frame.go 7.6 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335
  1. package lz4
  2. import (
  3. "encoding/binary"
  4. "fmt"
  5. "io"
  6. "io/ioutil"
  7. "github.com/pierrec/lz4/internal/xxh32"
  8. )
  9. //go:generate go run gen.go
  10. type Frame struct {
  11. Magic uint32
  12. Descriptor FrameDescriptor
  13. Blocks Blocks
  14. Checksum uint32
  15. checksum xxh32.XXHZero
  16. }
  17. func (f *Frame) initW(w *Writer) {
  18. f.Magic = frameMagic
  19. f.Descriptor.initW(w)
  20. f.Blocks.initW(w)
  21. f.checksum.Reset()
  22. }
  23. func (f *Frame) closeW(w *Writer) error {
  24. if err := f.Blocks.closeW(w); err != nil {
  25. return err
  26. }
  27. buf := w.buf[:0]
  28. // End mark (data block size of uint32(0)).
  29. buf = append(buf, 0, 0, 0, 0)
  30. if f.Descriptor.Flags.ContentChecksum() {
  31. buf = f.checksum.Sum(buf)
  32. }
  33. _, err := w.src.Write(buf)
  34. return err
  35. }
  36. func (f *Frame) initR(r *Reader) error {
  37. if f.Magic > 0 {
  38. // Header already read.
  39. return nil
  40. }
  41. newFrame:
  42. if err := readUint32(r.src, r.buf[:], &f.Magic); err != nil {
  43. return err
  44. }
  45. switch m := f.Magic; {
  46. case m == frameMagic:
  47. // All 16 values of frameSkipMagic are valid.
  48. case m>>8 == frameSkipMagic>>8:
  49. var skip uint32
  50. if err := binary.Read(r.src, binary.LittleEndian, &skip); err != nil {
  51. return err
  52. }
  53. if _, err := io.CopyN(ioutil.Discard, r.src, int64(skip)); err != nil {
  54. return err
  55. }
  56. goto newFrame
  57. default:
  58. return ErrInvalidFrame
  59. }
  60. if err := f.Descriptor.initR(r); err != nil {
  61. return err
  62. }
  63. f.Blocks.initR(r)
  64. f.checksum.Reset()
  65. return nil
  66. }
  67. func (f *Frame) closeR(r *Reader) error {
  68. f.Magic = 0
  69. if !f.Descriptor.Flags.ContentChecksum() {
  70. return nil
  71. }
  72. if err := readUint32(r.src, r.buf[:], &f.Checksum); err != nil {
  73. return err
  74. }
  75. if c := f.checksum.Sum32(); c != f.Checksum {
  76. return fmt.Errorf("%w: got %x; expected %x", ErrInvalidFrameChecksum, c, f.Checksum)
  77. }
  78. return nil
  79. }
  80. type FrameDescriptor struct {
  81. Flags DescriptorFlags
  82. ContentSize uint64
  83. Checksum uint8
  84. }
  85. func (fd *FrameDescriptor) initW(_ *Writer) {
  86. fd.Flags.VersionSet(1)
  87. fd.Flags.BlockIndependenceSet(true)
  88. }
  89. func (fd *FrameDescriptor) write(w *Writer) error {
  90. if fd.Checksum > 0 {
  91. // Header already written.
  92. return nil
  93. }
  94. buf := w.buf[:4+2]
  95. // Write the magic number here even though it belongs to the Frame.
  96. binary.LittleEndian.PutUint32(buf, w.frame.Magic)
  97. binary.LittleEndian.PutUint16(buf[4:], uint16(fd.Flags))
  98. if fd.Flags.Size() {
  99. buf = buf[:4+2+8]
  100. binary.LittleEndian.PutUint64(buf[4+2:], fd.ContentSize)
  101. }
  102. fd.Checksum = descriptorChecksum(buf[4:])
  103. buf = append(buf, fd.Checksum)
  104. _, err := w.src.Write(buf)
  105. return err
  106. }
  107. func (fd *FrameDescriptor) initR(r *Reader) error {
  108. // Read the flags and the checksum, hoping that there is not content size.
  109. buf := r.buf[:3]
  110. if _, err := io.ReadFull(r.src, buf); err != nil {
  111. return err
  112. }
  113. descr := binary.LittleEndian.Uint16(buf)
  114. fd.Flags = DescriptorFlags(descr)
  115. if fd.Flags.Size() {
  116. // Append the 8 missing bytes.
  117. buf = buf[:3+8]
  118. if _, err := io.ReadFull(r.src, buf[3:]); err != nil {
  119. return err
  120. }
  121. fd.ContentSize = binary.LittleEndian.Uint64(buf[2:])
  122. }
  123. fd.Checksum = buf[len(buf)-1] // the checksum is the last byte
  124. buf = buf[:len(buf)-1] // all descriptor fields except checksum
  125. if c := descriptorChecksum(buf); fd.Checksum != c {
  126. return fmt.Errorf("%w: got %x; expected %x", ErrInvalidHeaderChecksum, c, fd.Checksum)
  127. }
  128. // Validate the elements that can be.
  129. if !fd.Flags.BlockSizeIndex().isValid() {
  130. return ErrOptionInvalidBlockSize
  131. }
  132. return nil
  133. }
  134. func descriptorChecksum(buf []byte) byte {
  135. return byte(xxh32.ChecksumZero(buf) >> 8)
  136. }
  137. type Blocks struct {
  138. Block *FrameDataBlock
  139. Blocks chan chan *FrameDataBlock
  140. err error
  141. }
  142. func (b *Blocks) initW(w *Writer) {
  143. size := w.frame.Descriptor.Flags.BlockSizeIndex()
  144. if w.isNotConcurrent() {
  145. b.Blocks = nil
  146. b.Block = newFrameDataBlock(size)
  147. return
  148. }
  149. if cap(b.Blocks) != w.num {
  150. b.Blocks = make(chan chan *FrameDataBlock, w.num)
  151. }
  152. // goroutine managing concurrent block compression goroutines.
