标签：
neo4j 中对于字符串等变长值的保存策略是用一组定长的 block 来保存，block之间用单向链表链接。类 AbstractDynamicStore 实现了该功能，下面是其注释说明。
/**
* An abstract representation of a dynamic store. The difference between a
* normal AbstractStore and a AbstractDynamicStore is
* that the size of a record/entry can be dynamic.
* Instead of a fixed record this class uses blocks to store a record. If a
* record size is greater than the block size the record will use one or more
* blocks to store its data.
* A dynamic store don’t have a IdGenerator because the position of a
* record can’t be calculated just by knowing the id. Instead one should use a
* AbstractStore and store the start block of the record located in the
* dynamic store. Note: This class makes use of an id generator internally for
* managing free and non free blocks.
* Note, the first block of a dynamic store is reserved and contains information
* about the store.
*/
AbstractDynamicStore 类对应的存储文件格式如上图所示, 整个文件是有一个block_size=BLOCK_HEADER_SIZE(8Bytes)+block_content_size的定长数组和一个字符串“StringPropertyStore v0.A.2”或“ArrayPropertyStore v0.A.2”或“SchemaStore v0.A.2”(文件类型描述TYPE_DESCRIPTOR和 neo4j 的 ALL_STORES_VERSION构成)。访问时，可以通过 id 作为数组的下标进行访问。其中，文件的第1个 record 中前4 字节用来保存 block_size。文件的第2个 record开始保存实际的block数据,它由8个字节的block_header和定长的 block_content（可配置）构成. block_header 结构如下:
[x__ , ] 0: start record, 1: linked record
[ x, ] inUse
[ ,xxxx] high next block bits
下面看一下 AbstractDynamicStore.java 中 getRecord() 和readAndVerifyBlockSize() 成员函数，可以帮助理解 DynamicStore 的存储格式。
|
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
|
private DynamicRecord getRecord( long blockId, PersistenceWindow window, RecordLoad load )</pre><div>{DynamicRecord record = new DynamicRecord( blockId );Buffer buffer = window.getOffsettedBuffer( blockId );/** First 4b* [x , ][ , ][ , ][ , ] 0: start record, 1: linked record* [ x, ][ , ][ , ][ , ] inUse* [ ,xxxx][ , ][ , ][ , ] high next block bits* [ , ][xxxx,xxxx][xxxx,xxxx][xxxx,xxxx] nr of bytes in the data field in this record**/long firstInteger = buffer.getUnsignedInt();boolean isStartRecord = (firstInteger & 0x80000000) == 0;long maskedInteger = firstInteger & ~0x80000000;int highNibbleInMaskedInteger = (int) ( ( maskedInteger ) >> 28 );boolean inUse = highNibbleInMaskedInteger == Record.IN_USE.intValue();if ( !inUse && load != RecordLoad.FORCE ){throw new InvalidRecordException( "DynamicRecord Not in use, blockId[" + blockId + "]" );}int dataSize = getBlockSize() - BLOCK_HEADER_SIZE;int nrOfBytes = (int) ( firstInteger & 0xFFFFFF );/** Pointer to next block 4b (low bits of the pointer)*/long nextBlock = buffer.getUnsignedInt();long nextModifier = ( firstInteger & 0xF000000L ) << 8;long longNextBlock = longFromIntAndMod( nextBlock, nextModifier );boolean readData = load != RecordLoad.CHECK;if ( longNextBlock != Record.NO_NEXT_BLOCK.intValue()&& nrOfBytes < dataSize || nrOfBytes > dataSize ){readData = false;if ( load != RecordLoad.FORCE ){throw new InvalidRecordException( "Next block set[" + nextBlock+ "] current block illegal size[" + nrOfBytes + "/" + dataSize + "]" );}}record.setInUse( inUse );record.setStartRecord( isStartRecord );record.setLength( nrOfBytes );record.setNextBlock( longNextBlock );/** Data ‘nrOfBytes‘ bytes*/if ( readData ){byte byteArrayElement[] = new byte[nrOfBytes];buffer.get( byteArrayElement );record.setData( byteArrayElement );}return record;} |
|
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
|
</pre><div>protected void readAndVerifyBlockSize() throws IOException{ByteBuffer buffer = ByteBuffer.allocate( 4 );getFileChannel().position( 0 );getFileChannel().read( buffer );buffer.flip();blockSize = buffer.getInt();if ( blockSize <= 0 ){throw new InvalidRecordException( "Illegal block size: " +blockSize + " in " + getStorageFileName() );}} |
类SchemaStore，DynamicArrayStore(ArrayPropertyStore), DynamicStringStore(StringPropertyStore)都是继承成自类AbstractDynamicStore，所以与类DynamicArrayStore, DynamicStringStore和 SchemaStore对应文件的存储格式，都是遵循AbstractDynamicStore的存储格式，除了block块的大小（block_size）不同外。
| db 文件 | 存储类型 | block_size |
| neostore.labeltokenstore.db.names | StringPropertyStore | NAME_STORE_BLOCK_SIZE=30 |
