= lastpos; |
| 42 | DBUG_RETURN(0); /* No matching rows */ |
| 43 | } |
| 44 | |
| 45 | for (;;) |
| 46 | { |
| 47 | if (info->lastpos != disk_pos && |
| 48 | !(*info->s->compare_unique)(info,def,record,info->lastpos)) |
| 49 | { |
| 50 | my_errno=HA_ERR_FOUND_DUPP_UNIQUE; |
| 51 | info->errkey= (int) def->key; |
| 52 | info->dupp_key_pos= info->lastpos; |
| 53 | info->page_changed=1; /* Can't optimize read next */ |
| 54 | info->lastpos=lastpos; |
| 55 | DBUG_PRINT("info" ,("Found duplicate" )); |
| 56 | DBUG_RETURN(1); /* Found identical */ |
| 57 | } |
| 58 | if (_mi_search_next(info,info->s->keyinfo+def->key, info->lastkey, |
| 59 | MI_UNIQUE_HASH_LENGTH, SEARCH_BIGGER, |
| 60 | info->s->state.key_root[def->key]) || |
| 61 | memcmp(info->lastkey, key_buff, MI_UNIQUE_HASH_LENGTH)) |
| 62 | { |
| 63 | info->page_changed=1; /* Can't optimize read next */ |
| 64 | info->lastpos=lastpos; |
| 65 | DBUG_RETURN(0); /* end of tree */ |
| 66 | } |
| 67 | } |
| 68 | } |
| 69 | |
| 70 | |
| 71 | /* |
| 72 | Calculate a hash for a row |
| 73 | |
| 74 | TODO |
| 75 | Add support for bit fields |
| 76 | */ |
| 77 | |
| 78 | ha_checksum mi_unique_hash(MI_UNIQUEDEF *def, const uchar *record) |
| 79 | { |
| 80 | const uchar *pos, *end; |
| 81 | ha_checksum crc= 0; |
| 82 | ulong seed1=0, seed2= 4; |
| 83 | HA_KEYSEG *keyseg; |
| 84 | |
| 85 | for (keyseg=def->seg ; keyseg < def->end ; keyseg++) |
| 86 | { |
| 87 | enum ha_base_keytype type=(enum ha_base_keytype) keyseg->type; |
| 88 | uint length=keyseg->length; |
| 89 | |
| 90 | if (keyseg->null_bit) |
| 91 | { |
| 92 | if (record[keyseg->null_pos] & keyseg->null_bit) |
| 93 | { |
| 94 | /* |
| 95 | Change crc in a way different from an empty string or 0. |
| 96 | (This is an optimisation; The code will work even if this isn't |
| 97 | done) |
| 98 | */ |
| 99 | crc=((crc << 8) + 511+ |
| 100 | (crc >> (8*sizeof(ha_checksum)-8))); |
| 101 | continue; |
| 102 | } |
| 103 | } |
| 104 | pos= record+keyseg->start; |
| 105 | if (keyseg->flag & HA_VAR_LENGTH_PART) |
| 106 | { |
| 107 | uint pack_length= keyseg->bit_start; |
| 108 | uint tmp_length= (pack_length == 1 ? (uint) *(uchar*) pos : |
| 109 | uint2korr(pos)); |
| 110 | pos+= pack_length; /* Skip VARCHAR length */ |
| 111 | set_if_smaller(length,tmp_length); |
| 112 | } |
| 113 | else if (keyseg->flag & HA_BLOB_PART) |
| 114 | { |
| 115 | uint tmp_length=_mi_calc_blob_length(keyseg->bit_start,pos); |
| 116 | memcpy((char**) &pos, pos+keyseg->bit_start, sizeof(char*)); |
| 117 | if (!length || length > tmp_length) |
| 118 | length=tmp_length; /* The whole blob */ |
| 119 | } |
| 120 | end= pos+length; |
| 121 | if (type == HA_KEYTYPE_TEXT || type == HA_KEYTYPE_VARTEXT1 || |
| 122 | type == HA_KEYTYPE_VARTEXT2) |
| 123 | { |
| 124 | keyseg->charset->coll->hash_sort(keyseg->charset, |
| 125 | (const uchar*) pos, length, &seed1, |
| 126 | &seed2); |
| 127 | crc^= seed1; |
| 128 | } |
| 129 | else |
| 130 | while (pos != end) |
| 131 | crc=((crc << 8) + |
| 132 | (((uchar) *(uchar*) pos++))) + |
| 133 | (crc >> (8*sizeof(ha_checksum)-8)); |
| 134 | } |
| 135 | return crc; |
