summaryrefslogtreecommitdiff
path: root/src/mongo/db/index/index_access_method.h
blob: df62c4ae9371cdfc3fd0f720c0ebcf394282d79f (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
/**
*    Copyright (C) 2013-2014 MongoDB Inc.
*
*    This program is free software: you can redistribute it and/or  modify
*    it under the terms of the GNU Affero General Public License, version 3,
*    as published by the Free Software Foundation.
*
*    This program is distributed in the hope that it will be useful,
*    but WITHOUT ANY WARRANTY; without even the implied warranty of
*    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
*    GNU Affero General Public License for more details.
*
*    You should have received a copy of the GNU Affero General Public License
*    along with this program.  If not, see <http://www.gnu.org/licenses/>.
*
*    As a special exception, the copyright holders give permission to link the
*    code of portions of this program with the OpenSSL library under certain
*    conditions as described in each individual source file and distribute
*    linked combinations including the program with the OpenSSL library. You
*    must comply with the GNU Affero General Public License in all respects for
*    all of the code used other than as permitted herein. If you modify file(s)
*    with this exception, you may extend this exception to your version of the
*    file(s), but you are not obligated to do so. If you do not wish to do so,
*    delete this exception statement from your version. If you delete this
*    exception statement from all source files in the program, then also delete
*    it in the license file.
*/

#pragma once

#include <memory>

#include "mongo/base/disallow_copying.h"
#include "mongo/db/index/index_descriptor.h"
#include "mongo/db/jsobj.h"
#include "mongo/db/operation_context.h"
#include "mongo/db/record_id.h"
#include "mongo/db/sorter/sorter.h"
#include "mongo/db/storage/sorted_data_interface.h"

namespace mongo {

class BSONObjBuilder;
class MatchExpression;
class UpdateTicket;
struct InsertDeleteOptions;

/**
 * An IndexAccessMethod is the interface through which all the mutation, lookup, and
 * traversal of index entries is done. The class is designed so that the underlying index
 * data structure is opaque to the caller.
 *
 * IndexAccessMethods for existing indices are obtained through the system catalog.
 *
 * We assume the caller has whatever locks required.  This interface is not thread safe.
 *
 */
class IndexAccessMethod {
    MONGO_DISALLOW_COPYING(IndexAccessMethod);

public:
    IndexAccessMethod(IndexCatalogEntry* btreeState, SortedDataInterface* btree);
    virtual ~IndexAccessMethod() {}

    //
    // Lookup, traversal, and mutation support
    //

    /**
     * Internally generate the keys {k1, ..., kn} for 'obj'.  For each key k, insert (k ->
     * 'loc') into the index.  'obj' is the object at the location 'loc'.  If not NULL,
     * 'numInserted' will be set to the number of keys added to the index for the document.  If
     * there is more than one key for 'obj', either all keys will be inserted or none will.
     *
     * The behavior of the insertion can be specified through 'options'.
     */
    Status insert(OperationContext* txn,
                  const BSONObj& obj,
                  const RecordId& loc,
                  const InsertDeleteOptions& options,
                  int64_t* numInserted);

    /**
     * Analogous to above, but remove the records instead of inserting them.  If not NULL,
     * numDeleted will be set to the number of keys removed from the index for the document.
     */
    Status remove(OperationContext* txn,
                  const BSONObj& obj,
                  const RecordId& loc,
                  const InsertDeleteOptions& options,
                  int64_t* numDeleted);

    /**
     * Checks whether the index entries for the document 'from', which is placed at location
     * 'loc' on disk, can be changed to the index entries for the doc 'to'. Provides a ticket
     * for actually performing the update.
     *
     * Returns an error if the update is invalid.  The ticket will also be marked as invalid.
     * Returns OK if the update should proceed without error.  The ticket is marked as valid.
     *
     * There is no obligation to perform the update after performing validation.
     */
    Status validateUpdate(OperationContext* txn,
                          const BSONObj& from,
                          const BSONObj& to,
                          const RecordId& loc,
                          const InsertDeleteOptions& options,
                          UpdateTicket* ticket,
                          const MatchExpression* indexFilter);

    /**
     * Perform a validated update.  The keys for the 'from' object will be removed, and the keys
     * for the object 'to' will be added.  Returns OK if the update succeeded, failure if it did
     * not.  If an update does not succeed, the index will be unmodified, and the keys for
     * 'from' will remain.  Assumes that the index has not changed since validateUpdate was
     * called.  If the index was changed, we may return an error, as our ticket may have been
     * invalidated.
     */
    Status update(OperationContext* txn, const UpdateTicket& ticket, int64_t* numUpdated);

    /**
     * Returns an unpositioned cursor over 'this' index.
     */
    std::unique_ptr<SortedDataInterface::Cursor> newCursor(OperationContext* txn,
                                                           bool isForward = true) const;
    /**
     * Returns a pseudo-random cursor over 'this' index.
     */
    std::unique_ptr<SortedDataInterface::Cursor> newRandomCursor(OperationContext* txn) const;