| neostore.propertystore.db.index.keys | StringPropertyStore | NAME_STORE_BLOCK_SIZE=30 |
| neostore.relationshiptypestore.db.names | StringPropertyStore | NAME_STORE_BLOCK_SIZE=30 |
| neostore.propertystore.db.strings | StringPropertyStore | string_block_size=120 |
| neostore.nodestore.db.labels | ArrayPropertyStore | label_block_size=60 |
| neostore.propertystore.db.arrays | ArrayPropertyStore | array_block_size=120 |
| neostore.schemastore.db | SchemaStore | BLOCK_SIZE=56 |
block_size 通过配置文件或缺省值来设置的，下面的代码片段展示了neostore.propertystore.db.strings 文件的创建过程及block_size 的大小如何传入。
1) GraphDatabaseSettings.java
|
1
2
3
4
5
6
7
|
</pre><div>public static final Setting string_block_size = setting("string_block_size", INTEGER, "120",min(1));public static final Setting array_block_size = setting("array_block_size", INTEGER, "120",min(1));public static final Setting label_block_size = setting("label_block_size", INTEGER, "60",min(1));</div><pre> |
|
1
2
3
4
5
6
7
8
9
10
11
12
13
14
|
</pre><div>public static abstract class Configuration{public static final Setting string_block_size = GraphDatabaseSettings.string_block_size;public static final Setting array_block_size = GraphDatabaseSettings.array_block_size;public static final Setting label_block_size = GraphDatabaseSettings.label_block_size;public static final Setting dense_node_threshold = GraphDatabaseSettings.dense_node_threshold;} |
3) StoreFactory.java的createPropertyStore 函数
|
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
|
</pre><div>public void createPropertyStore( File fileName ){createEmptyStore( fileName, buildTypeDescriptorAndVersion( PropertyStore.TYPE_DESCRIPTOR ));int stringStoreBlockSize = config.get( Configuration.string_block_size );int arrayStoreBlockSize = config.get( Configuration.array_block_size )createDynamicStringStore(new File( fileName.getPath() + STRINGS_PART), stringStoreBlockSize, IdType.STRING_BLOCK);createPropertyKeyTokenStore( new File( fileName.getPath() + INDEX_PART ) );createDynamicArrayStore( new File( fileName.getPath() + ARRAYS_PART ), arrayStoreBlockSize );} |
4) StoreFactory.java的createDynamicStringStore函数
|
1
2
3
4
5
6
7
8
|
</pre><div>private void createDynamicStringStore( File fileName, int blockSize, IdType idType ){createEmptyDynamicStore(fileName, blockSize, DynamicStringStore.VERSION, idType);} |
5) StoreFactory.java的createEmptyDynamicStore 函数
|
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
|
</pre><div>/*** Creates a new empty store. A factory method returning an implementation* should make use of this method to initialize an empty store. Block size* must be greater than zero. Not that the first block will be marked as* reserved (contains info about the block size). There will be an overhead* for each block of <CODE>AbstractDynamicStore.BLOCK_HEADER_SIZE</CODE>bytes.*/public void createEmptyDynamicStore( File fileName, int baseBlockSize,String typeAndVersionDescriptor, IdType idType){int blockSize = baseBlockSize;// sanity checks…blockSize += AbstractDynamicStore.BLOCK_HEADER_SIZE;// write the headertry{FileChannel channel = fileSystemAbstraction.create(fileName);int endHeaderSize = blockSize+ UTF8.encode( typeAndVersionDescriptor ).length;ByteBuffer buffer = ByteBuffer.allocate( endHeaderSize );buffer.putInt( blockSize );buffer.position( endHeaderSize - typeAndVersionDescriptor.length() );buffer.put( UTF8.encode( typeAndVersionDescriptor ) ).flip();channel.write( buffer );channel.force( false );channel.close();}catch ( IOException e ){throw new UnderlyingStorageException( "Unable to create store "+ fileName, e );}idGeneratorFactory.create( fileSystemAbstraction, new File( fileName.getPath() + ".id"), 0 );// TODO highestIdInUse = 0 works now, but not when slave can create store files.IdGenerator idGenerator = idGeneratorFactory.open(fileSystemAbstraction,new File( fileName.getPath() + ".id"),idType.getGrabSize(), idType, 0 );idGenerator.nextId(); // reserve first for blockSizeidGenerator.close();} |
Graph database_neo4j 底层存储结构分析(4)
标签：
原文地址：http://www.cnblogs.com/gisblogs/p/4545754.html