| 136 | } |
| 137 | |
| 138 | |
| 139 | /* |
| 140 | compare unique key for two rows |
| 141 | |
| 142 | TODO |
| 143 | Add support for bit fields |
| 144 | |
| 145 | RETURN |
| 146 | 0 if both rows have equal unique value |
| 147 | # Rows are different |
| 148 | */ |
| 149 | |
| 150 | int mi_unique_comp(MI_UNIQUEDEF *def, const uchar *a, const uchar *b, |
| 151 | my_bool null_are_equal) |
| 152 | { |
| 153 | const uchar *pos_a, *pos_b, *end; |
| 154 | HA_KEYSEG *keyseg; |
| 155 | |
| 156 | for (keyseg=def->seg ; keyseg < def->end ; keyseg++) |
| 157 | { |
| 158 | enum ha_base_keytype type=(enum ha_base_keytype) keyseg->type; |
| 159 | uint a_length, b_length; |
| 160 | a_length= b_length= keyseg->length; |
| 161 | |
| 162 | /* If part is NULL it's regarded as different */ |
| 163 | if (keyseg->null_bit) |
| 164 | { |
| 165 | uint tmp; |
| 166 | if ((tmp=(a[keyseg->null_pos] & keyseg->null_bit)) != |
| 167 | (uint) (b[keyseg->null_pos] & keyseg->null_bit)) |
| 168 | return 1; |
| 169 | if (tmp) |
| 170 | { |
| 171 | if (!null_are_equal) |
| 172 | return 1; |
| 173 | continue; |
| 174 | } |
| 175 | } |
| 176 | pos_a= a+keyseg->start; |
| 177 | pos_b= b+keyseg->start; |
| 178 | if (keyseg->flag & HA_VAR_LENGTH_PART) |
| 179 | { |
| 180 | uint pack_length= keyseg->bit_start; |
| 181 | if (pack_length == 1) |
| 182 | { |
| 183 | a_length= (uint) *(uchar*) pos_a++; |
| 184 | b_length= (uint) *(uchar*) pos_b++; |
| 185 | } |
| 186 | else |
| 187 | { |
| 188 | a_length= uint2korr(pos_a); |
| 189 | b_length= uint2korr(pos_b); |
| 190 | pos_a+= 2; /* Skip VARCHAR length */ |
| 191 | pos_b+= 2; |
| 192 | } |
| 193 | set_if_smaller(a_length, keyseg->length); /* Safety */ |
| 194 | set_if_smaller(b_length, keyseg->length); /* safety */ |
| 195 | } |
| 196 | else if (keyseg->flag & HA_BLOB_PART) |
| 197 | { |
| 198 | /* Only compare 'length' characters if length != 0 */ |
| 199 | a_length= _mi_calc_blob_length(keyseg->bit_start,pos_a); |
| 200 | b_length= _mi_calc_blob_length(keyseg->bit_start,pos_b); |
| 201 | /* Check that a and b are of equal length */ |
| 202 | if (keyseg->length) |
| 203 | { |
| 204 | /* |
| 205 | This is used in some cases when we are not interested in comparing |
| 206 | the whole length of the blob. |
| 207 | */ |
| 208 | set_if_smaller(a_length, keyseg->length); |
| 209 | set_if_smaller(b_length, keyseg->length); |
| 210 | } |
| 211 | memcpy((char**) &pos_a, pos_a+keyseg->bit_start, sizeof(char*)); |
| 212 | memcpy((char**) &pos_b, pos_b+keyseg->bit_start, sizeof(char*)); |
| 213 | } |
| 214 | if (type == HA_KEYTYPE_TEXT || type == HA_KEYTYPE_VARTEXT1 || |
| 215 | type == HA_KEYTYPE_VARTEXT2) |
| 216 | { |
| 217 | if (ha_compare_text(keyseg->charset, (uchar *) pos_a, a_length, |
| 218 | (uchar *) pos_b, b_length, 0)) |
| 219 | return 1; |
| 220 | } |
| 221 | else |
| 222 | { |
| 223 | if (a_length != b_length) |
| 224 | return 1; |
| 225 | end= pos_a+a_length; |
| 226 | while (pos_a != end) |
| 227 | { |
| 228 | if (*pos_a++ != *pos_b++) |
| 229 | return 1; |
| 230 | } |
| 231 | } |
| 232 | } |
| 233 | return 0; |
| 234 | } |
| 235 | |