Merge pull request #543 from inikep/gz_compress

Gz compress

programs/Makefile

@@ -70,7 +70,7 @@ VOID = /dev/null
 HAVE_ZLIB := $(shell printf '\#include <zlib.h>\nint main(){}' | $(CC) -o have_zlib -x c - -lz 2> $(VOID) && echo 1 || echo 0)
 ifeq ($(HAVE_ZLIB), 1)
 TEMP := $(shell rm have_zlib$(EXT))
-ZLIBCPP = -DZSTD_GZDECOMPRESS
+ZLIBCPP = -DZSTD_GZCOMPRESS -DZSTD_GZDECOMPRESS
 ZLIBLD = -lz
 endif
 

programs/fileio.c

@@ -38,7 +38,7 @@
 #ifdef ZSTD_MULTITHREAD
 #  include "zstdmt_compress.h"
 #endif
-#ifdef ZSTD_GZDECOMPRESS
+#if defined(ZSTD_GZCOMPRESS) || defined(ZSTD_GZDECOMPRESS)
 #  include <zlib.h>
 #  if !defined(z_const)
 #    define z_const
@@ -121,6 +121,8 @@ static clock_t g_time = 0;
 /*-*************************************
 *  Local Parameters - Not thread safe
 ***************************************/
+static FIO_compressionType_t g_compressionType = FIO_zstdCompression;
+void FIO_setCompressionType(FIO_compressionType_t compressionType) { g_compressionType = compressionType; }
 static U32 g_overwrite = 0;
 void FIO_overwriteMode(void) { g_overwrite=1; }
 static U32 g_sparseFileSupport = 1;   /* 0 : no sparse allowed; 1: auto (file yes, stdout no); 2: force sparse */
@@ -361,13 +363,79 @@ static void FIO_freeCResources(cRess_t ress)
 }
 
 
+#ifdef ZSTD_GZCOMPRESS
+static unsigned long long FIO_compressGzFrame(cRess_t* ress, const char* srcFileName, U64 const srcFileSize, int compressionLevel, U64* readsize)
+{
+    unsigned long long inFileSize = 0, outFileSize = 0;
+    z_stream strm;
+    int ret;
+
+    if (compressionLevel > Z_BEST_COMPRESSION) compressionLevel = Z_BEST_COMPRESSION;
+
+    strm.zalloc = Z_NULL;
+    strm.zfree = Z_NULL;
+    strm.opaque = Z_NULL;
+
+    ret = deflateInit2(&strm, compressionLevel, Z_DEFLATED, 15 /* maxWindowLogSize */ + 16 /* gzip only */, 8, Z_DEFAULT_STRATEGY); /* see http://www.zlib.net/manual.html */
+    if (ret != Z_OK) EXM_THROW(71, "zstd: %s: deflateInit2 error %d \n", srcFileName, ret);
+
+    strm.next_in = 0;
+    strm.avail_in = Z_NULL;
+    strm.next_out = (Bytef*)ress->dstBuffer;
+    strm.avail_out = (uInt)ress->dstBufferSize;
+
+    while (1) {
+        if (strm.avail_in == 0) {
+            size_t const inSize = fread(ress->srcBuffer, 1, ress->srcBufferSize, ress->srcFile);
+            if (inSize == 0) break;
+            inFileSize += inSize;
+            strm.next_in = (z_const unsigned char*)ress->srcBuffer;
+            strm.avail_in = (uInt)inSize;
+        }
+        ret = deflate(&strm, Z_NO_FLUSH);
+        if (ret != Z_OK) EXM_THROW(72, "zstd: %s: deflate error %d \n", srcFileName, ret);
+        {   size_t const decompBytes = ress->dstBufferSize - strm.avail_out;
+            if (decompBytes) {
+                if (fwrite(ress->dstBuffer, 1, decompBytes, ress->dstFile) != decompBytes) EXM_THROW(73, "Write error : cannot write to output file");
+                outFileSize += decompBytes;
+                strm.next_out = (Bytef*)ress->dstBuffer;
+                strm.avail_out = (uInt)ress->dstBufferSize;
+            }
+        }
+        if (!srcFileSize) DISPLAYUPDATE(2, "\rRead : %u MB ==> %.2f%%", (U32)(inFileSize>>20), (double)outFileSize/inFileSize*100)
+        else DISPLAYUPDATE(2, "\rRead : %u / %u MB ==> %.2f%%", (U32)(inFileSize>>20), (U32)(srcFileSize>>20), (double)outFileSize/inFileSize*100);
+    }
+
+    while (1) {
+        ret = deflate(&strm, Z_FINISH);
+        {   size_t const decompBytes = ress->dstBufferSize - strm.avail_out;
+            if (decompBytes) {
+                if (fwrite(ress->dstBuffer, 1, decompBytes, ress->dstFile) != decompBytes) EXM_THROW(75, "Write error : cannot write to output file");
+                outFileSize += decompBytes;
+                strm.next_out = (Bytef*)ress->dstBuffer;
+                strm.avail_out = (uInt)ress->dstBufferSize;
+            }
+        }
+        if (ret == Z_STREAM_END) break;
+        if (ret != Z_BUF_ERROR) EXM_THROW(77, "zstd: %s: deflate error %d \n", srcFileName, ret);
+    }
+
+    ret = deflateEnd(&strm);
+    if (ret != Z_OK) EXM_THROW(79, "zstd: %s: deflateEnd error %d \n", srcFileName, ret);
+    *readsize = inFileSize;
+
+    return outFileSize;
+}
+#endif
+
+
 /*! FIO_compressFilename_internal() :
  *  same as FIO_compressFilename_extRess(), with `ress.desFile` already opened.
  *  @return : 0 : compression completed correctly,
  *            1 : missing or pb opening srcFileName
  */
 static int FIO_compressFilename_internal(cRess_t ress,
-                                         const char* dstFileName, const char* srcFileName)
+                                         const char* dstFileName, const char* srcFileName, int compressionLevel)
 {
     FILE* const srcFile = ress.srcFile;
     FILE* const dstFile = ress.dstFile;
@@ -375,6 +443,16 @@ static int FIO_compressFilename_internal(cRess_t ress,
     U64 compressedfilesize = 0;
     U64 const fileSize = UTIL_getFileSize(srcFileName);
 