  153. go func() {
  154. // Process next block compression item.
  155. for c := range b.Blocks {
  156. // Read the next compressed block result.
  157. // Waiting here ensures that the blocks are output in the order they were sent.
  158. // The incoming channel is always closed as it indicates to the caller that
  159. // the block has been processed.
  160. block := <-c
  161. if block == nil {
  162. // Notify the block compression routine that we are done with its result.
  163. // This is used when a sentinel block is sent to terminate the compression.
  164. close(c)
  165. return
  166. }
  167. // Do not attempt to write the block upon any previous failure.
  168. if b.err == nil {
  169. // Write the block.
  170. if err := block.write(w); err != nil && b.err == nil {
  171. // Keep the first error.
  172. b.err = err
  173. // All pending compression goroutines need to shut down, so we need to keep going.
  174. }
  175. }
  176. close(c)
  177. }
  178. }()
  179. }
  180. func (b *Blocks) closeW(w *Writer) error {
  181. if w.isNotConcurrent() {
  182. b.Block.closeW(w)
  183. b.Block = nil
  184. return nil
  185. }
  186. c := make(chan *FrameDataBlock)
  187. b.Blocks <- c
  188. c <- nil
  189. <-c
  190. err := b.err
  191. b.err = nil
  192. return err
  193. }
  194. func (b *Blocks) initR(r *Reader) {
  195. size := r.frame.Descriptor.Flags.BlockSizeIndex()
  196. b.Block = newFrameDataBlock(size)
  197. }
  198. func newFrameDataBlock(size BlockSizeIndex) *FrameDataBlock {
  199. return &FrameDataBlock{Data: size.get()}
  200. }
  201. type FrameDataBlock struct {
  202. Size DataBlockSize
  203. Data []byte
  204. Checksum uint32
  205. }
  206. func (b *FrameDataBlock) closeW(w *Writer) {
  207. size := w.frame.Descriptor.Flags.BlockSizeIndex()
  208. size.put(b.Data)
  209. }
  210. // Block compression errors are ignored since the buffer is sized appropriately.
  211. func (b *FrameDataBlock) compress(w *Writer, src []byte, ht []int) *FrameDataBlock {
  212. dst := b.Data[:len(src)] // trigger the incompressible flag in CompressBlock
  213. var n int
  214. switch w.level {
  215. case Fast:
  216. n, _ = CompressBlock(src, dst, ht)
  217. default:
  218. n, _ = CompressBlockHC(src, dst, w.level, ht)
  219. }
  220. if n == 0 {
  221. b.Size.uncompressedSet(true)
  222. dst = src
  223. } else {
  224. b.Size.uncompressedSet(false)
  225. dst = dst[:n]
  226. }
  227. b.Data = dst
  228. b.Size.sizeSet(len(dst))
  229. if w.frame.Descriptor.Flags.BlockChecksum() {
  230. b.Checksum = xxh32.ChecksumZero(src)
  231. }
  232. if w.frame.Descriptor.Flags.ContentChecksum() {
  233. _, _ = w.frame.checksum.Write(src)
  234. }
  235. return b
  236. }
  237. func (b *FrameDataBlock) write(w *Writer) error {
  238. buf := w.buf[:]
  239. out := w.src
  240. binary.LittleEndian.PutUint32(buf, uint32(b.Size))
  241. if _, err := out.Write(buf[:4]); err != nil {
  242. return err
  243. }
  244. if _, err := out.Write(b.Data); err != nil {
  245. return err
  246. }
  247. if b.Checksum == 0 {
  248. return nil
  249. }
  250. binary.LittleEndian.PutUint32(buf, b.Checksum)
  251. _, err := out.Write(buf[:4])
  252. return err
  253. }
  254. func (b *FrameDataBlock) uncompress(r *Reader, dst []byte) (int, error) {
  255. var x uint32
  256. if err := readUint32(r.src, r.buf[:], &x); err != nil {
  257. return 0, err
  258. }
  259. b.Size = DataBlockSize(x)
  260. if b.Size == 0 {
  261. // End of frame reached.
  262. return 0, io.EOF
  263. }
  264. isCompressed := !b.Size.uncompressed()
  265. size := b.Size.size()
  266. var data []byte
  267. if isCompressed {
  268. data = b.Data
  269. } else {
  270. data = dst
  271. }
  272. data = data[:size]
  273. if _, err := io.ReadFull(r.src, data); err != nil {
  274. return 0, err
  275. }
  276. if isCompressed {
  277. n, err := UncompressBlock(data, dst)
  278. if err != nil {
  279. return 0, err
  280. }
  281. data = dst[:n]
  282. }
  283. if r.frame.Descriptor.Flags.BlockChecksum() {
  284. if err := readUint32(r.src, r.buf[:], &b.Checksum); err != nil {
  285. return 0, err
  286. }
  287. if c := xxh32.ChecksumZero(data); c != b.Checksum {
  288. return 0, fmt.Errorf("%w: got %x; expected %x", ErrInvalidBlockChecksum, c, b.Checksum)
  289. }
  290. }
  291. if r.frame.Descriptor.Flags.ContentChecksum() {
  292. _, _ = r.frame.checksum.Write(data)
  293. }
  294. return len(data), nil
  295. }
  296. func readUint32(r io.Reader, buf []byte, x *uint32) error {
  297. if _, err := io.ReadFull(r, buf[:4]); err != nil {
  298. return err
  299. }
  300. *x = binary.LittleEndian.Uint32(buf)
  301. return nil
  302. }