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
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
|
<!-- Copyright Vespa.ai. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root. -->
<html lang="en">
<head>
<title>Developers guide to the serialized document format</title>
</head>
<body>
<h1>Developers guide to the serialized document format</h1>
<p>
When a Vespa document is stored or transferred from one application to
another, it is serialized. The serialization format tries to achieve
serialization robustness and speed. The most important fields are kept in
a header that is accessible at low cost. The other fields are located by
table look-ups.
</p>
<h2>Purpose</h2>
<p>The purpose of the serialized format is</p>
<ul>
<li><b>Robustness</b>. The format shall detect errors gracefully.</li>
<li>
<b>Speed</b>. Deserialization shall be fast, especially for basic fields
like <b>DocumentId</b>.
</li>
<li>
<b>Size</b>. The serialized format shall be compact and allow for
efficient storage and transfer.
</li>
</ul>
<p><strong>All fields are in network byte order.</strong></p>
<h2>Changelog</h2>
<h3>Current version: 8</h3>
<ul>
<li>
CRC removed from document format. There used to be a 4 byte CRC in the
end of a header or header + body serialization, calculated as a crc32 of
all the other data in the serialization. This CRC was included in the
document length.
</li>
</ul>
<h3>Version: 7</h3>
<ul>
<li>
The document length is now a static sized 4 byte value, instead of a
variable 2,4,8 byte value.
</li>
<li>
(Anything else? I wrote this changelog when bopping from 7 to 8. Dunno
if more was changed in 7.)
</li>
</ul>
<h3>Version: 6</h3>
This is the oldest version that we currently support. No known installation
stores documents with a version smaller than this.
<h2>Document Format</h2>
<p>This is the description of the serialized document format.</p>
<table border="1" cellspacing="0" cellpadding="1%" width="100%">
<caption>
<em>Document serialization format</em>
</caption>
<tr>
<td width="10%"><b>Field</b></td>
<td width="10%"><b>Type</b></td>
<td width="10%"><b>Length</b></td>
<td><b>Description</b></td>
</tr>
<tr>
<td>Version</td>
<td>Short integer</td>
<td>2</td>
<td>Version number. Current is 6.</td>
</tr>
<tr>
<td>Length</td>
<td>Integer</td>
<td>4 bytes</td>
<td>Total length of object (excluding this field and version).</td>
</tr>
<tr>
<td>Document ID</td>
<td>Bytes</td>
<td> </td>
<td>Unique ID for document. 0-terminated string, UTF-8 encoding.</td>
</tr>
<tr>
<td>Field Map</td>
<td>Bytes</td>
<td>See below</td>
<td>
Placeholder for fields. (Note: Fieldmaps may contain other fieldmaps)
</td>
</tr>
</table>
<p>Field maps are serialized like this</p>
<p></p>
<table border="1" cellspacing="0" cellpadding="1%" width="100%">
<caption>
<em>Fieldmap serialization format</em>
</caption>
<tr>
<td width="10%"><b>Field</b></td>
<td width="10%"><b>Type</b></td>
<td width="10%"><b>Length (bytes)</b></td>
<td><b>Description</b></td>
</tr>
<tr>
<td>Inventory bit mask</td>
<td>Byte</td>
<td>1</td>
<td>
Inventory bits describing the FieldMap element with data:<br />
Bit 0 set: FieldMap has document type <br />
Bit 1 set: FieldMap has header fields <br />
Bit 2 set: FieldMap has body fields <br />
Bit 3 set: FieldMap has external body fields<br />
</td>
</tr>
<tr>
<td colspan="4">
<b>Below section is present when bit 0 of inventory is set</b>
</td>
</tr>
<tr>
<td>Document Type</td>
<td>Bytes</td>
<td> </td>
<td>Document type. (0-terminated string, UTF-8 encoding.)</td>
</tr>
<tr>
<td>Version</td>
<td>Short integer</td>
<td>2</td>
<td>Document type version number.</td>
</tr>
<tr>
<td colspan="4">
<b>Below section is present when bit 1 of inventory is set</b>
</td>
</tr>
<tr>
<td>Header data</td>
<td>Data array</td>
<td>See below</td>
<td>Header data packed in data array</td>
</tr>
<tr>
<td colspan="4">
<b>Below section is present when bit 2 of inventory is set</b>
</td>
</tr>
<tr>
<td>Body data</td>
<td>Data array</td>
<td>See below</td>
<td>Body data packed in data array</td>
</tr>
</table>
<p></p>
<table border="1" cellspacing="0" cellpadding="1%" width="100%">
<caption>
<em>Data array serialization</em>
</caption>
<tr>
<td width="10%"><b>Field</b></td>
<td width="10%"><b>Type</b></td>
<td width="10%"><b>Length (bytes)</b></td>
<td><b>Description</b></td>
</tr>
<tr>
<td>Data length</td>
<td>Integer_2_4_8</td>
<td>2, 4 or 8</td>
<td>
Length of data block (see below). NOTE THAT THIS LENGTH INCLUDE
ITSELF.
