Merge pull request #1855 from shashank0791/shtavi-filter-precompressed-files

Added "--exclude-compressed" flag feature

programs/fileio.c

@@ -319,6 +319,8 @@ struct FIO_prefs_s {
     /* Computation resources preferences */
     unsigned memLimit;
     int nbWorkers;
+
+    int excludeCompressedFiles;
 };
 
 
@@ -359,6 +361,7 @@ FIO_prefs_t* FIO_createPreferences(void)
     ret->srcSizeHint = 0;
     ret->testMode = 0;
     ret->literalCompressionMode = ZSTD_lcm_auto;
+    ret->excludeCompressedFiles = 0;
     return ret;
 }
 
@@ -402,6 +405,8 @@ void FIO_setNbWorkers(FIO_prefs_t* const prefs, int nbWorkers) {
     prefs->nbWorkers = nbWorkers;
 }
 
+void FIO_setExcludeCompressedFile(FIO_prefs_t* const prefs, int excludeCompressedFiles) { prefs->excludeCompressedFiles = excludeCompressedFiles; }
+
 void FIO_setBlockSize(FIO_prefs_t* const prefs, int blockSize) {
     if (blockSize && prefs->nbWorkers==0)
         DISPLAYLEVEL(2, "Setting block size is useless in single-thread mode \n");
@@ -1425,6 +1430,21 @@ static int FIO_compressFilename_dstFile(FIO_prefs_t* const prefs,
     return result;
 }
 
+/* List used to compare file extensions (used with --exclude-compressed flag)
+* Different from the suffixList and should only apply to ZSTD compress operationResult
+*/
+static const char *compressedFileExtensions[] = {
+    ZSTD_EXTENSION,
+    TZSTD_EXTENSION,
+    GZ_EXTENSION,
+    TGZ_EXTENSION,
+    LZMA_EXTENSION,
+    XZ_EXTENSION,
+    TXZ_EXTENSION,
+    LZ4_EXTENSION,
+    TLZ4_EXTENSION,
+    NULL
+};
 
 /*! FIO_compressFilename_srcFile() :
  *  @return : 0 : compression completed correctly,
@@ -1451,6 +1471,15 @@ FIO_compressFilename_srcFile(FIO_prefs_t* const prefs,
         return 1;
     }
 
+    /* Check if "srcFile" is compressed. Only done if --exclude-compressed flag is used
+    * YES => ZSTD will skip compression of the file and will return 0.
+    * NO => ZSTD will resume with compress operation.
+    */
+    if (prefs->excludeCompressedFiles == 1 && UTIL_isCompressedFile(srcFileName, compressedFileExtensions)) {
+        DISPLAYLEVEL(4, "File is already compressed : %s \n", srcFileName);
+        return 0;
+    }
+
     ress.srcFile = FIO_openSrcFile(srcFileName);
     if (ress.srcFile == NULL) return 1;   /* srcFile could not be opened */
 

programs/fileio.h

@@ -93,6 +93,7 @@ void FIO_setLiteralCompressionMode(
 
 void FIO_setNoProgress(unsigned noProgress);
 void FIO_setNotificationLevel(int level);
+void FIO_setExcludeCompressedFile(FIO_prefs_t* const prefs, int excludeCompressedFiles);
 
 /*-*************************************
 *  Single File functions

programs/util.c

@@ -326,6 +326,27 @@ int UTIL_prepareFileList(const char *dirName, char** bufStart, size_t* pos, char
 
 #endif /* #ifdef _WIN32 */
 
+int UTIL_isCompressedFile(const char *inputName, const char *extensionList[])
+{
+  const char* ext = UTIL_getFileExtension(inputName);
+  while(*extensionList!=NULL)
+  {
+    const int isCompressedExtension = strcmp(ext,*extensionList);
+    if(isCompressedExtension==0)
+      return 1;
+    ++extensionList;
+  }
+   return 0;
+}
+
+/*Utility function to get file extension from file */
+const char* UTIL_getFileExtension(const char* infilename)
+{
+   const char* extension = strrchr(infilename, '.');
+   if(!extension || extension==infilename) return "";
+   return extension;
+}
+
 /*
  * UTIL_createFileList - takes a list of files and directories (params: inputNames, inputNamesNb), scans directories,
  *                       and returns a new list of files (params: return value, allocatedBuffer, allocatedNamesNb).

programs/util.h

@@ -40,7 +40,6 @@ extern "C" {
 #include <time.h>         /* clock_t, clock, CLOCKS_PER_SEC, nanosleep */
 #include "mem.h"          /* U32, U64 */
 
