compression.c 12 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349
  1. #include "rrdpush.h"
  2. #ifdef ENABLE_COMPRESSION
  3. #include "lz4.h"
  4. #define STREAM_COMPRESSION_MSG "STREAM_COMPRESSION"
  5. #define LZ4_MAX_MSG_SIZE 0x4000
  6. #define LZ4_STREAM_BUFFER_SIZE (0x10000 + LZ4_MAX_MSG_SIZE)
  7. #define SIGNATURE ((uint32_t)('z' | 0x80) | (0x80 << 8) | (0x80 << 16) | ('\n' << 24))
  8. #define SIGNATURE_MASK ((uint32_t)0xff | (0x80 << 8) | (0x80 << 16) | (0xff << 24))
  9. #define SIGNATURE_SIZE 4
  10. /*
  11. * LZ4 streaming API compressor specific data
  12. */
  13. struct compressor_data {
  14. LZ4_stream_t *stream;
  15. char *stream_buffer;
  16. size_t stream_buffer_pos;
  17. };
  18. /*
  19. * Reset compressor state for a new stream
  20. */
  21. static void lz4_compressor_reset(struct compressor_state *state)
  22. {
  23. if (state->data) {
  24. if (state->data->stream) {
  25. LZ4_resetStream_fast(state->data->stream);
  26. info("%s: Compressor Reset", STREAM_COMPRESSION_MSG);
  27. }
  28. state->data->stream_buffer_pos = 0;
  29. }
  30. }
  31. /*
  32. * Destroy compressor state and all related data
  33. */
  34. static void lz4_compressor_destroy(struct compressor_state **state)
  35. {
  36. if (state && *state) {
  37. struct compressor_state *s = *state;
  38. if (s->data) {
  39. if (s->data->stream)
  40. LZ4_freeStream(s->data->stream);
  41. freez(s->data->stream_buffer);
  42. freez(s->data);
  43. }
  44. freez(s->buffer);
  45. freez(s);
  46. *state = NULL;
  47. debug(D_STREAM, "%s: Compressor Destroyed.", STREAM_COMPRESSION_MSG);
  48. }
  49. }
  50. /*
  51. * Compress the given block of data
  52. * Compressed data will remain in the internal buffer until the next invocation
  53. * Return the size of compressed data block as result and the pointer to internal buffer using the last argument
  54. * or 0 in case of error
  55. */
  56. static size_t lz4_compressor_compress(struct compressor_state *state, const char *data, size_t size, char **out)
  57. {
  58. if (!state || !size || !out)
  59. return 0;
  60. if (size > LZ4_MAX_MSG_SIZE) {
  61. error("%s: Compression Failed - Message size %lu above compression buffer limit: %d", STREAM_COMPRESSION_MSG, size, LZ4_MAX_MSG_SIZE);
  62. return 0;
  63. }
  64. size_t max_dst_size = LZ4_COMPRESSBOUND(size);
  65. size_t data_size = max_dst_size + SIGNATURE_SIZE;
  66. if (!state->buffer) {
  67. state->buffer = mallocz(data_size);
  68. state->buffer_size = data_size;
  69. } else if (state->buffer_size < data_size) {
  70. state->buffer = reallocz(state->buffer, data_size);
  71. state->buffer_size = data_size;
  72. }
  73. memcpy(state->data->stream_buffer + state->data->stream_buffer_pos, data, size);
  74. long int compressed_data_size = LZ4_compress_fast_continue(state->data->stream,
  75. state->data->stream_buffer + state->data->stream_buffer_pos,
  76. state->buffer + SIGNATURE_SIZE, size, max_dst_size, 1);
  77. if (compressed_data_size < 0) {
  78. error("Data compression error: %ld", compressed_data_size);
  79. return 0;
  80. }
  81. state->data->stream_buffer_pos += size;
  82. if (state->data->stream_buffer_pos >= LZ4_STREAM_BUFFER_SIZE - LZ4_MAX_MSG_SIZE)
  83. state->data->stream_buffer_pos = 0;
  84. uint32_t len = ((compressed_data_size & 0x7f) | 0x80 | (((compressed_data_size & (0x7f << 7)) << 1) | 0x8000)) << 8;
  85. *(uint32_t *)state->buffer = len | SIGNATURE;
  86. *out = state->buffer;
  87. debug(D_STREAM, "%s: Compressed data header: %ld", STREAM_COMPRESSION_MSG, compressed_data_size);
  88. return compressed_data_size + SIGNATURE_SIZE;
  89. }
  90. /*
  91. * Create and initialize compressor state
  92. * Return the pointer to compressor_state structure created
  93. */
  94. struct compressor_state *create_compressor()
  95. {
  96. struct compressor_state *state = callocz(1, sizeof(struct compressor_state));
  97. state->reset = lz4_compressor_reset;
  98. state->compress = lz4_compressor_compress;
  99. state->destroy = lz4_compressor_destroy;
  100. state->data = callocz(1, sizeof(struct compressor_data));
  101. state->data->stream = LZ4_createStream();
  102. state->data->stream_buffer = callocz(1, LZ4_DECODER_RING_BUFFER_SIZE(LZ4_MAX_MSG_SIZE));
