writer.go 6.3 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267
  1. package lz4
  2. import (
  3. "encoding/binary"
  4. "fmt"
  5. "io"
  6. "github.com/pierrec/lz4/internal/xxh32"
  7. )
  8. // Writer implements the LZ4 frame encoder.
  9. type Writer struct {
  10. Header
  11. buf [19]byte // magic number(4) + header(flags(2)+[Size(8)+DictID(4)]+checksum(1)) does not exceed 19 bytes
  12. dst io.Writer // Destination.
  13. checksum xxh32.XXHZero // Frame checksum.
  14. zdata []byte // Compressed data.
  15. data []byte // Data to be compressed.
  16. idx int // Index into data.
  17. hashtable [winSize]int // Hash table used in CompressBlock().
  18. }
  19. // NewWriter returns a new LZ4 frame encoder.
  20. // No access to the underlying io.Writer is performed.
  21. // The supplied Header is checked at the first Write.
  22. // It is ok to change it before the first Write but then not until a Reset() is performed.
  23. func NewWriter(dst io.Writer) *Writer {
  24. return &Writer{dst: dst}
  25. }
  26. // writeHeader builds and writes the header (magic+header) to the underlying io.Writer.
  27. func (z *Writer) writeHeader() error {
  28. // Default to 4Mb if BlockMaxSize is not set.
  29. if z.Header.BlockMaxSize == 0 {
  30. z.Header.BlockMaxSize = bsMapID[7]
  31. }
  32. // The only option that needs to be validated.
  33. bSize := z.Header.BlockMaxSize
  34. bSizeID, ok := bsMapValue[bSize]
  35. if !ok {
  36. return fmt.Errorf("lz4: invalid block max size: %d", bSize)
  37. }
  38. // Allocate the compressed/uncompressed buffers.
  39. // The compressed buffer cannot exceed the uncompressed one.
  40. if n := 2 * bSize; cap(z.zdata) < n {
  41. z.zdata = make([]byte, n, n)
  42. }
  43. z.zdata = z.zdata[:bSize]
  44. z.data = z.zdata[:cap(z.zdata)][bSize:]
  45. z.idx = 0
  46. // Size is optional.
  47. buf := z.buf[:]
  48. // Set the fixed size data: magic number, block max size and flags.
  49. binary.LittleEndian.PutUint32(buf[0:], frameMagic)
  50. flg := byte(Version << 6)
  51. flg |= 1 << 5 // No block dependency.
  52. if z.Header.BlockChecksum {
  53. flg |= 1 << 4
  54. }
  55. if z.Header.Size > 0 {
  56. flg |= 1 << 3
  57. }
  58. if !z.Header.NoChecksum {
  59. flg |= 1 << 2
  60. }
  61. buf[4] = flg
  62. buf[5] = bSizeID << 4
  63. // Current buffer size: magic(4) + flags(1) + block max size (1).
  64. n := 6
  65. // Optional items.
  66. if z.Header.Size > 0 {
  67. binary.LittleEndian.PutUint64(buf[n:], z.Header.Size)
  68. n += 8
  69. }
  70. // The header checksum includes the flags, block max size and optional Size.
  71. buf[n] = byte(xxh32.ChecksumZero(buf[4:n]) >> 8 & 0xFF)
  72. z.checksum.Reset()
  73. // Header ready, write it out.
  74. if _, err := z.dst.Write(buf[0 : n+1]); err != nil {
  75. return err
  76. }
  77. z.Header.done = true
  78. if debugFlag {
  79. debug("wrote header %v", z.Header)
  80. }
  81. return nil
  82. }
  83. // Write compresses data from the supplied buffer into the underlying io.Writer.
  84. // Write does not return until the data has been written.
  85. func (z *Writer) Write(buf []byte) (int, error) {
  86. if !z.Header.done {
  87. if err := z.writeHeader(); err != nil {
  88. return 0, err
  89. }
  90. }
  91. if debugFlag {
  92. debug("input buffer len=%d index=%d", len(buf), z.idx)
  93. }
  94. zn := len(z.data)
  95. var n int
  96. for len(buf) > 0 {
  97. if z.idx == 0 && len(buf) >= zn {
  98. // Avoid a copy as there is enough data for a block.
  99. if err := z.compressBlock(buf[:zn]); err != nil {
  100. return n, err
  101. }
  102. n += zn
  103. buf = buf[zn:]
  104. continue
  105. }
  106. // Accumulate the data to be compressed.