    // ------ index level operations ------


    /**
     * initializes this index
     * only called once for the lifetime of the index
     * if called multiple times, is an error
     */
    Status initializeAsEmpty(OperationContext* txn);

    /**
     * Try to page-in the pages that contain the keys generated from 'obj'.
     * This can be used to speed up future accesses to an index by trying to ensure the
     * appropriate pages are not swapped out.
     * See prefetch.cpp.
     */
    Status touch(OperationContext* txn, const BSONObj& obj);

    /**
     * this pages in the entire index
     */
    Status touch(OperationContext* txn) const;

    /**
     * Walk the entire index, checking the internal structure for consistency.
     * Set numKeys to the number of keys in the index.
     *
     * 'output' is used to store results of validate when 'full' is true.
     * If 'full' is false, 'output' may be NULL.
     *
     * Return OK if the index is valid.
     *
     * Currently wasserts that the index is invalid.  This could/should be changed in
     * the future to return a Status.
     */
    Status validate(OperationContext* txn, bool full, int64_t* numKeys, BSONObjBuilder* output);

    /**
     * Add custom statistics about this index to BSON object builder, for display.
     *
     * 'scale' is a scaling factor to apply to all byte statistics.
     *
     * Returns true if stats were appended.
     */
    bool appendCustomStats(OperationContext* txn, BSONObjBuilder* result, double scale) const;

    /**
     * @return The number of bytes consumed by this index.
     *         Exactly what is counted is not defined based on padding, re-use, etc...
     */
    long long getSpaceUsedBytes(OperationContext* txn) const;

    RecordId findSingle(OperationContext* txn, const BSONObj& key) const;

    //
    // Bulk operations support
    //

    class BulkBuilder {
    public:
        /**
         * Insert into the BulkBuilder as-if inserting into an IndexAccessMethod.
         */
        Status insert(OperationContext* txn,
                      const BSONObj& obj,
                      const RecordId& loc,
                      const InsertDeleteOptions& options,
                      int64_t* numInserted);

    private:
        friend class IndexAccessMethod;

        using Sorter = mongo::Sorter<BSONObj, RecordId>;

        BulkBuilder(const IndexAccessMethod* index, const IndexDescriptor* descriptor);

        std::unique_ptr<Sorter> _sorter;
        const IndexAccessMethod* _real;
        int64_t _keysInserted = 0;
        bool _isMultiKey = false;
    };

    /**
     * Starts a bulk operation.
     * You work on the returned BulkBuilder and then call commitBulk.
     * This can return NULL, meaning bulk mode is not available.
     *
     * It is only legal to initiate bulk when the index is new and empty.
     */
    std::unique_ptr<BulkBuilder> initiateBulk();

    /**
     * Call this when you are ready to finish your bulk work.
     * Pass in the BulkBuilder returned from initiateBulk.
     * @param bulk - something created from initiateBulk
     * @param mayInterrupt - is this commit interruptable (will cancel)
     * @param dupsAllowed - if false, error or fill 'dups' if any duplicate values are found
     * @param dups - if NULL, error out on dups if not allowed
     *               if not NULL, put the bad RecordIds there
     */
    Status commitBulk(OperationContext* txn,
                      std::unique_ptr<BulkBuilder> bulk,
                      bool mayInterrupt,
                      bool dupsAllowed,
                      std::set<RecordId>* dups);

    /**
     * Fills 'keys' with the keys that should be generated for 'obj' on this index.
     */
    virtual void getKeys(const BSONObj& obj, BSONObjSet* keys) const = 0;

protected:
    // Determines whether it's OK to ignore ErrorCodes::KeyTooLong for this OperationContext
    bool ignoreKeyTooLong(OperationContext* txn);

    IndexCatalogEntry* _btreeState;  // owned by IndexCatalogEntry
    const IndexDescriptor* _descriptor;

private:
    void removeOneKey(OperationContext* txn,
                      const BSONObj& key,
                      const RecordId& loc,
                      bool dupsAllowed);

    const std::unique_ptr<SortedDataInterface> _newInterface;
};

/**
 * Updates are two steps: verify that it's a valid update, and perform it.
 * validateUpdate fills out the UpdateStatus and update actually applies it.
 */
class UpdateTicket {
    // No public interface
private:
    friend class IndexAccessMethod;

    bool _isValid;

    BSONObjSet oldKeys;
    BSONObjSet newKeys;

    // These point into the sets oldKeys and newKeys.
    std::vector<BSONObj*> removed;
    std::vector<BSONObj*> added;

    RecordId loc;
    bool dupsAllowed;
};

/**
 * Flags we can set for inserts and deletes (and updates, which are kind of both).
 */
struct InsertDeleteOptions {
    InsertDeleteOptions() : logIfError(false), dupsAllowed(false) {}

    // If there's an error, log() it.
    bool logIfError;

    // Are duplicate keys allowed in the index?
    bool dupsAllowed;
};

}  // namespace mongo