diff options
author | Yann Collet <yann.collet.73@gmail.com> | 2015-05-07 11:23:39 (GMT) |
---|---|---|
committer | Yann Collet <yann.collet.73@gmail.com> | 2015-05-07 11:23:39 (GMT) |
commit | 7d182b816ace89d6d6d16b7aee376a9962a05caa (patch) | |
tree | 5b2236b59cd0e42e3365b0765041a25c85519b39 /programs/lz4io.c | |
parent | 160661c7a4cbf805f4af74d2e3932a17a66e6ce7 (diff) | |
parent | fdd0029c3778768d0e8a80ce238d981f946c8b78 (diff) | |
download | lz4-7d182b816ace89d6d6d16b7aee376a9962a05caa.zip lz4-7d182b816ace89d6d6d16b7aee376a9962a05caa.tar.gz lz4-7d182b816ace89d6d6d16b7aee376a9962a05caa.tar.bz2 |
Dev
Diffstat (limited to 'programs/lz4io.c')
-rw-r--r-- | programs/lz4io.c | 672 |
1 files changed, 438 insertions, 234 deletions
diff --git a/programs/lz4io.c b/programs/lz4io.c index 02e03c8..209f5ed 100644 --- a/programs/lz4io.c +++ b/programs/lz4io.c @@ -46,22 +46,22 @@ /***************************** * Includes *****************************/ -#include <stdio.h> /* fprintf, fopen, fread, stdin, stdout */ +#include <stdio.h> /* fprintf, fopen, fread, stdin, stdout, fflush, getchar */ #include <stdlib.h> /* malloc, free */ #include <string.h> /* strcmp, strlen */ #include <time.h> /* clock */ #include <sys/types.h> /* stat64 */ #include <sys/stat.h> /* stat64 */ #include "lz4io.h" -#include "lz4.h" /* still required for legacy format */ -#include "lz4hc.h" /* still required for legacy format */ +#include "lz4.h" /* still required for legacy format */ +#include "lz4hc.h" /* still required for legacy format */ #include "lz4frame.h" /****************************** * OS-specific Includes ******************************/ -#if defined(MSDOS) || defined(OS2) || defined(WIN32) || defined(_WIN32) || defined(__CYGWIN__) +#if defined(MSDOS) || defined(OS2) || defined(WIN32) || defined(_WIN32) # include <fcntl.h> /* _O_BINARY */ # include <io.h> /* _setmode, _fileno, _get_osfhandle */ # define SET_BINARY_MODE(file) _setmode(_fileno(file), _O_BINARY) @@ -131,7 +131,7 @@ static int g_blockSizeId = LZ4IO_BLOCKSIZEID_DEFAULT; static int g_blockChecksum = 0; static int g_streamChecksum = 1; static int g_blockIndependence = 1; -static int g_sparseFileSupport = 0; +static int g_sparseFileSupport = 1; static int g_contentSizeFlag = 0; static const int minBlockSizeID = 4; @@ -159,7 +159,7 @@ static const int maxBlockSizeID = 7; #define EXTENDED_ARGUMENTS #define EXTENDED_HELP #define EXTENDED_FORMAT -#define DEFAULT_DECOMPRESSOR decodeLZ4S +#define DEFAULT_DECOMPRESSOR LZ4IO_decompressLZ4F /* ************************************************** */ @@ -240,7 +240,7 @@ static unsigned long long LZ4IO_GetFileSize(const char* infilename) struct stat statbuf; r = stat(infilename, &statbuf); #endif - if (r || !S_ISREG(statbuf.st_mode)) return 0; /* No good... */ + if (r || !S_ISREG(statbuf.st_mode)) return 0; /* failure, or is not a regular file */ return (unsigned long long)statbuf.st_size; } @@ -253,7 +253,7 @@ static int LZ4IO_GetBlockSize_FromBlockId (int id) { return (1 << (8 + (2 * id)) static int LZ4IO_isSkippableMagicNumber(unsigned int magic) { return (magic & LZ4IO_SKIPPABLEMASK) == LZ4IO_SKIPPABLE0; } -static int get_fileHandle(const char* input_filename, const char* output_filename, FILE** pfinput, FILE** pfoutput) +static int LZ4IO_getFiles(const char* input_filename, const char* output_filename, FILE** pfinput, FILE** pfoutput) { if (!strcmp (input_filename, stdinmark)) @@ -267,6 +267,12 @@ static int get_fileHandle(const char* input_filename, const char* output_filenam *pfinput = fopen(input_filename, "rb"); } + if ( *pfinput==0 ) + { + DISPLAYLEVEL(1, "Unable to access file for processing: %s\n", input_filename); + return 1; + } + if (!strcmp (output_filename, stdoutmark)) { DISPLAYLEVEL(4,"Using stdout for output\n"); @@ -283,18 +289,18 @@ static int get_fileHandle(const char* input_filename, const char* output_filenam fclose(*pfoutput); if (!g_overwrite) { - char ch; + int ch = 'Y'; DISPLAYLEVEL(2, "Warning : %s already exists\n", output_filename); - DISPLAYLEVEL(2, "Overwrite ? (Y/N) : "); - if (g_displayLevel <= 1) EXM_THROW(11, "Operation aborted : %s already exists", output_filename); /* No interaction possible */ - ch = (char)getchar(); - if ((ch!='Y') && (ch!='y')) EXM_THROW(11, "Operation aborted : %s already exists", output_filename); + if ((g_displayLevel <= 1) || (*pfinput == stdin)) + EXM_THROW(11, "Operation aborted : %s already exists", output_filename); /* No interaction possible */ + DISPLAYLEVEL(2, "Overwrite ? (Y/n) : "); + while((ch = getchar()) != '\n' && ch != EOF) /* flush integrated */ + if ((ch!