  103. state->buffer_size = LZ4_STREAM_BUFFER_SIZE;
  104. state->reset(state);
  105. debug(D_STREAM, "%s: Initialize streaming compression!", STREAM_COMPRESSION_MSG);
  106. return state;
  107. }
  108. /*
  109. * LZ4 streaming API decompressor specific data
  110. */
  111. struct decompressor_data {
  112. LZ4_streamDecode_t *stream;
  113. char *stream_buffer;
  114. size_t stream_buffer_size;
  115. size_t stream_buffer_pos;
  116. };
  117. /*
  118. * Reset decompressor state for a new stream
  119. */
  120. static void lz4_decompressor_reset(struct decompressor_state *state)
  121. {
  122. if (state->data) {
  123. if (state->data->stream)
  124. LZ4_setStreamDecode(state->data->stream, NULL, 0);
  125. state->data->stream_buffer_pos = 0;
  126. state->buffer_len = 0;
  127. state->out_buffer_len = 0;
  128. }
  129. }
  130. /*
  131. * Destroy decompressor state and all related data
  132. */
  133. static void lz4_decompressor_destroy(struct decompressor_state **state)
  134. {
  135. if (state && *state) {
  136. struct decompressor_state *s = *state;
  137. if (s->data) {
  138. debug(D_STREAM, "%s: Destroying decompressor.", STREAM_COMPRESSION_MSG);
  139. if (s->data->stream)
  140. LZ4_freeStreamDecode(s->data->stream);
  141. freez(s->data->stream_buffer);
  142. freez(s->data);
  143. }
  144. freez(s->buffer);
  145. freez(s);
  146. *state = NULL;
  147. }
  148. }
  149. static size_t decode_compress_header(const char *data, size_t data_size)
  150. {
  151. if (!data || !data_size)
  152. return 0;
  153. if (data_size < SIGNATURE_SIZE)
  154. return 0;
  155. uint32_t sign = *(uint32_t *)data;
  156. if ((sign & SIGNATURE_MASK) != SIGNATURE)
  157. return 0;
  158. size_t length = ((sign >> 8) & 0x7f) | ((sign >> 9) & (0x7f << 7));
  159. return length;
  160. }
  161. /*
  162. * Check input data for the compression header
  163. * Return the size of compressed data or 0 for uncompressed data
  164. */
  165. size_t is_compressed_data(const char *data, size_t data_size)
  166. {
  167. return decode_compress_header(data, data_size);
  168. }
  169. /*
  170. * Start the collection of compressed data in an internal buffer
  171. * Return the size of compressed data or 0 for uncompressed data
  172. */
  173. static size_t lz4_decompressor_start(struct decompressor_state *state, const char *header, size_t header_size)
  174. {
  175. size_t length = decode_compress_header(header, header_size);
  176. if (!length)
  177. return 0;
  178. if (!state->buffer) {
  179. state->buffer = mallocz(length);
  180. state->buffer_size = length;
  181. } else if (state->buffer_size < length) {
  182. state->buffer = reallocz(state->buffer, length);
  183. state->buffer_size = length;
  184. }
  185. state->buffer_len = length;
  186. state->buffer_pos = 0;
  187. state->out_buffer_pos = 0;
  188. state->out_buffer_len = 0;
  189. return length;
  190. }
  191. /*
  192. * Add a chunk of compressed data to the internal buffer
  193. * Return the current size of compressed data or 0 for error
  194. */
  195. static size_t lz4_decompressor_put(struct decompressor_state *state, const char *data, size_t size)
  196. {
  197. if (!state || !size || !data)
  198. return 0;
  199. if (!state->buffer)
  200. fatal("STREAM: No decompressor buffer allocated");
  201. if (state->buffer_pos + size > state->buffer_len) {
  202. error("STREAM: Decompressor buffer overflow %lu + %lu > %lu",
  203. state->buffer_pos, size, state->buffer_len);
  204. size = state->buffer_len - state->buffer_pos;
  205. }
  206. memcpy(state->buffer + state->buffer_pos, data, size);
  207. state->buffer_pos += size;
  208. return state->buffer_pos;
  209. }
  210. static size_t saving_percent(size_t comp_len, size_t src_len)
  211. {
  212. if (comp_len > src_len)
  213. comp_len = src_len;
  214. if (!src_len)
  215. return 0;
  216. return 100 - comp_len * 100 / src_len;
  217. }
  218. /*
  219. * Decompress the compressed data in the internal buffer
  220. * Return the size of uncompressed data or 0 for error
  221. */
  222. static size_t lz4_decompressor_decompress(struct decompressor_state *state)
  223. {
  224. if (!state)
  225. return 0;
  226. if (!state->buffer) {
