[File] [PATCH] Zstandard decompression using libzstd

[Martin Rodriguez Reboredo]

Uses the streaming decompression API of libzstd to obtain inflated info for
magic processing from Zstandard compressed data.
---
 configure.ac   | 19 ++++++++++++++
 src/compress.c | 68 ++++++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 87 insertions(+)

diff --git a/configure.ac b/configure.ac
index ffbe0f69..9172be9d 100644
--- a/configure.ac
+++ b/configure.ac
@@ -49,6 +49,11 @@ AC_ARG_ENABLE([xzlib],
 [AS_HELP_STRING([--disable-xzlib], [disable liblzma/xz compression support @<:@default=auto@:>@])])
 AC_MSG_RESULT($enable_xzlib)
 
+AC_MSG_CHECKING(for zstdlib support)
+AC_ARG_ENABLE([zstdlib],
+[AS_HELP_STRING([--disable-zstdlib], [disable zstdlib compression support @<:@default=auto@:>@])])
+AC_MSG_RESULT($enable_zstdlib)
+
 AC_MSG_CHECKING(for libseccomp support)
 AC_ARG_ENABLE([libseccomp],
 [AS_HELP_STRING([--disable-libseccomp], [disable libseccomp sandboxing @<:@default=auto@:>@])])
@@ -112,6 +117,9 @@ fi
 if test "$enable_xzlib" != "no"; then
   AC_CHECK_HEADERS(lzma.h)
 fi
+if test "$enable_zstdlib" != "no"; then
+  AC_CHECK_HEADERS(zstd.h zstd_errors.h)
+fi
 AC_CHECK_TYPE([sig_t],[AC_DEFINE([HAVE_SIG_T],1,[Have sig_t type])],,[#include <signal.h>])
 
 dnl Checks for typedefs, structures, and compiler characteristics.
@@ -180,6 +188,9 @@ fi
 if test "$enable_xzlib" != "no"; then
   AC_CHECK_LIB(lzma, lzma_stream_decoder)
 fi
+if test "$enable_zstdlib" != "no"; then
+  AC_CHECK_LIB(zstd, ZSTD_createDStream)
+fi
 if test "$enable_libseccomp" != "no"; then
     AC_CHECK_LIB(seccomp, seccomp_init)
 fi
@@ -215,6 +226,14 @@ fi
 if  test "$ac_cv_header_lzma_h$ac_cv_lib_lzma_lzma_stream_decoder" = "yesyes"; then
   AC_DEFINE([XZLIBSUPPORT], 1, [Enable xzlib compression support])
 fi
+if test "$enable_zstdlib" = "yes"; then
+  if test "$ac_cv_header_zstd_h$ac_cv_lib_zstd_ZSTD_createDStream" != "yesyes"; then
+    AC_MSG_ERROR([zstdlib support requested but not found])
+  fi
+fi
+if  test "$ac_cv_header_zstd_h$ac_cv_lib_zstd_ZSTD_createDStream" = "yesyes"; then
+  AC_DEFINE([ZSTDLIBSUPPORT], 1, [Enable zstdlib compression support])
+fi
 
 AC_CONFIG_FILES([Makefile src/Makefile magic/Makefile tests/Makefile doc/Makefile python/Makefile libmagic.pc])
 AC_OUTPUT
diff --git a/src/compress.c b/src/compress.c
index 113077ae..abb8e766 100644
--- a/src/compress.c
+++ b/src/compress.c
@@ -79,6 +79,12 @@ typedef void (*sig_t)(int);
 #include <lzma.h>
 #endif
 
+#if defined(HAVE_ZSTD_H) && defined(ZSTDLIBSUPPORT)
+#define BUILTIN_ZSTDLIB
+#include <zstd.h>
+#include <zstd_errors.h>
+#endif
+
 #ifdef DEBUG
 int tty = -1;
 #define DPRINTF(...)	do { \
@@ -175,6 +181,7 @@ private const struct {
 #define METH_FROZEN	2
 #define METH_BZIP	7
 #define METH_XZ		9
+#define METH_ZSTD	12
 #define METH_LZMA	13
 #define METH_ZLIB	14
     { { .magic = "\037\235" },	2, gzip_args, NULL },	/* 0, compressed */
@@ -223,6 +230,10 @@ private int uncompressbzlib(const unsigned char *, unsigned char **, size_t,
 private int uncompressxzlib(const unsigned char *, unsigned char **, size_t,
     size_t *);
 #endif
+#ifdef BUILTIN_ZSTDLIB
+private int uncompresszstd(const unsigned char *, unsigned char **, size_t,
+    size_t *);
+#endif
 
 static int makeerror(unsigned char **, size_t *, const char *, ...)
     __attribute__((__format__(__printf__, 3, 4)));
@@ -697,6 +708,55 @@ err:
 }
 #endif
 
+#ifdef BUILTIN_ZSTDLIB
+private int
+uncompresszstd(const unsigned char *old, unsigned char **newch,
+    size_t bytes_max, size_t *n)
+{
+	size_t rc;
+	ZSTD_DStream *zstd;
+	ZSTD_inBuffer in;
+	ZSTD_outBuffer out;
+
+	if ((zstd = ZSTD_createDStream()) == NULL)
+		return makeerror(newch, n, "No ZSTD decompression stream, %s", strerror(errno));
+
+	rc = ZSTD_DCtx_reset(zstd, ZSTD_reset_session_only);
+	if (ZSTD_isError(rc))
+		goto err;
+
+	if ((*newch = CAST(unsigned char *, malloc(bytes_max + 1))) == NULL) {
+		ZSTD_freeDStream(zstd);
+		return makeerror(newch, n, "No buffer, %s", strerror(errno));
+	}
+
+	in.src = CCAST(const void *, old);
+	in.size = *n;
+	in.pos = 0;
+	out.dst = RCAST(void *, *newch);
+	out.size = bytes_max;
+	out.pos = 0;
+
+	rc = ZSTD_decompressStream(zstd, &out, &in);
+	if (ZSTD_isError(rc))
+		goto err;
+
+	*n = out.pos;
+
+	ZSTD_freeDStream(zstd);
+
+	/* let's keep the nul-terminate tradition */
+	(*newch)[*n] = '\0';
+
+	return OKDATA;
+err:
+	ZSTD_freeDStream(zstd);
+	snprintf(RCAST(char *, *newch), bytes_max, "zstd error %d", ZSTD_getErrorCode(rc));
+	*n = strlen(RCAST(char *, *newch));
+	return ERRDATA;
+}
+#endif
+
 
 static int
 makeerror(unsigned char **buf, size_t *len, const char *fmt, ...)
@@ -863,6 +923,10 @@ methodname(size_t method)
 	case METH_XZ:
 	case METH_LZMA:
 		return "xzlib";
+#endif
+#ifdef BUILTIN_ZSTDLIB
+	case METH_ZSTD:
+		return "zstd";
 #endif
 	default:
 		return compr[method].argv[0];
@@ -899,6 +963,10 @@ uncompressbuf(int fd, size_t bytes_max, size_t method, const unsigned char *old,
 	case METH_XZ:
 	case METH_LZMA:
 		return uncompressxzlib(old, newch, bytes_max, n);
+#endif
+#ifdef BUILTIN_ZSTDLIB
+	case METH_ZSTD:
+		return uncompresszstd(old, newch, bytes_max, n);
 #endif
 	default:
 		break;
-- 
2.37.3