</td>
</tr>
<tr>
<td>Number of fields</td>
<td>Integer_1_4</td>
<td>1 or 4</td>
<td>Number of fields in data array</td>
</tr>
<tr>
<td colspan="4">
<b>Below block is repeated "Number of fields" times</b>
</td>
</tr>
<tr>
<td>Field ID</td>
<td>Integer_1_4</td>
<td>1 or 4</td>
<td>ID of field.</td>
</tr>
<tr>
<td>Field Size</td>
<td>Integer_1_2_4</td>
<td>1, 2 or 4</td>
<td>Length of field.</td>
</tr>
<tr>
<td colspan="4"><b>End of repeated block </b></td>
</tr>
<tr>
<td>Data block</td>
<td>Bytes</td>
<td> </td>
<td>
The data block.<br />
- Items are ordered the same way field array is sorted.<br />
- Use lengths from field array above to find item offset and
length.<br />
- If the block is compressed, lengths refer to decompressed
version
</td>
</tr>
</table>
<table border="1" cellspacing="0" cellpadding="1%" width="100%">
<caption>
<em>Data type serialization</em>
</caption>
<tr>
<td width="15%"><b>Data type</b></td>
<td width="10%"><b>Length</b></td>
<td><b>Serialization</b></td>
</tr>
<tr>
<td>Integer (ID 0)</td>
<td>4</td>
<td>Signed integer, two's complement notation, network byte order.</td>
</tr>
<tr>
<td>Floating point number (ID 1)</td>
<td>4</td>
<td>IEEE 754, single precision, network byte order.</td>
</tr>
<tr>
<td>String (ID 2)</td>
<td>1 + (1 or 4) + length + 1</td>
<td>
Strings are serialization format:<br />
- First byte represents coding. This has traditionally denoted
the maximum number of bits per character in the UTF-8 encoded string,
but has never been used in deserialization code.
<ul>
<li>Set to 32 if not used.</li>
<li>
Set to <32 if you know the UTF-8 string uses less bits per
character; e.g. ASCII could use 8.
</li>
<li>
Set bit 6 (decimal 64) if the string has an
<a href="#annotations">annotation tree</a>.