+    if (g_compressionType) {
+#ifdef ZSTD_GZCOMPRESS
+        compressedfilesize = FIO_compressGzFrame(&ress, srcFileName, fileSize, compressionLevel, &readsize);
+#else
+        (void)compressionLevel;
+        EXM_THROW(20, "zstd: %s: file cannot be compressed as gzip (zstd compiled without ZSTD_GZCOMPRESS) -- ignored \n", srcFileName);
+#endif
+        goto finish;
+    }
+
     /* init */
 #ifdef ZSTD_MULTITHREAD
     {   size_t const resetError = ZSTDMT_resetCStream(ress.cctx, fileSize);
@@ -432,6 +510,7 @@ static int FIO_compressFilename_internal(cRess_t ress,
         }
     }
 
+finish:
     /* Status */
     DISPLAYLEVEL(2, "\r%79s\r", "");
     DISPLAYLEVEL(2,"%-20s :%6.2f%%   (%6llu => %6llu bytes, %s) \n", srcFileName,
@@ -449,7 +528,7 @@ static int FIO_compressFilename_internal(cRess_t ress,
  *            1 : missing or pb opening srcFileName
  */
 static int FIO_compressFilename_srcFile(cRess_t ress,
-                                        const char* dstFileName, const char* srcFileName)
+                                        const char* dstFileName, const char* srcFileName, int compressionLevel)
 {
     int result;
 
@@ -462,7 +541,7 @@ static int FIO_compressFilename_srcFile(cRess_t ress,
     ress.srcFile = FIO_openSrcFile(srcFileName);
     if (!ress.srcFile) return 1;   /* srcFile could not be opened */
 
-    result = FIO_compressFilename_internal(ress, dstFileName, srcFileName);
+    result = FIO_compressFilename_internal(ress, dstFileName, srcFileName, compressionLevel);
 
     fclose(ress.srcFile);
     if (g_removeSrcFile && !result) { if (remove(srcFileName)) EXM_THROW(1, "zstd: %s: %s", srcFileName, strerror(errno)); } /* remove source file : --rm */
@@ -475,7 +554,7 @@ static int FIO_compressFilename_srcFile(cRess_t ress,
  *            1 : pb
  */
 static int FIO_compressFilename_dstFile(cRess_t ress,
-                                        const char* dstFileName, const char* srcFileName)
+                                        const char* dstFileName, const char* srcFileName, int compressionLevel)
 {
     int result;
     stat_t statbuf;
@@ -485,7 +564,7 @@ static int FIO_compressFilename_dstFile(cRess_t ress,
     if (ress.dstFile==NULL) return 1;  /* could not open dstFileName */
 
     if (strcmp (srcFileName, stdinmark) && UTIL_getFileStat(srcFileName, &statbuf)) stat_result = 1;
-    result = FIO_compressFilename_srcFile(ress, dstFileName, srcFileName);
+    result = FIO_compressFilename_srcFile(ress, dstFileName, srcFileName, compressionLevel);
 
     if (fclose(ress.dstFile)) { DISPLAYLEVEL(1, "zstd: %s: %s \n", dstFileName, strerror(errno)); result=1; }  /* error closing dstFile */
     if (result!=0) { if (remove(dstFileName)) EXM_THROW(1, "zstd: %s: %s", dstFileName, strerror(errno)); }  /* remove operation artefact */
@@ -502,7 +581,7 @@ int FIO_compressFilename(const char* dstFileName, const char* srcFileName,
     int const regFile = UTIL_isRegFile(srcFileName);
 
     cRess_t const ress = FIO_createCResources(dictFileName, compressionLevel, srcSize, regFile, comprParams);
-    int const result = FIO_compressFilename_dstFile(ress, dstFileName, srcFileName);
+    int const result = FIO_compressFilename_dstFile(ress, dstFileName, srcFileName, compressionLevel);
 
