/usr/share/gocode/src/github.com/influxdata/influxdb/cmd/influx_tsm/bz1/reader.go is in golang-github-influxdb-influxdb-dev 1.1.1+dfsg1-4.
This file is owned by root:root, with mode 0o644.
The actual contents of the file can be viewed below.
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 | package bz1 // import "github.com/influxdata/influxdb/cmd/influx_tsm/bz1"
import (
"bytes"
"encoding/binary"
"encoding/json"
"fmt"
"math"
"sort"
"time"
"github.com/boltdb/bolt"
"github.com/golang/snappy"
"github.com/influxdata/influxdb/cmd/influx_tsm/stats"
"github.com/influxdata/influxdb/cmd/influx_tsm/tsdb"
"github.com/influxdata/influxdb/tsdb/engine/tsm1"
)
// DefaultChunkSize is the size of chunks read from the bz1 shard
const DefaultChunkSize = 1000
// Reader is used to read all data from a bz1 shard.
type Reader struct {
path string
db *bolt.DB
tx *bolt.Tx
cursors []*cursor
currCursor int
keyBuf string
values []tsm1.Value
valuePos int
fields map[string]*tsdb.MeasurementFields
codecs map[string]*tsdb.FieldCodec
stats *stats.Stats
}
// NewReader returns a reader for the bz1 shard at path.
func NewReader(path string, stats *stats.Stats, chunkSize int) *Reader {
r := &Reader{
path: path,
fields: make(map[string]*tsdb.MeasurementFields),
codecs: make(map[string]*tsdb.FieldCodec),
stats: stats,
}
if chunkSize <= 0 {
chunkSize = DefaultChunkSize
}
r.values = make([]tsm1.Value, chunkSize)
return r
}
// Open opens the reader.
func (r *Reader) Open() error {
// Open underlying storage.
db, err := bolt.Open(r.path, 0666, &bolt.Options{Timeout: 1 * time.Second})
if err != nil {
return err
}
r.db = db
seriesSet := make(map[string]bool)
if err := r.db.View(func(tx *bolt.Tx) error {
var data []byte
meta := tx.Bucket([]byte("meta"))
if meta == nil {
// No data in this shard.
return nil
}
pointsBucket := tx.Bucket([]byte("points"))
if pointsBucket == nil {
return nil
}
if err := pointsBucket.ForEach(func(key, _ []byte) error {
seriesSet[string(key)] = true
return nil
}); err != nil {
return err
}
buf := meta.Get([]byte("fields"))
if buf == nil {
// No data in this shard.
return nil
}
data, err = snappy.Decode(nil, buf)
if err != nil {
return err
}
if err := json.Unmarshal(data, &r.fields); err != nil {
return err
}
return nil
}); err != nil {
return err
}
// Build the codec for each measurement.
for k, v := range r.fields {
r.codecs[k] = tsdb.NewFieldCodec(v.Fields)
}
r.tx, err = r.db.Begin(false)
if err != nil {
return err
}
// Create cursor for each field of each series.
for s := range seriesSet {
measurement := tsdb.MeasurementFromSeriesKey(s)
fields := r.fields[measurement]
if fields == nil {
r.stats.IncrFiltered()
continue
}
for _, f := range fields.Fields {
c := newCursor(r.tx, s, f.Name, r.codecs[measurement])
if c == nil {
continue
}
c.SeekTo(0)
r.cursors = append(r.cursors, c)
}
}
sort.Sort(cursors(r.cursors))
return nil
}
// Next returns whether there is any more data to be read.
func (r *Reader) Next() bool {
r.valuePos = 0
OUTER:
for {
if r.currCursor >= len(r.cursors) {
// All cursors drained. No more data remains.
return false
}
cc := r.cursors[r.currCursor]
r.keyBuf = tsm1.SeriesFieldKey(cc.series, cc.field)
for {
k, v := cc.Next()
if k == -1 {
// Go to next cursor and try again.
r.currCursor++
if r.valuePos == 0 {
// The previous cursor had no data. Instead of returning
// just go immediately to the next cursor.
continue OUTER
}
// There is some data available. Indicate that it should be read.
return true
}
if f, ok := v.(float64); ok {
if math.IsInf(f, 0) {
r.stats.AddPointsRead(1)
r.stats.IncrInf()
continue
}
if math.IsNaN(f) {
r.stats.AddPointsRead(1)
r.stats.IncrNaN()
continue
}
}
r.values[r.valuePos] = tsm1.NewValue(k, v)
r.valuePos++
if r.valuePos >= len(r.values) {
return true
}
}
}
}
// Read returns the next chunk of data in the shard, converted to tsm1 values. Data is
// emitted completely for every field, in every series, before the next field is processed.