='Y') && (ch!='y')) EXM_THROW(12, "No. Operation aborted : %s already exists", output_filename); } } *pfoutput = fopen( output_filename, "wb" ); } - if ( *pfinput==0 ) EXM_THROW(12, "Pb opening %s", input_filename); if ( *pfoutput==0) EXM_THROW(13, "Pb opening %s", output_filename); return 0; @@ -316,16 +322,23 @@ static void LZ4IO_writeLE32 (void* p, unsigned value32) dstPtr[3] = (unsigned char)(value32 >> 24); } +static int LZ4IO_LZ4_compress(const char* src, char* dst, int srcSize, int dstSize, int cLevel) +{ + (void)cLevel; + return LZ4_compress_fast(src, dst, srcSize, dstSize, 1); +} + /* LZ4IO_compressFilename_Legacy : * This function is intentionally "hidden" (not published in .h) * It generates compressed streams using the old 'legacy' format */ int LZ4IO_compressFilename_Legacy(const char* input_filename, const char* output_filename, int compressionlevel) { - int (*compressionFunction)(const char*, char*, int); + int (*compressionFunction)(const char* src, char* dst, int srcSize, int dstSize, int cLevel); unsigned long long filesize = 0; unsigned long long compressedfilesize = MAGICNUMBER_SIZE; char* in_buff; char* out_buff; + const int outBuffSize = LZ4_compressBound(LEGACY_BLOCKSIZE); FILE* finput; FILE* foutput; clock_t start, end; @@ -334,14 +347,14 @@ int LZ4IO_compressFilename_Legacy(const char* input_filename, const char* output /* Init */ start = clock(); - if (compressionlevel < 3) compressionFunction = LZ4_compress; else compressionFunction = LZ4_compressHC; + if (compressionlevel < 3) compressionFunction = LZ4IO_LZ4_compress; else compressionFunction = LZ4_compress_HC; - get_fileHandle(input_filename, output_filename, &finput, &foutput); - if ((g_displayLevel==2) && (compressionlevel==1)) g_displayLevel=3; + if (LZ4IO_getFiles(input_filename, output_filename, &finput, &foutput)) + EXM_THROW(20, "File error"); /* Allocate Memory */ in_buff = (char*)malloc(LEGACY_BLOCKSIZE); - out_buff = (char*)malloc(LZ4_compressBound(LEGACY_BLOCKSIZE)); + out_buff = (char*)malloc(outBuffSize); if (!in_buff || !out_buff) EXM_THROW(21, "Allocation error : not enough memory"); /* Write Archive Header */ @@ -359,9 +372,9 @@ int LZ4IO_compressFilename_Legacy(const char* input_filename, const char* output filesize += inSize; /* Compress Block */ - outSize = compressionFunction(in_buff, out_buff+4, inSize); + outSize = compressionFunction(in_buff, out_buff+4, inSize, outBuffSize, compressionlevel); compressedfilesize += outSize+4; - DISPLAYUPDATE(3, "\rRead : %i MB ==> %.2f%% ", (int)(filesize>>20), (double)compressedfilesize/filesize*100); + DISPLAYUPDATE(2, "\rRead : %i MB ==> %.2f%% ", (int)(filesize>>20), (double)compressedfilesize/filesize*100); /* Write Block */ LZ4IO_writeLE32(out_buff, outSize); @@ -372,6 +385,7 @@ int LZ4IO_compressFilename_Legacy(const char* input_filename, const char* output /* Status */ end = clock(); DISPLAYLEVEL(2, "\r%79s\r", ""); + filesize += !filesize; /* avoid divide by zero */ DISPLAYLEVEL(2,"Compressed %llu bytes into %llu bytes ==> %.2f%%\n", (unsigned long long) filesize, (unsigned long long) compressedfilesize, (double)compressedfilesize/filesize*100); { @@ -393,128 +407,210 @@ int LZ4IO_compressFilename_Legacy(const char* input_filename, const char* output * Compression using Frame format *********************************************/ -int LZ4IO_compressFilename(const char* input_filename, const char* output_filename, int compressionLevel) +typedef struct { + void* srcBuffer; + size_t srcBufferSize; + void* dstBuffer; + size_t dstBufferSize; + LZ4F_compressionContext_t ctx; +} cRess_t; + +static cRess_t LZ4IO_createCResources(void) +{ + const size_t blockSize = (size_t)LZ4IO_GetBlockSize_FromBlockId (g_blockSizeId); + cRess_t ress; + LZ4F_errorCode_t errorCode; + + errorCode = LZ4F_createCompressionContext(&(ress.ctx), LZ4F_VERSION); + if (LZ4F_isError(errorCode)) EXM_THROW(30, "Allocation error : can't create LZ4F context : %s", LZ4F_getErrorName(errorCode)); + + /* Allocate Memory */ + ress.srcBuffer = malloc(blockSize); + ress.srcBufferSize = blockSize; + ress.dstBufferSize = LZ4F_compressFrameBound(blockSize, NULL); /* cover worst case */ + ress.dstBuffer = malloc(ress.dstBufferSize); + if (!ress.srcBuffer || !ress.dstBuffer) EXM_THROW(31, "Allocation error : not enough memory"); + + return ress; +} + +static void LZ4IO_freeCResources(cRess_t ress) +{ + LZ4F_errorCode_t errorCode; + free(ress.srcBuffer); + free(ress.dstBuffer); + errorCode = LZ4F_freeCompressionContext(ress.ctx); + if (LZ4F_isError(errorCode)) EXM_THROW(38, "Error : can't free LZ4F context resource : %s", LZ4F_getErrorName(errorCode)); +} + +/* + * LZ4IO_compressFilename_extRess() + * result : 0 : compression completed correctly + * 1 : missing or pb opening srcFileName + */ +static int LZ4IO_compressFilename_extRess(cRess_t ress, const char* srcFileName, const char* dstFileName, int compressionLevel) { unsigned long long filesize = 0; unsigned long long compressedfilesize = 0; - char* in_buff; - char* out_buff; - FILE* finput; - FILE* foutput; - clock_t start, end; - int blockSize; - size_t sizeCheck, headerSize, readSize, outBuffSize; - LZ4F_compressionContext_t ctx; - LZ4F_errorCode_t errorCode; + FILE* srcFile; + FILE* dstFile; + void* const srcBuffer = ress.srcBuffer; + void* const dstBuffer = ress.dstBuffer; + const size_t dstBufferSize = ress.dstBufferSize; + const size_t blockSize = (size_t)LZ4IO_GetBlockSize_FromBlockId (g_blockSizeId); + size_t sizeCheck, headerSize, readSize; + LZ4F_compressionContext_t ctx = ress.ctx; /* just a pointer */ LZ4F_preferences_t prefs; /* Init */ - start = clock(); memset(&prefs, 0, sizeof(prefs)); - if ((g_displayLevel==2) && (compressionLevel>=3)) g_displayLevel=3; - errorCode = LZ4F_createCompressionContext(&ctx, LZ4F_VERSION); - if (LZ4F_isError(errorCode)) EXM_THROW(30, "Allocation error : can't create LZ4F context : %s", LZ4F_getErrorName(errorCode)); - get_fileHandle(input_filename, output_filename, &finput, &foutput); - blockSize = LZ4IO_GetBlockSize_FromBlockId (g_blockSizeId); + + /* File check */ + if (LZ4IO_getFiles(srcFileName, dstFileName, &srcFile, &dstFile)) return 1; /* Set compression parameters */ prefs.autoFlush = 1; prefs.compressionLevel = compressionLevel; - prefs.frameInfo.blockMode = (blockMode_t)g_blockIndependence; - prefs.frameInfo.blockSizeID = (blockSizeID_t)g_blockSizeId; - prefs.frameInfo.contentChecksumFlag = (contentChecksum_t)g_streamChecksum; + prefs.frameInfo.blockMode = (LZ4F_blockMode_t)g_blockIndependence; + prefs.frameInfo.blockSizeID = (LZ4F_blockSizeID_t)g_blockSizeId; + prefs.frameInfo.contentChecksumFlag = (LZ4F_contentChecksum_t)g_streamChecksum; if (g_contentSizeFlag) { - unsigned long long fileSize = LZ4IO_GetFileSize(input_filename); + unsigned long long fileSize = LZ4IO_GetFileSize(srcFileName); prefs.frameInfo.contentSize = fileSize; /* == 0 if input == stdin */ + if (fileSize==0) + DISPLAYLEVEL(3, "Warning : cannot determine uncompressed frame content size \n"); } - /* Allocate Memory */ - in_buff = (char*)malloc(blockSize); - outBuffSize = LZ4F_compressBound(blockSize, &prefs); - out_buff = (char*)malloc(outBuffSize); - if (!in_buff || !out_buff) EXM_THROW(31, "Allocation error : not enough memory"); - - /* Write Archive Header */ - headerSize = LZ4F_compressBegin(ctx, out_buff, outBuffSize, &prefs); - if (LZ4F_isError(headerSize)) EXM_THROW(32, "File header generation failed : %s", LZ4F_getErrorName(headerSize)); - sizeCheck = fwrite(out_buff, 1, headerSize, foutput); - if (sizeCheck!=headerSize) EXM_THROW(33, "Write error : cannot write header"); - compressedfilesize += headerSize; - /* read first block */ - readSize = fread(in_buff, (size_t)1, (size_t)blockSize, finput); + readSize = fread(srcBuffer, (size_t)1, blockSize, srcFile); filesize += readSize; - /* Main Loop */ - while (readSize>0) + /* single-block file */ + if (readSize < blockSize) { - size_t outSize; - - /* Compress Block */ - outSize = LZ4F_compressUpdate(ctx, out_buff, outBuffSize, in_buff, readSize, NULL); - if (LZ4F_isError(outSize)) EXM_THROW(34, "Compression failed : %s", LZ4F_getErrorName(outSize)); - compressedfilesize += outSize; - DISPLAYUPDATE(3, "\rRead : %i MB ==> %.2f%% ", (int)(filesize>>20), (double)compressedfilesize/filesize*100); + /* Compress in single pass */ + size_t cSize = LZ4F_compressFrame(dstBuffer, dstBufferSize, srcBuffer, readSize, &prefs); + if (LZ4F_isError(cSize)) EXM_THROW(34, "Compression failed : %s", LZ4F_getErrorName(cSize)); + compressedfilesize += cSize; + DISPLAYUPDATE(2, "\rRead : %u MB ==> %.2f%% ", + (unsigned)(filesize>>20), (double)compressedfilesize/(filesize+!filesize)*100); /* avoid division by zero */ /* Write Block */ - sizeCheck = fwrite(out_buff, 1, outSize, foutput); - if (sizeCheck!=outSize) EXM_THROW(35, "Write error : cannot write compressed block"); - - /* Read next block */ - readSize = fread(in_buff, (size_t)1, (size_t)blockSize, finput); - filesize += readSize; + sizeCheck = fwrite(dstBuffer, 1, cSize, dstFile); + if (sizeCheck!