</li>
</ul>
<br />
- Integer_1_4 with length of string. <br />
- The string (UTF-8 encoding), including 0-terminating byte.<br />
- An annotation tree, if bit 6 (decimal 64) of coding byte is
set:
<ul>
<li>
total length of all span trees excl. itself:
<strong>uint32</strong>
</li>
<li>number of span trees <strong>int_1_2_4</strong></li>
<li>
for each root node:
<ol>
<li>tree name serialized as String</li>
<li>
serialized SpanNode as given below, see
<a href="#annotations">annotation serialization</a>
</li>
</ol>
</li>
</ul>
</td>
</tr>
<tr>
<td>Raw bytes (ID 3)</td>
<td>Length of buffer</td>
<td>Byte for byte copy</td>
</tr>
<tr>
<td>Long integer (ID 4)</td>
<td>8</td>
<td>Signed integer, two's complement notation, network byte order.</td>
</tr>
<tr>
<td>Double floating point number (ID 5)</td>
<td>8</td>
<td>IEEE 754, double precision, network byte order.</td>
</tr>
<tr>
<td>Array (ID 6)</td>
<td>At least 8 bytes</td>
<td>
Arrays of any fields are serialized like this:<br />
- 4 bytes: Data type array consists of <br />
- 4 bytes: Number of elements in array<br />
Below sequence is repeated "number of element" times<br />
- 4 bytes: Length of element<br />
- Serialized element<br />
</td>
</tr>
<tr>
<td>Fieldmap (ID 7)</td>
<td> </td>
<td>Field maps (embedded or not) are defined above</td>
</tr>
<tr>
<td>Document (ID 8)</td>
<td> </td>
<td>Document objects (embedded or not) are defined above</td>
</tr>
<tr>
<td>Timestamp (ID 9)</td>
<td> </td>
<td>Same as long integer</td>
</tr>
<tr>
<td>Uri (ID 10)</td>
<td> </td>
<td>Same as string</td>
</tr>
<tr>
<td>Exact string (ID 11)</td>
<td> </td>
<td>Same as string</td>
</tr>
<tr>
<td>Content (ID 12)</td>
<td>At least 11 bytes</td>
<td>
Content fields are serialized like this:<br />
- Content type length (1 byte)<br />
- Content type (0 terminated string, UTF-8 encoding.)<br />
- Content encoding length (1 byte)<br />
- Content encoding (0 terminated string, UTF-8 encoding.)<br />
- Content language length (1 byte)<br />
- Content language (0 terminated string, UTF-8 encoding.)<br />
- Content length (Integer, 4 bytes)<br />
- Content (including 0-terminating char)<br />
</td>
</tr>
<tr>
<td>Content meta (ID 13)</td>
<td>At least 12 bytes</td>
<td>
Content (attachment) meta data are serialized like this:<br />
- Attachment size (Integer, 4 bytes)<br />
- Attachment name (0 terminated string, UTF-8 encoding.)<br />
- Attachment encoding (0 terminated string, UTF-8 encoding.)<br />
- Attachment content type (0 terminated string, UTF-8
encoding.)<br />
- Attachment part (0 terminated string, UTF-8 encoding.)<br />
- Attachment flag (Integer, 4 bytes)<br />
</td>
</tr>
<tr>
<td>Term boost (ID 15)</td>
<td> </td>
<td>Same as string</td>
</tr>
<tr>
<td>Byte (ID 16)</td>
<td>1</td>
<td>One single byte</td>
</tr>
<tr>
<td>Set (ID 17)</td>
<td>At least 8 bytes</td>
<td>
Set of any fields are serialized like this:<br />
- Integer (4 bytes): Data type set is made up of<br />
- Integer (4 bytes): Number of elements in set<br />
Below sequence is repeated "number of element" times<br />
- Serialized element<br />
</td>
</tr>
</table>
<table border="1" cellspacing="0" cellpadding="1%" width="100%">
<caption id="annotations">
<em>Annotation tree serialization</em>
</caption>
<tr>
<td width="15%"><b>Data type</b></td>
<td width="10%"><b>Length</b></td>
<td><b>Serialization</b></td>
</tr>
<tr>
<td>SpanNode (base class)</td>
<td>1 + (1, 2 or 4) + Annotation serialization + subclass payload</td>
<td>
<ul>
<li>
type <strong>byte</strong> (1: Span, 2: SpanList, 4:
AlternateSpanList)
</li>
<li>number of annotations <strong>int_1_2_4</strong></li>
<li>each annotation as given below</li>
<li>
(remaining payload serialized as given below by subclasses Span,
SpanList and AlternateSpanList)
</li>
</ul>
</td>
</tr>
<tr>
<td>Annotation</td>
<td>4 + (1, 2 or 4) + (possibly 4 + FieldValue serialization)</td>
<td>
<ul>
<li>
MD5 name hash (4 LSBytes) <strong>uint32</strong> (NOTE: 0-127
reserved for internal Vespa usage.)