// Data from Read() adheres to the requirements for writing to tsm1 shards
func (r *Reader) Read() (string, []tsm1.Value, error) {
return r.keyBuf, r.values[:r.valuePos], nil
}
// Close closes the reader.
func (r *Reader) Close() error {
r.tx.Rollback()
return r.db.Close()
}
// cursor provides ordered iteration across a series.
type cursor struct {
cursor *bolt.Cursor
buf []byte // uncompressed buffer
off int // buffer offset
fieldIndices []int
index int
series string
field string
dec *tsdb.FieldCodec
keyBuf int64
valBuf interface{}
}
// newCursor returns an instance of a bz1 cursor.
func newCursor(tx *bolt.Tx, series string, field string, dec *tsdb.FieldCodec) *cursor {
// Retrieve points bucket. Ignore if there is no bucket.
b := tx.Bucket([]byte("points")).Bucket([]byte(series))
if b == nil {
return nil
}
return &cursor{
cursor: b.Cursor(),
series: series,
field: field,
dec: dec,
keyBuf: -2,
}
}
// Seek moves the cursor to a position.
func (c *cursor) SeekTo(seek int64) {
var seekBytes [8]byte
binary.BigEndian.PutUint64(seekBytes[:], uint64(seek))
// Move cursor to appropriate block and set to buffer.
k, v := c.cursor.Seek(seekBytes[:])
if v == nil { // get the last block, it might have this time
_, v = c.cursor.Last()
} else if seek < int64(binary.BigEndian.Uint64(k)) { // the seek key is less than this block, go back one and check
_, v = c.cursor.Prev()
// if the previous block max time is less than the seek value, reset to where we were originally
if v == nil || seek > int64(binary.BigEndian.Uint64(v[0:8])) {
_, v = c.cursor.Seek(seekBytes[:])
}
}
c.setBuf(v)
// Read current block up to seek position.
c.seekBuf(seekBytes[:])
// Return current entry.
c.keyBuf, c.valBuf = c.read()
}
// seekBuf moves the cursor to a position within the current buffer.
func (c *cursor) seekBuf(seek []byte) (key, value []byte) {
for {
// Slice off the current entry.
buf := c.buf[c.off:]
// Exit if current entry's timestamp is on or after the seek.
if len(buf) == 0 {
return
}
if bytes.Compare(buf[0:8], seek) != -1 {
return
}
c.off += entryHeaderSize + entryDataSize(buf)
}
}
// Next returns the next key/value pair from the cursor. If there are no values
// remaining, -1 is returned.
func (c *cursor) Next() (int64, interface{}) {
for {
k, v := func() (int64, interface{}) {
if c.keyBuf != -2 {
k, v := c.keyBuf, c.valBuf
c.keyBuf = -2
return k, v
}
// Ignore if there is no buffer.
if len(c.buf) == 0 {
return -1, nil
}
// Move forward to next entry.
c.off += entryHeaderSize + entryDataSize(c.buf[c.off:])
// If no items left then read first item from next block.
if c.off >= len(c.buf) {
_, v := c.cursor.Next()
c.setBuf(v)
}
return c.read()
}()
if k != -1 && v == nil {
// There is a point in the series at the next timestamp,
// but not for this cursor's field. Go to the next point.
continue
}
return k, v
}
}
// setBuf saves a compressed block to the buffer.
func (c *cursor) setBuf(block []byte) {
// Clear if the block is empty.
if len(block) == 0 {
c.buf, c.off, c.fieldIndices, c.index = c.buf[0:0], 0, c.fieldIndices[0:0], 0
return
}
// Otherwise decode block into buffer.
// Skip over the first 8 bytes since they are the max timestamp.
buf, err := snappy.Decode(nil, block[8:])
if err != nil {
c.buf = c.buf[0:0]
fmt.Printf("block decode error: %s\n", err)
}
c.buf, c.off = buf, 0
}
// read reads the current key and value from the current block.
func (c *cursor) read() (key int64, value interface{}) {
// Return nil if the offset is at the end of the buffer.
if c.off >= len(c.buf) {
return -1, nil
}
// Otherwise read the current entry.
buf := c.buf[c.off:]
dataSize := entryDataSize(buf)
return tsdb.DecodeKeyValue(c.field, c.dec, buf[0:8], buf[entryHeaderSize:entryHeaderSize+dataSize])
}
// Sort bz1 cursors in correct order for writing to TSM files.
type cursors []*cursor
func (a cursors) Len() int { return len(a) }
func (a cursors) Swap(i, j int) { a[i], a[j] = a[j], a[i] }
func (a cursors) Less(i, j int) bool {
if a[i].series == a[j].series {
return a[i].field < a[j].field
}
return a[i].series < a[j].series
}
// entryHeaderSize is the number of bytes required for the header.
const entryHeaderSize = 8 + 4
// entryDataSize returns the size of an entry's data field, in bytes.
func entryDataSize(v []byte) int { return int(binary.BigEndian.Uint32(v[8:12])) }
|