=cSize) EXM_THROW(35, "Write error : cannot write compressed block"); } - /* End of Stream mark */ - headerSize = LZ4F_compressEnd(ctx, out_buff, outBuffSize, NULL); - if (LZ4F_isError(headerSize)) EXM_THROW(36, "End of file generation failed : %s", LZ4F_getErrorName(headerSize)); + else - sizeCheck = fwrite(out_buff, 1, headerSize, foutput); - if (sizeCheck!=headerSize) EXM_THROW(37, "Write error : cannot write end of stream"); - compressedfilesize += headerSize; + /* multiple-blocks file */ + { + /* Write Archive Header */ + headerSize = LZ4F_compressBegin(ctx, dstBuffer, dstBufferSize, &prefs); + if (LZ4F_isError(headerSize)) EXM_THROW(32, "File header generation failed : %s", LZ4F_getErrorName(headerSize)); + sizeCheck = fwrite(dstBuffer, 1, headerSize, dstFile); + if (sizeCheck!=headerSize) EXM_THROW(33, "Write error : cannot write header"); + compressedfilesize += headerSize; + + /* Main Loop */ + while (readSize>0) + { + size_t outSize; - /* Close & Free */ - free(in_buff); - free(out_buff); - fclose(finput); - fclose(foutput); - errorCode = LZ4F_freeCompressionContext(ctx); - if (LZ4F_isError(errorCode)) EXM_THROW(38, "Error : can't free LZ4F context resource : %s", LZ4F_getErrorName(errorCode)); + /* Compress Block */ + outSize = LZ4F_compressUpdate(ctx, dstBuffer, dstBufferSize, srcBuffer, readSize, NULL); + if (LZ4F_isError(outSize)) EXM_THROW(34, "Compression failed : %s", LZ4F_getErrorName(outSize)); + compressedfilesize += outSize; + DISPLAYUPDATE(2, "\rRead : %u MB ==> %.2f%% ", (unsigned)(filesize>>20), (double)compressedfilesize/filesize*100); + + /* Write Block */ + sizeCheck = fwrite(dstBuffer, 1, outSize, dstFile); + if (sizeCheck!=outSize) EXM_THROW(35, "Write error : cannot write compressed block"); + + /* Read next block */ + readSize = fread(srcBuffer, (size_t)1, (size_t)blockSize, srcFile); + filesize += readSize; + } + + /* End of Stream mark */ + headerSize = LZ4F_compressEnd(ctx, dstBuffer, dstBufferSize, NULL); + if (LZ4F_isError(headerSize)) EXM_THROW(36, "End of file generation failed : %s", LZ4F_getErrorName(headerSize)); + + sizeCheck = fwrite(dstBuffer, 1, headerSize, dstFile); + if (sizeCheck!=headerSize) EXM_THROW(37, "Write error : cannot write end of stream"); + compressedfilesize += headerSize; + } + + /* Release files */ + fclose (srcFile); + fclose (dstFile); /* Final Status */ - end = clock(); DISPLAYLEVEL(2, "\r%79s\r", ""); DISPLAYLEVEL(2, "Compressed %llu bytes into %llu bytes ==> %.2f%%\n", - (unsigned long long) filesize, (unsigned long long) compressedfilesize, (double)compressedfilesize/filesize*100); + filesize, compressedfilesize, (double)compressedfilesize/(filesize + !filesize)*100); /* avoid division by zero */ + + return 0; +} + + +int LZ4IO_compressFilename(const char* srcFileName, const char* dstFileName, int compressionLevel) +{ + clock_t start, end; + cRess_t ress; + int issueWithSrcFile = 0; + + /* Init */ + start = clock(); + ress = LZ4IO_createCResources(); + + /* Compress File */ + issueWithSrcFile += LZ4IO_compressFilename_extRess(ress, srcFileName, dstFileName, compressionLevel); + + /* Free resources */ + LZ4IO_freeCResources(ress); + + /* Final Status */ + end = clock(); { - double seconds = (double)(end - start)/CLOCKS_PER_SEC; - DISPLAYLEVEL(4, "Done in %.2f s ==> %.2f MB/s\n", seconds, (double)filesize / seconds / 1024 / 1024); + double seconds = (double)(end - start) / CLOCKS_PER_SEC; + DISPLAYLEVEL(4, "Completed in %.2f sec \n", seconds); } - return 0; + return issueWithSrcFile; } #define FNSPACE 30 -int LZ4IO_compressMultipleFilenames(const char** inFileNamesTable, int ifntSize, const char* suffix, int compressionlevel) +int LZ4IO_compressMultipleFilenames(const char** inFileNamesTable, int ifntSize, const char* suffix, int compressionLevel) { int i; - char* outFileName = (char*)malloc(FNSPACE); + int missed_files = 0; + char* dstFileName = (char*)malloc(FNSPACE); size_t ofnSize = FNSPACE; const size_t suffixSize = strlen(suffix); + cRess_t ress; + + /* init */ + ress = LZ4IO_createCResources(); + /* loop on each file */ for (i=0; i<ifntSize; i++) { size_t ifnSize = strlen(inFileNamesTable[i]); - if (ofnSize <= ifnSize+suffixSize+1) { free(outFileName); ofnSize = ifnSize + 20; outFileName = (char*)malloc(ofnSize); } - strcpy(outFileName, inFileNamesTable[i]); - strcat(outFileName, suffix); - LZ4IO_compressFilename(inFileNamesTable[i], outFileName, compressionlevel); + if (ofnSize <= ifnSize+suffixSize+1) { free(dstFileName); ofnSize = ifnSize + 20; dstFileName = (char*)malloc(ofnSize); } + strcpy(dstFileName, inFileNamesTable[i]); + strcat(dstFileName, suffix); + + missed_files += LZ4IO_compressFilename_extRess(ress, inFileNamesTable[i], dstFileName, compressionLevel); } - free(outFileName); - return 0; + + /* Close & Free */ + LZ4IO_freeCResources(ress); + free(dstFileName); + + return missed_files; } @@ -528,15 +624,105 @@ static unsigned LZ4IO_readLE32 (const void* s) unsigned value32 = srcPtr[0]; value32 += (srcPtr[1]<<8); value32 += (srcPtr[2]<<16); - value32 += (srcPtr[3]<<24); + value32 += ((unsigned)srcPtr[3])<<24; return value32; } -static unsigned long long decodeLegacyStream(FILE* finput, FILE* foutput) +static unsigned LZ4IO_fwriteSparse(FILE* file, const void* buffer, size_t bufferSize, unsigned storedSkips) +{ + const size_t* const bufferT = (const size_t*)buffer; /* Buffer is supposed malloc'ed, hence aligned on size_t */ + const size_t* ptrT = bufferT; + size_t bufferSizeT = bufferSize / sizeT; + const size_t* const bufferTEnd = bufferT + bufferSizeT; + static const size_t segmentSizeT = (32 KB) / sizeT; + + if (!g_sparseFileSupport) /* normal write */ + { + size_t sizeCheck = fwrite(buffer, 1, bufferSize, file); + if (sizeCheck != bufferSize) EXM_THROW(68, "Write error : cannot write decoded block"); + return 0; + } + + /* avoid int overflow */ + if (storedSkips > 1 GB) + { + int seekResult = fseek(file, 1 GB, SEEK_CUR); + if (seekResult != 0) EXM_THROW(68, "1 GB skip error (sparse file support)"); + storedSkips -= 1 GB; + } + + while (ptrT < bufferTEnd) + { + size_t seg0SizeT = segmentSizeT; + size_t nb0T; + int seekResult; + + /* count leading zeros */ + if (seg0SizeT > bufferSizeT) seg0SizeT = bufferSizeT; + bufferSizeT -= seg0SizeT; + for (nb0T=0; (nb0T < seg0SizeT) && (ptrT[nb0T] == 0); nb0T++) ; + storedSkips += (unsigned)(nb0T * sizeT); + + if (nb0T != seg0SizeT) /* not all 0s */ + { + size_t sizeCheck; + seekResult = fseek(file, storedSkips, SEEK_CUR); + if (seekResult) EXM_THROW(68, "Skip error (sparse file)"); + storedSkips = 0; + seg0SizeT -= nb0T; + ptrT += nb0T; + sizeCheck = fwrite(ptrT, sizeT, seg0SizeT, file); + if (sizeCheck != seg0SizeT) EXM_THROW(68, "Write error : cannot write decoded block"); + } + ptrT += seg0SizeT; + } + + if (bufferSize & maskT) /* size not multiple of sizeT : implies end of block */ + { + const char* const restStart = (const char*)bufferTEnd; + const char* restPtr = restStart; + size_t restSize = bufferSize & maskT; + const char* const restEnd = restStart + restSize; + for (; (restPtr < restEnd) && (*restPtr == 0); restPtr++) ; + storedSkips += (unsigned) (restPtr - restStart); + if (restPtr != restEnd) + { + size_t sizeCheck; + int seekResult = fseek(file, storedSkips, SEEK_CUR); + if (seekResult) EXM_THROW(68, "Skip error (end of block)"); + storedSkips = 0; + sizeCheck = fwrite(restPtr, 1, restEnd - restPtr, file); + if (sizeCheck != (size_t)(restEnd - restPtr)) EXM_THROW(68, "Write error : cannot write decoded end of block"); + } + } + + return storedSkips; +} + +static void LZ4IO_fwriteSparseEnd(FILE* file, unsigned storedSkips) +{ + char lastZeroByte[1] = { 0 }; + + if (storedSkips>0) /* implies g_sparseFileSupport */ + { + int seekResult; + size_t sizeCheck; + storedSkips --; + seekResult = fseek(file, storedSkips, SEEK_CUR); + if (seekResult != 0) EXM_THROW(69, "Final skip error (sparse file)\n"); + sizeCheck = fwrite(lastZeroByte, 1, 1, file); + if (sizeCheck != 1) EXM_THROW(69, "Write error : cannot write last zero\n"); + } +} + + +static unsigned g_magicRead = 0; +static unsigned long long LZ4IO_decodeLegacyStream(FILE* finput, FILE* foutput) { unsigned long long filesize = 0; char* in_buff; char* out_buff; + unsigned storedSkips = 0; /* Allocate Memory */ in_buff = (char*)malloc(LZ4_compressBound(LEGACY_BLOCKSIZE)); @@ -556,7 +742,7 @@ static unsigned long long decodeLegacyStream(FILE* finput, FILE* foutput) blockSize = LZ4IO_readLE32(in_buff); /* Convert to Little Endian */ if (blockSize > LZ4_COMPRESSBOUND(LEGACY_BLOCKSIZE)) { /* Cannot read next block : maybe new stream ? */ - fseek(finput, -4, SEEK_CUR); + g_magicRead = blockSize; break; } @@ -570,10 +756,11 @@ static unsigned long long decodeLegacyStream(FILE* finput, FILE* foutput) filesize += decodeSize; /* Write Block */ - sizeCheck = fwrite(out_buff, 1, decodeSize, foutput); - if (sizeCheck != (size_t)decodeSize) EXM_THROW(54, "Write error : cannot write decoded block into output\n"); + storedSkips = LZ4IO_fwriteSparse(foutput, out_buff, decodeSize, storedSkips); } + LZ4IO_fwriteSparseEnd(foutput, storedSkips); + /* Free */ free(in_buff); free(out_buff); @@ -582,172 +769,129 @@ static unsigned long long