     double const seconds = (double)(clock() - start) / CLOCKS_PER_SEC;
     DISPLAYLEVEL(4, "Completed in %.2f sec \n", seconds);
@@ -535,7 +614,7 @@ int FIO_compressMultipleFilenames(const char** inFileNamesTable, unsigned nbFile
         ress.dstFile = stdout;
         SET_BINARY_MODE(stdout);
         for (u=0; u<nbFiles; u++)
-            missed_files += FIO_compressFilename_srcFile(ress, stdoutmark, inFileNamesTable[u]);
+            missed_files += FIO_compressFilename_srcFile(ress, stdoutmark, inFileNamesTable[u], compressionLevel);
         if (fclose(ress.dstFile)) EXM_THROW(29, "Write error : cannot properly close stdout");
     } else {
         unsigned u;
@@ -544,7 +623,7 @@ int FIO_compressMultipleFilenames(const char** inFileNamesTable, unsigned nbFile
             if (dfnSize <= ifnSize+suffixSize+1) { free(dstFileName); dfnSize = ifnSize + 20; dstFileName = (char*)malloc(dfnSize); }
             strcpy(dstFileName, inFileNamesTable[u]);
             strcat(dstFileName, suffix);
-            missed_files += FIO_compressFilename_dstFile(ress, dstFileName, inFileNamesTable[u]);
+            missed_files += FIO_compressFilename_dstFile(ress, dstFileName, inFileNamesTable[u], compressionLevel);
     }   }
 
     /* Close & Free */

programs/fileio.h

@@ -31,9 +31,16 @@ extern "C" {
 #endif
 
 
+/*-*************************************
+*  Types
+***************************************/
+typedef enum { FIO_zstdCompression, FIO_gzipCompression } FIO_compressionType_t;
+
+
 /*-*************************************
 *  Parameters
 ***************************************/
+void FIO_setCompressionType(FIO_compressionType_t compressionType);
 void FIO_overwriteMode(void);
 void FIO_setNotificationLevel(unsigned level);
 void FIO_setSparseWrite(unsigned sparse);  /**< 0: no sparse; 1: disable on stdout; 2: always enabled */

programs/zstdcli.c

@@ -49,6 +49,7 @@
 #define AUTHOR "Yann Collet"
 #define WELCOME_MESSAGE "*** %s %i-bits %s, by %s ***\n", COMPRESSOR_NAME, (int)(sizeof(size_t)*8), ZSTD_VERSION, AUTHOR
 
+#define GZ_EXTENSION ".gz"
 #define ZSTD_EXTENSION ".zst"
 #define ZSTD_CAT "zstdcat"
 #define ZSTD_UNZSTD "unzstd"
@@ -123,6 +124,9 @@ static int usage_advanced(const char* programName)
     DISPLAY( " -T#    : use # threads for compression (default:1) \n");
     DISPLAY( " -B#    : select size of independent sections (default:0==automatic) \n");
 #endif
+#ifdef ZSTD_GZCOMPRESS
+    DISPLAY( "--format=gzip : compress files to the .gz format \n");
+#endif
 #endif
 #ifndef ZSTD_NODECOMPRESS
     DISPLAY( "--test  : test compressed file integrity \n");
@@ -283,6 +287,7 @@ int main(int argCount, const char* argv[])
     const char* programName = argv[0];
     const char* outFileName = NULL;
     const char* dictFileName = NULL;
+    const char* suffix = ZSTD_EXTENSION;
     unsigned maxDictSize = g_defaultMaxDictSize;
     unsigned dictID = 0;
     int dictCLevel = g_defaultDictCLevel;
@@ -359,6 +364,7 @@ int main(int argCount, const char* argv[])
                     if (!strcmp(argument, "--keep")) { FIO_setRemoveSrcFile(0); continue; }
                     if (!strcmp(argument, "--rm")) { FIO_setRemoveSrcFile(1); continue; }
                     if (!strcmp(argument, "--priority=rt")) { setRealTimePrio = 1; continue; }
+                    if (!strcmp(argument, "--format=gzip")) { suffix = GZ_EXTENSION; FIO_setCompressionType(FIO_gzipCompression); continue; }
 
                     /* long commands with arguments */
 #ifndef  ZSTD_NODICT
@@ -643,7 +649,7 @@ int main(int argCount, const char* argv[])
         if ((filenameIdx==1) && outFileName)
           operationResult = FIO_compressFilename(outFileName, filenameTable[0], dictFileName, cLevel, &compressionParams);
         else
-          operationResult = FIO_compressMultipleFilenames(filenameTable, filenameIdx, outFileName ? outFileName : ZSTD_EXTENSION, dictFileName, cLevel, &compressionParams);
+          operationResult = FIO_compressMultipleFilenames(filenameTable, filenameIdx, outFileName ? outFileName : suffix, dictFileName, cLevel, &compressionParams);
 #else
         DISPLAY("Compression not supported\n");
 #endif