  227. error("%s: No decompressor buffer allocated", STREAM_COMPRESSION_MSG);
  228. return 0;
  229. }
  230. long int decompressed_size = LZ4_decompress_safe_continue(state->data->stream, state->buffer,
  231. state->data->stream_buffer + state->data->stream_buffer_pos,
  232. state->buffer_len, state->data->stream_buffer_size - state->data->stream_buffer_pos);
  233. if (decompressed_size < 0) {
  234. error("%s: Decompressor error %ld", STREAM_COMPRESSION_MSG, decompressed_size);
  235. return 0;
  236. }
  237. state->out_buffer = state->data->stream_buffer + state->data->stream_buffer_pos;
  238. state->data->stream_buffer_pos += decompressed_size;
  239. if (state->data->stream_buffer_pos >= state->data->stream_buffer_size - LZ4_MAX_MSG_SIZE)
  240. state->data->stream_buffer_pos = 0;
  241. state->out_buffer_len = decompressed_size;
  242. state->out_buffer_pos = 0;
  243. // Some compression statistics
  244. size_t old_avg_saving = saving_percent(state->total_compressed, state->total_uncompressed);
  245. size_t old_avg_size = state->packet_count ? state->total_uncompressed / state->packet_count : 0;
  246. state->total_compressed += state->buffer_len + SIGNATURE_SIZE;
  247. state->total_uncompressed += decompressed_size;
  248. state->packet_count++;
  249. size_t saving = saving_percent(state->buffer_len, decompressed_size);
  250. size_t avg_saving = saving_percent(state->total_compressed, state->total_uncompressed);
  251. size_t avg_size = state->total_uncompressed / state->packet_count;
  252. if (old_avg_saving != avg_saving || old_avg_size != avg_size){
  253. debug(D_STREAM, "%s: Saving: %lu%% (avg. %lu%%), avg.size: %lu", STREAM_COMPRESSION_MSG, saving, avg_saving, avg_size);
  254. }
  255. return decompressed_size;
  256. }
  257. /*
  258. * Return the size of uncompressed data left in the internal buffer or 0 for error
  259. */
  260. static size_t lz4_decompressor_decompressed_bytes_in_buffer(struct decompressor_state *state)
  261. {
  262. return state->out_buffer_len ?
  263. state->out_buffer_len - state->out_buffer_pos : 0;
  264. }
  265. /*
  266. * Fill the buffer provided with uncompressed data from the internal buffer
  267. * Return the size of uncompressed data copied or 0 for error
  268. */
  269. static size_t lz4_decompressor_get(struct decompressor_state *state, char *data, size_t size)
  270. {
  271. if (!state || !size || !data)
  272. return 0;
  273. if (!state->out_buffer)
  274. fatal("%s: No decompressor output buffer allocated", STREAM_COMPRESSION_MSG);
  275. if (state->out_buffer_pos + size > state->out_buffer_len)
  276. size = state->out_buffer_len - state->out_buffer_pos;
  277. char *p = state->out_buffer + state->out_buffer_pos, *endp = p + size, *last_lf = NULL;
  278. for (; p < endp; ++p)
  279. if (*p == '\n' || *p == 0)
  280. last_lf = p;
  281. if (last_lf)
  282. size = last_lf + 1 - (state->out_buffer + state->out_buffer_pos);
  283. memcpy(data, state->out_buffer + state->out_buffer_pos, size);
  284. state->out_buffer_pos += size;
  285. return size;
  286. }
  287. /*
  288. * Create and initialize decompressor state
  289. * Return the pointer to decompressor_state structure created
  290. */
  291. struct decompressor_state *create_decompressor()
  292. {
  293. struct decompressor_state *state = callocz(1, sizeof(struct decompressor_state));
  294. state->reset = lz4_decompressor_reset;
  295. state->start = lz4_decompressor_start;
  296. state->put = lz4_decompressor_put;
  297. state->decompress = lz4_decompressor_decompress;
  298. state->get = lz4_decompressor_get;
  299. state->decompressed_bytes_in_buffer = lz4_decompressor_decompressed_bytes_in_buffer;
  300. state->destroy = lz4_decompressor_destroy;
  301. state->data = callocz(1, sizeof(struct decompressor_data));
  302. fatal_assert(state->data);
  303. state->data->stream = LZ4_createStreamDecode();
  304. state->data->stream_buffer_size = LZ4_decoderRingBufferSize(LZ4_MAX_MSG_SIZE);
  305. state->data->stream_buffer = mallocz(state->data->stream_buffer_size);
  306. fatal_assert(state->data->stream_buffer);
  307. state->reset(state);
  308. debug(D_STREAM, "%s: Initialize streaming decompression!", STREAM_COMPRESSION_MSG);
  309. return state;
  310. }
  311. #endif