decodeLegacyStream(FILE* finput, FILE* foutput) } -static unsigned long long decodeLZ4S(FILE* finput, FILE* foutput) + +typedef struct { + void* srcBuffer; + size_t srcBufferSize; + void* dstBuffer; + size_t dstBufferSize; + LZ4F_decompressionContext_t dCtx; +} dRess_t; + +static const size_t LZ4IO_dBufferSize = 64 KB; + +static dRess_t LZ4IO_createDResources(void) { - unsigned long long filesize = 0; - void* inBuff; - void* outBuff; -# define HEADERMAX 20 - char headerBuff[HEADERMAX]; - size_t sizeCheck; - const size_t inBuffSize = 256 KB; - const size_t outBuffSize = 256 KB; - LZ4F_decompressionContext_t ctx; + dRess_t ress; LZ4F_errorCode_t errorCode; - unsigned storedSkips = 0; /* init */ - errorCode = LZ4F_createDecompressionContext(&ctx, LZ4F_VERSION); - if (LZ4F_isError(errorCode)) EXM_THROW(60, "Can't create context : %s", LZ4F_getErrorName(errorCode)); - LZ4IO_writeLE32(headerBuff, LZ4IO_MAGICNUMBER); /* regenerated here, as it was already read from finput */ + errorCode = LZ4F_createDecompressionContext(&ress.dCtx, LZ4F_VERSION); + if (LZ4F_isError(errorCode)) EXM_THROW(60, "Can't create LZ4F context : %s", LZ4F_getErrorName(errorCode)); /* Allocate Memory */ - inBuff = malloc(256 KB); - outBuff = malloc(256 KB); - if (!inBuff || !outBuff) EXM_THROW(61, "Allocation error : not enough memory"); + ress.srcBufferSize = LZ4IO_dBufferSize; + ress.srcBuffer = malloc(ress.srcBufferSize); + ress.dstBufferSize = LZ4IO_dBufferSize; + ress.dstBuffer = malloc(ress.dstBufferSize); + if (!ress.srcBuffer || !ress.dstBuffer) EXM_THROW(61, "Allocation error : not enough memory"); + + return ress; +} + +static void LZ4IO_freeDResources(dRess_t ress) +{ + LZ4F_errorCode_t errorCode = LZ4F_freeDecompressionContext(ress.dCtx); + if (LZ4F_isError(errorCode)) EXM_THROW(69, "Error : can't free LZ4F context resource : %s", LZ4F_getErrorName(errorCode)); + free(ress.srcBuffer); + free(ress.dstBuffer); +} - /* Init feed with magic number (already consumed from FILE) */ + +static unsigned long long LZ4IO_decompressLZ4F(dRess_t ress, FILE* srcFile, FILE* dstFile) +{ + unsigned long long filesize = 0; + LZ4F_errorCode_t nextToLoad; + unsigned storedSkips = 0; + + /* Init feed with magic number (already consumed from FILE* sFile) */ { - size_t inSize = 4; - size_t outSize=0; - LZ4IO_writeLE32(inBuff, LZ4IO_MAGICNUMBER); - errorCode = LZ4F_decompress(ctx, outBuff, &outSize, inBuff, &inSize, NULL); - if (LZ4F_isError(errorCode)) EXM_THROW(62, "Header error : %s", LZ4F_getErrorName(errorCode)); + size_t inSize = MAGICNUMBER_SIZE; + size_t outSize= 0; + LZ4IO_writeLE32(ress.srcBuffer, LZ4IO_MAGICNUMBER); + nextToLoad = LZ4F_decompress(ress.dCtx, ress.dstBuffer, &outSize, ress.srcBuffer, &inSize, NULL); + if (LZ4F_isError(nextToLoad)) EXM_THROW(62, "Header error : %s", LZ4F_getErrorName(nextToLoad)); } - /* Main Loop */ - for (;;) + for (;nextToLoad;) { size_t readSize; size_t pos = 0; + size_t decodedBytes = ress.dstBufferSize; /* Read input */ - readSize = fread(inBuff, 1, inBuffSize, finput); - if (!readSize) break; /* empty file or stream */ + if (nextToLoad > ress.srcBufferSize) nextToLoad = ress.srcBufferSize; + readSize = fread(ress.srcBuffer, 1, nextToLoad, srcFile); + if (!readSize) + break; /* empty file or stream */ - while (pos < readSize) + while ((pos < readSize) || (decodedBytes == ress.dstBufferSize)) /* still to read, or still to flush */ { /* Decode Input (at least partially) */ size_t remaining = readSize - pos; - size_t decodedBytes = outBuffSize; - errorCode = LZ4F_decompress(ctx, outBuff, &decodedBytes, (char*)inBuff+pos, &remaining, NULL); - if (LZ4F_isError(errorCode)) EXM_THROW(66, "Decompression error : %s", LZ4F_getErrorName(errorCode)); + decodedBytes = ress.dstBufferSize; + nextToLoad = LZ4F_decompress(ress.dCtx, ress.dstBuffer, &decodedBytes, (char*)(ress.srcBuffer)+pos, &remaining, NULL); + if (LZ4F_isError(nextToLoad)) EXM_THROW(66, "Decompression error : %s", LZ4F_getErrorName(nextToLoad)); pos += remaining; + if (!nextToLoad) break; if (decodedBytes) { /* Write Block */ filesize += decodedBytes; - if (g_sparseFileSupport) - { - size_t* const oBuffStartT = (size_t*)outBuff; /* since outBuff is malloc'ed, it's aligned on size_t */ - size_t* oBuffPosT = oBuffStartT; - size_t oBuffSizeT = decodedBytes / sizeT; - size_t* const oBuffEndT = oBuffStartT + oBuffSizeT; - static const size_t bs0T = (32 KB) / sizeT; - while (oBuffPosT < oBuffEndT) - { - size_t seg0SizeT = bs0T; - size_t nb0T; - int seekResult; - if (seg0SizeT > oBuffSizeT) seg0SizeT = oBuffSizeT; - oBuffSizeT -= seg0SizeT; - for (nb0T=0; (nb0T < seg0SizeT) && (oBuffPosT[nb0T] == 0); nb0T++) ; - storedSkips += (unsigned)(nb0T * sizeT); - if (storedSkips > 1 GB) /* avoid int overflow */ - { - seekResult = fseek(foutput, 1 GB, SEEK_CUR); - if (seekResult != 0) EXM_THROW(68, "1 GB skip error (sparse file)"); - storedSkips -= 1 GB; - } - if (nb0T != seg0SizeT) /* not all 0s */ - { - seekResult = fseek(foutput, storedSkips, SEEK_CUR); - if (seekResult) EXM_THROW(68, "Skip error (sparse file)"); - storedSkips = 0; - seg0SizeT -= nb0T; - oBuffPosT += nb0T; - sizeCheck = fwrite(oBuffPosT, sizeT, seg0SizeT, foutput); - if (sizeCheck != seg0SizeT) EXM_THROW(68, "Write error : cannot write decoded block"); - } - oBuffPosT += seg0SizeT; - } - if (decodedBytes & maskT) /* size not multiple of sizeT (necessarily end of block) */ - { - const char* const restStart = (char*)oBuffEndT; - const char* restPtr = restStart; - size_t restSize = decodedBytes & maskT; - const char* const restEnd = restStart + restSize; - for (; (restPtr < restEnd) && (*restPtr == 0); restPtr++) ; - storedSkips += (unsigned) (restPtr - restStart); - if (restPtr != restEnd) - { - int seekResult = fseek(foutput, storedSkips, SEEK_CUR); - if (seekResult) EXM_THROW(68, "Skip error (end of block)"); - storedSkips = 0; - sizeCheck = fwrite(restPtr, 1, restEnd - restPtr, foutput); - if (sizeCheck != (size_t)(restEnd - restPtr)) EXM_THROW(68, "Write error : cannot write decoded end of block"); - } - } - } - else - { - sizeCheck = fwrite(outBuff, 1, decodedBytes, foutput); - if (sizeCheck != decodedBytes) EXM_THROW(68, "Write error : cannot write decoded block"); - } + DISPLAYUPDATE(2, "\rDecompressed : %u MB ", (unsigned)(filesize>>20)); + storedSkips = LZ4IO_fwriteSparse(dstFile, ress.dstBuffer, decodedBytes, storedSkips); } } - } - if ((g_sparseFileSupport) && (storedSkips>0)) - { - int seekResult; - storedSkips --; - seekResult = fseek(foutput, storedSkips, SEEK_CUR); - if (seekResult != 0) EXM_THROW(69, "Final skip error (sparse file)\n"); - memset(outBuff, 0, 1); - sizeCheck = fwrite(outBuff, 1, 1, foutput); - if (sizeCheck != 1) EXM_THROW(69, "Write error : cannot write last zero\n"); - } + LZ4IO_fwriteSparseEnd(dstFile, storedSkips); - /* Free */ - free(inBuff); - free(outBuff); - errorCode = LZ4F_freeDecompressionContext(ctx); - if (LZ4F_isError(errorCode)) EXM_THROW(69, "Error : can't free LZ4F context resource : %s", LZ4F_getErrorName(errorCode)); + if (nextToLoad!=0) + EXM_THROW(67, "Unfinished stream"); return filesize; } -static unsigned long long LZ4IO_passThrough(FILE* finput, FILE* foutput, unsigned char U32store[MAGICNUMBER_SIZE]) +static unsigned long long LZ4IO_passThrough(FILE* finput, FILE* foutput, unsigned char MNstore[MAGICNUMBER_SIZE]) { void* buffer = malloc(64 KB); size_t read = 1, sizeCheck; unsigned long long total = MAGICNUMBER_SIZE; + unsigned storedSkips = 0; - sizeCheck = fwrite(U32store, 1, MAGICNUMBER_SIZE, foutput); - if (sizeCheck != MAGICNUMBER_SIZE) EXM_THROW(50, "Pass-through error at start"); + sizeCheck = fwrite(MNstore, 1, MAGICNUMBER_SIZE, foutput); + if (sizeCheck != MAGICNUMBER_SIZE) EXM_THROW(50, "Pass-through write error"); while (read) { read = fread(buffer, 1, 64 KB, finput); total += read; - sizeCheck = fwrite(buffer, 1, read, foutput); - if (sizeCheck != read) EXM_THROW(50, "Pass-through error"); + storedSkips = LZ4IO_fwriteSparse(foutput, buffer, read, storedSkips); } + LZ4IO_fwriteSparseEnd(foutput, storedSkips); free(buffer); return total; } #define ENDOFSTREAM ((unsigned long long)-1) -static unsigned long long selectDecoder( FILE* finput, FILE* foutput) +static unsigned long long selectDecoder(dRess_t ress, FILE* finput, FILE* foutput) { - unsigned char U32store[MAGICNUMBER_SIZE]; + unsigned char MNstore[MAGICNUMBER_SIZE]; unsigned magicNumber, size; int errorNb; size_t nbReadBytes; @@ -757,33 +901,41 @@ static unsigned long long selectDecoder( FILE* finput, FILE* foutput) nbCalls++; /* Check Archive Header */ - nbReadBytes = fread(U32store, 1, MAGICNUMBER_SIZE, finput); - if (nbReadBytes==0) return ENDOFSTREAM; /* EOF */ - if (nbReadBytes != MAGICNUMBER_SIZE) EXM_THROW(40, "Unrecognized header : Magic Number unreadable"); - magicNumber = LZ4IO_readLE32(U32store); /* Little Endian format */ + if (g_magicRead) + { + magicNumber = g_magicRead; + g_magicRead = 0; + } + else + { + nbReadBytes = fread(MNstore, 1, MAGICNUMBER_SIZE, finput); + if (nbReadBytes==0) return ENDOFSTREAM; /* EOF */ + if (nbReadBytes != MAGICNUMBER_SIZE) EXM_THROW(40, "Unrecognized header : Magic Number