-
 /*-************************************************************
 * Avoid fseek()'s 2GiB barrier with MSVC, macOS, *BSD, MinGW
 ***************************************************************/
@@ -135,6 +134,8 @@ U32 UTIL_isDirectory(const char* infilename);
 int UTIL_getFileStat(const char* infilename, stat_t* statbuf);
 int UTIL_isSameFile(const char* file1, const char* file2);
 int UTIL_compareStr(const void *p1, const void *p2);
+int UTIL_isCompressedFile(const char* infilename, const char *extensionList[]);
+const char* UTIL_getFileExtension(const char* infilename);
 
 U32 UTIL_isFIFO(const char* infilename);
 U32 UTIL_isLink(const char* infilename);

programs/zstdcli.c

@@ -136,6 +136,7 @@ static int usage_advanced(const char* programName)
     DISPLAY( " -q     : suppress warnings; specify twice to suppress errors too\n");
     DISPLAY( " -c     : force write to standard output, even if it is the console\n");
     DISPLAY( " -l     : print information about zstd compressed files \n");
+    DISPLAY( "--exclude-compressed:  only compress files that are not previously compressed \n");
 #ifndef ZSTD_NOCOMPRESS
     DISPLAY( "--ultra : enable levels beyond %i, up to %i (requires more memory)\n", ZSTDCLI_CLEVEL_MAX, ZSTD_maxCLevel());
     DISPLAY( "--long[=#]: enable long distance matching with given window log (default: %u)\n", g_defaultMaxWindowLog);
@@ -708,7 +709,7 @@ int main(int argCount, const char* argv[])
                     if (!strcmp(argument, "--compress-literals")) { literalCompressionMode = ZSTD_lcm_huffman; continue; }
                     if (!strcmp(argument, "--no-compress-literals")) { literalCompressionMode = ZSTD_lcm_uncompressed; continue; }
                     if (!strcmp(argument, "--no-progress")) { FIO_setNoProgress(1); continue; }
-
+                    if (!strcmp(argument, "--exclude-compressed")) { FIO_setExcludeCompressedFile(prefs, 1); continue; }
                     /* long commands with arguments */
 #ifndef ZSTD_NODICT
                     if (longCommandWArg(&argument, "--train-cover")) {

tests/playTests.sh

@@ -215,6 +215,37 @@ $ZSTD tmp -c --compress-literals    -19      | $ZSTD -t
 $ZSTD -b --fast=1 -i0e1 tmp --compress-literals
 $ZSTD -b --fast=1 -i0e1 tmp --no-compress-literals
 
+println "test: --exclude-compressed flag"
+rm -rf precompressedFilterTestDir
+mkdir -p precompressedFilterTestDir
+./datagen $size > precompressedFilterTestDir/input.5
+./datagen $size > precompressedFilterTestDir/input.6
+$ZSTD --exclude-compressed --long --rm -r precompressedFilterTestDir
+sleep 5
+./datagen $size > precompressedFilterTestDir/input.7
+./datagen $size > precompressedFilterTestDir/input.8
+$ZSTD --exclude-compressed --long --rm -r precompressedFilterTestDir
+test ! -f precompressedFilterTestDir/input.5.zst.zst
+test ! -f precompressedFilterTestDir/input.6.zst.zst
+file1timestamp=`date -r precompressedFilterTestDir/input.5.zst +%s`
+file2timestamp=`date -r precompressedFilterTestDir/input.7.zst +%s`
+if [[ $file2timestamp -ge $file1timestamp ]]; then
+  println "Test is successful. input.5.zst is precompressed and therefore not compressed/modified again."
+else
+  println "Test is not successful"
+fi
+#File Extension check.
+./datagen $size > precompressedFilterTestDir/input.zstbar
+$ZSTD --exclude-compressed --long --rm -r precompressedFilterTestDir
+#ZSTD should compress input.zstbar
+test -f precompressedFilterTestDir/input.zstbar.zst
+#Check without the --exclude-compressed flag
+$ZSTD --long --rm -r precompressedFilterTestDir
+#Files should get compressed again without the --exclude-compressed flag.
+test -f precompressedFilterTestDir/input.5.zst.zst
+test -f precompressedFilterTestDir/input.6.zst.zst
+println "Test completed"
+
 println "test : file removal"
 $ZSTD -f --rm tmp
 test ! -f tmp  # tmp should no longer be present