</li>
<li>length <strong>int_1_2_4</strong></li>
<li>
the following fields are <em>only</em> present if length > 0:
<ul>
<li>data type id <strong>uint32</strong></li>
<li>
NOTE: no sequence id, as we will rely on annotations being
serialized/deserialized in particular order, so we don't need
to write this explicitly
</li>
<li>FieldValue as given by its own serialization</li>
</ul>
</li>
</ul>
</td>
</tr>
<tr>
<td>Span</td>
<td>SpanNode serialization + (1, 2 or 4) + (1, 2 or 4)</td>
<td>
<ul>
<li>serialization from SpanNode base class</li>
<li>
from index, as given by Java String (UTF-16)
<strong>int_1_2_4</strong>
</li>
<li>
length, as given by Java String (UTF-16)
<strong>int_1_2_4</strong>
</li>
</ul>
</td>
</tr>
<tr>
<td>SpanList</td>
<td>
SpanNode serialization + (1, 2 or 4) + n times SpanNode serialization
</td>
<td>
<ul>
<li>serialization from SpanNode base class</li>
<li>number of children <strong>int_1_2_4</strong></li>
<li>
each child node serialized as SpanNode (Span, SpanList,
AlternateSpanList)
</li>
</ul>
</td>
</tr>
<tr>
<td>AlternateSpanList</td>
<td>
SpanNode serialization + (1, 2 or 4) + n times (8 + SpanList
serialization)
</td>
<td>
<ul>
<li>serialization from SpanNode base class</li>
<li>number of child trees <strong>int_1_2_4</strong></li>
<li>
for each child tree:
<ul>
<li>probability <strong>double</strong></li>
<li>serialization as given by SpanList above</li>
</ul>
</li>
</ul>
</td>
</tr>
<tr>
<td>AnnotationRef</td>
<td>1, 2 or 4</td>
<td>
AnnotationRef serialization
<ul>
<li>
unique sequence id of annotation being referred to
<strong>int1_2_4</strong>
</li>
</ul>
</td>
</tr>
</table>
<table border="1" cellspacing="0" cellpadding="1%" width="100%">
<caption>
<em>Data types used in serialized format</em>
</caption>
<tr>
<td width="15%"><b>Data type</b></td>
<td><b>Serialization</b></td>
</tr>
<tr>
<td>Integer_1_4</td>
<td>
If bit 7 of first byte is unset, coded using 1 byte.<br />
If bit 7 of first byte is set, coded using 4 bytes (bit 7 of first
byte must be masked away).<br />
<em>Range: 0 - 2**31-1.</em>
</td>
</tr>
<tr>
<td>Integer_1_2_4</td>
<td>
If bit 7 of first byte is unset, coded using 1 byte.<br />
If bit 7 of first byte is set and bit 6 of first byte is unset, coded
using 2 bytes (bit 7 and 6 of first byte must be masked away).<br />
If bit 7 and 6 of first byte are set, coded using 4 bytes (bit 7 and 6
of first byte must be masked away).<br />
<em>Range: 0 - 2**30-1.</em>
</td>
</tr>
<tr>
<td>Integer_2_4_8</td>
<td>
If bit 7 of first byte is unset, coded using 2 byte.<br />
If bit 7 of first byte is set and bit 6 of first byte is unset, coded
using 4 bytes (bit 7 and 6 of first byte must be masked away).<br />
If bit 7 and 6 of first byte are set, coded using 8 bytes (bit 7 and 6
of first byte must be masked away).<br />
<em>Range: 0 - 2**62-1.</em>
</td>
</tr>
</table>
<h2>Document Update Format</h2>
<p>This is the description of the serialized document update format.</p>
<table border="1" cellspacing="0" cellpadding="1%" width="100%">
<caption>
<em>Document update serialization format</em>
</caption>
<tr>
<td width="10%"><b>Field</b></td>
<td width="10%"><b>Type</b></td>
<td width="10%"><b>Length</b></td>
<td><b>Description</b></td>
</tr>
<tr>
<td>Document ID</td>
<td>Bytes</td>
<td> </td>
<td>Unique ID for document. 0-terminated string, UTF-8 encoding.</td>
</tr>
<tr>
<td>Content byte</td>
<td>Byte</td>
<td>1 byte</td>
<td>Always set to 1</td>