unreadable"); + magicNumber = LZ4IO_readLE32(MNstore); /* Little Endian format */ + } if (LZ4IO_isSkippableMagicNumber(magicNumber)) magicNumber = LZ4IO_SKIPPABLE0; /* fold skippable magic numbers */ switch(magicNumber) { case LZ4IO_MAGICNUMBER: - return DEFAULT_DECOMPRESSOR(finput, foutput); + return LZ4IO_decompressLZ4F(ress, finput, foutput); case LEGACY_MAGICNUMBER: DISPLAYLEVEL(4, "Detected : Legacy format \n"); - return decodeLegacyStream(finput, foutput); + return LZ4IO_decodeLegacyStream(finput, foutput); case LZ4IO_SKIPPABLE0: DISPLAYLEVEL(4, "Skipping detected skippable area \n"); - nbReadBytes = fread(U32store, 1, 4, finput); + nbReadBytes = fread(MNstore, 1, 4, finput); if (nbReadBytes != 4) EXM_THROW(42, "Stream error : skippable size unreadable"); - size = LZ4IO_readLE32(U32store); /* Little Endian format */ + size = LZ4IO_readLE32(MNstore); /* Little Endian format */ errorNb = fseek(finput, size, SEEK_CUR); if (errorNb != 0) EXM_THROW(43, "Stream error : cannot skip skippable area"); - return selectDecoder(finput, foutput); + return selectDecoder(ress, finput, foutput); EXTENDED_FORMAT; default: if (nbCalls == 1) /* just started */ { if (g_overwrite) - return LZ4IO_passThrough(finput, foutput, U32store); + return LZ4IO_passThrough(finput, foutput, MNstore); EXM_THROW(44,"Unrecognized header : file cannot be decoded"); /* Wrong magic number at the beginning of 1st stream */ } DISPLAYLEVEL(2, "Stream followed by unrecognized data\n"); @@ -792,43 +944,95 @@ static unsigned long long selectDecoder( FILE* finput, FILE* foutput) } -int LZ4IO_decompressFilename(const char* input_filename, const char* output_filename) +static int LZ4IO_decompressFile_extRess(dRess_t ress, const char* input_filename, const char* output_filename) { unsigned long long filesize = 0, decodedSize=0; FILE* finput; FILE* foutput; - clock_t start, end; /* Init */ - start = clock(); - get_fileHandle(input_filename, output_filename, &finput, &foutput); + if (LZ4IO_getFiles(input_filename, output_filename, &finput, &foutput)) + return 1; /* sparse file */ - if (g_sparseFileSupport && foutput) { SET_SPARSE_FILE_MODE(foutput); } + if (g_sparseFileSupport) { SET_SPARSE_FILE_MODE(foutput); } /* Loop over multiple streams */ do { - decodedSize = selectDecoder(finput, foutput); + decodedSize = selectDecoder(ress, finput, foutput); if (decodedSize != ENDOFSTREAM) filesize += decodedSize; } while (decodedSize != ENDOFSTREAM); /* Final Status */ - end = clock(); DISPLAYLEVEL(2, "\r%79s\r", ""); DISPLAYLEVEL(2, "Successfully decoded %llu bytes \n", filesize); - { - double seconds = (double)(end - start)/CLOCKS_PER_SEC; - DISPLAYLEVEL(4, "Done in %.2f s ==> %.2f MB/s\n", seconds, (double)filesize / seconds / 1024 / 1024); - } /* Close */ fclose(finput); fclose(foutput); - /* Error status = OK */ return 0; } + +int LZ4IO_decompressFilename(const char* input_filename, const char* output_filename) +{ + dRess_t ress; + clock_t start, end; + int missingFiles = 0; + + start = clock(); + + ress = LZ4IO_createDResources(); + missingFiles += LZ4IO_decompressFile_extRess(ress, input_filename, output_filename); + LZ4IO_freeDResources(ress); + + end = clock(); + if (end==start) end=start+1; + { + double seconds = (double)(end - start)/CLOCKS_PER_SEC; + DISPLAYLEVEL(4, "Done in %.2f sec \n", seconds); + } + + return missingFiles; +} + + +int LZ4IO_decompressMultipleFilenames(const char** inFileNamesTable, int ifntSize, const char* suffix) +{ + int i; + int skippedFiles = 0; + int missingFiles = 0; + char* outFileName = (char*)malloc(FNSPACE); + size_t ofnSize = FNSPACE; + const size_t suffixSize = strlen(suffix); + char* ifnSuffix = (char*)malloc(suffixSize + 1); + dRess_t ress; + + ress = LZ4IO_createDResources(); + + for (i=0; i<ifntSize; i++) + { + size_t ifnSize = strlen(inFileNamesTable[i]); + strcpy(ifnSuffix, inFileNamesTable[i] + ifnSize - suffixSize); + if (ofnSize <= ifnSize-suffixSize+1) { free(outFileName); ofnSize = ifnSize + 20; outFileName = (char*)malloc(ofnSize); } + if (ifnSize <= suffixSize || strcmp(ifnSuffix, suffix) != 0) + { + DISPLAYLEVEL(1, "File extension doesn't match expected LZ4_EXTENSION (%4s); will not process file: %s\n", suffix, inFileNamesTable[i]); + skippedFiles++; + continue; + } + memcpy(outFileName, inFileNamesTable[i], ifnSize - suffixSize); + outFileName[ifnSize-suffixSize] = '\0'; + + missingFiles += LZ4IO_decompressFile_extRess(ress, inFileNamesTable[i], outFileName); + } + + LZ4IO_freeDResources(ress); + free(outFileName); + free(ifnSuffix); + return missingFiles + skippedFiles; +} |