  107. m := copy(z.data[z.idx:], buf)
  108. n += m
  109. z.idx += m
  110. buf = buf[m:]
  111. if debugFlag {
  112. debug("%d bytes copied to buf, current index %d", n, z.idx)
  113. }
  114. if z.idx < len(z.data) {
  115. // Buffer not filled.
  116. if debugFlag {
  117. debug("need more data for compression")
  118. }
  119. return n, nil
  120. }
  121. // Buffer full.
  122. if err := z.compressBlock(z.data); err != nil {
  123. return n, err
  124. }
  125. z.idx = 0
  126. }
  127. return n, nil
  128. }
  129. // compressBlock compresses a block.
  130. func (z *Writer) compressBlock(data []byte) error {
  131. if !z.NoChecksum {
  132. z.checksum.Write(data)
  133. }
  134. // The compressed block size cannot exceed the input's.
  135. var zn int
  136. var err error
  137. if level := z.Header.CompressionLevel; level != 0 {
  138. zn, err = CompressBlockHC(data, z.zdata, level)
  139. } else {
  140. zn, err = CompressBlock(data, z.zdata, z.hashtable[:])
  141. }
  142. var zdata []byte
  143. var bLen uint32
  144. if debugFlag {
  145. debug("block compression %d => %d", len(data), zn)
  146. }
  147. if err == nil && zn > 0 && zn < len(data) {
  148. // Compressible and compressed size smaller than uncompressed: ok!
  149. bLen = uint32(zn)
  150. zdata = z.zdata[:zn]
  151. } else {
  152. // Uncompressed block.
  153. bLen = uint32(len(data)) | compressedBlockFlag
  154. zdata = data
  155. }
  156. if debugFlag {
  157. debug("block compression to be written len=%d data len=%d", bLen, len(zdata))
  158. }
  159. // Write the block.
  160. if err := z.writeUint32(bLen); err != nil {
  161. return err
  162. }
  163. if _, err := z.dst.Write(zdata); err != nil {
  164. return err
  165. }
  166. if z.BlockChecksum {
  167. checksum := xxh32.ChecksumZero(zdata)
  168. if debugFlag {
  169. debug("block checksum %x", checksum)
  170. }
  171. if err := z.writeUint32(checksum); err != nil {
  172. return err
  173. }
  174. }
  175. if debugFlag {
  176. debug("current frame checksum %x", z.checksum.Sum32())
  177. }
  178. return nil
  179. }
  180. // Flush flushes any pending compressed data to the underlying writer.
  181. // Flush does not return until the data has been written.
  182. // If the underlying writer returns an error, Flush returns that error.
  183. func (z *Writer) Flush() error {
  184. if debugFlag {
  185. debug("flush with index %d", z.idx)
  186. }
  187. if z.idx == 0 {
  188. return nil
  189. }
  190. return z.compressBlock(z.data[:z.idx])
  191. }
  192. // Close closes the Writer, flushing any unwritten data to the underlying io.Writer, but does not close the underlying io.Writer.
  193. func (z *Writer) Close() error {
  194. if !z.Header.done {
  195. if err := z.writeHeader(); err != nil {
  196. return err
  197. }
  198. }
  199. if err := z.Flush(); err != nil {
  200. return err
  201. }
  202. if debugFlag {
  203. debug("writing last empty block")
  204. }
  205. if err := z.writeUint32(0); err != nil {
  206. return err
  207. }
  208. if !z.NoChecksum {
  209. checksum := z.checksum.Sum32()
  210. if debugFlag {
  211. debug("stream checksum %x", checksum)
  212. }
  213. if err := z.writeUint32(checksum); err != nil {
  214. return err
  215. }
  216. }
  217. return nil
  218. }
  219. // Reset clears the state of the Writer z such that it is equivalent to its
  220. // initial state from NewWriter, but instead writing to w.
  221. // No access to the underlying io.Writer is performed.
  222. func (z *Writer) Reset(w io.Writer) {
  223. z.Header = Header{}
  224. z.dst = w
  225. z.checksum.Reset()
  226. z.zdata = z.zdata[:0]
  227. z.data = z.data[:0]
  228. z.idx = 0
  229. }
  230. // writeUint32 writes a uint32 to the underlying writer.
  231. func (z *Writer) writeUint32(x uint32) error {
  232. buf := z.buf[:4]
  233. binary.LittleEndian.PutUint32(buf, x)
  234. _, err := z.dst.Write(buf)
  235. return err
  236. }