</tr>
<tr>
<td>Document Type</td>
<td>Bytes</td>
<td> </td>
<td>Document type. (0-terminated string, UTF-8 encoding.)</td>
</tr>
<tr>
<td>Number of fields to update</td>
<td>Integer</td>
<td>4 bytes</td>
<td>The number of fields to update</td>
</tr>
<tr>
<td>Serialized field updates</td>
<td>Field Update</td>
<td> </td>
<td>The serialized field updates. See below.</td>
</tr>
</table>
<table border="1" cellspacing="0" cellpadding="1%" width="100%">
<caption>
<em>Document update serialization format</em>
</caption>
<tr>
<td width="10%"><b>Field</b></td>
<td width="10%"><b>Type</b></td>
<td width="10%"><b>Length</b></td>
<td><b>Description</b></td>
</tr>
<tr>
<td>Field Id</td>
<td>Integer</td>
<td>4 bytes</td>
<td>Field id within document type.</td>
</tr>
<tr>
<td>Number of value updates</td>
<td>Integer</td>
<td>4 bytes</td>
<td>Numer of value updates to this field.</td>
</tr>
<tr>
<td>Serialized field update values</td>
<td>Bytes</td>
<td> </td>
<td>The serialized field update values. See below.</td>
</tr>
</table>
<table border="1" cellspacing="0" cellpadding="1%" width="100%">
<caption>
<em>Document update value serialization format</em>
</caption>
<tr>
<td width="10%"><b>Field</b></td>
<td width="10%"><b>Type</b></td>
<td width="10%"><b>Length</b></td>
<td><b>Description</b></td>
</tr>
<tr>
<th colspan="4">Add Value Update</th>
</tr>
<tr>
<td>Add Value Update ID</td>
<td>Integer</td>
<td>4 bytes</td>
<td>25 + 0x1000 for value updates.</td>
</tr>
<tr>
<td>Field serialization</td>
<td>FieldValue</td>
<td> </td>
<td>Serialization of the field to add.</td>
</tr>
<tr>
<td>Weight</td>
<td>Integer</td>
<td>4 bytes</td>
<td>Weight. Used if update applies to weighted set.</td>
</tr>
<tr>
<th colspan="4">Arithmetic Update</th>
</tr>
<tr>
<td>Arithmetic Update ID</td>
<td>Integer</td>
<td>4 bytes</td>
<td>26 + 0x1000 for arithmetic updates.</td>
</tr>
<tr>
<td>Operator ID</td>
<td>Integer</td>
<td>4 bytes</td>
<td>Identifies whether this does add, subtract, multiply or divide.</td>
</tr>
<tr>
<td>Operand</td>
<td>Double</td>
<td>8 bytes</td>
<td>The right operand to use in the arithmetic operation.</td>
</tr>
<tr>
<th colspan="4">Assign Update</th>
</tr>
<tr>
<td>Assign Update ID</td>
<td>Integer</td>
<td>4 bytes</td>
<td>27 + 0x1000 for assign updates.</td>
</tr>
<tr>
<td>Content flag</td>
<td>Byte</td>
<td>1 bytes</td>
<td>Contains 1 if we have content, 0 if not.</td>
</tr>
<tr>
<td>Field serialization</td>
<td>FieldValue</td>
<td> </td>
<td>Serialization of the field to assign.</td>
</tr>
<tr>
<th colspan="4">Clear Update</th>
</tr>
<tr>
<td>Clear Update ID</td>
<td>Integer</td>
<td>4 bytes</td>
<td>28 + 0x1000 for clear updates.</td>
</tr>
<tr>
<th colspan="4">Map Value Update</th>
</tr>
<tr>
<td>Map Value Update ID</td>
<td>Integer</td>
<td>4 bytes</td>
<td>29 + 0x1000 for map value updates.</td>
</tr>
<tr>
<td>Field serialization</td>
<td>FieldValue</td>
<td>4 bytes</td>
<td>The field indicating what entry to update.</td>
</tr>
<tr>
<td>Value Update</td>
<td>Document Value Update</td>
<td> </td>
<td>The update operation to apply to the field indicated above.</td>
</tr>
<tr>
<th colspan="4">Remove Value Update</th>
</tr>
<tr>
<td>Remove Update ID</td>
<td>Integer</td>
<td>4 bytes</td>
<td>30 + 0x1000 for remove updates.</td>
</tr>
<tr>
<td>Field serialization</td>
<td>FieldValue</td>
<td> </td>
<td>The field indicating what entry to update.</td>
</tr>
</table>
</body>
</html>
|
